2017-11-24 17:00:32 +03:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
2009-03-26 17:24:01 +03:00
|
|
|
* Machine check handler
|
2005-04-17 02:20:36 +04:00
|
|
|
*
|
2012-07-20 13:15:04 +04:00
|
|
|
* Copyright IBM Corp. 2000, 2009
|
2009-03-26 17:24:01 +03:00
|
|
|
* Author(s): Ingo Adlung <adlung@de.ibm.com>,
|
|
|
|
* Martin Schwidefsky <schwidefsky@de.ibm.com>,
|
|
|
|
* Cornelia Huck <cornelia.huck@de.ibm.com>,
|
|
|
|
* Heiko Carstens <heiko.carstens@de.ibm.com>,
|
2005-04-17 02:20:36 +04:00
|
|
|
*/
|
|
|
|
|
2011-01-05 14:47:28 +03:00
|
|
|
#include <linux/kernel_stat.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/errno.h>
|
2009-04-14 17:36:18 +04:00
|
|
|
#include <linux/hardirq.h>
|
2017-10-12 14:24:47 +03:00
|
|
|
#include <linux/log2.h>
|
2017-09-15 17:24:31 +03:00
|
|
|
#include <linux/kprobes.h>
|
2017-10-12 14:24:47 +03:00
|
|
|
#include <linux/slab.h>
|
2006-05-01 23:16:14 +04:00
|
|
|
#include <linux/time.h>
|
2017-02-08 20:51:30 +03:00
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/sched/signal.h>
|
|
|
|
|
2017-02-09 23:20:23 +03:00
|
|
|
#include <linux/export.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <asm/lowcore.h>
|
2009-03-26 17:24:01 +03:00
|
|
|
#include <asm/smp.h>
|
2016-05-31 16:06:51 +03:00
|
|
|
#include <asm/stp.h>
|
2009-06-12 12:26:21 +04:00
|
|
|
#include <asm/cputime.h>
|
2009-03-26 17:24:01 +03:00
|
|
|
#include <asm/nmi.h>
|
|
|
|
#include <asm/crw.h>
|
2014-10-06 19:53:53 +04:00
|
|
|
#include <asm/switch_to.h>
|
s390/nmi: fix vector register corruption
If a machine check happens, the machine has the vector facility installed
and the extended save area exists, the cpu will save vector register
contents into the extended save area. This is regardless of control
register 0 contents, which enables and disables the vector facility during
runtime.
On each machine check we should validate the vector registers. The current
code however tries to validate the registers only if the running task is
using vector registers in user space.
However even the current code is broken and causes vector register
corruption on machine checks, if user space uses them:
the prefix area contains a pointer (absolute address) to the machine check
extended save area. In order to save some space the save area was put into
an unused area of the second prefix page.
When validating vector register contents the code uses the absolute address
of the extended save area, which is wrong. Due to prefixing the vector
instructions will then access contents using absolute addresses instead
of real addresses, where the machine stored the contents.
If the above would work there is still the problem that register validition
would only happen if user space uses vector registers. If kernel space uses
them also, this may also lead to vector register content corruption:
if the kernel makes use of vector instructions, but the current running
user space context does not, the machine check handler will validate
floating point registers instead of vector registers.
Given the fact that writing to a floating point register may change the
upper halve of the corresponding vector register, we also experience vector
register corruption in this case.
Fix all of these issues, and always validate vector registers on each
machine check, if the machine has the vector facility installed and the
extended save area is defined.
Cc: <stable@vger.kernel.org> # 4.1+
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2015-07-07 09:40:49 +03:00
|
|
|
#include <asm/ctl_reg.h>
|
2017-06-07 12:30:42 +03:00
|
|
|
#include <asm/asm-offsets.h>
|
KVM: s390: Backup the guest's machine check info
When a machine check happens in the guest, related mcck info (mcic,
external damage code, ...) is stored in the vcpu's lowcore on the host.
Then the machine check handler's low-level part is executed, followed
by the high-level part.
If the high-level part's execution is interrupted by a new machine check
happening on the same vcpu on the host, the mcck info in the lowcore is
overwritten with the new machine check's data.
If the high-level part's execution is scheduled to a different cpu,
the mcck info in the lowcore is uncertain.
Therefore, for both cases, the further reinjection to the guest will use
the wrong data.
Let's backup the mcck info in the lowcore to the sie page
for further reinjection, so that the right data will be used.
Add new member into struct sie_page to store related machine check's
info of mcic, failing storage address and external damage code.
