WSL2-Linux-Kernel/drivers/target/target_core_transport.c

3437 строки
93 KiB
C
Исходник Обычный вид История

/*******************************************************************************
* Filename: target_core_transport.c
*
* This file contains the Generic Target Engine Core.
*
* (c) Copyright 2002-2013 Datera, Inc.
*
* Nicholas A. Bellinger <nab@kernel.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
******************************************************************************/
#include <linux/net.h>
#include <linux/delay.h>
#include <linux/string.h>
#include <linux/timer.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/kthread.h>
#include <linux/in.h>
#include <linux/cdrom.h>
#include <linux/module.h>
#include <linux/ratelimit.h>
#include <linux/vmalloc.h>
#include <asm/unaligned.h>
#include <net/sock.h>
#include <net/tcp.h>
#include <scsi/scsi_proto.h>
#include <scsi/scsi_common.h>
#include <target/target_core_base.h>
#include <target/target_core_backend.h>
#include <target/target_core_fabric.h>
#include "target_core_internal.h"
#include "target_core_alua.h"
#include "target_core_pr.h"
#include "target_core_ua.h"
#define CREATE_TRACE_POINTS
#include <trace/events/target.h>
static struct workqueue_struct *target_completion_wq;
static struct kmem_cache *se_sess_cache;
struct kmem_cache *se_ua_cache;
struct kmem_cache *t10_pr_reg_cache;
struct kmem_cache *t10_alua_lu_gp_cache;
struct kmem_cache *t10_alua_lu_gp_mem_cache;
struct kmem_cache *t10_alua_tg_pt_gp_cache;
struct kmem_cache *t10_alua_lba_map_cache;
struct kmem_cache *t10_alua_lba_map_mem_cache;
static void transport_complete_task_attr(struct se_cmd *cmd);
static int translate_sense_reason(struct se_cmd *cmd, sense_reason_t reason);
static void transport_handle_queue_full(struct se_cmd *cmd,
struct se_device *dev, int err, bool write_pending);
static void target_complete_ok_work(struct work_struct *work);
2011-07-19 12:55:10 +04:00
int init_se_kmem_caches(void)
{
se_sess_cache = kmem_cache_create("se_sess_cache",
sizeof(struct se_session), __alignof__(struct se_session),
0, NULL);
if (!se_sess_cache) {
pr_err("kmem_cache_create() for struct se_session"
" failed\n");
goto out;
}
se_ua_cache = kmem_cache_create("se_ua_cache",
sizeof(struct se_ua), __alignof__(struct se_ua),
0, NULL);
if (!se_ua_cache) {
pr_err("kmem_cache_create() for struct se_ua failed\n");
goto out_free_sess_cache;
}
t10_pr_reg_cache = kmem_cache_create("t10_pr_reg_cache",
sizeof(struct t10_pr_registration),
__alignof__(struct t10_pr_registration), 0, NULL);
if (!t10_pr_reg_cache) {
pr_err("kmem_cache_create() for struct t10_pr_registration"
" failed\n");
goto out_free_ua_cache;
}
t10_alua_lu_gp_cache = kmem_cache_create("t10_alua_lu_gp_cache",
sizeof(struct t10_alua_lu_gp), __alignof__(struct t10_alua_lu_gp),
0, NULL);
if (!t10_alua_lu_gp_cache) {
pr_err("kmem_cache_create() for t10_alua_lu_gp_cache"
" failed\n");
goto out_free_pr_reg_cache;
}
t10_alua_lu_gp_mem_cache = kmem_cache_create("t10_alua_lu_gp_mem_cache",
sizeof(struct t10_alua_lu_gp_member),
__alignof__(struct t10_alua_lu_gp_member), 0, NULL);
if (!t10_alua_lu_gp_mem_cache) {
pr_err("kmem_cache_create() for t10_alua_lu_gp_mem_"
"cache failed\n");
goto out_free_lu_gp_cache;
}
t10_alua_tg_pt_gp_cache = kmem_cache_create("t10_alua_tg_pt_gp_cache",
sizeof(struct t10_alua_tg_pt_gp),
__alignof__(struct t10_alua_tg_pt_gp), 0, NULL);
if (!t10_alua_tg_pt_gp_cache) {
pr_err("kmem_cache_create() for t10_alua_tg_pt_gp_"
"cache failed\n");
goto out_free_lu_gp_mem_cache;
}
t10_alua_lba_map_cache = kmem_cache_create(
"t10_alua_lba_map_cache",
sizeof(struct t10_alua_lba_map),
__alignof__(struct t10_alua_lba_map), 0, NULL);
if (!t10_alua_lba_map_cache) {
pr_err("kmem_cache_create() for t10_alua_lba_map_"
"cache failed\n");
goto out_free_tg_pt_gp_cache;
}
t10_alua_lba_map_mem_cache = kmem_cache_create(
"t10_alua_lba_map_mem_cache",
sizeof(struct t10_alua_lba_map_member),
__alignof__(struct t10_alua_lba_map_member), 0, NULL);
if (!t10_alua_lba_map_mem_cache) {
pr_err("kmem_cache_create() for t10_alua_lba_map_mem_"
"cache failed\n");
goto out_free_lba_map_cache;
}
target_completion_wq = alloc_workqueue("target_completion",
WQ_MEM_RECLAIM, 0);
if (!target_completion_wq)
goto out_free_lba_map_mem_cache;
return 0;
out_free_lba_map_mem_cache:
kmem_cache_destroy(t10_alua_lba_map_mem_cache);
out_free_lba_map_cache:
kmem_cache_destroy(t10_alua_lba_map_cache);
out_free_tg_pt_gp_cache:
kmem_cache_destroy(t10_alua_tg_pt_gp_cache);
out_free_lu_gp_mem_cache:
kmem_cache_destroy(t10_alua_lu_gp_mem_cache);
out_free_lu_gp_cache:
kmem_cache_destroy(t10_alua_lu_gp_cache);
out_free_pr_reg_cache:
kmem_cache_destroy(t10_pr_reg_cache);
out_free_ua_cache:
kmem_cache_destroy(se_ua_cache);
out_free_sess_cache:
kmem_cache_destroy(se_sess_cache);
out:
2011-07-19 12:55:10 +04:00
return -ENOMEM;
}
2011-07-19 12:55:10 +04:00
void release_se_kmem_caches(void)
{
destroy_workqueue(target_completion_wq);
kmem_cache_destroy(se_sess_cache);
kmem_cache_destroy(se_ua_cache);
kmem_cache_destroy(t10_pr_reg_cache);
kmem_cache_destroy(t10_alua_lu_gp_cache);
kmem_cache_destroy(t10_alua_lu_gp_mem_cache);
kmem_cache_destroy(t10_alua_tg_pt_gp_cache);
kmem_cache_destroy(t10_alua_lba_map_cache);
kmem_cache_destroy(t10_alua_lba_map_mem_cache);
}
2011-07-19 12:55:10 +04:00
/* This code ensures unique mib indexes are handed out. */
static DEFINE_SPINLOCK(scsi_mib_index_lock);
static u32 scsi_mib_index[SCSI_INDEX_TYPE_MAX];
/*
* Allocate a new row index for the entry type specified
*/
u32 scsi_get_new_index(scsi_index_t type)
{
u32 new_index;
2011-07-19 12:55:10 +04:00
BUG_ON((type < 0) || (type >= SCSI_INDEX_TYPE_MAX));
2011-07-19 12:55:10 +04:00
spin_lock(&scsi_mib_index_lock);
new_index = ++scsi_mib_index[type];
spin_unlock(&scsi_mib_index_lock);
return new_index;
}
void transport_subsystem_check_init(void)
{
int ret;
static int sub_api_initialized;
if (sub_api_initialized)
return;
ret = request_module("target_core_iblock");
if (ret != 0)
pr_err("Unable to load target_core_iblock\n");
ret = request_module("target_core_file");
if (ret != 0)
pr_err("Unable to load target_core_file\n");
ret = request_module("target_core_pscsi");
if (ret != 0)
pr_err("Unable to load target_core_pscsi\n");
target: Add a user-passthrough backstore Add a LIO storage engine that presents commands to userspace for execution. This would allow more complex backstores to be implemented out-of-kernel, and also make experimentation a-la FUSE (but at the SCSI level -- "SUSE"?) possible. It uses a mmap()able UIO device per LUN to share a command ring and data area. The commands are raw SCSI CDBs and iovs for in/out data. The command ring is also reused for returning scsi command status and optional sense data. This implementation is based on Shaohua Li's earlier version but heavily modified. Differences include: * Shared memory allocated by kernel, not locked-down user pages * Single ring for command request and response * Offsets instead of embedded pointers * Generic SCSI CDB passthrough instead of per-cmd specialization in ring format. * Uses UIO device instead of anon_file passed in mailbox. * Optional in-kernel handling of some commands. The main reason for these differences is to permit greater resiliency if the user process dies or hangs. Things not yet implemented (on purpose): * Zero copy. The data area is flexible enough to allow page flipping or backend-allocated pages to be used by fabrics, but it's not clear these are performance wins. Can come later. * Out-of-order command completion by userspace. Possible to add by just allowing userspace to change cmd_id in rsp cmd entries, but currently not supported. * No locks between kernel cmd submission and completion routines. Sounds like it's possible, but this can come later. * Sparse allocation of mmaped area. Current code vmallocs the whole thing. If the mapped area was larger and not fully mapped then the driver would have more freedom to change cmd and data area sizes based on demand. Current code open issues: * The use of idrs may be overkill -- we maybe can replace them with a simple counter to generate cmd_ids, and a hash table to get a cmd_id's associated pointer. * Use of a free-running counter for cmd ring instead of explicit modulo math. This would require power-of-2 cmd ring size. (Add kconfig depends NET - Randy) Signed-off-by: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2014-10-02 03:07:05 +04:00
ret = request_module("target_core_user");
if (ret != 0)
pr_err("Unable to load target_core_user\n");
2011-07-19 12:55:10 +04:00
sub_api_initialized = 1;
}
struct se_session *transport_init_session(enum target_prot_op sup_prot_ops)
{
struct se_session *se_sess;
se_sess = kmem_cache_zalloc(se_sess_cache, GFP_KERNEL);
if (!se_sess) {
pr_err("Unable to allocate struct se_session from"
" se_sess_cache\n");
return ERR_PTR(-ENOMEM);
}
INIT_LIST_HEAD(&se_sess->sess_list);
INIT_LIST_HEAD(&se_sess->sess_acl_list);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
INIT_LIST_HEAD(&se_sess->sess_cmd_list);
INIT_LIST_HEAD(&se_sess->sess_wait_list);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
spin_lock_init(&se_sess->sess_cmd_lock);
se_sess->sup_prot_ops = sup_prot_ops;
return se_sess;
}
EXPORT_SYMBOL(transport_init_session);
int transport_alloc_session_tags(struct se_session *se_sess,
unsigned int tag_num, unsigned int tag_size)
{
int rc;
treewide: kzalloc() -> kcalloc() The kzalloc() function has a 2-factor argument form, kcalloc(). This patch replaces cases of: kzalloc(a * b, gfp) with: kcalloc(a * b, gfp) as well as handling cases of: kzalloc(a * b * c, gfp) with: kzalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kzalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kzalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kzalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kzalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kzalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(char) * COUNT + COUNT , ...) | kzalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kzalloc + kcalloc ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kzalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kzalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kzalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kzalloc(C1 * C2 * C3, ...) | kzalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kzalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kzalloc(sizeof(THING) * C2, ...) | kzalloc(sizeof(TYPE) * C2, ...) | kzalloc(C1 * C2 * C3, ...) | kzalloc(C1 * C2, ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - (E1) * E2 + E1, E2 , ...) | - kzalloc + kcalloc ( - (E1) * (E2) + E1, E2 , ...) | - kzalloc + kcalloc ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-13 00:03:40 +03:00
se_sess->sess_cmd_map = kcalloc(tag_size, tag_num,
mm, tree wide: replace __GFP_REPEAT by __GFP_RETRY_MAYFAIL with more useful semantic __GFP_REPEAT was designed to allow retry-but-eventually-fail semantic to the page allocator. This has been true but only for allocations requests larger than PAGE_ALLOC_COSTLY_ORDER. It has been always ignored for smaller sizes. This is a bit unfortunate because there is no way to express the same semantic for those requests and they are considered too important to fail so they might end up looping in the page allocator for ever, similarly to GFP_NOFAIL requests. Now that the whole tree has been cleaned up and accidental or misled usage of __GFP_REPEAT flag has been removed for !costly requests we can give the original flag a better name and more importantly a more useful semantic. Let's rename it to __GFP_RETRY_MAYFAIL which tells the user that the allocator would try really hard but there is no promise of a success. This will work independent of the order and overrides the default allocator behavior. Page allocator users have several levels of guarantee vs. cost options (take GFP_KERNEL as an example) - GFP_KERNEL & ~__GFP_RECLAIM - optimistic allocation without _any_ attempt to free memory at all. The most light weight mode which even doesn't kick the background reclaim. Should be used carefully because it might deplete the memory and the next user might hit the more aggressive reclaim - GFP_KERNEL & ~__GFP_DIRECT_RECLAIM (or GFP_NOWAIT)- optimistic allocation without any attempt to free memory from the current context but can wake kswapd to reclaim memory if the zone is below the low watermark. Can be used from either atomic contexts or when the request is a performance optimization and there is another fallback for a slow path. - (GFP_KERNEL|__GFP_HIGH) & ~__GFP_DIRECT_RECLAIM (aka GFP_ATOMIC) - non sleeping allocation with an expensive fallback so it can access some portion of memory reserves. Usually used from interrupt/bh context with an expensive slow path fallback. - GFP_KERNEL - both background and direct reclaim are allowed and the _default_ page allocator behavior is used. That means that !costly allocation requests are basically nofail but there is no guarantee of that behavior so failures have to be checked properly by callers (e.g. OOM killer victim is allowed to fail currently). - GFP_KERNEL | __GFP_NORETRY - overrides the default allocator behavior and all allocation requests fail early rather than cause disruptive reclaim (one round of reclaim in this implementation). The OOM killer is not invoked. - GFP_KERNEL | __GFP_RETRY_MAYFAIL - overrides the default allocator behavior and all allocation requests try really hard. The request will fail if the reclaim cannot make any progress. The OOM killer won't be triggered. - GFP_KERNEL | __GFP_NOFAIL - overrides the default allocator behavior and all allocation requests will loop endlessly until they succeed. This might be really dangerous especially for larger orders. Existing users of __GFP_REPEAT are changed to __GFP_RETRY_MAYFAIL because they already had their semantic. No new users are added. __alloc_pages_slowpath is changed to bail out for __GFP_RETRY_MAYFAIL if there is no progress and we have already passed the OOM point. This means that all the reclaim opportunities have been exhausted except the most disruptive one (the OOM killer) and a user defined fallback behavior is more sensible than keep retrying in the page allocator. [akpm@linux-foundation.org: fix arch/sparc/kernel/mdesc.c] [mhocko@suse.com: semantic fix] Link: http://lkml.kernel.org/r/20170626123847.GM11534@dhcp22.suse.cz [mhocko@kernel.org: address other thing spotted by Vlastimil] Link: http://lkml.kernel.org/r/20170626124233.GN11534@dhcp22.suse.cz Link: http://lkml.kernel.org/r/20170623085345.11304-3-mhocko@kernel.org Signed-off-by: Michal Hocko <mhocko@suse.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Alex Belits <alex.belits@cavium.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Christoph Hellwig <hch@infradead.org> Cc: Darrick J. Wong <darrick.wong@oracle.com> Cc: David Daney <david.daney@cavium.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Mel Gorman <mgorman@suse.de> Cc: NeilBrown <neilb@suse.com> Cc: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-13 00:36:45 +03:00
GFP_KERNEL | __GFP_NOWARN | __GFP_RETRY_MAYFAIL);
if (!se_sess->sess_cmd_map) {
se_sess->sess_cmd_map = vzalloc(tag_num * tag_size);
if (!se_sess->sess_cmd_map) {
pr_err("Unable to allocate se_sess->sess_cmd_map\n");
return -ENOMEM;
}
}
rc = percpu_ida_init(&se_sess->sess_tag_pool, tag_num);
if (rc < 0) {
pr_err("Unable to init se_sess->sess_tag_pool,"
" tag_num: %u\n", tag_num);
kvfree(se_sess->sess_cmd_map);
se_sess->sess_cmd_map = NULL;
return -ENOMEM;
}
return 0;
}
EXPORT_SYMBOL(transport_alloc_session_tags);
struct se_session *transport_init_session_tags(unsigned int tag_num,
unsigned int tag_size,
enum target_prot_op sup_prot_ops)
{
struct se_session *se_sess;
int rc;
if (tag_num != 0 && !tag_size) {
pr_err("init_session_tags called with percpu-ida tag_num:"
" %u, but zero tag_size\n", tag_num);
return ERR_PTR(-EINVAL);
}
if (!tag_num && tag_size) {
pr_err("init_session_tags called with percpu-ida tag_size:"
" %u, but zero tag_num\n", tag_size);
return ERR_PTR(-EINVAL);
}
se_sess = transport_init_session(sup_prot_ops);
if (IS_ERR(se_sess))
return se_sess;
rc = transport_alloc_session_tags(se_sess, tag_num, tag_size);
if (rc < 0) {
transport_free_session(se_sess);
return ERR_PTR(-ENOMEM);
}
return se_sess;
}
EXPORT_SYMBOL(transport_init_session_tags);
/*
target: Convert session_lock to irqsave This patch converts the remaining struct se_portal_group->session_lock usage to use irqsave+irqrestore to address the following warnings for hardware target mode interrupt context usage. This change generate other warnings for current iscsi-target mode still using ->session_lock with spin_lock_bh, which will need to be converted in a seperate patch. [ 492.480728] [ INFO: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected ] [ 492.488194] 3.0.0+ #23 [ 492.490820] ------------------------------------------------------ [ 492.497704] sh/7162 [HC0[0]:SC0[2]:HE0:SE0] is trying to acquire: [ 492.504493] (&(&se_tpg->session_lock)->rlock){+.....}, at: [<ffffffffa022364d>] transport_deregister_session+0x2d/0x163 [target_core_mod] 492.518390] [ 492.518390] and this task is already holding: [ 492.524897] (&(&ha->hardware_lock)->rlock){-.-...}, at: [<ffffffffa00b9146>] qla_tgt_stop_phase1+0x5e/0x27e [qla2xxx] [ 492.536856] which would create a new lock dependency: [ 492.542481] (&(&ha->hardware_lock)->rlock){-.-...} -> (&(&se_tpg->session_lock)->rlock){+.....} [ 492.552321] [ 492.552321] but this new dependency connects a HARDIRQ-irq-safe lock: [ 492.561149] (&(&ha->hardware_lock)->rlock){-.-...} [ 492.566400] ... which became HARDIRQ-irq-safe at: [ 492.571841] [<ffffffff81064720>] __lock_acquire+0x68f/0x921 [ 492.578247] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.584367] [<ffffffff813a74c6>] _raw_spin_lock_irqsave+0x44/0x56 [ 492.591358] [<ffffffffa009b1be>] qla24xx_msix_default+0x5c/0x2aa [qla2xxx] [ 492.599227] [<ffffffff81088582>] handle_irq_event_percpu+0x5a/0x197 [ 492.606413] [<ffffffff810886fb>] handle_irq_event+0x3c/0x5c [ 492.612822] [<ffffffff8108a6dc>] handle_edge_irq+0xcc/0xf1 [ 492.619138] [<ffffffff810039b9>] handle_irq+0x83/0x8e [ 492.624971] [<ffffffff8100333e>] do_IRQ+0x48/0xaf [ 492.630413] [<ffffffff813a7cd3>] ret_from_intr+0x0/0x1a [ 492.636437] [<ffffffff81001dc1>] cpu_idle+0x5b/0x8d [ 492.642073] [<ffffffff81392709>] rest_init+0xad/0xb4 [ 492.647809] [<ffffffff81a1cbbc>] start_kernel+0x366/0x371 [ 492.654030] [<ffffffff81a1c2b1>] x86_64_start_reservations+0xb8/0xbc [ 492.661311] [<ffffffff81a1c3b6>] x86_64_start_kernel+0x101/0x110 [ 492.668204] [ 492.668205] to a HARDIRQ-irq-unsafe lock: [ 492.674324] (&(&se_tpg->session_lock)->rlock){+.....} [ 492.679862] ... which became HARDIRQ-irq-unsafe at: [ 492.685497] ... [<ffffffff8106479a>] __lock_acquire+0x709/0x921 [ 492.692209] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.698330] [<ffffffff813a75ed>] _raw_spin_lock_bh+0x31/0x40 [ 492.704836] [<ffffffffa021c208>] core_tpg_del_initiator_node_acl+0x89/0x336 [target_core_mod] [ 492.714546] [<ffffffffa02fb075>] tcm_qla2xxx_drop_nodeacl+0x20/0x2d [tcm_qla2xxx] [ 492.723087] [<ffffffffa02108d9>] target_fabric_nacl_base_release+0x22/0x24 [target_core_mod] [ 492.732698] [<ffffffffa01661c8>] config_item_release+0x7d/0xa3 [configfs] [ 492.740465] [<ffffffff811d48fe>] kref_put+0x43/0x4d [ 492.746101] [<ffffffffa0166149>] config_item_put+0x19/0x1b [configfs] [ 492.753481] [<ffffffffa0164987>] configfs_rmdir+0x1eb/0x258 [configfs] [ 492.760957] [<ffffffff810ecc54>] vfs_rmdir+0x79/0xd0 [ 492.766690] [<ffffffff810eec4a>] do_rmdir+0xc2/0x111 [ 492.772423] [<ffffffff810eecd0>] sys_rmdir+0x11/0x13 [ 492.778156] [<ffffffff813ae4d2>] system_call_fastpath+0x16/0x1b [ 492.784953] Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-08-31 23:34:39 +04:00
* Called with spin_lock_irqsave(&struct se_portal_group->session_lock called.
*/
void __transport_register_session(
struct se_portal_group *se_tpg,
struct se_node_acl *se_nacl,
struct se_session *se_sess,
void *fabric_sess_ptr)
{
const struct target_core_fabric_ops *tfo = se_tpg->se_tpg_tfo;
unsigned char buf[PR_REG_ISID_LEN];
se_sess->se_tpg = se_tpg;
se_sess->fabric_sess_ptr = fabric_sess_ptr;
/*
* Used by struct se_node_acl's under ConfigFS to locate active se_session-t
*
* Only set for struct se_session's that will actually be moving I/O.
* eg: *NOT* discovery sessions.
*/
if (se_nacl) {
/*
*
* Determine if fabric allows for T10-PI feature bits exposed to
* initiators for device backends with !dev->dev_attrib.pi_prot_type.
*
* If so, then always save prot_type on a per se_node_acl node
* basis and re-instate the previous sess_prot_type to avoid
* disabling PI from below any previously initiator side
* registered LUNs.
*/
if (se_nacl->saved_prot_type)
se_sess->sess_prot_type = se_nacl->saved_prot_type;
else if (tfo->tpg_check_prot_fabric_only)
se_sess->sess_prot_type = se_nacl->saved_prot_type =
tfo->tpg_check_prot_fabric_only(se_tpg);
/*
* If the fabric module supports an ISID based TransportID,
* save this value in binary from the fabric I_T Nexus now.
*/
2011-07-19 12:55:10 +04:00
if (se_tpg->se_tpg_tfo->sess_get_initiator_sid != NULL) {
memset(&buf[0], 0, PR_REG_ISID_LEN);
2011-07-19 12:55:10 +04:00
se_tpg->se_tpg_tfo->sess_get_initiator_sid(se_sess,
&buf[0], PR_REG_ISID_LEN);
se_sess->sess_bin_isid = get_unaligned_be64(&buf[0]);
}
spin_lock_irq(&se_nacl->nacl_sess_lock);
/*
* The se_nacl->nacl_sess pointer will be set to the
* last active I_T Nexus for each struct se_node_acl.
