2005-04-17 02:20:36 +04:00
|
|
|
/*
|
|
|
|
* drivers/s390/s390mach.c
|
|
|
|
* S/390 machine check handler
|
|
|
|
*
|
2008-07-14 11:58:46 +04:00
|
|
|
* Copyright IBM Corp. 2000,2008
|
2005-04-17 02:20:36 +04:00
|
|
|
* Author(s): Ingo Adlung (adlung@de.ibm.com)
|
|
|
|
* Martin Schwidefsky (schwidefsky@de.ibm.com)
|
2008-07-14 11:58:46 +04:00
|
|
|
* Cornelia Huck <cornelia.huck@de.ibm.com>
|
2005-04-17 02:20:36 +04:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/errno.h>
|
|
|
|
#include <linux/workqueue.h>
|
2006-05-01 23:16:14 +04:00
|
|
|
#include <linux/time.h>
|
2007-02-05 23:16:47 +03:00
|
|
|
#include <linux/device.h>
|
2006-06-29 17:03:17 +04:00
|
|
|
#include <linux/kthread.h>
|
2007-02-05 23:18:19 +03:00
|
|
|
#include <asm/etr.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <asm/lowcore.h>
|
2007-02-05 23:16:47 +03:00
|
|
|
#include <asm/cio.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include "s390mach.h"
|
|
|
|
|
|
|
|
static struct semaphore m_sem;
|
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
static NORET_TYPE void
|
2005-04-17 02:20:36 +04:00
|
|
|
s390_handle_damage(char *msg)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_SMP
|
|
|
|
smp_send_stop();
|
|
|
|
#endif
|
|
|
|
disabled_wait((unsigned long) __builtin_return_address(0));
|
2005-06-26 01:55:30 +04:00
|
|
|
for(;;);
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
2008-07-14 11:58:46 +04:00
|
|
|
static crw_handler_t crw_handlers[NR_RSCS];
|
|
|
|
|
|
|
|
/**
|
|
|
|
* s390_register_crw_handler() - register a channel report word handler
|
|
|
|
* @rsc: reporting source code to handle
|
|
|
|
* @handler: handler to be registered
|
|
|
|
*
|
|
|
|
* Returns %0 on success and a negative error value otherwise.
|
|
|
|
*/
|
|
|
|
int s390_register_crw_handler(int rsc, crw_handler_t handler)
|
|
|
|
{
|
|
|
|
if ((rsc < 0) || (rsc >= NR_RSCS))
|
|
|
|
return -EINVAL;
|
|
|
|
if (!cmpxchg(&crw_handlers[rsc], NULL, handler))
|
|
|
|
return 0;
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* s390_unregister_crw_handler() - unregister a channel report word handler
|
|
|
|
* @rsc: reporting source code to handle
|
|
|
|
*/
|
|
|
|
void s390_unregister_crw_handler(int rsc)
|
|
|
|
{
|
|
|
|
if ((rsc < 0) || (rsc >= NR_RSCS))
|
|
|
|
return;
|
|
|
|
xchg(&crw_handlers[rsc], NULL);
|
|
|
|
synchronize_sched();
|
|
|
|
}
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
|
|
|
* Retrieve CRWs and call function to handle event.