Signed-off-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
2017-06-07 12:41:19 +03:00
|
|
|
#include <linux/kvm_host.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
struct mcck_struct {
|
2015-10-12 14:04:12 +03:00
|
|
|
unsigned int kill_task : 1;
|
|
|
|
unsigned int channel_report : 1;
|
|
|
|
unsigned int warning : 1;
|
2015-10-09 14:48:03 +03:00
|
|
|
unsigned int stp_queue : 1;
|
2015-10-12 14:00:39 +03:00
|
|
|
unsigned long mcck_code;
|
2005-06-26 01:55:30 +04:00
|
|
|
};
|
|
|
|
|
|
|
|
static DEFINE_PER_CPU(struct mcck_struct, cpu_mcck);
|
2017-10-12 14:24:47 +03:00
|
|
|
static struct kmem_cache *mcesa_cache;
|
|
|
|
static unsigned long mcesa_origin_lc;
|
|
|
|
|
|
|
|
static inline int nmi_needs_mcesa(void)
|
|
|
|
{
|
|
|
|
return MACHINE_HAS_VX || MACHINE_HAS_GS;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned long nmi_get_mcesa_size(void)
|
|
|
|
{
|
|
|
|
if (MACHINE_HAS_GS)
|
|
|
|
return MCESA_MAX_SIZE;
|
|
|
|
return MCESA_MIN_SIZE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The initial machine check extended save area for the boot CPU.
|
|
|
|
* It will be replaced by nmi_init() with an allocated structure.
|
|
|
|
* The structure is required for machine check happening early in
|
|
|
|
* the boot process.
|
|
|
|
*/
|
|
|
|
static struct mcesa boot_mcesa __initdata __aligned(MCESA_MAX_SIZE);
|
|
|
|
|
|
|
|
void __init nmi_alloc_boot_cpu(struct lowcore *lc)
|
|
|
|
{
|
|
|
|
if (!nmi_needs_mcesa())
|
|
|
|
return;
|
|
|
|
lc->mcesad = (unsigned long) &boot_mcesa;
|
|
|
|
if (MACHINE_HAS_GS)
|
|
|
|
lc->mcesad |= ilog2(MCESA_MAX_SIZE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int __init nmi_init(void)
|
|
|
|
{
|
|
|
|
unsigned long origin, cr0, size;
|
|
|
|
|
|
|
|
if (!nmi_needs_mcesa())
|
|
|
|
return 0;
|
|
|
|
size = nmi_get_mcesa_size();
|
|
|
|
if (size > MCESA_MIN_SIZE)
|
|
|
|
mcesa_origin_lc = ilog2(size);
|
|
|
|
/* create slab cache for the machine-check-extended-save-areas */
|
|
|
|
mcesa_cache = kmem_cache_create("nmi_save_areas", size, size, 0, NULL);
|
|
|
|
if (!mcesa_cache)
|
|
|
|
panic("Couldn't create nmi save area cache");
|
|
|
|
origin = (unsigned long) kmem_cache_alloc(mcesa_cache, GFP_KERNEL);
|
|
|
|
if (!origin)
|
|
|
|
panic("Couldn't allocate nmi save area");
|
|
|
|
/* The pointer is stored with mcesa_bits ORed in */
|
|
|
|
kmemleak_not_leak((void *) origin);
|
|
|
|
__ctl_store(cr0, 0, 0);
|
|
|
|
__ctl_clear_bit(0, 28); /* disable lowcore protection */
|
|
|
|
/* Replace boot_mcesa on the boot CPU */
|
|
|
|
S390_lowcore.mcesad = origin | mcesa_origin_lc;
|
|
|
|
__ctl_load(cr0, 0, 0);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
early_initcall(nmi_init);
|
|
|
|
|
|
|
|
int nmi_alloc_per_cpu(struct lowcore *lc)
|
|
|
|
{
|
|
|
|
unsigned long origin;
|
|
|
|
|
|
|
|
if (!nmi_needs_mcesa())
|
|
|
|
return 0;
|
|
|
|
origin = (unsigned long) kmem_cache_alloc(mcesa_cache, GFP_KERNEL);
|
|
|
|
if (!origin)
|
|
|
|
return -ENOMEM;
|
|
|
|
/* The pointer is stored with mcesa_bits ORed in */
|
|
|
|
kmemleak_not_leak((void *) origin);
|
|
|
|
lc->mcesad = origin | mcesa_origin_lc;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void nmi_free_per_cpu(struct lowcore *lc)
|
|
|
|
{
|
|
|
|
if (!nmi_needs_mcesa())
|
|
|
|
return;
|
|
|
|
kmem_cache_free(mcesa_cache, (void *)(lc->mcesad & MCESA_ORIGIN_MASK));
|
|
|
|
}
|
2005-06-26 01:55:30 +04:00
|
|
|
|
2017-09-15 17:24:31 +03:00
|
|
|
static notrace void s390_handle_damage(void)
|
2009-03-26 17:24:01 +03:00
|
|
|
{
|
2017-09-15 17:24:31 +03:00
|
|
|
smp_emergency_stop();
|
2009-03-26 17:24:01 +03:00
|
|
|
disabled_wait((unsigned long) __builtin_return_address(0));
|
|
|
|
while (1);
|
|
|
|
}
|
2017-09-15 17:24:31 +03:00
|
|
|
NOKPROBE_SYMBOL(s390_handle_damage);
|
2009-03-26 17:24:01 +03:00
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
2005-06-26 01:55:30 +04:00
|
|
|
* Main machine check handler function. Will be called with interrupts enabled
|
|
|
|
* or disabled and machine checks enabled or disabled.