*/
se_nacl->nacl_sess = se_sess;
list_add_tail(&se_sess->sess_acl_list,
&se_nacl->acl_sess_list);
spin_unlock_irq(&se_nacl->nacl_sess_lock);
}
list_add_tail(&se_sess->sess_list, &se_tpg->tpg_sess_list);
pr_debug("TARGET_CORE[%s]: Registered fabric_sess_ptr: %p\n",
2011-07-19 12:55:10 +04:00
se_tpg->se_tpg_tfo->get_fabric_name(), se_sess->fabric_sess_ptr);
}
EXPORT_SYMBOL(__transport_register_session);
void transport_register_session(
struct se_portal_group *se_tpg,
struct se_node_acl *se_nacl,
struct se_session *se_sess,
void *fabric_sess_ptr)
{
target: Convert session_lock to irqsave This patch converts the remaining struct se_portal_group->session_lock usage to use irqsave+irqrestore to address the following warnings for hardware target mode interrupt context usage. This change generate other warnings for current iscsi-target mode still using ->session_lock with spin_lock_bh, which will need to be converted in a seperate patch. [ 492.480728] [ INFO: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected ] [ 492.488194] 3.0.0+ #23 [ 492.490820] ------------------------------------------------------ [ 492.497704] sh/7162 [HC0[0]:SC0[2]:HE0:SE0] is trying to acquire: [ 492.504493] (&(&se_tpg->session_lock)->rlock){+.....}, at: [<ffffffffa022364d>] transport_deregister_session+0x2d/0x163 [target_core_mod] 492.518390] [ 492.518390] and this task is already holding: [ 492.524897] (&(&ha->hardware_lock)->rlock){-.-...}, at: [<ffffffffa00b9146>] qla_tgt_stop_phase1+0x5e/0x27e [qla2xxx] [ 492.536856] which would create a new lock dependency: [ 492.542481] (&(&ha->hardware_lock)->rlock){-.-...} -> (&(&se_tpg->session_lock)->rlock){+.....} [ 492.552321] [ 492.552321] but this new dependency connects a HARDIRQ-irq-safe lock: [ 492.561149] (&(&ha->hardware_lock)->rlock){-.-...} [ 492.566400] ... which became HARDIRQ-irq-safe at: [ 492.571841] [<ffffffff81064720>] __lock_acquire+0x68f/0x921 [ 492.578247] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.584367] [<ffffffff813a74c6>] _raw_spin_lock_irqsave+0x44/0x56 [ 492.591358] [<ffffffffa009b1be>] qla24xx_msix_default+0x5c/0x2aa [qla2xxx] [ 492.599227] [<ffffffff81088582>] handle_irq_event_percpu+0x5a/0x197 [ 492.606413] [<ffffffff810886fb>] handle_irq_event+0x3c/0x5c [ 492.612822] [<ffffffff8108a6dc>] handle_edge_irq+0xcc/0xf1 [ 492.619138] [<ffffffff810039b9>] handle_irq+0x83/0x8e [ 492.624971] [<ffffffff8100333e>] do_IRQ+0x48/0xaf [ 492.630413] [<ffffffff813a7cd3>] ret_from_intr+0x0/0x1a [ 492.636437] [<ffffffff81001dc1>] cpu_idle+0x5b/0x8d [ 492.642073] [<ffffffff81392709>] rest_init+0xad/0xb4 [ 492.647809] [<ffffffff81a1cbbc>] start_kernel+0x366/0x371 [ 492.654030] [<ffffffff81a1c2b1>] x86_64_start_reservations+0xb8/0xbc [ 492.661311] [<ffffffff81a1c3b6>] x86_64_start_kernel+0x101/0x110 [ 492.668204] [ 492.668205] to a HARDIRQ-irq-unsafe lock: [ 492.674324] (&(&se_tpg->session_lock)->rlock){+.....} [ 492.679862] ... which became HARDIRQ-irq-unsafe at: [ 492.685497] ... [<ffffffff8106479a>] __lock_acquire+0x709/0x921 [ 492.692209] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.698330] [<ffffffff813a75ed>] _raw_spin_lock_bh+0x31/0x40 [ 492.704836] [<ffffffffa021c208>] core_tpg_del_initiator_node_acl+0x89/0x336 [target_core_mod] [ 492.714546] [<ffffffffa02fb075>] tcm_qla2xxx_drop_nodeacl+0x20/0x2d [tcm_qla2xxx] [ 492.723087] [<ffffffffa02108d9>] target_fabric_nacl_base_release+0x22/0x24 [target_core_mod] [ 492.732698] [<ffffffffa01661c8>] config_item_release+0x7d/0xa3 [configfs] [ 492.740465] [<ffffffff811d48fe>] kref_put+0x43/0x4d [ 492.746101] [<ffffffffa0166149>] config_item_put+0x19/0x1b [configfs] [ 492.753481] [<ffffffffa0164987>] configfs_rmdir+0x1eb/0x258 [configfs] [ 492.760957] [<ffffffff810ecc54>] vfs_rmdir+0x79/0xd0 [ 492.766690] [<ffffffff810eec4a>] do_rmdir+0xc2/0x111 [ 492.772423] [<ffffffff810eecd0>] sys_rmdir+0x11/0x13 [ 492.778156] [<ffffffff813ae4d2>] system_call_fastpath+0x16/0x1b [ 492.784953] Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-08-31 23:34:39 +04:00
unsigned long flags;
spin_lock_irqsave(&se_tpg->session_lock, flags);
__transport_register_session(se_tpg, se_nacl, se_sess, fabric_sess_ptr);
target: Convert session_lock to irqsave This patch converts the remaining struct se_portal_group->session_lock usage to use irqsave+irqrestore to address the following warnings for hardware target mode interrupt context usage. This change generate other warnings for current iscsi-target mode still using ->session_lock with spin_lock_bh, which will need to be converted in a seperate patch. [ 492.480728] [ INFO: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected ] [ 492.488194] 3.0.0+ #23 [ 492.490820] ------------------------------------------------------ [ 492.497704] sh/7162 [HC0[0]:SC0[2]:HE0:SE0] is trying to acquire: [ 492.504493] (&(&se_tpg->session_lock)->rlock){+.....}, at: [<ffffffffa022364d>] transport_deregister_session+0x2d/0x163 [target_core_mod] 492.518390] [ 492.518390] and this task is already holding: [ 492.524897] (&(&ha->hardware_lock)->rlock){-.-...}, at: [<ffffffffa00b9146>] qla_tgt_stop_phase1+0x5e/0x27e [qla2xxx] [ 492.536856] which would create a new lock dependency: [ 492.542481] (&(&ha->hardware_lock)->rlock){-.-...} -> (&(&se_tpg->session_lock)->rlock){+.....} [ 492.552321] [ 492.552321] but this new dependency connects a HARDIRQ-irq-safe lock: [ 492.561149] (&(&ha->hardware_lock)->rlock){-.-...} [ 492.566400] ... which became HARDIRQ-irq-safe at: [ 492.571841] [<ffffffff81064720>] __lock_acquire+0x68f/0x921 [ 492.578247] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.584367] [<ffffffff813a74c6>] _raw_spin_lock_irqsave+0x44/0x56 [ 492.591358] [<ffffffffa009b1be>] qla24xx_msix_default+0x5c/0x2aa [qla2xxx] [ 492.599227] [<ffffffff81088582>] handle_irq_event_percpu+0x5a/0x197 [ 492.606413] [<ffffffff810886fb>] handle_irq_event+0x3c/0x5c [ 492.612822] [<ffffffff8108a6dc>] handle_edge_irq+0xcc/0xf1 [ 492.619138] [<ffffffff810039b9>] handle_irq+0x83/0x8e [ 492.624971] [<ffffffff8100333e>] do_IRQ+0x48/0xaf [ 492.630413] [<ffffffff813a7cd3>] ret_from_intr+0x0/0x1a [ 492.636437] [<ffffffff81001dc1>] cpu_idle+0x5b/0x8d [ 492.642073] [<ffffffff81392709>] rest_init+0xad/0xb4 [ 492.647809] [<ffffffff81a1cbbc>] start_kernel+0x366/0x371 [ 492.654030] [<ffffffff81a1c2b1>] x86_64_start_reservations+0xb8/0xbc [ 492.661311] [<ffffffff81a1c3b6>] x86_64_start_kernel+0x101/0x110 [ 492.668204] [ 492.668205] to a HARDIRQ-irq-unsafe lock: [ 492.674324] (&(&se_tpg->session_lock)->rlock){+.....} [ 492.679862] ... which became HARDIRQ-irq-unsafe at: [ 492.685497] ... [<ffffffff8106479a>] __lock_acquire+0x709/0x921 [ 492.692209] [<ffffffff81064eff>] lock_acquire+0xe0/0x10d [ 492.698330] [<ffffffff813a75ed>] _raw_spin_lock_bh+0x31/0x40 [ 492.704836] [<ffffffffa021c208>] core_tpg_del_initiator_node_acl+0x89/0x336 [target_core_mod] [ 492.714546] [<ffffffffa02fb075>] tcm_qla2xxx_drop_nodeacl+0x20/0x2d [tcm_qla2xxx] [ 492.723087] [<ffffffffa02108d9>] target_fabric_nacl_base_release+0x22/0x24 [target_core_mod] [ 492.732698] [<ffffffffa01661c8>] config_item_release+0x7d/0xa3 [configfs] [ 492.740465] [<ffffffff811d48fe>] kref_put+0x43/0x4d [ 492.746101] [<ffffffffa0166149>] config_item_put+0x19/0x1b [configfs] [ 492.753481] [<ffffffffa0164987>] configfs_rmdir+0x1eb/0x258 [configfs] [ 492.760957] [<ffffffff810ecc54>] vfs_rmdir+0x79/0xd0 [ 492.766690] [<ffffffff810eec4a>] do_rmdir+0xc2/0x111 [ 492.772423] [<ffffffff810eecd0>] sys_rmdir+0x11/0x13 [ 492.778156] [<ffffffff813ae4d2>] system_call_fastpath+0x16/0x1b [ 492.784953] Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-08-31 23:34:39 +04:00
spin_unlock_irqrestore(&se_tpg->session_lock, flags);
}
EXPORT_SYMBOL(transport_register_session);
struct se_session *
target_alloc_session(struct se_portal_group *tpg,
unsigned int tag_num, unsigned int tag_size,
enum target_prot_op prot_op,
const char *initiatorname, void *private,
int (*callback)(struct se_portal_group *,
struct se_session *, void *))
{
struct se_session *sess;
/*
* If the fabric driver is using percpu-ida based pre allocation
* of I/O descriptor tags, go ahead and perform that setup now..
*/
if (tag_num != 0)
sess = transport_init_session_tags(tag_num, tag_size, prot_op);
else
sess = transport_init_session(prot_op);
if (IS_ERR(sess))
return sess;
sess->se_node_acl = core_tpg_check_initiator_node_acl(tpg,
(unsigned char *)initiatorname);
if (!sess->se_node_acl) {
transport_free_session(sess);
return ERR_PTR(-EACCES);
}
/*
* Go ahead and perform any remaining fabric setup that is
* required before transport_register_session().
*/
if (callback != NULL) {
int rc = callback(tpg, sess, private);
if (rc) {
transport_free_session(sess);
return ERR_PTR(rc);
}
}
transport_register_session(tpg, sess->se_node_acl, sess, private);
return sess;
}
EXPORT_SYMBOL(target_alloc_session);
ssize_t target_show_dynamic_sessions(struct se_portal_group *se_tpg, char *page)
{
struct se_session *se_sess;
ssize_t len = 0;
spin_lock_bh(&se_tpg->session_lock);
list_for_each_entry(se_sess, &se_tpg->tpg_sess_list, sess_list) {
if (!se_sess->se_node_acl)
continue;
if (!se_sess->se_node_acl->dynamic_node_acl)
continue;
if (strlen(se_sess->se_node_acl->initiatorname) + 1 + len > PAGE_SIZE)
break;
len += snprintf(page + len, PAGE_SIZE - len, "%s\n",
se_sess->se_node_acl->initiatorname);
len += 1; /* Include NULL terminator */
}
spin_unlock_bh(&se_tpg->session_lock);
return len;
}
EXPORT_SYMBOL(target_show_dynamic_sessions);
static void target_complete_nacl(struct kref *kref)
{
struct se_node_acl *nacl = container_of(kref,
struct se_node_acl, acl_kref);
struct se_portal_group *se_tpg = nacl->se_tpg;
if (!nacl->dynamic_stop) {
complete(&nacl->acl_free_comp);
return;
}
mutex_lock(&se_tpg->acl_node_mutex);
target: Fix node_acl demo-mode + uncached dynamic shutdown regression This patch fixes a generate_node_acls = 1 + cache_dynamic_acls = 0 regression, that was introduced by commit 01d4d673558985d9a118e1e05026633c3e2ade9b Author: Nicholas Bellinger <nab@linux-iscsi.org> Date: Wed Dec 7 12:55:54 2016 -0800 which originally had the proper list_del_init() usage, but was dropped during list review as it was thought unnecessary by HCH. However, list_del_init() usage is required during the special generate_node_acls = 1 + cache_dynamic_acls = 0 case when transport_free_session() does a list_del(&se_nacl->acl_list), followed by target_complete_nacl() doing the same thing. This was manifesting as a general protection fault as reported by Justin: kernel: general protection fault: 0000 [#1] SMP kernel: Modules linked in: kernel: CPU: 0 PID: 11047 Comm: iscsi_ttx Not tainted 4.13.0-rc2.x86_64.1+ #20 kernel: Hardware name: Intel Corporation S5500BC/S5500BC, BIOS S5500.86B.01.00.0064.050520141428 05/05/2014 kernel: task: ffff88026939e800 task.stack: ffffc90007884000 kernel: RIP: 0010:target_put_nacl+0x49/0xb0 kernel: RSP: 0018:ffffc90007887d70 EFLAGS: 00010246 kernel: RAX: dead000000000200 RBX: ffff8802556ca000 RCX: 0000000000000000 kernel: RDX: dead000000000100 RSI: 0000000000000246 RDI: ffff8802556ce028 kernel: RBP: ffffc90007887d88 R08: 0000000000000001 R09: 0000000000000000 kernel: R10: ffffc90007887df8 R11: ffffea0009986900 R12: ffff8802556ce020 kernel: R13: ffff8802556ce028 R14: ffff8802556ce028 R15: ffffffff88d85540 kernel: FS: 0000000000000000(0000) GS:ffff88027fc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fffe36f5f94 CR3: 0000000009209000 CR4: 00000000003406f0 kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 kernel: DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 kernel: Call Trace: kernel: transport_free_session+0x67/0x140 kernel: transport_deregister_session+0x7a/0xc0 kernel: iscsit_close_session+0x92/0x210 kernel: iscsit_close_connection+0x5f9/0x840 kernel: iscsit_take_action_for_connection_exit+0xfe/0x110 kernel: iscsi_target_tx_thread+0x140/0x1e0 kernel: ? wait_woken+0x90/0x90 kernel: kthread+0x124/0x160 kernel: ? iscsit_thread_get_cpumask+0x90/0x90 kernel: ? kthread_create_on_node+0x40/0x40 kernel: ret_from_fork+0x22/0x30 kernel: Code: 00 48 89 fb 4c 8b a7 48 01 00 00 74 68 4d 8d 6c 24 08 4c 89 ef e8 e8 28 43 00 48 8b 93 20 04 00 00 48 8b 83 28 04 00 00 4c 89 ef <48> 89 42 08 48 89 10 48 b8 00 01 00 00 00 00 ad de 48 89 83 20 kernel: RIP: target_put_nacl+0x49/0xb0 RSP: ffffc90007887d70 kernel: ---[ end trace f12821adbfd46fed ]--- To address this, go ahead and use proper list_del_list() for all cases of se_nacl->acl_list deletion. Reported-by: Justin Maggard <jmaggard01@gmail.com> Tested-by: Justin Maggard <jmaggard01@gmail.com> Cc: Justin Maggard <jmaggard01@gmail.com> Cc: stable@vger.kernel.org # 4.1+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-08-07 02:10:03 +03:00
list_del_init(&nacl->acl_list);
mutex_unlock(&se_tpg->acl_node_mutex);
core_tpg_wait_for_nacl_pr_ref(nacl);
core_free_device_list_for_node(nacl, se_tpg);
kfree(nacl);
}
void target_put_nacl(struct se_node_acl *nacl)
{
kref_put(&nacl->acl_kref, target_complete_nacl);
}
EXPORT_SYMBOL(target_put_nacl);
void transport_deregister_session_configfs(struct se_session *se_sess)
{
struct se_node_acl *se_nacl;
unsigned long flags;
/*
* Used by struct se_node_acl's under ConfigFS to locate active struct se_session
*/
se_nacl = se_sess->se_node_acl;
if (se_nacl) {
spin_lock_irqsave(&se_nacl->nacl_sess_lock, flags);
if (!list_empty(&se_sess->sess_acl_list))
list_del_init(&se_sess->sess_acl_list);
/*
* If the session list is empty, then clear the pointer.
* Otherwise, set the struct se_session pointer from the tail
* element of the per struct se_node_acl active session list.
*/
if (list_empty(&se_nacl->acl_sess_list))
se_nacl->nacl_sess = NULL;
else {
se_nacl->nacl_sess = container_of(
se_nacl->acl_sess_list.prev,
struct se_session, sess_acl_list);
}
spin_unlock_irqrestore(&se_nacl->nacl_sess_lock, flags);
}
}
EXPORT_SYMBOL(transport_deregister_session_configfs);
void transport_free_session(struct se_session *se_sess)
{
struct se_node_acl *se_nacl = se_sess->se_node_acl;
/*
* Drop the se_node_acl->nacl_kref obtained from within
* core_tpg_get_initiator_node_acl().
*/
if (se_nacl) {
struct se_portal_group *se_tpg = se_nacl->se_tpg;
const struct target_core_fabric_ops *se_tfo = se_tpg->se_tpg_tfo;
unsigned long flags;
se_sess->se_node_acl = NULL;
/*
* Also determine if we need to drop the extra ->cmd_kref if
* it had been previously dynamically generated, and
* the endpoint is not caching dynamic ACLs.
*/
mutex_lock(&se_tpg->acl_node_mutex);
if (se_nacl->dynamic_node_acl &&
!se_tfo->tpg_check_demo_mode_cache(se_tpg)) {
spin_lock_irqsave(&se_nacl->nacl_sess_lock, flags);
if (list_empty(&se_nacl->acl_sess_list))
se_nacl->dynamic_stop = true;
spin_unlock_irqrestore(&se_nacl->nacl_sess_lock, flags);
if (se_nacl->dynamic_stop)
target: Fix node_acl demo-mode + uncached dynamic shutdown regression This patch fixes a generate_node_acls = 1 + cache_dynamic_acls = 0 regression, that was introduced by commit 01d4d673558985d9a118e1e05026633c3e2ade9b Author: Nicholas Bellinger <nab@linux-iscsi.org> Date: Wed Dec 7 12:55:54 2016 -0800 which originally had the proper list_del_init() usage, but was dropped during list review as it was thought unnecessary by HCH. However, list_del_init() usage is required during the special generate_node_acls = 1 + cache_dynamic_acls = 0 case when transport_free_session() does a list_del(&se_nacl->acl_list), followed by target_complete_nacl() doing the same thing. This was manifesting as a general protection fault as reported by Justin: kernel: general protection fault: 0000 [#1] SMP kernel: Modules linked in: kernel: CPU: 0 PID: 11047 Comm: iscsi_ttx Not tainted 4.13.0-rc2.x86_64.1+ #20 kernel: Hardware name: Intel Corporation S5500BC/S5500BC, BIOS S5500.86B.01.00.0064.050520141428 05/05/2014 kernel: task: ffff88026939e800 task.stack: ffffc90007884000 kernel: RIP: 0010:target_put_nacl+0x49/0xb0 kernel: RSP: 0018:ffffc90007887d70 EFLAGS: 00010246 kernel: RAX: dead000000000200 RBX: ffff8802556ca000 RCX: 0000000000000000 kernel: RDX: dead000000000100 RSI: 0000000000000246 RDI: ffff8802556ce028 kernel: RBP: ffffc90007887d88 R08: 0000000000000001 R09: 0000000000000000 kernel: R10: ffffc90007887df8 R11: ffffea0009986900 R12: ffff8802556ce020 kernel: R13: ffff8802556ce028 R14: ffff8802556ce028 R15: ffffffff88d85540 kernel: FS: 0000000000000000(0000) GS:ffff88027fc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fffe36f5f94 CR3: 0000000009209000 CR4: 00000000003406f0 kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 kernel: DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 kernel: Call Trace: kernel: transport_free_session+0x67/0x140 kernel: transport_deregister_session+0x7a/0xc0 kernel: iscsit_close_session+0x92/0x210 kernel: iscsit_close_connection+0x5f9/0x840 kernel: iscsit_take_action_for_connection_exit+0xfe/0x110 kernel: iscsi_target_tx_thread+0x140/0x1e0 kernel: ? wait_woken+0x90/0x90 kernel: kthread+0x124/0x160 kernel: ? iscsit_thread_get_cpumask+0x90/0x90 kernel: ? kthread_create_on_node+0x40/0x40 kernel: ret_from_fork+0x22/0x30 kernel: Code: 00 48 89 fb 4c 8b a7 48 01 00 00 74 68 4d 8d 6c 24 08 4c 89 ef e8 e8 28 43 00 48 8b 93 20 04 00 00 48 8b 83 28 04 00 00 4c 89 ef <48> 89 42 08 48 89 10 48 b8 00 01 00 00 00 00 ad de 48 89 83 20 kernel: RIP: target_put_nacl+0x49/0xb0 RSP: ffffc90007887d70 kernel: ---[ end trace f12821adbfd46fed ]--- To address this, go ahead and use proper list_del_list() for all cases of se_nacl->acl_list deletion. Reported-by: Justin Maggard <jmaggard01@gmail.com> Tested-by: Justin Maggard <jmaggard01@gmail.com> Cc: Justin Maggard <jmaggard01@gmail.com> Cc: stable@vger.kernel.org # 4.1+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-08-07 02:10:03 +03:00
list_del_init(&se_nacl->acl_list);
}
mutex_unlock(&se_tpg->acl_node_mutex);
if (se_nacl->dynamic_stop)
target_put_nacl(se_nacl);
target_put_nacl(se_nacl);
}
if (se_sess->sess_cmd_map) {
percpu_ida_destroy(&se_sess->sess_tag_pool);
kvfree(se_sess->sess_cmd_map);
}
kmem_cache_free(se_sess_cache, se_sess);
}
EXPORT_SYMBOL(transport_free_session);
void transport_deregister_session(struct se_session *se_sess)
{
struct se_portal_group *se_tpg = se_sess->se_tpg;
unsigned long flags;
if (!se_tpg) {
transport_free_session(se_sess);
return;
}
spin_lock_irqsave(&se_tpg->session_lock, flags);
list_del(&se_sess->sess_list);
se_sess->se_tpg = NULL;
se_sess->fabric_sess_ptr = NULL;
spin_unlock_irqrestore(&se_tpg->session_lock, flags);
pr_debug("TARGET_CORE[%s]: Deregistered fabric_sess\n",
2011-07-19 12:55:10 +04:00
se_tpg->se_tpg_tfo->get_fabric_name());
/*
* If last kref is dropping now for an explicit NodeACL, awake sleeping
* ->acl_free_comp caller to wakeup configfs se_node_acl->acl_group
* removal context from within transport_free_session() code.
*
* For dynamic ACL, target_put_nacl() uses target_complete_nacl()
* to release all remaining generate_node_acl=1 created ACL resources.
*/
transport_free_session(se_sess);
}
EXPORT_SYMBOL(transport_deregister_session);
static void target_remove_from_state_list(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
unsigned long flags;
if (!dev)
return;
spin_lock_irqsave(&dev->execute_task_lock, flags);
if (cmd->state_active) {
list_del(&cmd->state_list);
cmd->state_active = false;
}
spin_unlock_irqrestore(&dev->execute_task_lock, flags);
}
static int transport_cmd_check_stop_to_fabric(struct se_cmd *cmd)
{
unsigned long flags;
target_remove_from_state_list(cmd);
/*
* Clear struct se_cmd->se_lun before the handoff to FE.
*/
cmd->se_lun = NULL;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
spin_lock_irqsave(&cmd->t_state_lock, flags);
/*
* Determine if frontend context caller is requesting the stopping of
2011-07-19 12:55:10 +04:00
* this command for frontend exceptions.
*/
if (cmd->transport_state & CMD_T_STOP) {
pr_debug("%s:%d CMD_T_STOP for ITT: 0x%08llx\n",
__func__, __LINE__, cmd->tag);
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
complete_all(&cmd->t_transport_stop_comp);
return 1;
}
cmd->transport_state &= ~CMD_T_ACTIVE;
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
/*
* Some fabric modules like tcm_loop can release their internally
* allocated I/O reference and struct se_cmd now.
*
* Fabric modules are expected to return '1' here if the se_cmd being
* passed is released at this point, or zero if not being released.
*/
return cmd->se_tfo->check_stop_free(cmd);
}
static void transport_lun_remove_cmd(struct se_cmd *cmd)
{
2011-07-19 12:55:10 +04:00
struct se_lun *lun = cmd->se_lun;
if (!lun)
return;
if (cmpxchg(&cmd->lun_ref_active, true, false))
percpu_ref_put(&lun->lun_ref);
}
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 06:00:17 +03:00
int transport_cmd_finish_abort(struct se_cmd *cmd, int remove)
{
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
bool ack_kref = (cmd->se_cmd_flags & SCF_ACK_KREF);
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 06:00:17 +03:00
int ret = 0;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
if (cmd->se_cmd_flags & SCF_SE_LUN_CMD)
transport_lun_remove_cmd(cmd);
target: Add TFO->abort_task for aborted task resources release Now that TASK_ABORTED status is not generated for all cases by TMR ABORT_TASK + LUN_RESET, a new TFO->abort_task() caller is necessary in order to give fabric drivers a chance to unmap hardware / software resources before the se_cmd descriptor is released via the normal TFO->release_cmd() codepath. This patch adds TFO->aborted_task() in core_tmr_abort_task() in place of the original transport_send_task_abort(), and also updates all fabric drivers to implement this caller. The fabric drivers that include changes to perform cleanup via ->aborted_task() are: - iscsi-target - iser-target - srpt - tcm_qla2xxx The fabric drivers that currently set ->aborted_task() to NOPs are: - loopback - tcm_fc - usb-gadget - sbp-target - vhost-scsi For the latter five, there appears to be no additional cleanup required before invoking TFO->release_cmd() to release the se_cmd descriptor. v2 changes: - Move ->aborted_task() call into transport_cmd_finish_abort (Alex) Cc: Alex Leung <amleung21@yahoo.com> Cc: Mark Rustad <mark.d.rustad@intel.com> Cc: Roland Dreier <roland@kernel.org> Cc: Vu Pham <vu@mellanox.com> Cc: Chris Boot <bootc@bootc.net> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Giridhar Malavali <giridhar.malavali@qlogic.com> Cc: Saurav Kashyap <saurav.kashyap@qlogic.com> Cc: Quinn Tran <quinn.tran@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2014-03-23 01:55:56 +04:00
/*
* Allow the fabric driver to unmap any resources before
* releasing the descriptor via TFO->release_cmd()
*/
if (remove)
cmd->se_tfo->aborted_task(cmd);
if (transport_cmd_check_stop_to_fabric(cmd))
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 06:00:17 +03:00
return 1;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
if (remove && ack_kref)
ret = target_put_sess_cmd(cmd);
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 06:00:17 +03:00
return ret;
}
static void target_complete_failure_work(struct work_struct *work)
{
struct se_cmd *cmd = container_of(work, struct se_cmd, work);
transport_generic_request_failure(cmd,
TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE);
}
/*
* Used when asking transport to copy Sense Data from the underlying
* Linux/SCSI struct scsi_cmnd
*/
static unsigned char *transport_get_sense_buffer(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
WARN_ON(!cmd->se_lun);
if (!dev)
return NULL;
if (cmd->se_cmd_flags & SCF_SENT_CHECK_CONDITION)
return NULL;
cmd->scsi_sense_length = TRANSPORT_SENSE_BUFFER;
pr_debug("HBA_[%u]_PLUG[%s]: Requesting sense for SAM STATUS: 0x%02x\n",
dev->se_hba->hba_id, dev->transport->name, cmd->scsi_status);
return cmd->sense_buffer;
}
void transport_copy_sense_to_cmd(struct se_cmd *cmd, unsigned char *sense)
{
unsigned char *cmd_sense_buf;
unsigned long flags;
spin_lock_irqsave(&cmd->t_state_lock, flags);
cmd_sense_buf = transport_get_sense_buffer(cmd);
if (!cmd_sense_buf) {
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
return;
}
cmd->se_cmd_flags |= SCF_TRANSPORT_TASK_SENSE;
memcpy(cmd_sense_buf, sense, cmd->scsi_sense_length);
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
}
EXPORT_SYMBOL(transport_copy_sense_to_cmd);
void target_complete_cmd(struct se_cmd *cmd, u8 scsi_status)
{
struct se_device *dev = cmd->se_dev;
int success;
unsigned long flags;
cmd->scsi_status = scsi_status;
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_lock_irqsave(&cmd->t_state_lock, flags);
switch (cmd->scsi_status) {
case SAM_STAT_CHECK_CONDITION:
if (cmd->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE)
success = 1;
else
success = 0;
break;
default:
success = 1;
break;
}
/*
* Check for case where an explicit ABORT_TASK has been received
* and transport_wait_for_tasks() will be waiting for completion..
*/
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
if (cmd->transport_state & CMD_T_ABORTED ||
cmd->transport_state & CMD_T_STOP) {
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
/*
* If COMPARE_AND_WRITE was stopped by __transport_wait_for_tasks(),
* release se_device->caw_sem obtained by sbc_compare_and_write()
* since target_complete_ok_work() or target_complete_failure_work()
* won't be called to invoke the normal CAW completion callbacks.
*/
if (cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) {
up(&dev->caw_sem);
}
complete_all(&cmd->t_transport_stop_comp);
return;
} else if (!success) {
INIT_WORK(&cmd->work, target_complete_failure_work);
} else {
INIT_WORK(&cmd->work, target_complete_ok_work);
}
cmd->t_state = TRANSPORT_COMPLETE;
cmd->transport_state |= (CMD_T_COMPLETE | CMD_T_ACTIVE);
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
if (cmd->se_cmd_flags & SCF_USE_CPUID)
queue_work_on(cmd->cpuid, target_completion_wq, &cmd->work);
else
queue_work(target_completion_wq, &cmd->work);
}
EXPORT_SYMBOL(target_complete_cmd);
void target_complete_cmd_with_length(struct se_cmd *cmd, u8 scsi_status, int length)
{
if ((scsi_status == SAM_STAT_GOOD ||
cmd->se_cmd_flags & SCF_TREAT_READ_AS_NORMAL) &&
length < cmd->data_length) {
if (cmd->se_cmd_flags & SCF_UNDERFLOW_BIT) {
cmd->residual_count += cmd->data_length - length;
} else {
cmd->se_cmd_flags |= SCF_UNDERFLOW_BIT;
cmd->residual_count = cmd->data_length - length;
}
cmd->data_length = length;
}
target_complete_cmd(cmd, scsi_status);
}
EXPORT_SYMBOL(target_complete_cmd_with_length);
static void target_add_to_state_list(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
unsigned long flags;
spin_lock_irqsave(&dev->execute_task_lock, flags);
if (!cmd->state_active) {
list_add_tail(&cmd->state_list, &dev->state_list);
cmd->state_active = true;
}
spin_unlock_irqrestore(&dev->execute_task_lock, flags);
}
/*
* Handle QUEUE_FULL / -EAGAIN and -ENOMEM status
*/
static void transport_write_pending_qf(struct se_cmd *cmd);
static void transport_complete_qf(struct se_cmd *cmd);
void target_qf_do_work(struct work_struct *work)
{
struct se_device *dev = container_of(work, struct se_device,
qf_work_queue);
LIST_HEAD(qf_cmd_list);
struct se_cmd *cmd, *cmd_tmp;
spin_lock_irq(&dev->qf_cmd_lock);
list_splice_init(&dev->qf_cmd_list, &qf_cmd_list);
spin_unlock_irq(&dev->qf_cmd_lock);
list_for_each_entry_safe(cmd, cmd_tmp, &qf_cmd_list, se_qf_node) {
list_del(&cmd->se_qf_node);
atomic_dec_mb(&dev->dev_qf_count);
pr_debug("Processing %s cmd: %p QUEUE_FULL in work queue"
" context: %s\n", cmd->se_tfo->get_fabric_name(), cmd,
(cmd->t_state == TRANSPORT_COMPLETE_QF_OK) ? "COMPLETE_OK" :
(cmd->t_state == TRANSPORT_COMPLETE_QF_WP) ? "WRITE_PENDING"
: "UNKNOWN");
if (cmd->t_state == TRANSPORT_COMPLETE_QF_WP)
transport_write_pending_qf(cmd);
else if (cmd->t_state == TRANSPORT_COMPLETE_QF_OK ||
cmd->t_state == TRANSPORT_COMPLETE_QF_ERR)
transport_complete_qf(cmd);
}
}
unsigned char *transport_dump_cmd_direction(struct se_cmd *cmd)
{
switch (cmd->data_direction) {
case DMA_NONE:
return "NONE";
case DMA_FROM_DEVICE:
return "READ";
case DMA_TO_DEVICE:
return "WRITE";
case DMA_BIDIRECTIONAL:
return "BIDI";
default:
break;
}
return "UNKNOWN";
}
void transport_dump_dev_state(
struct se_device *dev,
char *b,
int *bl)
{
*bl += sprintf(b + *bl, "Status: ");
if (dev->export_count)
*bl += sprintf(b + *bl, "ACTIVATED");
else
*bl += sprintf(b + *bl, "DEACTIVATED");
*bl += sprintf(b + *bl, " Max Queue Depth: %d", dev->queue_depth);
*bl += sprintf(b + *bl, " SectorSize: %u HwMaxSectors: %u\n",
dev->dev_attrib.block_size,
dev->dev_attrib.hw_max_sectors);
*bl += sprintf(b + *bl, " ");
}
void transport_dump_vpd_proto_id(
struct t10_vpd *vpd,
unsigned char *p_buf,
int p_buf_len)
{
unsigned char buf[VPD_TMP_BUF_SIZE];
int len;
memset(buf, 0, VPD_TMP_BUF_SIZE);
len = sprintf(buf, "T10 VPD Protocol Identifier: ");
switch (vpd->protocol_identifier) {
case 0x00:
sprintf(buf+len, "Fibre Channel\n");
break;
case 0x10:
sprintf(buf+len, "Parallel SCSI\n");
break;
case 0x20:
sprintf(buf+len, "SSA\n");
break;
case 0x30:
sprintf(buf+len, "IEEE 1394\n");
break;
case 0x40:
sprintf(buf+len, "SCSI Remote Direct Memory Access"
" Protocol\n");
break;
case 0x50:
sprintf(buf+len, "Internet SCSI (iSCSI)\n");
break;
case 0x60:
sprintf(buf+len, "SAS Serial SCSI Protocol\n");
break;
case 0x70:
sprintf(buf+len, "Automation/Drive Interface Transport"
" Protocol\n");
break;
case 0x80:
sprintf(buf+len, "AT Attachment Interface ATA/ATAPI\n");
break;
default:
sprintf(buf+len, "Unknown 0x%02x\n",
vpd->protocol_identifier);
break;
}
if (p_buf)
strncpy(p_buf, buf, p_buf_len);
else
pr_debug("%s", buf);
}
void
transport_set_vpd_proto_id(struct t10_vpd *vpd, unsigned char *page_83)
{
/*
* Check if the Protocol Identifier Valid (PIV) bit is set..
*
* from spc3r23.pdf section 7.5.1
*/
if (page_83[1] & 0x80) {
vpd->protocol_identifier = (page_83[0] & 0xf0);
vpd->protocol_identifier_set = 1;
transport_dump_vpd_proto_id(vpd, NULL, 0);
}
}
EXPORT_SYMBOL(transport_set_vpd_proto_id);
int transport_dump_vpd_assoc(
struct t10_vpd *vpd,
unsigned char *p_buf,
int p_buf_len)
{
unsigned char buf[VPD_TMP_BUF_SIZE];
2011-07-19 12:55:10 +04:00
int ret = 0;
int len;
memset(buf, 0, VPD_TMP_BUF_SIZE);
len = sprintf(buf, "T10 VPD Identifier Association: ");
switch (vpd->association) {
case 0x00:
sprintf(buf+len, "addressed logical unit\n");
break;
case 0x10:
sprintf(buf+len, "target port\n");
break;
case 0x20:
sprintf(buf+len, "SCSI target device\n");
break;
default:
sprintf(buf+len, "Unknown 0x%02x\n", vpd->association);
2011-07-19 12:55:10 +04:00
ret = -EINVAL;
break;
}
if (p_buf)
strncpy(p_buf, buf, p_buf_len);
else
pr_debug("%s", buf);
return ret;
}
int transport_set_vpd_assoc(struct t10_vpd *vpd, unsigned char *page_83)
{
/*
* The VPD identification association..
*
* from spc3r23.pdf Section 7.6.3.1 Table 297
*/
vpd->association = (page_83[1] & 0x30);
return transport_dump_vpd_assoc(vpd, NULL, 0);
}
EXPORT_SYMBOL(transport_set_vpd_assoc);
int transport_dump_vpd_ident_type(
struct t10_vpd *vpd,
unsigned char *p_buf,
int p_buf_len)
{
unsigned char buf[VPD_TMP_BUF_SIZE];
2011-07-19 12:55:10 +04:00
int ret = 0;
int len;
memset(buf, 0, VPD_TMP_BUF_SIZE);
len = sprintf(buf, "T10 VPD Identifier Type: ");
switch (vpd->device_identifier_type) {
case 0x00:
sprintf(buf+len, "Vendor specific\n");
break;
case 0x01:
sprintf(buf+len, "T10 Vendor ID based\n");
break;
case 0x02:
sprintf(buf+len, "EUI-64 based\n");
break;
case 0x03:
sprintf(buf+len, "NAA\n");
break;
case 0x04:
sprintf(buf+len, "Relative target port identifier\n");
break;
case 0x08:
sprintf(buf+len, "SCSI name string\n");
break;
default:
sprintf(buf+len, "Unsupported: 0x%02x\n",
vpd->device_identifier_type);
2011-07-19 12:55:10 +04:00
ret = -EINVAL;
break;
}
2011-07-19 12:55:10 +04:00
if (p_buf) {
if (p_buf_len < strlen(buf)+1)
return -EINVAL;
strncpy(p_buf, buf, p_buf_len);
2011-07-19 12:55:10 +04:00
} else {
pr_debug("%s", buf);
2011-07-19 12:55:10 +04:00
}
return ret;
}
int transport_set_vpd_ident_type(struct t10_vpd *vpd, unsigned char *page_83)
{
/*
* The VPD identifier type..
*
* from spc3r23.pdf Section 7.6.3.1 Table 298
*/
vpd->device_identifier_type = (page_83[1] & 0x0f);
return transport_dump_vpd_ident_type(vpd, NULL, 0);
}
EXPORT_SYMBOL(transport_set_vpd_ident_type);
int transport_dump_vpd_ident(
struct t10_vpd *vpd,
unsigned char *p_buf,
int p_buf_len)
{
unsigned char buf[VPD_TMP_BUF_SIZE];
int ret = 0;
memset(buf, 0, VPD_TMP_BUF_SIZE);
switch (vpd->device_identifier_code_set) {
case 0x01: /* Binary */
snprintf(buf, sizeof(buf),
"T10 VPD Binary Device Identifier: %s\n",
&vpd->device_identifier[0]);
break;
case 0x02: /* ASCII */
snprintf(buf, sizeof(buf),
"T10 VPD ASCII Device Identifier: %s\n",
&vpd->device_identifier[0]);
break;
case 0x03: /* UTF-8 */
snprintf(buf, sizeof(buf),
"T10 VPD UTF-8 Device Identifier: %s\n",
&vpd->device_identifier[0]);
break;
default:
sprintf(buf, "T10 VPD Device Identifier encoding unsupported:"
" 0x%02x", vpd->device_identifier_code_set);
2011-07-19 12:55:10 +04:00
ret = -EINVAL;
break;
}
if (p_buf)
strncpy(p_buf, buf, p_buf_len);
else
pr_debug("%s", buf);
return ret;
}
int
transport_set_vpd_ident(struct t10_vpd *vpd, unsigned char *page_83)
{
static const char hex_str[] = "0123456789abcdef";
int j = 0, i = 4; /* offset to start of the identifier */
/*
* The VPD Code Set (encoding)
*
* from spc3r23.pdf Section 7.6.3.1 Table 296
*/
vpd->device_identifier_code_set = (page_83[0] & 0x0f);
switch (vpd->device_identifier_code_set) {
case 0x01: /* Binary */
vpd->device_identifier[j++] =
hex_str[vpd->device_identifier_type];
while (i < (4 + page_83[3])) {
vpd->device_identifier[j++] =
hex_str[(page_83[i] & 0xf0) >> 4];
vpd->device_identifier[j++] =
hex_str[page_83[i] & 0x0f];
i++;
}
break;
case 0x02: /* ASCII */
case 0x03: /* UTF-8 */
while (i < (4 + page_83[3]))
vpd->device_identifier[j++] = page_83[i++];
break;
default:
break;
}
return transport_dump_vpd_ident(vpd, NULL, 0);
}
EXPORT_SYMBOL(transport_set_vpd_ident);
target/qla2xxx: Honor max_data_sg_nents I/O transfer limit This patch adds an optional fabric driver provided SGL limit that target-core will honor as it's own internal I/O maximum transfer length limit, as exposed by EVPD=0xb0 block limits parameters. This is required for handling cases when host I/O transfer length exceeds the requested EVPD block limits maximum transfer length. The initial user of this logic is qla2xxx, so that we can avoid having to reject I/Os from some legacy FC hosts where EVPD=0xb0 parameters are not honored. When se_cmd payload length exceeds the provided limit in target_check_max_data_sg_nents() code, se_cmd->data_length + se_cmd->prot_length are reset with se_cmd->residual_count plus underflow bit for outgoing TFO response callbacks. It also checks for existing CDB level underflow + overflow and recalculates final residual_count as necessary. Note this patch currently assumes 1:1 mapping of PAGE_SIZE per struct scatterlist entry. Reported-by: Craig Watson <craig.watson@vanguard-rugged.com> Cc: Craig Watson <craig.watson@vanguard-rugged.com> Tested-by: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Roland Dreier <roland@purestorage.com> Cc: Arun Easi <arun.easi@qlogic.com> Cc: Giridhar Malavali <giridhar.malavali@qlogic.com> Cc: Andrew Vasquez <andrew.vasquez@qlogic.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2015-07-31 04:28:13 +03:00
static sense_reason_t
target_check_max_data_sg_nents(struct se_cmd *cmd, struct se_device *dev,
unsigned int size)
{
u32 mtl;
if (!cmd->se_tfo->max_data_sg_nents)
return TCM_NO_SENSE;
/*
* Check if fabric enforced maximum SGL entries per I/O descriptor
* exceeds se_cmd->data_length. If true, set SCF_UNDERFLOW_BIT +
* residual_count and reduce original cmd->data_length to maximum
* length based on single PAGE_SIZE entry scatter-lists.
*/
mtl = (cmd->se_tfo->max_data_sg_nents * PAGE_SIZE);
if (cmd->data_length > mtl) {
/*
* If an existing CDB overflow is present, calculate new residual
* based on CDB size minus fabric maximum transfer length.
*
* If an existing CDB underflow is present, calculate new residual
* based on original cmd->data_length minus fabric maximum transfer
* length.
*
* Otherwise, set the underflow residual based on cmd->data_length
* minus fabric maximum transfer length.
*/
if (cmd->se_cmd_flags & SCF_OVERFLOW_BIT) {
cmd->residual_count = (size - mtl);
} else if (cmd->se_cmd_flags & SCF_UNDERFLOW_BIT) {
u32 orig_dl = size + cmd->residual_count;
cmd->residual_count = (orig_dl - mtl);
} else {
cmd->se_cmd_flags |= SCF_UNDERFLOW_BIT;
cmd->residual_count = (cmd->data_length - mtl);
}
cmd->data_length = mtl;
/*
* Reset sbc_check_prot() calculated protection payload
* length based upon the new smaller MTL.
*/
if (cmd->prot_length) {
u32 sectors = (mtl / dev->dev_attrib.block_size);
cmd->prot_length = dev->prot_length * sectors;
}
}
return TCM_NO_SENSE;
}
sense_reason_t
target_cmd_size_check(struct se_cmd *cmd, unsigned int size)
{
struct se_device *dev = cmd->se_dev;
if (cmd->unknown_data_length) {
cmd->data_length = size;
} else if (size != cmd->data_length) {
pr_warn_ratelimited("TARGET_CORE[%s]: Expected Transfer Length:"
" %u does not match SCSI CDB Length: %u for SAM Opcode:"
" 0x%02x\n", cmd->se_tfo->get_fabric_name(),
cmd->data_length, size, cmd->t_task_cdb[0]);
if (cmd->data_direction == DMA_TO_DEVICE) {
if (cmd->se_cmd_flags & SCF_SCSI_DATA_CDB) {
pr_err_ratelimited("Rejecting underflow/overflow"
" for WRITE data CDB\n");
return TCM_INVALID_CDB_FIELD;
}
/*
* Some fabric drivers like iscsi-target still expect to
* always reject overflow writes. Reject this case until
* full fabric driver level support for overflow writes
* is introduced tree-wide.
*/
if (size > cmd->data_length) {
pr_err_ratelimited("Rejecting overflow for"
" WRITE control CDB\n");
return TCM_INVALID_CDB_FIELD;
}
}
/*
* Reject READ_* or WRITE_* with overflow/underflow for
* type SCF_SCSI_DATA_CDB.
*/
if (dev->dev_attrib.block_size != 512) {
pr_err("Failing OVERFLOW/UNDERFLOW for LBA op"
" CDB on non 512-byte sector setup subsystem"
" plugin: %s\n", dev->transport->name);
/* Returns CHECK_CONDITION + INVALID_CDB_FIELD */
return TCM_INVALID_CDB_FIELD;
}
/*
* For the overflow case keep the existing fabric provided
* ->data_length. Otherwise for the underflow case, reset
* ->data_length to the smaller SCSI expected data transfer
* length.
*/
if (size > cmd->data_length) {
cmd->se_cmd_flags |= SCF_OVERFLOW_BIT;
cmd->residual_count = (size - cmd->data_length);
} else {
cmd->se_cmd_flags |= SCF_UNDERFLOW_BIT;
cmd->residual_count = (cmd->data_length - size);
cmd->data_length = size;
}
}
target/qla2xxx: Honor max_data_sg_nents I/O transfer limit This patch adds an optional fabric driver provided SGL limit that target-core will honor as it's own internal I/O maximum transfer length limit, as exposed by EVPD=0xb0 block limits parameters. This is required for handling cases when host I/O transfer length exceeds the requested EVPD block limits maximum transfer length. The initial user of this logic is qla2xxx, so that we can avoid having to reject I/Os from some legacy FC hosts where EVPD=0xb0 parameters are not honored. When se_cmd payload length exceeds the provided limit in target_check_max_data_sg_nents() code, se_cmd->data_length + se_cmd->prot_length are reset with se_cmd->residual_count plus underflow bit for outgoing TFO response callbacks. It also checks for existing CDB level underflow + overflow and recalculates final residual_count as necessary. Note this patch currently assumes 1:1 mapping of PAGE_SIZE per struct scatterlist entry. Reported-by: Craig Watson <craig.watson@vanguard-rugged.com> Cc: Craig Watson <craig.watson@vanguard-rugged.com> Tested-by: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Roland Dreier <roland@purestorage.com> Cc: Arun Easi <arun.easi@qlogic.com> Cc: Giridhar Malavali <giridhar.malavali@qlogic.com> Cc: Andrew Vasquez <andrew.vasquez@qlogic.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2015-07-31 04:28:13 +03:00
return target_check_max_data_sg_nents(cmd, dev, size);
}
/*
* Used by fabric modules containing a local struct se_cmd within their
* fabric dependent per I/O descriptor.
*
* Preserves the value of @cmd->tag.