|
2005-04-17 02:20:36 +04:00
|
|
|
*/
|
2009-03-26 17:24:01 +03:00
|
|
|
void s390_handle_mcck(void)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2005-06-26 01:55:30 +04:00
|
|
|
unsigned long flags;
|
|
|
|
struct mcck_struct mcck;
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* Disable machine checks and get the current state of accumulated
|
|
|
|
* machine checks. Afterwards delete the old state and enable machine
|
|
|
|
* checks again.
|
|
|
|
*/
|
|
|
|
local_irq_save(flags);
|
|
|
|
local_mcck_disable();
|
2014-11-28 17:40:57 +03:00
|
|
|
mcck = *this_cpu_ptr(&cpu_mcck);
|
|
|
|
memset(this_cpu_ptr(&cpu_mcck), 0, sizeof(mcck));
|
2014-04-15 14:55:07 +04:00
|
|
|
clear_cpu_flag(CIF_MCCK_PENDING);
|
2005-06-26 01:55:30 +04:00
|
|
|
local_mcck_enable();
|
|
|
|
local_irq_restore(flags);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
if (mcck.channel_report)
|
2009-03-26 17:24:01 +03:00
|
|
|
crw_handle_channel_report();
|
2009-03-26 17:24:02 +03:00
|
|
|
/*
|
|
|
|
* A warning may remain for a prolonged period on the bare iron.
|
|
|
|
* (actually until the machine is powered off, or the problem is gone)
|
|
|
|
* So we just stop listening for the WARNING MCH and avoid continuously
|
|
|
|
* being interrupted. One caveat is however, that we must do this per
|
|
|
|
* processor and cannot use the smp version of ctl_clear_bit().
|
|
|
|
* On VM we only get one interrupt per virtally presented machinecheck.
|
|
|
|
* Though one suffices, we may get one interrupt per (virtual) cpu.
|
|
|
|
*/
|
2005-06-26 01:55:30 +04:00
|
|
|
if (mcck.warning) { /* WARNING pending ? */
|
2005-04-17 02:20:36 +04:00
|
|
|
static int mchchk_wng_posted = 0;
|
2009-03-26 17:24:02 +03:00
|
|
|
|
|
|
|
/* Use single cpu clear, as we cannot handle smp here. */
|
2005-04-17 02:20:36 +04:00
|
|
|
__ctl_clear_bit(14, 24); /* Disable WARNING MCH */
|
|
|
|
if (xchg(&mchchk_wng_posted, 1) == 0)
|
2006-10-02 13:19:00 +04:00
|
|
|
kill_cad_pid(SIGPWR, 1);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2015-10-09 14:48:03 +03:00
|
|
|
if (mcck.stp_queue)
|
|
|
|
stp_queue_work();
|
2005-06-26 01:55:30 +04:00
|
|
|
if (mcck.kill_task) {
|
|
|
|
local_irq_enable();
|
|
|
|
printk(KERN_EMERG "mcck: Terminating task because of machine "
|
2015-10-12 14:00:39 +03:00
|
|
|
"malfunction (code 0x%016lx).\n", mcck.mcck_code);
|
2005-06-26 01:55:30 +04:00
|
|
|
printk(KERN_EMERG "mcck: task: %s, pid: %d.\n",
|
|
|
|
current->comm, current->pid);
|
|
|
|
do_exit(SIGSEGV);
|
|
|
|
}
|
|
|
|
}
|
2008-05-21 15:37:34 +04:00
|
|
|
EXPORT_SYMBOL_GPL(s390_handle_mcck);
|
2005-06-26 01:55:30 +04:00
|
|
|
|
|
|
|
/*
|
2017-10-12 14:24:48 +03:00
|
|
|
* returns 0 if all required registers are available
|
2005-06-26 01:55:30 +04:00
|
|
|
* returns 1 otherwise
|
|
|
|
*/
|
2017-10-12 14:24:48 +03:00
|
|
|
static int notrace s390_check_registers(union mci mci, int umode)
|
2005-06-26 01:55:30 +04:00
|
|
|
{
|
2017-10-12 14:24:45 +03:00
|
|
|
union ctlreg2 cr2;
|
2005-06-26 01:55:30 +04:00
|
|
|
int kill_task;
|
|
|
|
|
|
|
|
kill_task = 0;
|
2009-03-26 17:24:01 +03:00
|
|
|
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.gr) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* General purpose registers couldn't be restored and have
|
2016-08-22 15:40:06 +03:00
|
|
|
* unknown contents. Stop system or terminate process.
|
2005-06-26 01:55:30 +04:00
|
|
|
*/
|
2016-08-22 15:40:06 +03:00
|
|
|
if (!umode)
|
|
|
|
s390_handle_damage();
|
2005-06-26 01:55:30 +04:00
|
|
|
kill_task = 1;
|
2009-03-26 17:24:01 +03:00
|
|
|
}
|
2017-10-12 14:24:48 +03:00
|
|
|
/* Check control registers */
|
2017-02-21 12:51:55 +03:00
|
|
|
if (!mci.cr) {
|
|
|
|
/*
|
|
|
|
* Control registers have unknown contents.
|
|
|
|
* Can't recover and therefore stopping machine.
|
|
|
|
*/
|
|
|
|
s390_handle_damage();
|
|
|
|
}
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.fp) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
2016-08-22 15:40:06 +03:00
|
|
|
* Floating point registers can't be restored. If the
|
|
|
|
* kernel currently uses floating point registers the
|
|
|
|
* system is stopped. If the process has its floating
|
|
|
|
* pointer registers loaded it is terminated.
|
2005-06-26 01:55:30 +04:00
|
|
|
*/
|
2016-08-22 15:40:06 +03:00
|
|
|
if (S390_lowcore.fpu_flags & KERNEL_VXR_V0V7)
|
|
|
|
s390_handle_damage();
|
|
|
|
if (!test_cpu_flag(CIF_FPU))
|
|
|
|
kill_task = 1;
|
2009-03-26 17:24:01 +03:00
|
|
|
}
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.fc) {
|
2015-02-12 15:08:27 +03:00
|
|
|
/*
|
|
|
|
* Floating point control register can't be restored.
|
2016-08-22 15:40:06 +03:00
|
|
|
* If the kernel currently uses the floating pointer
|
|
|
|
* registers and needs the FPC register the system is
|
|
|
|
* stopped. If the process has its floating pointer
|
2017-10-12 14:24:48 +03:00
|
|
|
* registers loaded it is terminated.
|
2015-02-12 15:08:27 +03:00
|
|
|
*/
|
2016-08-22 15:40:06 +03:00
|
|
|
if (S390_lowcore.fpu_flags & KERNEL_FPC)
|
|
|
|
s390_handle_damage();
|
|
|
|
if (!test_cpu_flag(CIF_FPU))
|
|
|
|
kill_task = 1;
|
2016-12-13 14:42:07 +03:00
|
|
|
}
|
2015-02-12 15:08:27 +03:00
|
|
|
|
2017-10-12 14:24:48 +03:00
|
|
|
if (MACHINE_HAS_VX) {
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.vr) {
|
2014-10-06 19:53:53 +04:00
|
|
|
/*
|
2016-08-22 15:40:06 +03:00
|
|
|
* Vector registers can't be restored. If the kernel
|
|
|
|
* currently uses vector registers the system is
|
|
|
|
* stopped. If the process has its vector registers
|
2017-10-12 14:24:48 +03:00
|
|
|
* loaded it is terminated.
|
2014-10-06 19:53:53 +04:00
|
|
|
*/
|
2016-08-22 15:40:06 +03:00
|
|
|
if (S390_lowcore.fpu_flags & KERNEL_VXR)
|
|
|
|
s390_handle_damage();
|
|
|
|
if (!test_cpu_flag(CIF_FPU))
|
|
|
|
kill_task = 1;
|
2014-10-06 19:53:53 +04:00
|
|
|
}
|
|
|
|
}
|
2017-10-12 14:24:48 +03:00
|
|
|
/* Check if access registers are valid */
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.ar) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* Access registers have unknown contents.
|
|
|
|
* Terminating task.