*/
void transport_init_se_cmd(
struct se_cmd *cmd,
const struct target_core_fabric_ops *tfo,
struct se_session *se_sess,
u32 data_length,
int data_direction,
int task_attr,
unsigned char *sense_buffer)
{
INIT_LIST_HEAD(&cmd->se_delayed_node);
INIT_LIST_HEAD(&cmd->se_qf_node);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
INIT_LIST_HEAD(&cmd->se_cmd_list);
INIT_LIST_HEAD(&cmd->state_list);
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
init_completion(&cmd->t_transport_stop_comp);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
init_completion(&cmd->cmd_wait_comp);
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_lock_init(&cmd->t_state_lock);
INIT_WORK(&cmd->work, NULL);
kref_init(&cmd->cmd_kref);
cmd->se_tfo = tfo;
cmd->se_sess = se_sess;
cmd->data_length = data_length;
cmd->data_direction = data_direction;
cmd->sam_task_attr = task_attr;
cmd->sense_buffer = sense_buffer;
cmd->state_active = false;
}
EXPORT_SYMBOL(transport_init_se_cmd);
static sense_reason_t
transport_check_alloc_task_attr(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
/*
* Check if SAM Task Attribute emulation is enabled for this
* struct se_device storage object
*/
if (dev->transport->transport_flags & TRANSPORT_FLAG_PASSTHROUGH)
return 0;
if (cmd->sam_task_attr == TCM_ACA_TAG) {
pr_debug("SAM Task Attribute ACA"
" emulation is not supported\n");
return TCM_INVALID_CDB_FIELD;
}
return 0;
}
sense_reason_t
target_setup_cmd_from_cdb(struct se_cmd *cmd, unsigned char *cdb)
{
struct se_device *dev = cmd->se_dev;
sense_reason_t ret;
/*
* Ensure that the received CDB is less than the max (252 + 8) bytes
* for VARIABLE_LENGTH_CMD
*/
if (scsi_command_size(cdb) > SCSI_MAX_VARLEN_CDB_SIZE) {
pr_err("Received SCSI CDB with command_size: %d that"
" exceeds SCSI_MAX_VARLEN_CDB_SIZE: %d\n",
scsi_command_size(cdb), SCSI_MAX_VARLEN_CDB_SIZE);
return TCM_INVALID_CDB_FIELD;
}
/*
* If the received CDB is larger than TCM_MAX_COMMAND_SIZE,
* allocate the additional extended CDB buffer now.. Otherwise
* setup the pointer from __t_task_cdb to t_task_cdb.
*/
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
if (scsi_command_size(cdb) > sizeof(cmd->__t_task_cdb)) {
cmd->t_task_cdb = kzalloc(scsi_command_size(cdb),
GFP_KERNEL);
if (!cmd->t_task_cdb) {
pr_err("Unable to allocate cmd->t_task_cdb"
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
" %u > sizeof(cmd->__t_task_cdb): %lu ops\n",
scsi_command_size(cdb),
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
(unsigned long)sizeof(cmd->__t_task_cdb));
return TCM_OUT_OF_RESOURCES;
}
} else
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
cmd->t_task_cdb = &cmd->__t_task_cdb[0];
/*
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
* Copy the original CDB into cmd->
*/
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
memcpy(cmd->t_task_cdb, cdb, scsi_command_size(cdb));
trace_target_sequencer_start(cmd);
ret = dev->transport->parse_cdb(cmd);
if (ret == TCM_UNSUPPORTED_SCSI_OPCODE)
pr_warn_ratelimited("%s/%s: Unsupported SCSI Opcode 0x%02x, sending CHECK_CONDITION.\n",
cmd->se_tfo->get_fabric_name(),
cmd->se_sess->se_node_acl->initiatorname,
cmd->t_task_cdb[0]);
if (ret)
return ret;
ret = transport_check_alloc_task_attr(cmd);
if (ret)
return ret;
cmd->se_cmd_flags |= SCF_SUPPORTED_SAM_OPCODE;
atomic_long_inc(&cmd->se_lun->lun_stats.cmd_pdus);
return 0;
}
EXPORT_SYMBOL(target_setup_cmd_from_cdb);
/*
* Used by fabric module frontends to queue tasks directly.
* May only be used from process context.
*/
int transport_handle_cdb_direct(
struct se_cmd *cmd)
{
sense_reason_t ret;
target: Fix bug for transport_generic_wait_for_tasks with direct operation This patch fixes a bug in transport_handle_cdb_direct() usage with target_core where transport_generic_wait_for_tasks() was bypassing active I/O + usage of cmd->t_transport_stop_comp because cmd->t_transport_active=1 was not being set before dispatching with transport_generic_new_cmd(). The fix follows existing usage in transport_generic_handle_cdb*() -> transport_add_cmd_to_queue() and set these directly, as well as handle transport_generic_new_cmd() exceptions for QUEUE_FULL and CHECK_CONDITION instead of propigating up to RX context fabric code. The bug was manifesting itself with the following SLUB poison overwritten warnings with iscsi-target v4.1 LUNs using the new process context direct operation during session reinstatement with active I/O exception handling: [885410.498267] ============================================================================= [885410.621622] BUG lio_cmd_cache: Poison overwritten [885410.621791] ----------------------------------------------------------------------------- [885410.621792] [885410.623420] INFO: 0xffff880000cf3750-0xffff880000cf378d. First byte 0x6a instead of 0x6b [885410.626332] INFO: Allocated in iscsit_allocate_cmd+0x1c/0xd4 [iscsi_target_mod] age=345 cpu=1 pid=22554 [885411.855189] INFO: Freed in iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] age=1410 cpu=1 pid=22554 [885411.856048] INFO: Slab 0xffffea000002d480 objects=22 used=0 fp=0xffff880000cf7300 flags=0x4080 [885411.856368] INFO: Object 0xffff880000cf33c0 @offset=13248 fp=0xffff880000cf6780 <SNIP> [885411.955678] Pid: 22554, comm: iscsi_trx Not tainted 3.0.0-rc7+ #30 [885411.956040] Call Trace: [885411.957029] [<ffffffff810e5cf9>] print_trailer+0x12e/0x137 [885412.752879] [<ffffffff810e61d9>] check_bytes_and_report+0xb9/0xfd [885412.754933] [<ffffffff810e62d2>] check_object+0xb5/0x192 [885412.755099] [<ffffffff810e6445>] __free_slab+0x96/0x13a [885412.757008] [<ffffffff810e652a>] discard_slab+0x41/0x43 [885412.758171] [<ffffffff810e7a4c>] __slab_free+0xf3/0xfe [885412.761027] [<ffffffffa030a536>] ? iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.761354] [<ffffffff810e7e95>] kmem_cache_free+0x6f/0xac [885412.761536] [<ffffffffa030a536>] iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.762056] [<ffffffffa020e467>] ? iblock_free_task+0x34/0x39 [target_core_iblock] [885412.762368] [<ffffffffa0314131>] lio_release_cmd+0x10/0x12 [iscsi_target_mod] [885412.764129] [<ffffffffa02c2254>] transport_release_cmd+0x2f/0x33 [target_core_mod] [885412.805024] [<ffffffffa02c230e>] transport_generic_remove+0xb6/0xc3 [target_core_mod] [885412.806424] [<ffffffff81035b5f>] ? try_to_wake_up+0x1bd/0x1bd [885412.809033] [<ffffffffa02c241f>] transport_generic_free_cmd+0x75/0x7d [target_core_mod] [885412.810066] [<ffffffffa02c2643>] transport_generic_wait_for_tasks+0x21c/0x22b [target_core_mod] [885412.811056] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813059] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813200] [<ffffffffa030b81d>] iscsit_close_connection+0x1d5/0x63a [iscsi_target_mod] [885412.813517] [<ffffffffa0300a82>] iscsit_take_action_for_connection_exit+0xdb/0xe0 [iscsi_target_mod] [885412.813851] [<ffffffffa03111e9>] iscsi_target_rx_thread+0x11f6/0x1221 [iscsi_target_mod] [885412.829024] [<ffffffff81033e8d>] ? pick_next_task_fair+0xbe/0x10e [885412.831010] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.833011] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.835010] [<ffffffff8105388a>] kthread+0x7d/0x85 [885412.837022] [<ffffffff813a7124>] kernel_thread_helper+0x4/0x10 [885412.838008] [<ffffffff8105380d>] ? kthread_worker_fn+0x145/0x145 [885412.840047] [<ffffffff813a7120>] ? gs_change+0x13/0x13 [885412.842007] FIX lio_cmd_cache: Restoring 0xffff880000cf3750-0xffff880000cf378d=0x6 Cc: Christoph Hellwig <hch@lst.de> Cc: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-07-30 16:03:58 +04:00
if (!cmd->se_lun) {
dump_stack();
pr_err("cmd->se_lun is NULL\n");
return -EINVAL;
}
if (in_interrupt()) {
dump_stack();
pr_err("transport_generic_handle_cdb cannot be called"
" from interrupt context\n");
return -EINVAL;
}
target: Fix bug for transport_generic_wait_for_tasks with direct operation This patch fixes a bug in transport_handle_cdb_direct() usage with target_core where transport_generic_wait_for_tasks() was bypassing active I/O + usage of cmd->t_transport_stop_comp because cmd->t_transport_active=1 was not being set before dispatching with transport_generic_new_cmd(). The fix follows existing usage in transport_generic_handle_cdb*() -> transport_add_cmd_to_queue() and set these directly, as well as handle transport_generic_new_cmd() exceptions for QUEUE_FULL and CHECK_CONDITION instead of propigating up to RX context fabric code. The bug was manifesting itself with the following SLUB poison overwritten warnings with iscsi-target v4.1 LUNs using the new process context direct operation during session reinstatement with active I/O exception handling: [885410.498267] ============================================================================= [885410.621622] BUG lio_cmd_cache: Poison overwritten [885410.621791] ----------------------------------------------------------------------------- [885410.621792] [885410.623420] INFO: 0xffff880000cf3750-0xffff880000cf378d. First byte 0x6a instead of 0x6b [885410.626332] INFO: Allocated in iscsit_allocate_cmd+0x1c/0xd4 [iscsi_target_mod] age=345 cpu=1 pid=22554 [885411.855189] INFO: Freed in iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] age=1410 cpu=1 pid=22554 [885411.856048] INFO: Slab 0xffffea000002d480 objects=22 used=0 fp=0xffff880000cf7300 flags=0x4080 [885411.856368] INFO: Object 0xffff880000cf33c0 @offset=13248 fp=0xffff880000cf6780 <SNIP> [885411.955678] Pid: 22554, comm: iscsi_trx Not tainted 3.0.0-rc7+ #30 [885411.956040] Call Trace: [885411.957029] [<ffffffff810e5cf9>] print_trailer+0x12e/0x137 [885412.752879] [<ffffffff810e61d9>] check_bytes_and_report+0xb9/0xfd [885412.754933] [<ffffffff810e62d2>] check_object+0xb5/0x192 [885412.755099] [<ffffffff810e6445>] __free_slab+0x96/0x13a [885412.757008] [<ffffffff810e652a>] discard_slab+0x41/0x43 [885412.758171] [<ffffffff810e7a4c>] __slab_free+0xf3/0xfe [885412.761027] [<ffffffffa030a536>] ? iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.761354] [<ffffffff810e7e95>] kmem_cache_free+0x6f/0xac [885412.761536] [<ffffffffa030a536>] iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.762056] [<ffffffffa020e467>] ? iblock_free_task+0x34/0x39 [target_core_iblock] [885412.762368] [<ffffffffa0314131>] lio_release_cmd+0x10/0x12 [iscsi_target_mod] [885412.764129] [<ffffffffa02c2254>] transport_release_cmd+0x2f/0x33 [target_core_mod] [885412.805024] [<ffffffffa02c230e>] transport_generic_remove+0xb6/0xc3 [target_core_mod] [885412.806424] [<ffffffff81035b5f>] ? try_to_wake_up+0x1bd/0x1bd [885412.809033] [<ffffffffa02c241f>] transport_generic_free_cmd+0x75/0x7d [target_core_mod] [885412.810066] [<ffffffffa02c2643>] transport_generic_wait_for_tasks+0x21c/0x22b [target_core_mod] [885412.811056] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813059] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813200] [<ffffffffa030b81d>] iscsit_close_connection+0x1d5/0x63a [iscsi_target_mod] [885412.813517] [<ffffffffa0300a82>] iscsit_take_action_for_connection_exit+0xdb/0xe0 [iscsi_target_mod] [885412.813851] [<ffffffffa03111e9>] iscsi_target_rx_thread+0x11f6/0x1221 [iscsi_target_mod] [885412.829024] [<ffffffff81033e8d>] ? pick_next_task_fair+0xbe/0x10e [885412.831010] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.833011] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.835010] [<ffffffff8105388a>] kthread+0x7d/0x85 [885412.837022] [<ffffffff813a7124>] kernel_thread_helper+0x4/0x10 [885412.838008] [<ffffffff8105380d>] ? kthread_worker_fn+0x145/0x145 [885412.840047] [<ffffffff813a7120>] ? gs_change+0x13/0x13 [885412.842007] FIX lio_cmd_cache: Restoring 0xffff880000cf3750-0xffff880000cf378d=0x6 Cc: Christoph Hellwig <hch@lst.de> Cc: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-07-30 16:03:58 +04:00
/*
* Set TRANSPORT_NEW_CMD state and CMD_T_ACTIVE to ensure that
* outstanding descriptors are handled correctly during shutdown via
* transport_wait_for_tasks()
target: Fix bug for transport_generic_wait_for_tasks with direct operation This patch fixes a bug in transport_handle_cdb_direct() usage with target_core where transport_generic_wait_for_tasks() was bypassing active I/O + usage of cmd->t_transport_stop_comp because cmd->t_transport_active=1 was not being set before dispatching with transport_generic_new_cmd(). The fix follows existing usage in transport_generic_handle_cdb*() -> transport_add_cmd_to_queue() and set these directly, as well as handle transport_generic_new_cmd() exceptions for QUEUE_FULL and CHECK_CONDITION instead of propigating up to RX context fabric code. The bug was manifesting itself with the following SLUB poison overwritten warnings with iscsi-target v4.1 LUNs using the new process context direct operation during session reinstatement with active I/O exception handling: [885410.498267] ============================================================================= [885410.621622] BUG lio_cmd_cache: Poison overwritten [885410.621791] ----------------------------------------------------------------------------- [885410.621792] [885410.623420] INFO: 0xffff880000cf3750-0xffff880000cf378d. First byte 0x6a instead of 0x6b [885410.626332] INFO: Allocated in iscsit_allocate_cmd+0x1c/0xd4 [iscsi_target_mod] age=345 cpu=1 pid=22554 [885411.855189] INFO: Freed in iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] age=1410 cpu=1 pid=22554 [885411.856048] INFO: Slab 0xffffea000002d480 objects=22 used=0 fp=0xffff880000cf7300 flags=0x4080 [885411.856368] INFO: Object 0xffff880000cf33c0 @offset=13248 fp=0xffff880000cf6780 <SNIP> [885411.955678] Pid: 22554, comm: iscsi_trx Not tainted 3.0.0-rc7+ #30 [885411.956040] Call Trace: [885411.957029] [<ffffffff810e5cf9>] print_trailer+0x12e/0x137 [885412.752879] [<ffffffff810e61d9>] check_bytes_and_report+0xb9/0xfd [885412.754933] [<ffffffff810e62d2>] check_object+0xb5/0x192 [885412.755099] [<ffffffff810e6445>] __free_slab+0x96/0x13a [885412.757008] [<ffffffff810e652a>] discard_slab+0x41/0x43 [885412.758171] [<ffffffff810e7a4c>] __slab_free+0xf3/0xfe [885412.761027] [<ffffffffa030a536>] ? iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.761354] [<ffffffff810e7e95>] kmem_cache_free+0x6f/0xac [885412.761536] [<ffffffffa030a536>] iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.762056] [<ffffffffa020e467>] ? iblock_free_task+0x34/0x39 [target_core_iblock] [885412.762368] [<ffffffffa0314131>] lio_release_cmd+0x10/0x12 [iscsi_target_mod] [885412.764129] [<ffffffffa02c2254>] transport_release_cmd+0x2f/0x33 [target_core_mod] [885412.805024] [<ffffffffa02c230e>] transport_generic_remove+0xb6/0xc3 [target_core_mod] [885412.806424] [<ffffffff81035b5f>] ? try_to_wake_up+0x1bd/0x1bd [885412.809033] [<ffffffffa02c241f>] transport_generic_free_cmd+0x75/0x7d [target_core_mod] [885412.810066] [<ffffffffa02c2643>] transport_generic_wait_for_tasks+0x21c/0x22b [target_core_mod] [885412.811056] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813059] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813200] [<ffffffffa030b81d>] iscsit_close_connection+0x1d5/0x63a [iscsi_target_mod] [885412.813517] [<ffffffffa0300a82>] iscsit_take_action_for_connection_exit+0xdb/0xe0 [iscsi_target_mod] [885412.813851] [<ffffffffa03111e9>] iscsi_target_rx_thread+0x11f6/0x1221 [iscsi_target_mod] [885412.829024] [<ffffffff81033e8d>] ? pick_next_task_fair+0xbe/0x10e [885412.831010] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.833011] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.835010] [<ffffffff8105388a>] kthread+0x7d/0x85 [885412.837022] [<ffffffff813a7124>] kernel_thread_helper+0x4/0x10 [885412.838008] [<ffffffff8105380d>] ? kthread_worker_fn+0x145/0x145 [885412.840047] [<ffffffff813a7120>] ? gs_change+0x13/0x13 [885412.842007] FIX lio_cmd_cache: Restoring 0xffff880000cf3750-0xffff880000cf378d=0x6 Cc: Christoph Hellwig <hch@lst.de> Cc: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-07-30 16:03:58 +04:00
*
* Also, we don't take cmd->t_state_lock here as we only expect
* this to be called for initial descriptor submission.
*/
cmd->t_state = TRANSPORT_NEW_CMD;
cmd->transport_state |= CMD_T_ACTIVE;
target: Fix bug for transport_generic_wait_for_tasks with direct operation This patch fixes a bug in transport_handle_cdb_direct() usage with target_core where transport_generic_wait_for_tasks() was bypassing active I/O + usage of cmd->t_transport_stop_comp because cmd->t_transport_active=1 was not being set before dispatching with transport_generic_new_cmd(). The fix follows existing usage in transport_generic_handle_cdb*() -> transport_add_cmd_to_queue() and set these directly, as well as handle transport_generic_new_cmd() exceptions for QUEUE_FULL and CHECK_CONDITION instead of propigating up to RX context fabric code. The bug was manifesting itself with the following SLUB poison overwritten warnings with iscsi-target v4.1 LUNs using the new process context direct operation during session reinstatement with active I/O exception handling: [885410.498267] ============================================================================= [885410.621622] BUG lio_cmd_cache: Poison overwritten [885410.621791] ----------------------------------------------------------------------------- [885410.621792] [885410.623420] INFO: 0xffff880000cf3750-0xffff880000cf378d. First byte 0x6a instead of 0x6b [885410.626332] INFO: Allocated in iscsit_allocate_cmd+0x1c/0xd4 [iscsi_target_mod] age=345 cpu=1 pid=22554 [885411.855189] INFO: Freed in iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] age=1410 cpu=1 pid=22554 [885411.856048] INFO: Slab 0xffffea000002d480 objects=22 used=0 fp=0xffff880000cf7300 flags=0x4080 [885411.856368] INFO: Object 0xffff880000cf33c0 @offset=13248 fp=0xffff880000cf6780 <SNIP> [885411.955678] Pid: 22554, comm: iscsi_trx Not tainted 3.0.0-rc7+ #30 [885411.956040] Call Trace: [885411.957029] [<ffffffff810e5cf9>] print_trailer+0x12e/0x137 [885412.752879] [<ffffffff810e61d9>] check_bytes_and_report+0xb9/0xfd [885412.754933] [<ffffffff810e62d2>] check_object+0xb5/0x192 [885412.755099] [<ffffffff810e6445>] __free_slab+0x96/0x13a [885412.757008] [<ffffffff810e652a>] discard_slab+0x41/0x43 [885412.758171] [<ffffffff810e7a4c>] __slab_free+0xf3/0xfe [885412.761027] [<ffffffffa030a536>] ? iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.761354] [<ffffffff810e7e95>] kmem_cache_free+0x6f/0xac [885412.761536] [<ffffffffa030a536>] iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.762056] [<ffffffffa020e467>] ? iblock_free_task+0x34/0x39 [target_core_iblock] [885412.762368] [<ffffffffa0314131>] lio_release_cmd+0x10/0x12 [iscsi_target_mod] [885412.764129] [<ffffffffa02c2254>] transport_release_cmd+0x2f/0x33 [target_core_mod] [885412.805024] [<ffffffffa02c230e>] transport_generic_remove+0xb6/0xc3 [target_core_mod] [885412.806424] [<ffffffff81035b5f>] ? try_to_wake_up+0x1bd/0x1bd [885412.809033] [<ffffffffa02c241f>] transport_generic_free_cmd+0x75/0x7d [target_core_mod] [885412.810066] [<ffffffffa02c2643>] transport_generic_wait_for_tasks+0x21c/0x22b [target_core_mod] [885412.811056] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813059] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813200] [<ffffffffa030b81d>] iscsit_close_connection+0x1d5/0x63a [iscsi_target_mod] [885412.813517] [<ffffffffa0300a82>] iscsit_take_action_for_connection_exit+0xdb/0xe0 [iscsi_target_mod] [885412.813851] [<ffffffffa03111e9>] iscsi_target_rx_thread+0x11f6/0x1221 [iscsi_target_mod] [885412.829024] [<ffffffff81033e8d>] ? pick_next_task_fair+0xbe/0x10e [885412.831010] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.833011] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.835010] [<ffffffff8105388a>] kthread+0x7d/0x85 [885412.837022] [<ffffffff813a7124>] kernel_thread_helper+0x4/0x10 [885412.838008] [<ffffffff8105380d>] ? kthread_worker_fn+0x145/0x145 [885412.840047] [<ffffffff813a7120>] ? gs_change+0x13/0x13 [885412.842007] FIX lio_cmd_cache: Restoring 0xffff880000cf3750-0xffff880000cf378d=0x6 Cc: Christoph Hellwig <hch@lst.de> Cc: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-07-30 16:03:58 +04:00
/*
* transport_generic_new_cmd() is already handling QUEUE_FULL,
* so follow TRANSPORT_NEW_CMD processing thread context usage
* and call transport_generic_request_failure() if necessary..
*/
ret = transport_generic_new_cmd(cmd);
if (ret)
transport_generic_request_failure(cmd, ret);
target: Fix bug for transport_generic_wait_for_tasks with direct operation This patch fixes a bug in transport_handle_cdb_direct() usage with target_core where transport_generic_wait_for_tasks() was bypassing active I/O + usage of cmd->t_transport_stop_comp because cmd->t_transport_active=1 was not being set before dispatching with transport_generic_new_cmd(). The fix follows existing usage in transport_generic_handle_cdb*() -> transport_add_cmd_to_queue() and set these directly, as well as handle transport_generic_new_cmd() exceptions for QUEUE_FULL and CHECK_CONDITION instead of propigating up to RX context fabric code. The bug was manifesting itself with the following SLUB poison overwritten warnings with iscsi-target v4.1 LUNs using the new process context direct operation during session reinstatement with active I/O exception handling: [885410.498267] ============================================================================= [885410.621622] BUG lio_cmd_cache: Poison overwritten [885410.621791] ----------------------------------------------------------------------------- [885410.621792] [885410.623420] INFO: 0xffff880000cf3750-0xffff880000cf378d. First byte 0x6a instead of 0x6b [885410.626332] INFO: Allocated in iscsit_allocate_cmd+0x1c/0xd4 [iscsi_target_mod] age=345 cpu=1 pid=22554 [885411.855189] INFO: Freed in iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] age=1410 cpu=1 pid=22554 [885411.856048] INFO: Slab 0xffffea000002d480 objects=22 used=0 fp=0xffff880000cf7300 flags=0x4080 [885411.856368] INFO: Object 0xffff880000cf33c0 @offset=13248 fp=0xffff880000cf6780 <SNIP> [885411.955678] Pid: 22554, comm: iscsi_trx Not tainted 3.0.0-rc7+ #30 [885411.956040] Call Trace: [885411.957029] [<ffffffff810e5cf9>] print_trailer+0x12e/0x137 [885412.752879] [<ffffffff810e61d9>] check_bytes_and_report+0xb9/0xfd [885412.754933] [<ffffffff810e62d2>] check_object+0xb5/0x192 [885412.755099] [<ffffffff810e6445>] __free_slab+0x96/0x13a [885412.757008] [<ffffffff810e652a>] discard_slab+0x41/0x43 [885412.758171] [<ffffffff810e7a4c>] __slab_free+0xf3/0xfe [885412.761027] [<ffffffffa030a536>] ? iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.761354] [<ffffffff810e7e95>] kmem_cache_free+0x6f/0xac [885412.761536] [<ffffffffa030a536>] iscsit_release_cmd+0x208/0x217 [iscsi_target_mod] [885412.762056] [<ffffffffa020e467>] ? iblock_free_task+0x34/0x39 [target_core_iblock] [885412.762368] [<ffffffffa0314131>] lio_release_cmd+0x10/0x12 [iscsi_target_mod] [885412.764129] [<ffffffffa02c2254>] transport_release_cmd+0x2f/0x33 [target_core_mod] [885412.805024] [<ffffffffa02c230e>] transport_generic_remove+0xb6/0xc3 [target_core_mod] [885412.806424] [<ffffffff81035b5f>] ? try_to_wake_up+0x1bd/0x1bd [885412.809033] [<ffffffffa02c241f>] transport_generic_free_cmd+0x75/0x7d [target_core_mod] [885412.810066] [<ffffffffa02c2643>] transport_generic_wait_for_tasks+0x21c/0x22b [target_core_mod] [885412.811056] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813059] [<ffffffff8139f0b1>] ? mutex_lock+0x11/0x32 [885412.813200] [<ffffffffa030b81d>] iscsit_close_connection+0x1d5/0x63a [iscsi_target_mod] [885412.813517] [<ffffffffa0300a82>] iscsit_take_action_for_connection_exit+0xdb/0xe0 [iscsi_target_mod] [885412.813851] [<ffffffffa03111e9>] iscsi_target_rx_thread+0x11f6/0x1221 [iscsi_target_mod] [885412.829024] [<ffffffff81033e8d>] ? pick_next_task_fair+0xbe/0x10e [885412.831010] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.833011] [<ffffffffa030fff3>] ? iscsit_handle_scsi_cmd+0x91d/0x91d [iscsi_target_mod] [885412.835010] [<ffffffff8105388a>] kthread+0x7d/0x85 [885412.837022] [<ffffffff813a7124>] kernel_thread_helper+0x4/0x10 [885412.838008] [<ffffffff8105380d>] ? kthread_worker_fn+0x145/0x145 [885412.840047] [<ffffffff813a7120>] ? gs_change+0x13/0x13 [885412.842007] FIX lio_cmd_cache: Restoring 0xffff880000cf3750-0xffff880000cf378d=0x6 Cc: Christoph Hellwig <hch@lst.de> Cc: Andy Grover <agrover@redhat.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-07-30 16:03:58 +04:00
return 0;
}
EXPORT_SYMBOL(transport_handle_cdb_direct);
sense_reason_t
transport_generic_map_mem_to_cmd(struct se_cmd *cmd, struct scatterlist *sgl,
u32 sgl_count, struct scatterlist *sgl_bidi, u32 sgl_bidi_count)
{
if (!sgl || !sgl_count)
return 0;
/*
* Reject SCSI data overflow with map_mem_to_cmd() as incoming
* scatterlists already have been set to follow what the fabric
* passes for the original expected data transfer length.
*/
if (cmd->se_cmd_flags & SCF_OVERFLOW_BIT) {
pr_warn("Rejecting SCSI DATA overflow for fabric using"
" SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC\n");
return TCM_INVALID_CDB_FIELD;
}
cmd->t_data_sg = sgl;
cmd->t_data_nents = sgl_count;
cmd->t_bidi_data_sg = sgl_bidi;
cmd->t_bidi_data_nents = sgl_bidi_count;
cmd->se_cmd_flags |= SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC;
return 0;
}
/**
* target_submit_cmd_map_sgls - lookup unpacked lun and submit uninitialized
* se_cmd + use pre-allocated SGL memory.
*
* @se_cmd: command descriptor to submit
* @se_sess: associated se_sess for endpoint
* @cdb: pointer to SCSI CDB
* @sense: pointer to SCSI sense buffer
* @unpacked_lun: unpacked LUN to reference for struct se_lun
* @data_length: fabric expected data transfer length
* @task_attr: SAM task attribute
* @data_dir: DMA data direction
* @flags: flags for command submission from target_sc_flags_tables
* @sgl: struct scatterlist memory for unidirectional mapping
* @sgl_count: scatterlist count for unidirectional mapping
* @sgl_bidi: struct scatterlist memory for bidirectional READ mapping
* @sgl_bidi_count: scatterlist count for bidirectional READ mapping
* @sgl_prot: struct scatterlist memory protection information
* @sgl_prot_count: scatterlist count for protection information
*
* Task tags are supported if the caller has set @se_cmd->tag.
*
* Returns non zero to signal active I/O shutdown failure. All other
* setup exceptions will be returned as a SCSI CHECK_CONDITION response,
* but still return zero here.
*
* This may only be called from process context, and also currently
* assumes internal allocation of fabric payload buffer by target-core.
*/
int target_submit_cmd_map_sgls(struct se_cmd *se_cmd, struct se_session *se_sess,
unsigned char *cdb, unsigned char *sense, u64 unpacked_lun,
u32 data_length, int task_attr, int data_dir, int flags,
struct scatterlist *sgl, u32 sgl_count,
struct scatterlist *sgl_bidi, u32 sgl_bidi_count,
struct scatterlist *sgl_prot, u32 sgl_prot_count)
{
struct se_portal_group *se_tpg;
sense_reason_t rc;
int ret;
se_tpg = se_sess->se_tpg;
BUG_ON(!se_tpg);
BUG_ON(se_cmd->se_tfo || se_cmd->se_sess);
BUG_ON(in_interrupt());
/*
* Initialize se_cmd for target operation. From this point
* exceptions are handled by sending exception status via
* target_core_fabric_ops->queue_status() callback
*/
transport_init_se_cmd(se_cmd, se_tpg->se_tpg_tfo, se_sess,
data_length, data_dir, task_attr, sense);
if (flags & TARGET_SCF_USE_CPUID)
se_cmd->se_cmd_flags |= SCF_USE_CPUID;
else
se_cmd->cpuid = WORK_CPU_UNBOUND;
if (flags & TARGET_SCF_UNKNOWN_SIZE)
se_cmd->unknown_data_length = 1;
/*
* Obtain struct se_cmd->cmd_kref reference and add new cmd to
* se_sess->sess_cmd_list. A second kref_get here is necessary
* for fabrics using TARGET_SCF_ACK_KREF that expect a second
* kref_put() to happen during fabric packet acknowledgement.
*/
ret = target_get_sess_cmd(se_cmd, flags & TARGET_SCF_ACK_KREF);
if (ret)
return ret;
/*
* Signal bidirectional data payloads to target-core
*/
if (flags & TARGET_SCF_BIDI_OP)
se_cmd->se_cmd_flags |= SCF_BIDI;
/*
* Locate se_lun pointer and attach it to struct se_cmd
*/
rc = transport_lookup_cmd_lun(se_cmd, unpacked_lun);
if (rc) {
transport_send_check_condition_and_sense(se_cmd, rc, 0);
target_put_sess_cmd(se_cmd);
return 0;
}
rc = target_setup_cmd_from_cdb(se_cmd, cdb);
if (rc != 0) {
transport_generic_request_failure(se_cmd, rc);
return 0;
}
/*
* Save pointers for SGLs containing protection information,
* if present.
*/
if (sgl_prot_count) {
se_cmd->t_prot_sg = sgl_prot;
se_cmd->t_prot_nents = sgl_prot_count;
se_cmd->se_cmd_flags |= SCF_PASSTHROUGH_PROT_SG_TO_MEM_NOALLOC;
}
/*
* When a non zero sgl_count has been passed perform SGL passthrough
* mapping for pre-allocated fabric memory instead of having target
* core perform an internal SGL allocation..
*/
if (sgl_count != 0) {
BUG_ON(!sgl);
/*
* A work-around for tcm_loop as some userspace code via
* scsi-generic do not memset their associated read buffers,
* so go ahead and do that here for type non-data CDBs. Also
* note that this is currently guaranteed to be a single SGL
* for this case by target core in target_setup_cmd_from_cdb()
* -> transport_generic_cmd_sequencer().
*/
if (!(se_cmd->se_cmd_flags & SCF_SCSI_DATA_CDB) &&
se_cmd->data_direction == DMA_FROM_DEVICE) {
unsigned char *buf = NULL;
if (sgl)
buf = kmap(sg_page(sgl)) + sgl->offset;
if (buf) {
memset(buf, 0, sgl->length);
kunmap(sg_page(sgl));
}
}
rc = transport_generic_map_mem_to_cmd(se_cmd, sgl, sgl_count,
sgl_bidi, sgl_bidi_count);
if (rc != 0) {
transport_generic_request_failure(se_cmd, rc);
return 0;
}
}
/*
* Check if we need to delay processing because of ALUA
* Active/NonOptimized primary access state..
*/
core_alua_check_nonop_delay(se_cmd);
transport_handle_cdb_direct(se_cmd);
return 0;
}
EXPORT_SYMBOL(target_submit_cmd_map_sgls);
/**
* target_submit_cmd - lookup unpacked lun and submit uninitialized se_cmd
*
* @se_cmd: command descriptor to submit
* @se_sess: associated se_sess for endpoint
* @cdb: pointer to SCSI CDB
* @sense: pointer to SCSI sense buffer
* @unpacked_lun: unpacked LUN to reference for struct se_lun
* @data_length: fabric expected data transfer length
* @task_attr: SAM task attribute
* @data_dir: DMA data direction
* @flags: flags for command submission from target_sc_flags_tables
*
* Task tags are supported if the caller has set @se_cmd->tag.
*
* Returns non zero to signal active I/O shutdown failure. All other
* setup exceptions will be returned as a SCSI CHECK_CONDITION response,
* but still return zero here.
*
* This may only be called from process context, and also currently
* assumes internal allocation of fabric payload buffer by target-core.
*
* It also assumes interal target core SGL memory allocation.
*/
int target_submit_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
unsigned char *cdb, unsigned char *sense, u64 unpacked_lun,
u32 data_length, int task_attr, int data_dir, int flags)
{
return target_submit_cmd_map_sgls(se_cmd, se_sess, cdb, sense,
unpacked_lun, data_length, task_attr, data_dir,
flags, NULL, 0, NULL, 0, NULL, 0);
}
EXPORT_SYMBOL(target_submit_cmd);
static void target_complete_tmr_failure(struct work_struct *work)
{
struct se_cmd *se_cmd = container_of(work, struct se_cmd, work);
se_cmd->se_tmr_req->response = TMR_LUN_DOES_NOT_EXIST;
se_cmd->se_tfo->queue_tm_rsp(se_cmd);
transport_lun_remove_cmd(se_cmd);
transport_cmd_check_stop_to_fabric(se_cmd);
}
static bool target_lookup_lun_from_tag(struct se_session *se_sess, u64 tag,
u64 *unpacked_lun)
{
struct se_cmd *se_cmd;
unsigned long flags;
bool ret = false;
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
list_for_each_entry(se_cmd, &se_sess->sess_cmd_list, se_cmd_list) {
if (se_cmd->se_cmd_flags & SCF_SCSI_TMR_CDB)
continue;
if (se_cmd->tag == tag) {
*unpacked_lun = se_cmd->orig_fe_lun;
ret = true;
break;
}
}
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
return ret;
}
/**
* target_submit_tmr - lookup unpacked lun and submit uninitialized se_cmd
* for TMR CDBs
*
* @se_cmd: command descriptor to submit
* @se_sess: associated se_sess for endpoint
* @sense: pointer to SCSI sense buffer
* @unpacked_lun: unpacked LUN to reference for struct se_lun
* @fabric_tmr_ptr: fabric context for TMR req
* @tm_type: Type of TM request
* @gfp: gfp type for caller
* @tag: referenced task tag for TMR_ABORT_TASK
* @flags: submit cmd flags
*
* Callable from all contexts.
**/
int target_submit_tmr(struct se_cmd *se_cmd, struct se_session *se_sess,
unsigned char *sense, u64 unpacked_lun,
void *fabric_tmr_ptr, unsigned char tm_type,
gfp_t gfp, u64 tag, int flags)
{
struct se_portal_group *se_tpg;
int ret;
se_tpg = se_sess->se_tpg;
BUG_ON(!se_tpg);
transport_init_se_cmd(se_cmd, se_tpg->se_tpg_tfo, se_sess,
0, DMA_NONE, TCM_SIMPLE_TAG, sense);
/*
* FIXME: Currently expect caller to handle se_cmd->se_tmr_req
* allocation failure.
*/
ret = core_tmr_alloc_req(se_cmd, fabric_tmr_ptr, tm_type, gfp);
if (ret < 0)
return -ENOMEM;
if (tm_type == TMR_ABORT_TASK)
se_cmd->se_tmr_req->ref_task_tag = tag;
/* See target_submit_cmd for commentary */
ret = target_get_sess_cmd(se_cmd, flags & TARGET_SCF_ACK_KREF);
target: Check sess_tearing_down in target_get_sess_cmd() Target core code assumes that target_splice_sess_cmd_list() has set sess_tearing_down and moved the list of pending commands to sess_wait_list, no more commands will be added to the session; if any are added, nothing keeps the se_session from being freed while the command is still in flight, which e.g. leads to use-after-free of se_cmd->se_sess in target_release_cmd_kref(). To enforce this invariant, put a check of sess_tearing_down inside of sess_cmd_lock in target_get_sess_cmd(); any checks before this are racy and can lead to the use-after-free described above. For example, the qla_target check in qlt_do_work() checks sess_tearing_down from work thread context but then drops all locks before calling target_submit_cmd() (as it must, since that is a sleeping function). However, since no locks are held, anything can happen with respect to the session it has looked up -- although it does correctly get sess_kref within its lock, so the memory won't be freed while target_submit_cmd() is actually running, nothing stops eg an ACL from being dropped and calling ->shutdown_session() (which calls into target_splice_sess_cmd_list()) before we get to target_get_sess_cmd(). Once this happens, the se_session memory can be freed as soon as target_submit_cmd() returns and qlt_do_work() drops its reference, even though we've just added a command to sess_cmd_list. To prevent this use-after-free, check sess_tearing_down inside of sess_cmd_lock right before target_get_sess_cmd() adds a command to sess_cmd_list; this is synchronized with target_splice_sess_cmd_list() so that every command is either waited for or not added to the queue. (nab: Keep target_submit_cmd() returning void for now..) Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2012-07-16 22:04:40 +04:00
if (ret) {
core_tmr_release_req(se_cmd->se_tmr_req);
return ret;
}
/*
* If this is ABORT_TASK with no explicit fabric provided LUN,
* go ahead and search active session tags for a match to figure
* out unpacked_lun for the original se_cmd.
*/
if (tm_type == TMR_ABORT_TASK && (flags & TARGET_SCF_LOOKUP_LUN_FROM_TAG)) {
if (!target_lookup_lun_from_tag(se_sess, tag, &unpacked_lun))
goto failure;
}
ret = transport_lookup_tmr_lun(se_cmd, unpacked_lun);
if (ret)
goto failure;
transport_generic_handle_tmr(se_cmd);
return 0;
/*
* For callback during failure handling, push this work off
* to process context with TMR_LUN_DOES_NOT_EXIST status.
*/
failure:
INIT_WORK(&se_cmd->work, target_complete_tmr_failure);
schedule_work(&se_cmd->work);
return 0;
}
EXPORT_SYMBOL(target_submit_tmr);
/*
* Handle SAM-esque emulation for generic transport request failures.
*/
void transport_generic_request_failure(struct se_cmd *cmd,
sense_reason_t sense_reason)
{
int ret = 0, post_ret = 0;
pr_debug("-----[ Storage Engine Exception; sense_reason %d\n",
sense_reason);
target_show_cmd("-----[ ", cmd);
/*
* For SAM Task Attribute emulation for failed struct se_cmd
*/
transport_complete_task_attr(cmd);
/*
* Handle special case for COMPARE_AND_WRITE failure, where the
* callback is expected to drop the per device ->caw_sem.
*/
if ((cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) &&
cmd->transport_complete_callback)
cmd->transport_complete_callback(cmd, false, &post_ret);
if (transport_check_aborted_status(cmd, 1))
return;
switch (sense_reason) {
target: Address legacy PYX_TRANSPORT_* return code breakage This patch removes legacy usage of PYX_TRANSPORT_* return codes in a number of locations and addresses cases where transport_generic_request_failure() was returning the incorrect sense upon CHECK_CONDITION status after the v3.1 converson to use errno return codes. This includes the conversion of transport_generic_request_failure() to process cmd->scsi_sense_reason and handle extra TCM_RESERVATION_CONFLICT before calling transport_send_check_condition_and_sense() to queue up response status. It also drops PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES legacy usgae, and returns TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE w/ a response for these cases. transport_generic_allocate_tasks(), transport_generic_new_cmd(), backend SCF_SCSI_DATA_SG_IO_CDB ->do_task(), and emulated ->execute_task() have all been updated to set se_cmd->scsi_sense_reason and return errno codes universally upon failure. This includes cmd->scsi_sense_reason assignment in target_core_alua.c, target_core_pr.c and target_core_cdb.c emulation code. Finally it updates fabric modules to remove the legacy usage, and for TFO->new_cmd_map() callers forwards return values outside of fabric code. iscsi-target has also been updated to remove a handful of special cases related to the cleanup and signaling QUEUE_FULL handling w/ ft_write_pending() (v2: Drop extra SCF_SCSI_CDB_EXCEPTION check during failure from transport_generic_new_cmd, and re-add missing task->task_error_status assignment in transport_complete_task) Cc: Christoph Hellwig <hch@lst.de> Cc: stable@kernel.org Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-11-04 13:36:16 +04:00
case TCM_NON_EXISTENT_LUN:
case TCM_UNSUPPORTED_SCSI_OPCODE:
case TCM_INVALID_CDB_FIELD:
case TCM_INVALID_PARAMETER_LIST:
case TCM_PARAMETER_LIST_LENGTH_ERROR:
target: Address legacy PYX_TRANSPORT_* return code breakage This patch removes legacy usage of PYX_TRANSPORT_* return codes in a number of locations and addresses cases where transport_generic_request_failure() was returning the incorrect sense upon CHECK_CONDITION status after the v3.1 converson to use errno return codes. This includes the conversion of transport_generic_request_failure() to process cmd->scsi_sense_reason and handle extra TCM_RESERVATION_CONFLICT before calling transport_send_check_condition_and_sense() to queue up response status. It also drops PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES legacy usgae, and returns TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE w/ a response for these cases. transport_generic_allocate_tasks(), transport_generic_new_cmd(), backend SCF_SCSI_DATA_SG_IO_CDB ->do_task(), and emulated ->execute_task() have all been updated to set se_cmd->scsi_sense_reason and return errno codes universally upon failure. This includes cmd->scsi_sense_reason assignment in target_core_alua.c, target_core_pr.c and target_core_cdb.c emulation code. Finally it updates fabric modules to remove the legacy usage, and for TFO->new_cmd_map() callers forwards return values outside of fabric code. iscsi-target has also been updated to remove a handful of special cases related to the cleanup and signaling QUEUE_FULL handling w/ ft_write_pending() (v2: Drop extra SCF_SCSI_CDB_EXCEPTION check during failure from transport_generic_new_cmd, and re-add missing task->task_error_status assignment in transport_complete_task) Cc: Christoph Hellwig <hch@lst.de> Cc: stable@kernel.org Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-11-04 13:36:16 +04:00
case TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE:
case TCM_UNKNOWN_MODE_PAGE:
case TCM_WRITE_PROTECTED:
case TCM_ADDRESS_OUT_OF_RANGE:
target: Address legacy PYX_TRANSPORT_* return code breakage This patch removes legacy usage of PYX_TRANSPORT_* return codes in a number of locations and addresses cases where transport_generic_request_failure() was returning the incorrect sense upon CHECK_CONDITION status after the v3.1 converson to use errno return codes. This includes the conversion of transport_generic_request_failure() to process cmd->scsi_sense_reason and handle extra TCM_RESERVATION_CONFLICT before calling transport_send_check_condition_and_sense() to queue up response status. It also drops PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES legacy usgae, and returns TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE w/ a response for these cases. transport_generic_allocate_tasks(), transport_generic_new_cmd(), backend SCF_SCSI_DATA_SG_IO_CDB ->do_task(), and emulated ->execute_task() have all been updated to set se_cmd->scsi_sense_reason and return errno codes universally upon failure. This includes cmd->scsi_sense_reason assignment in target_core_alua.c, target_core_pr.c and target_core_cdb.c emulation code. Finally it updates fabric modules to remove the legacy usage, and for TFO->new_cmd_map() callers forwards return values outside of fabric code. iscsi-target has also been updated to remove a handful of special cases related to the cleanup and signaling QUEUE_FULL handling w/ ft_write_pending() (v2: Drop extra SCF_SCSI_CDB_EXCEPTION check during failure from transport_generic_new_cmd, and re-add missing task->task_error_status assignment in transport_complete_task) Cc: Christoph Hellwig <hch@lst.de> Cc: stable@kernel.org Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-11-04 13:36:16 +04:00
case TCM_CHECK_CONDITION_ABORT_CMD:
case TCM_CHECK_CONDITION_UNIT_ATTENTION:
case TCM_CHECK_CONDITION_NOT_READY:
case TCM_LOGICAL_BLOCK_GUARD_CHECK_FAILED:
case TCM_LOGICAL_BLOCK_APP_TAG_CHECK_FAILED:
case TCM_LOGICAL_BLOCK_REF_TAG_CHECK_FAILED:
case TCM_COPY_TARGET_DEVICE_NOT_REACHABLE:
case TCM_TOO_MANY_TARGET_DESCS:
case TCM_UNSUPPORTED_TARGET_DESC_TYPE_CODE:
case TCM_TOO_MANY_SEGMENT_DESCS:
case TCM_UNSUPPORTED_SEGMENT_DESC_TYPE_CODE:
break;
case TCM_OUT_OF_RESOURCES:
cmd->scsi_status = SAM_STAT_TASK_SET_FULL;
goto queue_status;
case TCM_LUN_BUSY:
cmd->scsi_status = SAM_STAT_BUSY;
goto queue_status;
target: Address legacy PYX_TRANSPORT_* return code breakage This patch removes legacy usage of PYX_TRANSPORT_* return codes in a number of locations and addresses cases where transport_generic_request_failure() was returning the incorrect sense upon CHECK_CONDITION status after the v3.1 converson to use errno return codes. This includes the conversion of transport_generic_request_failure() to process cmd->scsi_sense_reason and handle extra TCM_RESERVATION_CONFLICT before calling transport_send_check_condition_and_sense() to queue up response status. It also drops PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES legacy usgae, and returns TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE w/ a response for these cases. transport_generic_allocate_tasks(), transport_generic_new_cmd(), backend SCF_SCSI_DATA_SG_IO_CDB ->do_task(), and emulated ->execute_task() have all been updated to set se_cmd->scsi_sense_reason and return errno codes universally upon failure. This includes cmd->scsi_sense_reason assignment in target_core_alua.c, target_core_pr.c and target_core_cdb.c emulation code. Finally it updates fabric modules to remove the legacy usage, and for TFO->new_cmd_map() callers forwards return values outside of fabric code. iscsi-target has also been updated to remove a handful of special cases related to the cleanup and signaling QUEUE_FULL handling w/ ft_write_pending() (v2: Drop extra SCF_SCSI_CDB_EXCEPTION check during failure from transport_generic_new_cmd, and re-add missing task->task_error_status assignment in transport_complete_task) Cc: Christoph Hellwig <hch@lst.de> Cc: stable@kernel.org Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-11-04 13:36:16 +04:00
case TCM_RESERVATION_CONFLICT:
/*
* No SENSE Data payload for this case, set SCSI Status
* and queue the response to $FABRIC_MOD.
*
* Uses linux/include/scsi/scsi.h SAM status codes defs
*/
cmd->scsi_status = SAM_STAT_RESERVATION_CONFLICT;
/*
* For UA Interlock Code 11b, a RESERVATION CONFLICT will
* establish a UNIT ATTENTION with PREVIOUS RESERVATION
* CONFLICT STATUS.