|
|
|
|
*/
|
|
|
|
kill_task = 1;
|
2009-03-26 17:24:01 +03:00
|
|
|
}
|
2017-10-12 14:24:48 +03:00
|
|
|
/* Check guarded storage registers */
|
2017-10-12 14:24:45 +03:00
|
|
|
cr2.val = S390_lowcore.cregs_save_area[2];
|
|
|
|
if (cr2.gse) {
|
2017-10-12 14:24:48 +03:00
|
|
|
if (!mci.gs) {
|
2016-01-26 16:10:34 +03:00
|
|
|
/*
|
|
|
|
* Guarded storage register can't be restored and
|
|
|
|
* the current processes uses guarded storage.
|
|
|
|
* It has to be terminated.
|
|
|
|
*/
|
|
|
|
kill_task = 1;
|
2017-10-12 14:24:48 +03:00
|
|
|
}
|
2016-01-26 16:10:34 +03:00
|
|
|
}
|
2005-06-26 01:55:30 +04:00
|
|
|
/* Check if old PSW is valid */
|
2017-10-12 14:24:48 +03:00
|
|
|
if (!mci.wp) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* Can't tell if we come from user or kernel mode
|
|
|
|
* -> stopping machine.
|
|
|
|
*/
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2017-10-12 14:24:48 +03:00
|
|
|
}
|
|
|
|
/* Check for invalid kernel instruction address */
|
|
|
|
if (!mci.ia && !umode) {
|
|
|
|
/*
|
|
|
|
* The instruction address got lost while running
|
|
|
|
* in the kernel -> stopping machine.
|
|
|
|
*/
|
|
|
|
s390_handle_damage();
|
|
|
|
}
|
2005-06-26 01:55:30 +04:00
|
|
|
|
2015-10-12 14:00:39 +03:00
|
|
|
if (!mci.ms || !mci.pm || !mci.ia)
|
2005-06-26 01:55:30 +04:00
|
|
|
kill_task = 1;
|
|
|
|
|
|
|
|
return kill_task;
|
|
|
|
}
|
2017-10-12 14:24:48 +03:00
|
|
|
NOKPROBE_SYMBOL(s390_check_registers);
|
2005-06-26 01:55:30 +04:00
|
|
|
|
KVM: s390: Backup the guest's machine check info
When a machine check happens in the guest, related mcck info (mcic,
external damage code, ...) is stored in the vcpu's lowcore on the host.
Then the machine check handler's low-level part is executed, followed
by the high-level part.
If the high-level part's execution is interrupted by a new machine check
happening on the same vcpu on the host, the mcck info in the lowcore is
overwritten with the new machine check's data.
If the high-level part's execution is scheduled to a different cpu,
the mcck info in the lowcore is uncertain.
Therefore, for both cases, the further reinjection to the guest will use
the wrong data.
Let's backup the mcck info in the lowcore to the sie page
for further reinjection, so that the right data will be used.
Add new member into struct sie_page to store related machine check's
info of mcic, failing storage address and external damage code.
Signed-off-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
2017-06-07 12:41:19 +03:00
|
|
|
/*
|
|
|
|
* Backup the guest's machine check info to its description block
|
|
|
|
*/
|
|
|
|
static void notrace s390_backup_mcck_info(struct pt_regs *regs)
|
|
|
|
{
|
|
|
|
struct mcck_volatile_info *mcck_backup;
|
|
|
|
struct sie_page *sie_page;
|
|
|
|
|
|
|
|
/* r14 contains the sie block, which was set in sie64a */
|
|
|
|
struct kvm_s390_sie_block *sie_block =
|
|
|
|
(struct kvm_s390_sie_block *) regs->gprs[14];
|
|
|
|
|
|
|
|
if (sie_block == NULL)
|
|
|
|
/* Something's seriously wrong, stop system. */
|
|
|
|
s390_handle_damage();
|
|
|
|
|
|
|
|
sie_page = container_of(sie_block, struct sie_page, sie_block);
|
|
|
|
mcck_backup = &sie_page->mcck_info;
|
|
|
|
mcck_backup->mcic = S390_lowcore.mcck_interruption_code &
|
|
|
|
~(MCCK_CODE_CP | MCCK_CODE_EXT_DAMAGE);
|
|
|
|
mcck_backup->ext_damage_code = S390_lowcore.external_damage_code;
|
|
|
|
mcck_backup->failing_storage_address
|
|
|
|
= S390_lowcore.failing_storage_address;
|
|
|
|
}
|
2017-09-15 17:24:31 +03:00
|
|
|
NOKPROBE_SYMBOL(s390_backup_mcck_info);
|
KVM: s390: Backup the guest's machine check info
When a machine check happens in the guest, related mcck info (mcic,
external damage code, ...) is stored in the vcpu's lowcore on the host.