*
* See spc4r17, section 7.4.6 Control Mode Page, Table 349
*/
2011-07-19 12:55:10 +04:00
if (cmd->se_sess &&
cmd->se_dev->dev_attrib.emulate_ua_intlck_ctrl == 2) {
target_ua_allocate_lun(cmd->se_sess->se_node_acl,
cmd->orig_fe_lun, 0x2C,
ASCQ_2CH_PREVIOUS_RESERVATION_CONFLICT_STATUS);
}
goto queue_status;
default:
pr_err("Unknown transport error for CDB 0x%02x: %d\n",
cmd->t_task_cdb[0], sense_reason);
sense_reason = TCM_UNSUPPORTED_SCSI_OPCODE;
break;
}
ret = transport_send_check_condition_and_sense(cmd, sense_reason, 0);
if (ret)
target: Address legacy PYX_TRANSPORT_* return code breakage This patch removes legacy usage of PYX_TRANSPORT_* return codes in a number of locations and addresses cases where transport_generic_request_failure() was returning the incorrect sense upon CHECK_CONDITION status after the v3.1 converson to use errno return codes. This includes the conversion of transport_generic_request_failure() to process cmd->scsi_sense_reason and handle extra TCM_RESERVATION_CONFLICT before calling transport_send_check_condition_and_sense() to queue up response status. It also drops PYX_TRANSPORT_OUT_OF_MEMORY_RESOURCES legacy usgae, and returns TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE w/ a response for these cases. transport_generic_allocate_tasks(), transport_generic_new_cmd(), backend SCF_SCSI_DATA_SG_IO_CDB ->do_task(), and emulated ->execute_task() have all been updated to set se_cmd->scsi_sense_reason and return errno codes universally upon failure. This includes cmd->scsi_sense_reason assignment in target_core_alua.c, target_core_pr.c and target_core_cdb.c emulation code. Finally it updates fabric modules to remove the legacy usage, and for TFO->new_cmd_map() callers forwards return values outside of fabric code. iscsi-target has also been updated to remove a handful of special cases related to the cleanup and signaling QUEUE_FULL handling w/ ft_write_pending() (v2: Drop extra SCF_SCSI_CDB_EXCEPTION check during failure from transport_generic_new_cmd, and re-add missing task->task_error_status assignment in transport_complete_task) Cc: Christoph Hellwig <hch@lst.de> Cc: stable@kernel.org Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-11-04 13:36:16 +04:00
goto queue_full;
check_stop:
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return;
queue_status:
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_status(cmd);
if (!ret)
goto check_stop;
queue_full:
transport_handle_queue_full(cmd, cmd->se_dev, ret, false);
}
EXPORT_SYMBOL(transport_generic_request_failure);
void __target_execute_cmd(struct se_cmd *cmd, bool do_checks)
{
sense_reason_t ret;
if (!cmd->execute_cmd) {
ret = TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
goto err;
}
if (do_checks) {
/*
* Check for an existing UNIT ATTENTION condition after
* target_handle_task_attr() has done SAM task attr
* checking, and possibly have already defered execution
* out to target_restart_delayed_cmds() context.
*/
ret = target_scsi3_ua_check(cmd);
if (ret)
goto err;
ret = target_alua_state_check(cmd);
if (ret)
goto err;
ret = target_check_reservation(cmd);
if (ret) {
cmd->scsi_status = SAM_STAT_RESERVATION_CONFLICT;
goto err;
}
}
ret = cmd->execute_cmd(cmd);
if (!ret)
return;
err:
spin_lock_irq(&cmd->t_state_lock);
cmd->transport_state &= ~CMD_T_SENT;
spin_unlock_irq(&cmd->t_state_lock);
transport_generic_request_failure(cmd, ret);
}
static int target_write_prot_action(struct se_cmd *cmd)
{
u32 sectors;
/*
* Perform WRITE_INSERT of PI using software emulation when backend
* device has PI enabled, if the transport has not already generated
* PI using hardware WRITE_INSERT offload.
*/
switch (cmd->prot_op) {
case TARGET_PROT_DOUT_INSERT:
if (!(cmd->se_sess->sup_prot_ops & TARGET_PROT_DOUT_INSERT))
sbc_dif_generate(cmd);
break;
case TARGET_PROT_DOUT_STRIP:
if (cmd->se_sess->sup_prot_ops & TARGET_PROT_DOUT_STRIP)
break;
sectors = cmd->data_length >> ilog2(cmd->se_dev->dev_attrib.block_size);
cmd->pi_err = sbc_dif_verify(cmd, cmd->t_task_lba,
sectors, 0, cmd->t_prot_sg, 0);
if (unlikely(cmd->pi_err)) {
spin_lock_irq(&cmd->t_state_lock);
cmd->transport_state &= ~CMD_T_SENT;
spin_unlock_irq(&cmd->t_state_lock);
transport_generic_request_failure(cmd, cmd->pi_err);
return -1;
}
break;
default:
break;
}
return 0;
}
static bool target_handle_task_attr(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
if (dev->transport->transport_flags & TRANSPORT_FLAG_PASSTHROUGH)
return false;
cmd->se_cmd_flags |= SCF_TASK_ATTR_SET;
/*
* Check for the existence of HEAD_OF_QUEUE, and if true return 1
* to allow the passed struct se_cmd list of tasks to the front of the list.
*/
switch (cmd->sam_task_attr) {
case TCM_HEAD_TAG:
pr_debug("Added HEAD_OF_QUEUE for CDB: 0x%02x\n",
cmd->t_task_cdb[0]);
return false;
case TCM_ORDERED_TAG:
atomic_inc_mb(&dev->dev_ordered_sync);
pr_debug("Added ORDERED for CDB: 0x%02x to ordered list\n",
cmd->t_task_cdb[0]);
/*
* Execute an ORDERED command if no other older commands
* exist that need to be completed first.
*/
if (!atomic_read(&dev->simple_cmds))
return false;
break;
default:
/*
* For SIMPLE and UNTAGGED Task Attribute commands
*/
atomic_inc_mb(&dev->simple_cmds);
break;
}
if (atomic_read(&dev->dev_ordered_sync) == 0)
return false;
spin_lock(&dev->delayed_cmd_lock);
list_add_tail(&cmd->se_delayed_node, &dev->delayed_cmd_list);
spin_unlock(&dev->delayed_cmd_lock);
pr_debug("Added CDB: 0x%02x Task Attr: 0x%02x to delayed CMD listn",
cmd->t_task_cdb[0], cmd->sam_task_attr);
return true;
}
static int __transport_check_aborted_status(struct se_cmd *, int);
void target_execute_cmd(struct se_cmd *cmd)
{
/*
* Determine if frontend context caller is requesting the stopping of
* this command for frontend exceptions.
*
* If the received CDB has aleady been aborted stop processing it here.
*/
spin_lock_irq(&cmd->t_state_lock);
if (__transport_check_aborted_status(cmd, 1)) {
spin_unlock_irq(&cmd->t_state_lock);
return;
}
if (cmd->transport_state & CMD_T_STOP) {
pr_debug("%s:%d CMD_T_STOP for ITT: 0x%08llx\n",
__func__, __LINE__, cmd->tag);
spin_unlock_irq(&cmd->t_state_lock);
complete_all(&cmd->t_transport_stop_comp);
return;
}
cmd->t_state = TRANSPORT_PROCESSING;
cmd->transport_state &= ~CMD_T_PRE_EXECUTE;
cmd->transport_state |= CMD_T_ACTIVE | CMD_T_SENT;
spin_unlock_irq(&cmd->t_state_lock);
if (target_write_prot_action(cmd))
return;
if (target_handle_task_attr(cmd)) {
spin_lock_irq(&cmd->t_state_lock);
cmd->transport_state &= ~CMD_T_SENT;
spin_unlock_irq(&cmd->t_state_lock);
return;
}
__target_execute_cmd(cmd, true);
}
EXPORT_SYMBOL(target_execute_cmd);
/*
* Process all commands up to the last received ORDERED task attribute which
* requires another blocking boundary
*/
static void target_restart_delayed_cmds(struct se_device *dev)
{
for (;;) {
struct se_cmd *cmd;
spin_lock(&dev->delayed_cmd_lock);
if (list_empty(&dev->delayed_cmd_list)) {
spin_unlock(&dev->delayed_cmd_lock);
break;
}
cmd = list_entry(dev->delayed_cmd_list.next,
struct se_cmd, se_delayed_node);
list_del(&cmd->se_delayed_node);
spin_unlock(&dev->delayed_cmd_lock);
cmd->transport_state |= CMD_T_SENT;
__target_execute_cmd(cmd, true);
if (cmd->sam_task_attr == TCM_ORDERED_TAG)
break;
}
}
/*
* Called from I/O completion to determine which dormant/delayed
* and ordered cmds need to have their tasks added to the execution queue.
*/
static void transport_complete_task_attr(struct se_cmd *cmd)
{
struct se_device *dev = cmd->se_dev;
if (dev->transport->transport_flags & TRANSPORT_FLAG_PASSTHROUGH)
return;
if (!(cmd->se_cmd_flags & SCF_TASK_ATTR_SET))
goto restart;
if (cmd->sam_task_attr == TCM_SIMPLE_TAG) {
atomic_dec_mb(&dev->simple_cmds);
dev->dev_cur_ordered_id++;
} else if (cmd->sam_task_attr == TCM_HEAD_TAG) {
dev->dev_cur_ordered_id++;
pr_debug("Incremented dev_cur_ordered_id: %u for HEAD_OF_QUEUE\n",
dev->dev_cur_ordered_id);
} else if (cmd->sam_task_attr == TCM_ORDERED_TAG) {
atomic_dec_mb(&dev->dev_ordered_sync);
dev->dev_cur_ordered_id++;
pr_debug("Incremented dev_cur_ordered_id: %u for ORDERED\n",
dev->dev_cur_ordered_id);
}
cmd->se_cmd_flags &= ~SCF_TASK_ATTR_SET;
restart:
target_restart_delayed_cmds(dev);
}
static void transport_complete_qf(struct se_cmd *cmd)
{
int ret = 0;
transport_complete_task_attr(cmd);
/*
* If a fabric driver ->write_pending() or ->queue_data_in() callback
* has returned neither -ENOMEM or -EAGAIN, assume it's fatal and
* the same callbacks should not be retried. Return CHECK_CONDITION
* if a scsi_status is not already set.
*
* If a fabric driver ->queue_status() has returned non zero, always
* keep retrying no matter what..
*/
if (cmd->t_state == TRANSPORT_COMPLETE_QF_ERR) {
if (cmd->scsi_status)
goto queue_status;
cmd->se_cmd_flags |= SCF_EMULATED_TASK_SENSE;
cmd->scsi_status = SAM_STAT_CHECK_CONDITION;
cmd->scsi_sense_length = TRANSPORT_SENSE_BUFFER;
translate_sense_reason(cmd, TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE);
goto queue_status;
}
/*
* Check if we need to send a sense buffer from
* the struct se_cmd in question. We do NOT want
* to take this path of the IO has been marked as
* needing to be treated like a "normal read". This
* is the case if it's a tape read, and either the
* FM, EOM, or ILI bits are set, but there is no
* sense data.
*/
if (!(cmd->se_cmd_flags & SCF_TREAT_READ_AS_NORMAL) &&
cmd->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE)
goto queue_status;
switch (cmd->data_direction) {
case DMA_FROM_DEVICE:
/* queue status if not treating this as a normal read */
if (cmd->scsi_status &&
!(cmd->se_cmd_flags & SCF_TREAT_READ_AS_NORMAL))
goto queue_status;
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_data_in(cmd);
break;
case DMA_TO_DEVICE:
if (cmd->se_cmd_flags & SCF_BIDI) {
ret = cmd->se_tfo->queue_data_in(cmd);
break;
}
/* fall through */
case DMA_NONE:
queue_status:
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_status(cmd);
break;
default:
break;
}
if (ret < 0) {
transport_handle_queue_full(cmd, cmd->se_dev, ret, false);
return;
}
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
}
static void transport_handle_queue_full(struct se_cmd *cmd, struct se_device *dev,
int err, bool write_pending)
{
/*
* -EAGAIN or -ENOMEM signals retry of ->write_pending() and/or
* ->queue_data_in() callbacks from new process context.
*
* Otherwise for other errors, transport_complete_qf() will send
* CHECK_CONDITION via ->queue_status() instead of attempting to
* retry associated fabric driver data-transfer callbacks.
*/
if (err == -EAGAIN || err == -ENOMEM) {
cmd->t_state = (write_pending) ? TRANSPORT_COMPLETE_QF_WP :
TRANSPORT_COMPLETE_QF_OK;
} else {
pr_warn_ratelimited("Got unknown fabric queue status: %d\n", err);
cmd->t_state = TRANSPORT_COMPLETE_QF_ERR;
}
spin_lock_irq(&dev->qf_cmd_lock);
list_add_tail(&cmd->se_qf_node, &cmd->se_dev->qf_cmd_list);
atomic_inc_mb(&dev->dev_qf_count);
spin_unlock_irq(&cmd->se_dev->qf_cmd_lock);
schedule_work(&cmd->se_dev->qf_work_queue);
}
static bool target_read_prot_action(struct se_cmd *cmd)
{
switch (cmd->prot_op) {
case TARGET_PROT_DIN_STRIP:
if (!(cmd->se_sess->sup_prot_ops & TARGET_PROT_DIN_STRIP)) {
u32 sectors = cmd->data_length >>
ilog2(cmd->se_dev->dev_attrib.block_size);
cmd->pi_err = sbc_dif_verify(cmd, cmd->t_task_lba,
sectors, 0, cmd->t_prot_sg,
0);
if (cmd->pi_err)
return true;
}
break;
case TARGET_PROT_DIN_INSERT:
if (cmd->se_sess->sup_prot_ops & TARGET_PROT_DIN_INSERT)
break;
sbc_dif_generate(cmd);
break;
default:
break;
}
return false;
}
static void target_complete_ok_work(struct work_struct *work)
{
struct se_cmd *cmd = container_of(work, struct se_cmd, work);
int ret;
/*
* Check if we need to move delayed/dormant tasks from cmds on the
* delayed execution list after a HEAD_OF_QUEUE or ORDERED Task
* Attribute.
*/
transport_complete_task_attr(cmd);
/*
* Check to schedule QUEUE_FULL work, or execute an existing
* cmd->transport_qf_callback()
*/
if (atomic_read(&cmd->se_dev->dev_qf_count) != 0)
schedule_work(&cmd->se_dev->qf_work_queue);
/*
* Check if we need to send a sense buffer from
* the struct se_cmd in question. We do NOT want
* to take this path of the IO has been marked as
* needing to be treated like a "normal read". This
* is the case if it's a tape read, and either the
* FM, EOM, or ILI bits are set, but there is no
* sense data.
*/
if (!(cmd->se_cmd_flags & SCF_TREAT_READ_AS_NORMAL) &&
cmd->se_cmd_flags & SCF_TRANSPORT_TASK_SENSE) {
WARN_ON(!cmd->scsi_status);
ret = transport_send_check_condition_and_sense(
cmd, 0, 1);
if (ret)
goto queue_full;
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return;
}
/*
* Check for a callback, used by amongst other things
* XDWRITE_READ_10 and COMPARE_AND_WRITE emulation.
*/
if (cmd->transport_complete_callback) {
sense_reason_t rc;
bool caw = (cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE);
bool zero_dl = !(cmd->data_length);
int post_ret = 0;
rc = cmd->transport_complete_callback(cmd, true, &post_ret);
if (!rc && !post_ret) {
if (caw && zero_dl)
goto queue_rsp;
return;
} else if (rc) {
ret = transport_send_check_condition_and_sense(cmd,
rc, 0);
if (ret)
goto queue_full;
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return;
}
}
queue_rsp:
switch (cmd->data_direction) {
case DMA_FROM_DEVICE:
/*
* if this is a READ-type IO, but SCSI status
* is set, then skip returning data and just
* return the status -- unless this IO is marked
* as needing to be treated as a normal read,
* in which case we want to go ahead and return
* the data. This happens, for example, for tape
* reads with the FM, EOM, or ILI bits set, with
* no sense data.
*/
if (cmd->scsi_status &&
!(cmd->se_cmd_flags & SCF_TREAT_READ_AS_NORMAL))
goto queue_status;
atomic_long_add(cmd->data_length,
&cmd->se_lun->lun_stats.tx_data_octets);
/*
* Perform READ_STRIP of PI using software emulation when
* backend had PI enabled, if the transport will not be
* performing hardware READ_STRIP offload.
*/
if (target_read_prot_action(cmd)) {
ret = transport_send_check_condition_and_sense(cmd,
cmd->pi_err, 0);
if (ret)
goto queue_full;
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return;
}
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_data_in(cmd);
if (ret)
goto queue_full;
break;
case DMA_TO_DEVICE:
atomic_long_add(cmd->data_length,
&cmd->se_lun->lun_stats.rx_data_octets);
/*
* Check if we need to send READ payload for BIDI-COMMAND
*/
if (cmd->se_cmd_flags & SCF_BIDI) {
atomic_long_add(cmd->data_length,
&cmd->se_lun->lun_stats.tx_data_octets);
ret = cmd->se_tfo->queue_data_in(cmd);
if (ret)
goto queue_full;
break;
}
/* fall through */
case DMA_NONE:
queue_status:
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_status(cmd);
if (ret)
goto queue_full;
break;
default:
break;
}
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return;
queue_full:
pr_debug("Handling complete_ok QUEUE_FULL: se_cmd: %p,"
" data_direction: %d\n", cmd, cmd->data_direction);
transport_handle_queue_full(cmd, cmd->se_dev, ret, false);
}
void target_free_sgl(struct scatterlist *sgl, int nents)
{
lib/scatterlist: Fix chaining support in sgl_alloc_order() This patch avoids that workloads with large block sizes (megabytes) can trigger the following call stack with the ib_srpt driver (that driver is the only driver that chains scatterlists allocated by sgl_alloc_order()): BUG: Bad page state in process kworker/0:1H pfn:2423a78 page:fffffb03d08e9e00 count:-3 mapcount:0 mapping: (null) index:0x0 flags: 0x57ffffc0000000() raw: 0057ffffc0000000 0000000000000000 0000000000000000 fffffffdffffffff raw: dead000000000100 dead000000000200 0000000000000000 0000000000000000 page dumped because: nonzero _count CPU: 0 PID: 733 Comm: kworker/0:1H Tainted: G I 4.15.0-rc7.bart+ #1 Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 Workqueue: ib-comp-wq ib_cq_poll_work [ib_core] Call Trace: dump_stack+0x5c/0x83 bad_page+0xf5/0x10f get_page_from_freelist+0xa46/0x11b0 __alloc_pages_nodemask+0x103/0x290 sgl_alloc_order+0x101/0x180 target_alloc_sgl+0x2c/0x40 [target_core_mod] srpt_alloc_rw_ctxs+0x173/0x2d0 [ib_srpt] srpt_handle_new_iu+0x61e/0x7f0 [ib_srpt] __ib_process_cq+0x55/0xa0 [ib_core] ib_cq_poll_work+0x1b/0x60 [ib_core] process_one_work+0x141/0x340 worker_thread+0x47/0x3e0 kthread+0xf5/0x130 ret_from_fork+0x1f/0x30 Fixes: e80a0af4759a ("lib/scatterlist: Introduce sgl_alloc() and sgl_free()") Reported-by: Laurence Oberman <loberman@redhat.com> Tested-by: Laurence Oberman <loberman@redhat.com> Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Cc: Nicholas A. Bellinger <nab@linux-iscsi.org> Cc: Laurence Oberman <loberman@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-01-19 22:00:54 +03:00
sgl_free_n_order(sgl, nents, 0);
}
EXPORT_SYMBOL(target_free_sgl);
static inline void transport_reset_sgl_orig(struct se_cmd *cmd)
{
/*
* Check for saved t_data_sg that may be used for COMPARE_AND_WRITE
* emulation, and free + reset pointers if necessary..
*/
if (!cmd->t_data_sg_orig)
return;
kfree(cmd->t_data_sg);
cmd->t_data_sg = cmd->t_data_sg_orig;
cmd->t_data_sg_orig = NULL;
cmd->t_data_nents = cmd->t_data_nents_orig;
cmd->t_data_nents_orig = 0;
}
static inline void transport_free_pages(struct se_cmd *cmd)
{
if (!(cmd->se_cmd_flags & SCF_PASSTHROUGH_PROT_SG_TO_MEM_NOALLOC)) {
target_free_sgl(cmd->t_prot_sg, cmd->t_prot_nents);
cmd->t_prot_sg = NULL;
cmd->t_prot_nents = 0;
}
if (cmd->se_cmd_flags & SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC) {
/*
* Release special case READ buffer payload required for
* SG_TO_MEM_NOALLOC to function with COMPARE_AND_WRITE
*/
if (cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) {
target_free_sgl(cmd->t_bidi_data_sg,
cmd->t_bidi_data_nents);
cmd->t_bidi_data_sg = NULL;
cmd->t_bidi_data_nents = 0;
}
transport_reset_sgl_orig(cmd);
return;
}
transport_reset_sgl_orig(cmd);
target_free_sgl(cmd->t_data_sg, cmd->t_data_nents);
cmd->t_data_sg = NULL;
cmd->t_data_nents = 0;
target_free_sgl(cmd->t_bidi_data_sg, cmd->t_bidi_data_nents);
cmd->t_bidi_data_sg = NULL;
cmd->t_bidi_data_nents = 0;
}
void *transport_kmap_data_sg(struct se_cmd *cmd)
{
struct scatterlist *sg = cmd->t_data_sg;
struct page **pages;
int i;
/*
* We need to take into account a possible offset here for fabrics like
* tcm_loop who may be using a contig buffer from the SCSI midlayer for
* control CDBs passed as SGLs via transport_generic_map_mem_to_cmd()
*/
if (!cmd->t_data_nents)
return NULL;
BUG_ON(!sg);
if (cmd->t_data_nents == 1)
return kmap(sg_page(sg)) + sg->offset;
/* >1 page. use vmap */
pages = kmalloc_array(cmd->t_data_nents, sizeof(*pages), GFP_KERNEL);
if (!pages)
return NULL;
/* convert sg[] to pages[] */
for_each_sg(cmd->t_data_sg, sg, cmd->t_data_nents, i) {
pages[i] = sg_page(sg);
}
cmd->t_data_vmap = vmap(pages, cmd->t_data_nents, VM_MAP, PAGE_KERNEL);
kfree(pages);
if (!cmd->t_data_vmap)
return NULL;
return cmd->t_data_vmap + cmd->t_data_sg[0].offset;
}
EXPORT_SYMBOL(transport_kmap_data_sg);
void transport_kunmap_data_sg(struct se_cmd *cmd)
{
if (!cmd->t_data_nents) {
return;
} else if (cmd->t_data_nents == 1) {
kunmap(sg_page(cmd->t_data_sg));
return;
}
vunmap(cmd->t_data_vmap);
cmd->t_data_vmap = NULL;
}
EXPORT_SYMBOL(transport_kunmap_data_sg);
int
target_alloc_sgl(struct scatterlist **sgl, unsigned int *nents, u32 length,
bool zero_page, bool chainable)
{
gfp_t gfp = GFP_KERNEL | (zero_page ? __GFP_ZERO : 0);
*sgl = sgl_alloc_order(length, 0, chainable, gfp, nents);
return *sgl ? 0 : -ENOMEM;
}
EXPORT_SYMBOL(target_alloc_sgl);
/*
* Allocate any required resources to execute the command. For writes we
* might not have the payload yet, so notify the fabric via a call to
* ->write_pending instead. Otherwise place it on the execution queue.
*/
sense_reason_t
transport_generic_new_cmd(struct se_cmd *cmd)
{
unsigned long flags;
int ret = 0;
bool zero_flag = !(cmd->se_cmd_flags & SCF_SCSI_DATA_CDB);
if (cmd->prot_op != TARGET_PROT_NORMAL &&
!(cmd->se_cmd_flags & SCF_PASSTHROUGH_PROT_SG_TO_MEM_NOALLOC)) {
ret = target_alloc_sgl(&cmd->t_prot_sg, &cmd->t_prot_nents,
cmd->prot_length, true, false);
if (ret < 0)
return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
}
/*
* Determine is the TCM fabric module has already allocated physical
* memory, and is directly calling transport_generic_map_mem_to_cmd()
* beforehand.