Then the machine check handler's low-level part is executed, followed
by the high-level part.
If the high-level part's execution is interrupted by a new machine check
happening on the same vcpu on the host, the mcck info in the lowcore is
overwritten with the new machine check's data.
If the high-level part's execution is scheduled to a different cpu,
the mcck info in the lowcore is uncertain.
Therefore, for both cases, the further reinjection to the guest will use
the wrong data.
Let's backup the mcck info in the lowcore to the sie page
for further reinjection, so that the right data will be used.
Add new member into struct sie_page to store related machine check's
info of mcic, failing storage address and external damage code.
Signed-off-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
2017-06-07 12:41:19 +03:00
|
|
|
|
2006-04-28 05:40:23 +04:00
|
|
|
#define MAX_IPD_COUNT 29
|
2006-05-01 23:16:14 +04:00
|
|
|
#define MAX_IPD_TIME (5 * 60 * USEC_PER_SEC) /* 5 minutes */
|
2006-04-28 05:40:23 +04:00
|
|
|
|
2009-03-26 17:24:01 +03:00
|
|
|
#define ED_STP_ISLAND 6 /* External damage STP island check */
|
|
|
|
#define ED_STP_SYNC 7 /* External damage STP sync check */
|
|
|
|
|
2017-06-07 12:30:42 +03:00
|
|
|
#define MCCK_CODE_NO_GUEST (MCCK_CODE_CP | MCCK_CODE_EXT_DAMAGE)
|
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* machine check handler.
|
|
|
|
*/
|
2009-03-26 17:23:59 +03:00
|
|
|
void notrace s390_do_machine_check(struct pt_regs *regs)
|
2005-06-26 01:55:30 +04:00
|
|
|
{
|
2009-03-26 17:24:01 +03:00
|
|
|
static int ipd_count;
|
2006-04-28 05:40:23 +04:00
|
|
|
static DEFINE_SPINLOCK(ipd_lock);
|
|
|
|
static unsigned long long last_ipd;
|
2009-03-26 17:24:01 +03:00
|
|
|
struct mcck_struct *mcck;
|
2006-04-28 05:40:23 +04:00
|
|
|
unsigned long long tmp;
|
2015-10-12 14:00:39 +03:00
|
|
|
union mci mci;
|
2017-06-07 12:30:42 +03:00
|
|
|
unsigned long mcck_dam_code;
|
2005-06-26 01:55:30 +04:00
|
|
|
|
2009-04-14 17:36:18 +04:00
|
|
|
nmi_enter();
|
2013-01-02 18:18:18 +04:00
|
|
|
inc_irq_stat(NMI_NMI);
|
2015-10-12 14:00:39 +03:00
|
|
|
mci.val = S390_lowcore.mcck_interruption_code;
|
s390: Replace __get_cpu_var uses
__get_cpu_var() is used for multiple purposes in the kernel source. One of
them is address calculation via the form &__get_cpu_var(x). This calculates
the address for the instance of the percpu variable of the current processor
based on an offset.
Other use cases are for storing and retrieving data from the current
processors percpu area. __get_cpu_var() can be used as an lvalue when
writing data or on the right side of an assignment.
__get_cpu_var() is defined as :
#define __get_cpu_var(var) (*this_cpu_ptr(&(var)))
__get_cpu_var() always only does an address determination. However, store
and retrieve operations could use a segment prefix (or global register on
other platforms) to avoid the address calculation.
this_cpu_write() and this_cpu_read() can directly take an offset into a
percpu area and use optimized assembly code to read and write per cpu
variables.
This patch converts __get_cpu_var into either an explicit address
calculation using this_cpu_ptr() or into a use of this_cpu operations that
use the offset. Thereby address calculations are avoided and less registers
are used when code is generated.
At the end of the patch set all uses of __get_cpu_var have been removed so
the macro is removed too.
The patch set includes passes over all arches as well. Once these operations
are used throughout then specialized macros can be defined in non -x86
arches as well in order to optimize per cpu access by f.e. using a global
register that may be set to the per cpu base.
Transformations done to __get_cpu_var()
1. Determine the address of the percpu instance of the current processor.
DEFINE_PER_CPU(int, y);
int *x = &__get_cpu_var(y);
Converts to
int *x = this_cpu_ptr(&y);
2. Same as #1 but this time an array structure is involved.
DEFINE_PER_CPU(int, y[20]);
int *x = __get_cpu_var(y);
Converts to
int *x = this_cpu_ptr(y);
3. Retrieve the content of the current processors instance of a per cpu
variable.