*/
if (!(cmd->se_cmd_flags & SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC) &&
cmd->data_length) {
if ((cmd->se_cmd_flags & SCF_BIDI) ||
(cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE)) {
u32 bidi_length;
if (cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE)
bidi_length = cmd->t_task_nolb *
cmd->se_dev->dev_attrib.block_size;
else
bidi_length = cmd->data_length;
ret = target_alloc_sgl(&cmd->t_bidi_data_sg,
&cmd->t_bidi_data_nents,
bidi_length, zero_flag, false);
if (ret < 0)
return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
}
ret = target_alloc_sgl(&cmd->t_data_sg, &cmd->t_data_nents,
cmd->data_length, zero_flag, false);
if (ret < 0)
return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
} else if ((cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) &&
cmd->data_length) {
/*
* Special case for COMPARE_AND_WRITE with fabrics
* using SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC.
*/
u32 caw_length = cmd->t_task_nolb *
cmd->se_dev->dev_attrib.block_size;
ret = target_alloc_sgl(&cmd->t_bidi_data_sg,
&cmd->t_bidi_data_nents,
caw_length, zero_flag, false);
if (ret < 0)
return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
}
/*
* If this command is not a write we can execute it right here,
* for write buffers we need to notify the fabric driver first
* and let it call back once the write buffers are ready.
*/
target_add_to_state_list(cmd);
if (cmd->data_direction != DMA_TO_DEVICE || cmd->data_length == 0) {
target_execute_cmd(cmd);
return 0;
}
spin_lock_irqsave(&cmd->t_state_lock, flags);
cmd->t_state = TRANSPORT_WRITE_PENDING;
/*
* Determine if frontend context caller is requesting the stopping of
* this command for frontend exceptions.
*/
if (cmd->transport_state & CMD_T_STOP) {
pr_debug("%s:%d CMD_T_STOP for ITT: 0x%08llx\n",
__func__, __LINE__, cmd->tag);
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
complete_all(&cmd->t_transport_stop_comp);
return 0;
}
cmd->transport_state &= ~CMD_T_ACTIVE;
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
ret = cmd->se_tfo->write_pending(cmd);
if (ret)
goto queue_full;
return 0;
queue_full:
pr_debug("Handling write_pending QUEUE__FULL: se_cmd: %p\n", cmd);
transport_handle_queue_full(cmd, cmd->se_dev, ret, true);
return 0;
}
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
EXPORT_SYMBOL(transport_generic_new_cmd);
static void transport_write_pending_qf(struct se_cmd *cmd)
{
unsigned long flags;
int ret;
bool stop;
spin_lock_irqsave(&cmd->t_state_lock, flags);
stop = (cmd->transport_state & (CMD_T_STOP | CMD_T_ABORTED));
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
if (stop) {
pr_debug("%s:%d CMD_T_STOP|CMD_T_ABORTED for ITT: 0x%08llx\n",
__func__, __LINE__, cmd->tag);
complete_all(&cmd->t_transport_stop_comp);
return;
}
ret = cmd->se_tfo->write_pending(cmd);
if (ret) {
pr_debug("Handling write_pending QUEUE__FULL: se_cmd: %p\n",
cmd);
transport_handle_queue_full(cmd, cmd->se_dev, ret, true);
}
}
static bool
__transport_wait_for_tasks(struct se_cmd *, bool, bool *, bool *,
unsigned long *flags);
static void target_wait_free_cmd(struct se_cmd *cmd, bool *aborted, bool *tas)
{
unsigned long flags;
spin_lock_irqsave(&cmd->t_state_lock, flags);
__transport_wait_for_tasks(cmd, true, aborted, tas, &flags);
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
}
int transport_generic_free_cmd(struct se_cmd *cmd, int wait_for_tasks)
{
int ret = 0;
bool aborted = false, tas = false;
2011-10-09 12:00:58 +04:00
if (!(cmd->se_cmd_flags & SCF_SE_LUN_CMD)) {
if (wait_for_tasks && (cmd->se_cmd_flags & SCF_SCSI_TMR_CDB))
target_wait_free_cmd(cmd, &aborted, &tas);
2011-10-09 12:00:58 +04:00
if (!aborted || tas)
ret = target_put_sess_cmd(cmd);
2011-10-09 12:00:58 +04:00
} else {
if (wait_for_tasks)
target_wait_free_cmd(cmd, &aborted, &tas);
/*
* Handle WRITE failure case where transport_generic_new_cmd()
* has already added se_cmd to state_list, but fabric has
* failed command before I/O submission.
*/
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
if (cmd->state_active)
target_remove_from_state_list(cmd);
2011-10-09 12:00:58 +04:00
if (cmd->se_lun)
transport_lun_remove_cmd(cmd);
if (!aborted || tas)
ret = target_put_sess_cmd(cmd);
}
/*
* If the task has been internally aborted due to TMR ABORT_TASK
* or LUN_RESET, target_core_tmr.c is responsible for performing
* the remaining calls to target_put_sess_cmd(), and not the
* callers of this function.
*/
if (aborted) {
pr_debug("Detected CMD_T_ABORTED for ITT: %llu\n", cmd->tag);
wait_for_completion(&cmd->cmd_wait_comp);
cmd->se_tfo->release_cmd(cmd);
ret = 1;
}
return ret;
}
EXPORT_SYMBOL(transport_generic_free_cmd);
/**
* target_get_sess_cmd - Add command to active ->sess_cmd_list
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
* @se_cmd: command descriptor to add
* @ack_kref: Signal that fabric will perform an ack target_put_sess_cmd()
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
*/
int target_get_sess_cmd(struct se_cmd *se_cmd, bool ack_kref)
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
{
struct se_session *se_sess = se_cmd->se_sess;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
unsigned long flags;
target: Check sess_tearing_down in target_get_sess_cmd() Target core code assumes that target_splice_sess_cmd_list() has set sess_tearing_down and moved the list of pending commands to sess_wait_list, no more commands will be added to the session; if any are added, nothing keeps the se_session from being freed while the command is still in flight, which e.g. leads to use-after-free of se_cmd->se_sess in target_release_cmd_kref(). To enforce this invariant, put a check of sess_tearing_down inside of sess_cmd_lock in target_get_sess_cmd(); any checks before this are racy and can lead to the use-after-free described above. For example, the qla_target check in qlt_do_work() checks sess_tearing_down from work thread context but then drops all locks before calling target_submit_cmd() (as it must, since that is a sleeping function). However, since no locks are held, anything can happen with respect to the session it has looked up -- although it does correctly get sess_kref within its lock, so the memory won't be freed while target_submit_cmd() is actually running, nothing stops eg an ACL from being dropped and calling ->shutdown_session() (which calls into target_splice_sess_cmd_list()) before we get to target_get_sess_cmd(). Once this happens, the se_session memory can be freed as soon as target_submit_cmd() returns and qlt_do_work() drops its reference, even though we've just added a command to sess_cmd_list. To prevent this use-after-free, check sess_tearing_down inside of sess_cmd_lock right before target_get_sess_cmd() adds a command to sess_cmd_list; this is synchronized with target_splice_sess_cmd_list() so that every command is either waited for or not added to the queue. (nab: Keep target_submit_cmd() returning void for now..) Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2012-07-16 22:04:40 +04:00
int ret = 0;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
/*
* Add a second kref if the fabric caller is expecting to handle
* fabric acknowledgement that requires two target_put_sess_cmd()
* invocations before se_cmd descriptor release.
*/
if (ack_kref) {
if (!kref_get_unless_zero(&se_cmd->cmd_kref))
return -EINVAL;
se_cmd->se_cmd_flags |= SCF_ACK_KREF;
}
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
target: Check sess_tearing_down in target_get_sess_cmd() Target core code assumes that target_splice_sess_cmd_list() has set sess_tearing_down and moved the list of pending commands to sess_wait_list, no more commands will be added to the session; if any are added, nothing keeps the se_session from being freed while the command is still in flight, which e.g. leads to use-after-free of se_cmd->se_sess in target_release_cmd_kref(). To enforce this invariant, put a check of sess_tearing_down inside of sess_cmd_lock in target_get_sess_cmd(); any checks before this are racy and can lead to the use-after-free described above. For example, the qla_target check in qlt_do_work() checks sess_tearing_down from work thread context but then drops all locks before calling target_submit_cmd() (as it must, since that is a sleeping function). However, since no locks are held, anything can happen with respect to the session it has looked up -- although it does correctly get sess_kref within its lock, so the memory won't be freed while target_submit_cmd() is actually running, nothing stops eg an ACL from being dropped and calling ->shutdown_session() (which calls into target_splice_sess_cmd_list()) before we get to target_get_sess_cmd(). Once this happens, the se_session memory can be freed as soon as target_submit_cmd() returns and qlt_do_work() drops its reference, even though we've just added a command to sess_cmd_list. To prevent this use-after-free, check sess_tearing_down inside of sess_cmd_lock right before target_get_sess_cmd() adds a command to sess_cmd_list; this is synchronized with target_splice_sess_cmd_list() so that every command is either waited for or not added to the queue. (nab: Keep target_submit_cmd() returning void for now..) Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2012-07-16 22:04:40 +04:00
if (se_sess->sess_tearing_down) {
ret = -ESHUTDOWN;
goto out;
}
se_cmd->transport_state |= CMD_T_PRE_EXECUTE;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
list_add_tail(&se_cmd->se_cmd_list, &se_sess->sess_cmd_list);
target: Check sess_tearing_down in target_get_sess_cmd() Target core code assumes that target_splice_sess_cmd_list() has set sess_tearing_down and moved the list of pending commands to sess_wait_list, no more commands will be added to the session; if any are added, nothing keeps the se_session from being freed while the command is still in flight, which e.g. leads to use-after-free of se_cmd->se_sess in target_release_cmd_kref(). To enforce this invariant, put a check of sess_tearing_down inside of sess_cmd_lock in target_get_sess_cmd(); any checks before this are racy and can lead to the use-after-free described above. For example, the qla_target check in qlt_do_work() checks sess_tearing_down from work thread context but then drops all locks before calling target_submit_cmd() (as it must, since that is a sleeping function). However, since no locks are held, anything can happen with respect to the session it has looked up -- although it does correctly get sess_kref within its lock, so the memory won't be freed while target_submit_cmd() is actually running, nothing stops eg an ACL from being dropped and calling ->shutdown_session() (which calls into target_splice_sess_cmd_list()) before we get to target_get_sess_cmd(). Once this happens, the se_session memory can be freed as soon as target_submit_cmd() returns and qlt_do_work() drops its reference, even though we've just added a command to sess_cmd_list. To prevent this use-after-free, check sess_tearing_down inside of sess_cmd_lock right before target_get_sess_cmd() adds a command to sess_cmd_list; this is synchronized with target_splice_sess_cmd_list() so that every command is either waited for or not added to the queue. (nab: Keep target_submit_cmd() returning void for now..) Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2012-07-16 22:04:40 +04:00
out:
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
if (ret && ack_kref)
target_put_sess_cmd(se_cmd);
target: Check sess_tearing_down in target_get_sess_cmd() Target core code assumes that target_splice_sess_cmd_list() has set sess_tearing_down and moved the list of pending commands to sess_wait_list, no more commands will be added to the session; if any are added, nothing keeps the se_session from being freed while the command is still in flight, which e.g. leads to use-after-free of se_cmd->se_sess in target_release_cmd_kref(). To enforce this invariant, put a check of sess_tearing_down inside of sess_cmd_lock in target_get_sess_cmd(); any checks before this are racy and can lead to the use-after-free described above. For example, the qla_target check in qlt_do_work() checks sess_tearing_down from work thread context but then drops all locks before calling target_submit_cmd() (as it must, since that is a sleeping function). However, since no locks are held, anything can happen with respect to the session it has looked up -- although it does correctly get sess_kref within its lock, so the memory won't be freed while target_submit_cmd() is actually running, nothing stops eg an ACL from being dropped and calling ->shutdown_session() (which calls into target_splice_sess_cmd_list()) before we get to target_get_sess_cmd(). Once this happens, the se_session memory can be freed as soon as target_submit_cmd() returns and qlt_do_work() drops its reference, even though we've just added a command to sess_cmd_list. To prevent this use-after-free, check sess_tearing_down inside of sess_cmd_lock right before target_get_sess_cmd() adds a command to sess_cmd_list; this is synchronized with target_splice_sess_cmd_list() so that every command is either waited for or not added to the queue. (nab: Keep target_submit_cmd() returning void for now..) Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2012-07-16 22:04:40 +04:00
return ret;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
}
EXPORT_SYMBOL(target_get_sess_cmd);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
static void target_free_cmd_mem(struct se_cmd *cmd)
{
transport_free_pages(cmd);
if (cmd->se_cmd_flags & SCF_SCSI_TMR_CDB)
core_tmr_release_req(cmd->se_tmr_req);
if (cmd->t_task_cdb != cmd->__t_task_cdb)
kfree(cmd->t_task_cdb);
}
static void target_release_cmd_kref(struct kref *kref)
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
{
struct se_cmd *se_cmd = container_of(kref, struct se_cmd, cmd_kref);
struct se_session *se_sess = se_cmd->se_sess;
unsigned long flags;
bool fabric_stop;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
if (se_sess) {
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
spin_lock(&se_cmd->t_state_lock);
fabric_stop = (se_cmd->transport_state & CMD_T_FABRIC_STOP) &&
(se_cmd->transport_state & CMD_T_ABORTED);
spin_unlock(&se_cmd->t_state_lock);
if (se_cmd->cmd_wait_set || fabric_stop) {
list_del_init(&se_cmd->se_cmd_list);
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
target_free_cmd_mem(se_cmd);
complete(&se_cmd->cmd_wait_comp);
return;
}
list_del_init(&se_cmd->se_cmd_list);
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
}
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 08:31:09 +03:00
target_free_cmd_mem(se_cmd);
se_cmd->se_tfo->release_cmd(se_cmd);
}
/**
* target_put_sess_cmd - decrease the command reference count
* @se_cmd: command to drop a reference from
*
* Returns 1 if and only if this target_put_sess_cmd() call caused the
* refcount to drop to zero. Returns zero otherwise.
*/
int target_put_sess_cmd(struct se_cmd *se_cmd)
{
return kref_put(&se_cmd->cmd_kref, target_release_cmd_kref);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
}
EXPORT_SYMBOL(target_put_sess_cmd);
static const char *data_dir_name(enum dma_data_direction d)
{
switch (d) {
case DMA_BIDIRECTIONAL: return "BIDI";
case DMA_TO_DEVICE: return "WRITE";
case DMA_FROM_DEVICE: return "READ";
case DMA_NONE: return "NONE";
}
return "(?)";
}
static const char *cmd_state_name(enum transport_state_table t)
{
switch (t) {
case TRANSPORT_NO_STATE: return "NO_STATE";
case TRANSPORT_NEW_CMD: return "NEW_CMD";
case TRANSPORT_WRITE_PENDING: return "WRITE_PENDING";
case TRANSPORT_PROCESSING: return "PROCESSING";
case TRANSPORT_COMPLETE: return "COMPLETE";
case TRANSPORT_ISTATE_PROCESSING:
return "ISTATE_PROCESSING";
case TRANSPORT_COMPLETE_QF_WP: return "COMPLETE_QF_WP";
case TRANSPORT_COMPLETE_QF_OK: return "COMPLETE_QF_OK";
case TRANSPORT_COMPLETE_QF_ERR: return "COMPLETE_QF_ERR";
}
return "(?)";
}
static void target_append_str(char **str, const char *txt)
{
char *prev = *str;
*str = *str ? kasprintf(GFP_ATOMIC, "%s,%s", *str, txt) :
kstrdup(txt, GFP_ATOMIC);
kfree(prev);
}
/*
* Convert a transport state bitmask into a string. The caller is
* responsible for freeing the returned pointer.
*/
static char *target_ts_to_str(u32 ts)
{
char *str = NULL;
if (ts & CMD_T_ABORTED)
target_append_str(&str, "aborted");
if (ts & CMD_T_ACTIVE)
target_append_str(&str, "active");
if (ts & CMD_T_COMPLETE)
target_append_str(&str, "complete");
if (ts & CMD_T_SENT)
target_append_str(&str, "sent");
if (ts & CMD_T_STOP)
target_append_str(&str, "stop");
if (ts & CMD_T_FABRIC_STOP)
target_append_str(&str, "fabric_stop");
return str;
}
static const char *target_tmf_name(enum tcm_tmreq_table tmf)
{
switch (tmf) {
case TMR_ABORT_TASK: return "ABORT_TASK";
case TMR_ABORT_TASK_SET: return "ABORT_TASK_SET";
case TMR_CLEAR_ACA: return "CLEAR_ACA";
case TMR_CLEAR_TASK_SET: return "CLEAR_TASK_SET";
case TMR_LUN_RESET: return "LUN_RESET";
case TMR_TARGET_WARM_RESET: return "TARGET_WARM_RESET";
case TMR_TARGET_COLD_RESET: return "TARGET_COLD_RESET";
case TMR_UNKNOWN: break;
}
return "(?)";
}
void target_show_cmd(const char *pfx, struct se_cmd *cmd)
{
char *ts_str = target_ts_to_str(cmd->transport_state);
const u8 *cdb = cmd->t_task_cdb;
struct se_tmr_req *tmf = cmd->se_tmr_req;
if (!(cmd->se_cmd_flags & SCF_SCSI_TMR_CDB)) {
pr_debug("%scmd %#02x:%#02x with tag %#llx dir %s i_state %d t_state %s len %d refcnt %d transport_state %s\n",
pfx, cdb[0], cdb[1], cmd->tag,
data_dir_name(cmd->data_direction),
cmd->se_tfo->get_cmd_state(cmd),
cmd_state_name(cmd->t_state), cmd->data_length,
kref_read(&cmd->cmd_kref), ts_str);
} else {
pr_debug("%stmf %s with tag %#llx ref_task_tag %#llx i_state %d t_state %s refcnt %d transport_state %s\n",
pfx, target_tmf_name(tmf->function), cmd->tag,
tmf->ref_task_tag, cmd->se_tfo->get_cmd_state(cmd),
cmd_state_name(cmd->t_state),
kref_read(&cmd->cmd_kref), ts_str);
}
kfree(ts_str);
}
EXPORT_SYMBOL(target_show_cmd);
/**
* target_sess_cmd_list_set_waiting - Flag all commands in
* sess_cmd_list to complete cmd_wait_comp. Set
* sess_tearing_down so no more commands are queued.
* @se_sess: session to flag
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
*/
void target_sess_cmd_list_set_waiting(struct se_session *se_sess)
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
{
struct se_cmd *se_cmd, *tmp_cmd;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
unsigned long flags;
int rc;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
if (se_sess->sess_tearing_down) {
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
return;
}
se_sess->sess_tearing_down = 1;
list_splice_init(&se_sess->sess_cmd_list, &se_sess->sess_wait_list);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
list_for_each_entry_safe(se_cmd, tmp_cmd,
&se_sess->sess_wait_list, se_cmd_list) {
rc = kref_get_unless_zero(&se_cmd->cmd_kref);
if (rc) {
se_cmd->cmd_wait_set = 1;
spin_lock(&se_cmd->t_state_lock);
se_cmd->transport_state |= CMD_T_FABRIC_STOP;
spin_unlock(&se_cmd->t_state_lock);
} else
list_del_init(&se_cmd->se_cmd_list);
}
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
}
EXPORT_SYMBOL(target_sess_cmd_list_set_waiting);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
/**
* target_wait_for_sess_cmds - Wait for outstanding descriptors
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
* @se_sess: session to wait for active I/O
*/
void target_wait_for_sess_cmds(struct se_session *se_sess)
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
{
struct se_cmd *se_cmd, *tmp_cmd;
unsigned long flags;
bool tas;
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
list_for_each_entry_safe(se_cmd, tmp_cmd,
&se_sess->sess_wait_list, se_cmd_list) {
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
pr_debug("Waiting for se_cmd: %p t_state: %d, fabric state:"
" %d\n", se_cmd, se_cmd->t_state,
se_cmd->se_tfo->get_cmd_state(se_cmd));
spin_lock_irqsave(&se_cmd->t_state_lock, flags);
tas = (se_cmd->transport_state & CMD_T_TAS);
spin_unlock_irqrestore(&se_cmd->t_state_lock, flags);
if (!target_put_sess_cmd(se_cmd)) {
if (tas)
target_put_sess_cmd(se_cmd);
}
wait_for_completion(&se_cmd->cmd_wait_comp);
pr_debug("After cmd_wait_comp: se_cmd: %p t_state: %d"
" fabric state: %d\n", se_cmd, se_cmd->t_state,
se_cmd->se_tfo->get_cmd_state(se_cmd));
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
se_cmd->se_tfo->release_cmd(se_cmd);
}
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
WARN_ON(!list_empty(&se_sess->sess_cmd_list));
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
}
EXPORT_SYMBOL(target_wait_for_sess_cmds);
target: Fix NULL dereference during LUN lookup + active I/O shutdown When transport_clear_lun_ref() is shutting down a se_lun via configfs with new I/O in-flight, it's possible to trigger a NULL pointer dereference in transport_lookup_cmd_lun() due to the fact percpu_ref_get() doesn't do any __PERCPU_REF_DEAD checking before incrementing lun->lun_ref.count after lun->lun_ref has switched to atomic_t mode. This results in a NULL pointer dereference as LUN shutdown code in core_tpg_remove_lun() continues running after the existing ->release() -> core_tpg_lun_ref_release() callback completes, and clears the RCU protected se_lun->lun_se_dev pointer. During the OOPs, the state of lun->lun_ref in the process which triggered the NULL pointer dereference looks like the following on v4.1.y stable code: struct se_lun { lun_link_magic = 4294932337, lun_status = TRANSPORT_LUN_STATUS_FREE, ..... lun_se_dev = 0x0, lun_sep = 0x0, ..... lun_ref = { count = { counter = 1 }, percpu_count_ptr = 3, release = 0xffffffffa02fa1e0 <core_tpg_lun_ref_release>, confirm_switch = 0x0, force_atomic = false, rcu = { next = 0xffff88154fa1a5d0, func = 0xffffffff8137c4c0 <percpu_ref_switch_to_atomic_rcu> } } } To address this bug, use percpu_ref_tryget_live() to ensure once __PERCPU_REF_DEAD is visable on all CPUs and ->lun_ref has switched to atomic_t, all new I/Os will fail to obtain a new lun->lun_ref reference. Also use an explicit percpu_ref_kill_and_confirm() callback to block on ->lun_ref_comp to allow the first stage and associated RCU grace period to complete, and then block on ->lun_ref_shutdown waiting for the final percpu_ref_put() to drop the last reference via transport_lun_remove_cmd() before continuing with core_tpg_remove_lun() shutdown. Reported-by: Rob Millner <rlm@daterainc.com> Tested-by: Rob Millner <rlm@daterainc.com> Cc: Rob Millner <rlm@daterainc.com> Tested-by: Vaibhav Tandon <vst@datera.io> Cc: Vaibhav Tandon <vst@datera.io> Tested-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com> Cc: <stable@vger.kernel.org> # v3.14+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-02-23 09:06:32 +03:00
static void target_lun_confirm(struct percpu_ref *ref)
{
struct se_lun *lun = container_of(ref, struct se_lun, lun_ref);
complete(&lun->lun_ref_comp);
}
void transport_clear_lun_ref(struct se_lun *lun)
{
target: Fix NULL dereference during LUN lookup + active I/O shutdown When transport_clear_lun_ref() is shutting down a se_lun via configfs with new I/O in-flight, it's possible to trigger a NULL pointer dereference in transport_lookup_cmd_lun() due to the fact percpu_ref_get() doesn't do any __PERCPU_REF_DEAD checking before incrementing lun->lun_ref.count after lun->lun_ref has switched to atomic_t mode. This results in a NULL pointer dereference as LUN shutdown code in core_tpg_remove_lun() continues running after the existing ->release() -> core_tpg_lun_ref_release() callback completes, and clears the RCU protected se_lun->lun_se_dev pointer. During the OOPs, the state of lun->lun_ref in the process which triggered the NULL pointer dereference looks like the following on v4.1.y stable code: struct se_lun { lun_link_magic = 4294932337, lun_status = TRANSPORT_LUN_STATUS_FREE, ..... lun_se_dev = 0x0, lun_sep = 0x0, ..... lun_ref = { count = { counter = 1 }, percpu_count_ptr = 3, release = 0xffffffffa02fa1e0 <core_tpg_lun_ref_release>, confirm_switch = 0x0, force_atomic = false, rcu = { next = 0xffff88154fa1a5d0, func = 0xffffffff8137c4c0 <percpu_ref_switch_to_atomic_rcu> } } } To address this bug, use percpu_ref_tryget_live() to ensure once __PERCPU_REF_DEAD is visable on all CPUs and ->lun_ref has switched to atomic_t, all new I/Os will fail to obtain a new lun->lun_ref reference. Also use an explicit percpu_ref_kill_and_confirm() callback to block on ->lun_ref_comp to allow the first stage and associated RCU grace period to complete, and then block on ->lun_ref_shutdown waiting for the final percpu_ref_put() to drop the last reference via transport_lun_remove_cmd() before continuing with core_tpg_remove_lun() shutdown. Reported-by: Rob Millner <rlm@daterainc.com> Tested-by: Rob Millner <rlm@daterainc.com> Cc: Rob Millner <rlm@daterainc.com> Tested-by: Vaibhav Tandon <vst@datera.io> Cc: Vaibhav Tandon <vst@datera.io> Tested-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com> Cc: <stable@vger.kernel.org> # v3.14+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-02-23 09:06:32 +03:00
/*
* Mark the percpu-ref as DEAD, switch to atomic_t mode, drop
* the initial reference and schedule confirm kill to be
* executed after one full RCU grace period has completed.