DEFINE_PER_CPU(int, y);
int x = __get_cpu_var(y)
Converts to
int x = __this_cpu_read(y);
4. Retrieve the content of a percpu struct
DEFINE_PER_CPU(struct mystruct, y);
struct mystruct x = __get_cpu_var(y);
Converts to
memcpy(&x, this_cpu_ptr(&y), sizeof(x));
5. Assignment to a per cpu variable
DEFINE_PER_CPU(int, y)
__get_cpu_var(y) = x;
Converts to
this_cpu_write(y, x);
6. Increment/Decrement etc of a per cpu variable
DEFINE_PER_CPU(int, y);
__get_cpu_var(y)++
Converts to
this_cpu_inc(y)
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
CC: linux390@de.ibm.com
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
2014-08-17 21:30:45 +04:00
|
|
|
mcck = this_cpu_ptr(&cpu_mcck);
|
2005-06-26 01:55:30 +04:00
|
|
|
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.sd) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/* System damage -> stopping machine */
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2009-03-26 17:24:01 +03:00
|
|
|
}
|
2017-06-07 12:30:42 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Reinject the instruction processing damages' machine checks
|
|
|
|
* including Delayed Access Exception into the guest
|
|
|
|
* instead of damaging the host if they happen in the guest.
|
|
|
|
*/
|
|
|
|
if (mci.pd && !test_cpu_flag(CIF_MCCK_GUEST)) {
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.b) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/* Processing backup -> verify if we can survive this */
|
|
|
|
u64 z_mcic, o_mcic, t_mcic;
|
|
|
|
z_mcic = (1ULL<<63 | 1ULL<<59 | 1ULL<<29);
|
|
|
|
o_mcic = (1ULL<<43 | 1ULL<<42 | 1ULL<<41 | 1ULL<<40 |
|
|
|
|
1ULL<<36 | 1ULL<<35 | 1ULL<<34 | 1ULL<<32 |
|
|
|
|
1ULL<<30 | 1ULL<<21 | 1ULL<<20 | 1ULL<<17 |
|
|
|
|
1ULL<<16);
|
2015-10-12 14:00:39 +03:00
|
|
|
t_mcic = mci.val;
|
2005-06-26 01:55:30 +04:00
|
|
|
|
|
|
|
if (((t_mcic & z_mcic) != 0) ||
|
|
|
|
((t_mcic & o_mcic) != o_mcic)) {
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
2006-04-28 05:40:23 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Nullifying exigent condition, therefore we might
|
|
|
|
* retry this instruction.
|
|
|
|
*/
|
|
|
|
spin_lock(&ipd_lock);
|
2013-01-30 12:49:40 +04:00
|
|
|
tmp = get_tod_clock();
|
2006-04-28 05:40:23 +04:00
|
|
|
if (((tmp - last_ipd) >> 12) < MAX_IPD_TIME)
|
|
|
|
ipd_count++;
|
|
|
|
else
|
|
|
|
ipd_count = 1;
|
|
|
|
last_ipd = tmp;
|
|
|
|
if (ipd_count == MAX_IPD_COUNT)
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2006-04-28 05:40:23 +04:00
|
|
|
spin_unlock(&ipd_lock);
|
2009-03-26 17:24:01 +03:00
|
|
|
} else {
|
2005-06-26 01:55:30 +04:00
|
|
|
/* Processing damage -> stopping machine */
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
|
|
|
}
|
2017-10-12 14:24:48 +03:00
|
|
|
if (s390_check_registers(mci, user_mode(regs))) {
|
2016-08-22 15:40:06 +03:00
|
|
|
/*
|
|
|
|
* Couldn't restore all register contents for the
|
|
|
|
* user space process -> mark task for termination.
|
|
|
|
*/
|
|
|
|
mcck->kill_task = 1;
|
|
|
|
mcck->mcck_code = mci.val;
|
|
|
|
set_cpu_flag(CIF_MCCK_PENDING);
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
KVM: s390: Backup the guest's machine check info
When a machine check happens in the guest, related mcck info (mcic,
external damage code, ...) is stored in the vcpu's lowcore on the host.
Then the machine check handler's low-level part is executed, followed
by the high-level part.
If the high-level part's execution is interrupted by a new machine check
happening on the same vcpu on the host, the mcck info in the lowcore is
overwritten with the new machine check's data.
If the high-level part's execution is scheduled to a different cpu,
the mcck info in the lowcore is uncertain.
Therefore, for both cases, the further reinjection to the guest will use
the wrong data.