*/
percpu_ref_kill_and_confirm(&lun->lun_ref, target_lun_confirm);
/*
* The first completion waits for percpu_ref_switch_to_atomic_rcu()
* to call target_lun_confirm after lun->lun_ref has been marked
* as __PERCPU_REF_DEAD on all CPUs, and switches to atomic_t
* mode so that percpu_ref_tryget_live() lookup of lun->lun_ref
* fails for all new incoming I/O.
*/
wait_for_completion(&lun->lun_ref_comp);
target: Fix NULL dereference during LUN lookup + active I/O shutdown When transport_clear_lun_ref() is shutting down a se_lun via configfs with new I/O in-flight, it's possible to trigger a NULL pointer dereference in transport_lookup_cmd_lun() due to the fact percpu_ref_get() doesn't do any __PERCPU_REF_DEAD checking before incrementing lun->lun_ref.count after lun->lun_ref has switched to atomic_t mode. This results in a NULL pointer dereference as LUN shutdown code in core_tpg_remove_lun() continues running after the existing ->release() -> core_tpg_lun_ref_release() callback completes, and clears the RCU protected se_lun->lun_se_dev pointer. During the OOPs, the state of lun->lun_ref in the process which triggered the NULL pointer dereference looks like the following on v4.1.y stable code: struct se_lun { lun_link_magic = 4294932337, lun_status = TRANSPORT_LUN_STATUS_FREE, ..... lun_se_dev = 0x0, lun_sep = 0x0, ..... lun_ref = { count = { counter = 1 }, percpu_count_ptr = 3, release = 0xffffffffa02fa1e0 <core_tpg_lun_ref_release>, confirm_switch = 0x0, force_atomic = false, rcu = { next = 0xffff88154fa1a5d0, func = 0xffffffff8137c4c0 <percpu_ref_switch_to_atomic_rcu> } } } To address this bug, use percpu_ref_tryget_live() to ensure once __PERCPU_REF_DEAD is visable on all CPUs and ->lun_ref has switched to atomic_t, all new I/Os will fail to obtain a new lun->lun_ref reference. Also use an explicit percpu_ref_kill_and_confirm() callback to block on ->lun_ref_comp to allow the first stage and associated RCU grace period to complete, and then block on ->lun_ref_shutdown waiting for the final percpu_ref_put() to drop the last reference via transport_lun_remove_cmd() before continuing with core_tpg_remove_lun() shutdown. Reported-by: Rob Millner <rlm@daterainc.com> Tested-by: Rob Millner <rlm@daterainc.com> Cc: Rob Millner <rlm@daterainc.com> Tested-by: Vaibhav Tandon <vst@datera.io> Cc: Vaibhav Tandon <vst@datera.io> Tested-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com> Cc: <stable@vger.kernel.org> # v3.14+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-02-23 09:06:32 +03:00
/*
* The second completion waits for percpu_ref_put_many() to
* invoke ->release() after lun->lun_ref has switched to
* atomic_t mode, and lun->lun_ref.count has reached zero.
*
* At this point all target-core lun->lun_ref references have
* been dropped via transport_lun_remove_cmd(), and it's safe
* to proceed with the remaining LUN shutdown.
*/
wait_for_completion(&lun->lun_shutdown_comp);
}
static bool
__transport_wait_for_tasks(struct se_cmd *cmd, bool fabric_stop,
bool *aborted, bool *tas, unsigned long *flags)
__releases(&cmd->t_state_lock)
__acquires(&cmd->t_state_lock)
{
assert_spin_locked(&cmd->t_state_lock);
WARN_ON_ONCE(!irqs_disabled());
if (fabric_stop)
cmd->transport_state |= CMD_T_FABRIC_STOP;
if (cmd->transport_state & CMD_T_ABORTED)
*aborted = true;
if (cmd->transport_state & CMD_T_TAS)
*tas = true;
if (!(cmd->se_cmd_flags & SCF_SE_LUN_CMD) &&
!(cmd->se_cmd_flags & SCF_SCSI_TMR_CDB))
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
return false;
if (!(cmd->se_cmd_flags & SCF_SUPPORTED_SAM_OPCODE) &&
!(cmd->se_cmd_flags & SCF_SCSI_TMR_CDB))
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
return false;
if (!(cmd->transport_state & CMD_T_ACTIVE))
return false;
if (fabric_stop && *aborted)
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
return false;
cmd->transport_state |= CMD_T_STOP;
target_show_cmd("wait_for_tasks: Stopping ", cmd);
spin_unlock_irqrestore(&cmd->t_state_lock, *flags);
while (!wait_for_completion_timeout(&cmd->t_transport_stop_comp,
180 * HZ))
target_show_cmd("wait for tasks: ", cmd);
spin_lock_irqsave(&cmd->t_state_lock, *flags);
cmd->transport_state &= ~(CMD_T_ACTIVE | CMD_T_STOP);
pr_debug("wait_for_tasks: Stopped wait_for_completion(&cmd->"
"t_transport_stop_comp) for ITT: 0x%08llx\n", cmd->tag);
return true;
}
/**
* transport_wait_for_tasks - set CMD_T_STOP and wait for t_transport_stop_comp
* @cmd: command to wait on
*/
bool transport_wait_for_tasks(struct se_cmd *cmd)
{
unsigned long flags;
bool ret, aborted = false, tas = false;
spin_lock_irqsave(&cmd->t_state_lock, flags);
ret = __transport_wait_for_tasks(cmd, false, &aborted, &tas, &flags);
2011-10-09 12:00:58 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
target: Add generic active I/O shutdown logic This patch adds the initial pieces of generic active I/O shutdown logic. This is intended to be a 'opt-in' feature for fabric modules that includes the following functions to provide a mechinism for fabric modules to track se_cmd via se_session->sess_cmd_list: *) target_get_sess_cmd() - Add se_cmd to sess->sess_cmd_list, called from fabric module incoming I/O path. *) target_put_sess_cmd() - Check for completion or drop se_cmd from ->sess_cmd_list *) target_splice_sess_cmd_list() - Splice active I/O list from ->sess_cmd_list to ->sess_wait_list, can called with HW fabric lock held. *) target_wait_for_sess_cmds() - Walk ->sess_wait_list waiting on individual ->cmd_wait_comp. Optional transport_wait_for_tasks() call. target_splice_sess_cmd_list() is allowed to be called under HW fabric lock, and performs the splice into se_sess->sess_wait_list and set se_cmd->cmd_wait_set. Then target_wait_for_sess_cmds() walks the list waiting for individual target_put_sess_cmd() fabric callbacks to complete. It also adds TFO->check_release_cmd() to split the completion and memory release calls, where a fabric module uses target_put_sess_cmd() to check for I/O completion during session shutdown. This is currently pushed out into fabric modules as current fabric code may sleep here waiting for TFO->check_stop_free() to complete in main response path, and because target_wait_for_sess_cmds() calling TFO->release_cmd() to free fabric descriptor memory directly. Cc: Christoph Hellwig <hch@lst.de> Cc: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas A. Bellinger <nab@linux-iscsi.org>
2011-11-03 08:52:08 +04:00
return ret;
}
2011-10-09 12:00:58 +04:00
EXPORT_SYMBOL(transport_wait_for_tasks);
struct sense_info {
u8 key;
u8 asc;
u8 ascq;
bool add_sector_info;
};
static const struct sense_info sense_info_table[] = {
[TCM_NO_SENSE] = {
.key = NOT_READY
},
[TCM_NON_EXISTENT_LUN] = {
.key = ILLEGAL_REQUEST,
.asc = 0x25 /* LOGICAL UNIT NOT SUPPORTED */
},
[TCM_UNSUPPORTED_SCSI_OPCODE] = {
.key = ILLEGAL_REQUEST,
.asc = 0x20, /* INVALID COMMAND OPERATION CODE */
},
[TCM_SECTOR_COUNT_TOO_MANY] = {
.key = ILLEGAL_REQUEST,
.asc = 0x20, /* INVALID COMMAND OPERATION CODE */
},
[TCM_UNKNOWN_MODE_PAGE] = {
.key = ILLEGAL_REQUEST,
.asc = 0x24, /* INVALID FIELD IN CDB */
},
[TCM_CHECK_CONDITION_ABORT_CMD] = {
.key = ABORTED_COMMAND,
.asc = 0x29, /* BUS DEVICE RESET FUNCTION OCCURRED */
.ascq = 0x03,
},
[TCM_INCORRECT_AMOUNT_OF_DATA] = {
.key = ABORTED_COMMAND,
.asc = 0x0c, /* WRITE ERROR */
.ascq = 0x0d, /* NOT ENOUGH UNSOLICITED DATA */
},
[TCM_INVALID_CDB_FIELD] = {
.key = ILLEGAL_REQUEST,
.asc = 0x24, /* INVALID FIELD IN CDB */
},
[TCM_INVALID_PARAMETER_LIST] = {
.key = ILLEGAL_REQUEST,
.asc = 0x26, /* INVALID FIELD IN PARAMETER LIST */
},
[TCM_TOO_MANY_TARGET_DESCS] = {
.key = ILLEGAL_REQUEST,
.asc = 0x26,
.ascq = 0x06, /* TOO MANY TARGET DESCRIPTORS */
},
[TCM_UNSUPPORTED_TARGET_DESC_TYPE_CODE] = {
.key = ILLEGAL_REQUEST,
.asc = 0x26,
.ascq = 0x07, /* UNSUPPORTED TARGET DESCRIPTOR TYPE CODE */
},
[TCM_TOO_MANY_SEGMENT_DESCS] = {
.key = ILLEGAL_REQUEST,
.asc = 0x26,
.ascq = 0x08, /* TOO MANY SEGMENT DESCRIPTORS */
},
[TCM_UNSUPPORTED_SEGMENT_DESC_TYPE_CODE] = {
.key = ILLEGAL_REQUEST,
.asc = 0x26,
.ascq = 0x09, /* UNSUPPORTED SEGMENT DESCRIPTOR TYPE CODE */
},
[TCM_PARAMETER_LIST_LENGTH_ERROR] = {
.key = ILLEGAL_REQUEST,
.asc = 0x1a, /* PARAMETER LIST LENGTH ERROR */
},
[TCM_UNEXPECTED_UNSOLICITED_DATA] = {
.key = ILLEGAL_REQUEST,
.asc = 0x0c, /* WRITE ERROR */
.ascq = 0x0c, /* UNEXPECTED_UNSOLICITED_DATA */
},
[TCM_SERVICE_CRC_ERROR] = {
.key = ABORTED_COMMAND,
.asc = 0x47, /* PROTOCOL SERVICE CRC ERROR */
.ascq = 0x05, /* N/A */
},
[TCM_SNACK_REJECTED] = {
.key = ABORTED_COMMAND,
.asc = 0x11, /* READ ERROR */
.ascq = 0x13, /* FAILED RETRANSMISSION REQUEST */
},
[TCM_WRITE_PROTECTED] = {
.key = DATA_PROTECT,
.asc = 0x27, /* WRITE PROTECTED */
},
[TCM_ADDRESS_OUT_OF_RANGE] = {
.key = ILLEGAL_REQUEST,
.asc = 0x21, /* LOGICAL BLOCK ADDRESS OUT OF RANGE */
},
[TCM_CHECK_CONDITION_UNIT_ATTENTION] = {
.key = UNIT_ATTENTION,
},
[TCM_CHECK_CONDITION_NOT_READY] = {
.key = NOT_READY,
},
[TCM_MISCOMPARE_VERIFY] = {
.key = MISCOMPARE,
.asc = 0x1d, /* MISCOMPARE DURING VERIFY OPERATION */
.ascq = 0x00,
},
[TCM_LOGICAL_BLOCK_GUARD_CHECK_FAILED] = {
.key = ABORTED_COMMAND,
.asc = 0x10,
.ascq = 0x01, /* LOGICAL BLOCK GUARD CHECK FAILED */
.add_sector_info = true,
},
[TCM_LOGICAL_BLOCK_APP_TAG_CHECK_FAILED] = {
.key = ABORTED_COMMAND,
.asc = 0x10,
.ascq = 0x02, /* LOGICAL BLOCK APPLICATION TAG CHECK FAILED */
.add_sector_info = true,
},
[TCM_LOGICAL_BLOCK_REF_TAG_CHECK_FAILED] = {
.key = ABORTED_COMMAND,
.asc = 0x10,
.ascq = 0x03, /* LOGICAL BLOCK REFERENCE TAG CHECK FAILED */
.add_sector_info = true,
},
[TCM_COPY_TARGET_DEVICE_NOT_REACHABLE] = {
.key = COPY_ABORTED,
.asc = 0x0d,
.ascq = 0x02, /* COPY TARGET DEVICE NOT REACHABLE */
},
[TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE] = {
/*
* Returning ILLEGAL REQUEST would cause immediate IO errors on
* Solaris initiators. Returning NOT READY instead means the
* operations will be retried a finite number of times and we
* can survive intermittent errors.
*/
.key = NOT_READY,
.asc = 0x08, /* LOGICAL UNIT COMMUNICATION FAILURE */
},
[TCM_INSUFFICIENT_REGISTRATION_RESOURCES] = {
/*
* From spc4r22 section5.7.7,5.7.8
* If a PERSISTENT RESERVE OUT command with a REGISTER service action
* or a REGISTER AND IGNORE EXISTING KEY service action or
* REGISTER AND MOVE service actionis attempted,
* but there are insufficient device server resources to complete the
* operation, then the command shall be terminated with CHECK CONDITION
* status, with the sense key set to ILLEGAL REQUEST,and the additonal
* sense code set to INSUFFICIENT REGISTRATION RESOURCES.
*/
.key = ILLEGAL_REQUEST,
.asc = 0x55,
.ascq = 0x04, /* INSUFFICIENT REGISTRATION RESOURCES */
},
};
static int translate_sense_reason(struct se_cmd *cmd, sense_reason_t reason)
{
const struct sense_info *si;
u8 *buffer = cmd->sense_buffer;
int r = (__force int)reason;
u8 asc, ascq;
bool desc_format = target_sense_desc_format(cmd->se_dev);
if (r < ARRAY_SIZE(sense_info_table) && sense_info_table[r].key)
si = &sense_info_table[r];
else
si = &sense_info_table[(__force int)
TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE];
if (reason == TCM_CHECK_CONDITION_UNIT_ATTENTION) {
core_scsi3_ua_for_check_condition(cmd, &asc, &ascq);
WARN_ON_ONCE(asc == 0);
} else if (si->asc == 0) {
WARN_ON_ONCE(cmd->scsi_asc == 0);
asc = cmd->scsi_asc;
ascq = cmd->scsi_ascq;
} else {
asc = si->asc;
ascq = si->ascq;
}
scsi_build_sense_buffer(desc_format, buffer, si->key, asc, ascq);
if (si->add_sector_info)
return scsi_set_sense_information(buffer,
cmd->scsi_sense_length,
cmd->bad_sector);
return 0;
}
int
transport_send_check_condition_and_sense(struct se_cmd *cmd,
sense_reason_t reason, int from_transport)
{
unsigned long flags;
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->se_cmd_flags & SCF_SENT_CHECK_CONDITION) {
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
return 0;
}
cmd->se_cmd_flags |= SCF_SENT_CHECK_CONDITION;
target: Updates from AGrover and HCH (round 3) This patch contains a squashed version of third round series cleanups, improvements ,and simplfications from Andy and Christoph ahead of the heavy lifting between round 3 -> 4 for the target core SGL conversion. This include cleanups to the main target I/O path and other miscellaneous updates. target: Replace custom sg<->buf functions with lib funcs target: Simplify sector limiting code target: get_cdb should never return NULL target: Simplify transport_memcpy_se_mem_read_contig target: Use assignment rather than increment for t_task_cdbs target: Don't pass dma_size to generic_get_mem target: Pass sg with type scatterlist in transport_map_sg_to_mem target: Move task_sg_num next to task_sg in struct se_task target: inline struct se_transport_task into struct se_cmd target: Change name & semantics of transport_get_sectors() target: Remove unused members of se_cmd target: Rename se_cmd.t_task_cdbs to t_task_list_num target: Fix some spelling target: Remove unused var from transport_generic_do_tmr target: map_sg_to_mem: return sg_count in return value target/pscsi: Use min_t for sector limits target/pscsi: Unused param for pscsi_get_bio() target: Rename get_cdb_count to allocate_tasks target: Make transport_generic_new_cmd() available for iscsi-target target: Remove fabric callback to allocate iovecs target: Fix transport_generic_new_cmd WRITE comment (hch: Use __GFP_ZERO usage for alloc_pages() usage) Signed-off-by: Andy Grover <agrover@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-05-03 04:12:10 +04:00
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
if (!from_transport) {
int rc;
cmd->se_cmd_flags |= SCF_EMULATED_TASK_SENSE;
cmd->scsi_status = SAM_STAT_CHECK_CONDITION;
cmd->scsi_sense_length = TRANSPORT_SENSE_BUFFER;
rc = translate_sense_reason(cmd, reason);
if (rc)
return rc;
}
trace_target_cmd_complete(cmd);
return cmd->se_tfo->queue_status(cmd);
}
EXPORT_SYMBOL(transport_send_check_condition_and_sense);
static int __transport_check_aborted_status(struct se_cmd *cmd, int send_status)
__releases(&cmd->t_state_lock)
__acquires(&cmd->t_state_lock)
{
int ret;
assert_spin_locked(&cmd->t_state_lock);
WARN_ON_ONCE(!irqs_disabled());
if (!(cmd->transport_state & CMD_T_ABORTED))
return 0;
/*
* If cmd has been aborted but either no status is to be sent or it has
* already been sent, just return
*/
if (!send_status || !(cmd->se_cmd_flags & SCF_SEND_DELAYED_TAS)) {
if (send_status)
cmd->se_cmd_flags |= SCF_SEND_DELAYED_TAS;
return 1;
}
pr_debug("Sending delayed SAM_STAT_TASK_ABORTED status for CDB:"
" 0x%02x ITT: 0x%08llx\n", cmd->t_task_cdb[0], cmd->tag);
cmd->se_cmd_flags &= ~SCF_SEND_DELAYED_TAS;
cmd->scsi_status = SAM_STAT_TASK_ABORTED;
trace_target_cmd_complete(cmd);
spin_unlock_irq(&cmd->t_state_lock);
ret = cmd->se_tfo->queue_status(cmd);
if (ret)
transport_handle_queue_full(cmd, cmd->se_dev, ret, false);
spin_lock_irq(&cmd->t_state_lock);
return 1;
}
int transport_check_aborted_status(struct se_cmd *cmd, int send_status)
{
int ret;
spin_lock_irq(&cmd->t_state_lock);
ret = __transport_check_aborted_status(cmd, send_status);
spin_unlock_irq(&cmd->t_state_lock);
return ret;
}
EXPORT_SYMBOL(transport_check_aborted_status);
void transport_send_task_abort(struct se_cmd *cmd)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->se_cmd_flags & (SCF_SENT_CHECK_CONDITION)) {
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
return;
}
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
/*
* If there are still expected incoming fabric WRITEs, we wait
* until until they have completed before sending a TASK_ABORTED
* response. This response with TASK_ABORTED status will be
* queued back to fabric module by transport_check_aborted_status().
*/
if (cmd->data_direction == DMA_TO_DEVICE) {
2011-07-19 12:55:10 +04:00
if (cmd->se_tfo->write_pending_status(cmd) != 0) {
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->se_cmd_flags & SCF_SEND_DELAYED_TAS) {
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
goto send_abort;
}
cmd->se_cmd_flags |= SCF_SEND_DELAYED_TAS;
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
return;
}
}
send_abort:
cmd->scsi_status = SAM_STAT_TASK_ABORTED;
transport_lun_remove_cmd(cmd);
pr_debug("Setting SAM_STAT_TASK_ABORTED status for CDB: 0x%02x, ITT: 0x%08llx\n",
cmd->t_task_cdb[0], cmd->tag);
trace_target_cmd_complete(cmd);
ret = cmd->se_tfo->queue_status(cmd);
if (ret)
transport_handle_queue_full(cmd, cmd->se_dev, ret, false);
}
static void target_tmr_work(struct work_struct *work)
{
struct se_cmd *cmd = container_of(work, struct se_cmd, work);
struct se_device *dev = cmd->se_dev;
struct se_tmr_req *tmr = cmd->se_tmr_req;
unsigned long flags;
int ret;
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->transport_state & CMD_T_ABORTED) {
tmr->response = TMR_FUNCTION_REJECTED;
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
goto check_stop;
}
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
switch (tmr->function) {
case TMR_ABORT_TASK:
core_tmr_abort_task(dev, tmr, cmd->se_sess);
break;
case TMR_ABORT_TASK_SET:
case TMR_CLEAR_ACA:
case TMR_CLEAR_TASK_SET:
tmr->response = TMR_TASK_MGMT_FUNCTION_NOT_SUPPORTED;
break;
case TMR_LUN_RESET:
ret = core_tmr_lun_reset(dev, tmr, NULL, NULL);
tmr->response = (!ret) ? TMR_FUNCTION_COMPLETE :
TMR_FUNCTION_REJECTED;
if (tmr->response == TMR_FUNCTION_COMPLETE) {
target_ua_allocate_lun(cmd->se_sess->se_node_acl,
cmd->orig_fe_lun, 0x29,
ASCQ_29H_BUS_DEVICE_RESET_FUNCTION_OCCURRED);
}
break;
case TMR_TARGET_WARM_RESET:
tmr->response = TMR_FUNCTION_REJECTED;
break;
case TMR_TARGET_COLD_RESET:
tmr->response = TMR_FUNCTION_REJECTED;
break;
default:
pr_err("Unknown TMR function: 0x%02x.\n",
tmr->function);
tmr->response = TMR_FUNCTION_REJECTED;
break;
}
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->transport_state & CMD_T_ABORTED) {
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
goto check_stop;
}
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
2011-07-19 12:55:10 +04:00
cmd->se_tfo->queue_tm_rsp(cmd);
check_stop:
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
}
int transport_generic_handle_tmr(
struct se_cmd *cmd)
{
unsigned long flags;
bool aborted = false;
spin_lock_irqsave(&cmd->t_state_lock, flags);
if (cmd->transport_state & CMD_T_ABORTED) {
aborted = true;
} else {
cmd->t_state = TRANSPORT_ISTATE_PROCESSING;
cmd->transport_state |= CMD_T_ACTIVE;
}
spin_unlock_irqrestore(&cmd->t_state_lock, flags);
if (aborted) {
pr_warn_ratelimited("handle_tmr caught CMD_T_ABORTED TMR %d"
"ref_tag: %llu tag: %llu\n", cmd->se_tmr_req->function,
cmd->se_tmr_req->ref_task_tag, cmd->tag);
transport_lun_remove_cmd(cmd);
transport_cmd_check_stop_to_fabric(cmd);
return 0;
}
INIT_WORK(&cmd->work, target_tmr_work);
queue_work(cmd->se_dev->tmr_wq, &cmd->work);
return 0;
}
EXPORT_SYMBOL(transport_generic_handle_tmr);
bool
target_check_wce(struct se_device *dev)
{
bool wce = false;
if (dev->transport->get_write_cache)
wce = dev->transport->get_write_cache(dev);
else if (dev->dev_attrib.emulate_write_cache > 0)
wce = true;
return wce;
}
bool
target_check_fua(struct se_device *dev)
{
return target_check_wce(dev) && dev->dev_attrib.emulate_fua_write > 0;
}