Let's backup the mcck info in the lowcore to the sie page
for further reinjection, so that the right data will be used.
Add new member into struct sie_page to store related machine check's
info of mcic, failing storage address and external damage code.
Signed-off-by: QingFeng Hao <haoqf@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
2017-06-07 12:41:19 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Backup the machine check's info if it happens when the guest
|
|
|
|
* is running.
|
|
|
|
*/
|
|
|
|
if (test_cpu_flag(CIF_MCCK_GUEST))
|
|
|
|
s390_backup_mcck_info(regs);
|
|
|
|
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.cd) {
|
2007-02-05 23:18:19 +03:00
|
|
|
/* Timing facility damage */
|
2015-10-12 13:39:09 +03:00
|
|
|
s390_handle_damage();
|
2007-02-05 23:18:19 +03:00
|
|
|
}
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.ed && mci.ec) {
|
2007-02-05 23:18:19 +03:00
|
|
|
/* External damage */
|
2008-07-14 11:58:56 +04:00
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_STP_SYNC))
|
2015-10-09 14:48:03 +03:00
|
|
|
mcck->stp_queue |= stp_sync_check();
|
2008-07-14 11:58:56 +04:00
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_STP_ISLAND))
|
2015-10-09 14:48:03 +03:00
|
|
|
mcck->stp_queue |= stp_island_check();
|
2016-05-31 16:06:51 +03:00
|
|
|
if (mcck->stp_queue)
|
2015-10-09 14:48:03 +03:00
|
|
|
set_cpu_flag(CIF_MCCK_PENDING);
|
2007-02-05 23:18:19 +03:00
|
|
|
}
|
2017-06-07 12:30:42 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Reinject storage related machine checks into the guest if they
|
|
|
|
* happen when the guest is running.
|
|
|
|
*/
|
|
|
|
if (!test_cpu_flag(CIF_MCCK_GUEST)) {
|
|
|
|
if (mci.se)
|
|
|
|
/* Storage error uncorrected */
|
|
|
|
s390_handle_damage();
|
|
|
|
if (mci.ke)
|
|
|
|
/* Storage key-error uncorrected */
|
|
|
|
s390_handle_damage();
|
|
|
|
if (mci.ds && mci.fa)
|
|
|
|
/* Storage degradation */
|
|
|
|
s390_handle_damage();
|
|
|
|
}
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.cp) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/* Channel report word pending */
|
|
|
|
mcck->channel_report = 1;
|
2014-04-15 14:55:07 +04:00
|
|
|
set_cpu_flag(CIF_MCCK_PENDING);
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
2015-10-12 14:00:39 +03:00
|
|
|
if (mci.w) {
|
2005-06-26 01:55:30 +04:00
|
|
|
/* Warning pending */
|
|
|
|
mcck->warning = 1;
|
2014-04-15 14:55:07 +04:00
|
|
|
set_cpu_flag(CIF_MCCK_PENDING);
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
2017-06-07 12:30:42 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If there are only Channel Report Pending and External Damage
|
|
|
|
* machine checks, they will not be reinjected into the guest
|
|
|
|
* because they refer to host conditions only.
|
|
|
|
*/
|
|
|
|
mcck_dam_code = (mci.val & MCIC_SUBCLASS_MASK);
|
|
|
|
if (test_cpu_flag(CIF_MCCK_GUEST) &&
|
|
|
|
(mcck_dam_code & MCCK_CODE_NO_GUEST) != mcck_dam_code) {
|
|
|
|
/* Set exit reason code for host's later handling */
|
|
|
|
*((long *)(regs->gprs[15] + __SF_SIE_REASON)) = -EINTR;
|
|
|
|
}
|
|
|
|
clear_cpu_flag(CIF_MCCK_GUEST);
|
2009-04-14 17:36:18 +04:00
|
|
|
nmi_exit();
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2017-09-15 17:24:31 +03:00
|
|
|
NOKPROBE_SYMBOL(s390_do_machine_check);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2009-03-26 17:24:01 +03:00
|
|
|
static int __init machine_check_init(void)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2007-02-05 23:18:19 +03:00
|
|
|
ctl_set_bit(14, 25); /* enable external damage MCH */
|
2009-03-26 17:24:01 +03:00
|
|
|
ctl_set_bit(14, 27); /* enable system recovery MCH */
|
2005-04-17 02:20:36 +04:00
|
|
|
ctl_set_bit(14, 24); /* enable warning MCH */
|
|
|
|
return 0;
|
|
|
|
}
|
2015-08-17 09:09:17 +03:00
|
|
|
early_initcall(machine_check_init);
|