|
|
|
|
*/
|
2008-07-14 11:58:46 +04:00
|
|
|
static int s390_collect_crw_info(void *param)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
2006-01-06 11:19:25 +03:00
|
|
|
struct crw crw[2];
|
2007-04-27 18:01:34 +04:00
|
|
|
int ccode;
|
2005-04-17 02:20:36 +04:00
|
|
|
struct semaphore *sem;
|
2006-01-06 11:19:25 +03:00
|
|
|
unsigned int chain;
|
2008-05-07 11:22:56 +04:00
|
|
|
int ignore;
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
sem = (struct semaphore *)param;
|
|
|
|
repeat:
|
2008-05-07 11:22:56 +04:00
|
|
|
ignore = down_interruptible(sem);
|
2006-01-06 11:19:25 +03:00
|
|
|
chain = 0;
|
2005-04-17 02:20:36 +04:00
|
|
|
while (1) {
|
2006-01-06 11:19:25 +03:00
|
|
|
if (unlikely(chain > 1)) {
|
|
|
|
struct crw tmp_crw;
|
|
|
|
|
|
|
|
printk(KERN_WARNING"%s: Code does not support more "
|
|
|
|
"than two chained crws; please report to "
|
2008-04-17 09:46:21 +04:00
|
|
|
"linux390@de.ibm.com!\n", __func__);
|
2006-01-06 11:19:25 +03:00
|
|
|
ccode = stcrw(&tmp_crw);
|
|
|
|
printk(KERN_WARNING"%s: crw reports slct=%d, oflw=%d, "
|
|
|
|
"chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
|
2008-04-17 09:46:21 +04:00
|
|
|
__func__, tmp_crw.slct, tmp_crw.oflw,
|
2006-01-06 11:19:25 +03:00
|
|
|
tmp_crw.chn, tmp_crw.rsc, tmp_crw.anc,
|
|
|
|
tmp_crw.erc, tmp_crw.rsid);
|
|
|
|
printk(KERN_WARNING"%s: This was crw number %x in the "
|
2008-04-17 09:46:21 +04:00
|
|
|
"chain\n", __func__, chain);
|
2006-01-06 11:19:25 +03:00
|
|
|
if (ccode != 0)
|
|
|
|
break;
|
|
|
|
chain = tmp_crw.chn ? chain + 1 : 0;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
ccode = stcrw(&crw[chain]);
|
2005-04-17 02:20:36 +04:00
|
|
|
if (ccode != 0)
|
|
|
|
break;
|
2006-09-20 17:59:47 +04:00
|
|
|
printk(KERN_DEBUG "crw_info : CRW reports slct=%d, oflw=%d, "
|
|
|
|
"chn=%d, rsc=%X, anc=%d, erc=%X, rsid=%X\n",
|
|
|
|
crw[chain].slct, crw[chain].oflw, crw[chain].chn,
|
|
|
|
crw[chain].rsc, crw[chain].anc, crw[chain].erc,
|
|
|
|
crw[chain].rsid);
|
2005-04-17 02:20:36 +04:00
|
|
|
/* Check for overflows. */
|
2006-01-06 11:19:25 +03:00
|
|
|
if (crw[chain].oflw) {
|
2008-07-14 11:58:46 +04:00
|
|
|
int i;
|
|
|
|
|
2008-04-17 09:46:21 +04:00
|
|
|
pr_debug("%s: crw overflow detected!\n", __func__);
|
2008-07-14 11:58:46 +04:00
|
|
|
for (i = 0; i < NR_RSCS; i++) {
|
|
|
|
if (crw_handlers[i])
|
|
|
|
crw_handlers[i](NULL, NULL, 1);
|
|
|
|
}
|
2006-01-06 11:19:25 +03:00
|
|
|
chain = 0;
|
2005-04-17 02:20:36 +04:00
|
|
|
continue;
|
|
|
|
}
|
2008-07-14 11:58:46 +04:00
|
|
|
if (crw[0].chn && !chain) {
|
|
|
|
chain++;
|
|
|
|
continue;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
2008-07-14 11:58:46 +04:00
|
|
|
if (crw_handlers[crw[chain].rsc])
|
|
|
|
crw_handlers[crw[chain].rsc](&crw[0],
|
|
|
|
chain ? &crw[1] : NULL,
|
|
|
|
0);
|
2006-01-06 11:19:25 +03:00
|
|
|
/* chain is always 0 or 1 here. */
|
|
|
|
chain = crw[chain].chn ? chain + 1 : 0;
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
goto repeat;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
struct mcck_struct {
|
|
|
|
int kill_task;
|
|
|
|
int channel_report;
|
|
|
|
int warning;
|
|
|
|
unsigned long long mcck_code;
|
|
|
|
};
|
|
|
|
|
|
|
|
static DEFINE_PER_CPU(struct mcck_struct, cpu_mcck);
|
|
|
|
|
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
|
|
|
*/
|
|
|
|
void
|
2005-06-26 01:55:30 +04:00
|
|
|
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();
|
|
|
|
mcck = __get_cpu_var(cpu_mcck);
|
|
|
|
memset(&__get_cpu_var(cpu_mcck), 0, sizeof(struct mcck_struct));
|
|
|
|
clear_thread_flag(TIF_MCCK_PENDING);
|
|
|
|
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)
|
2005-04-17 02:20:36 +04:00
|
|
|
up(&m_sem);
|
|
|
|
|
|
|
|
#ifdef CONFIG_MACHCHK_WARNING
|
|
|
|
/*
|
|
|
|
* The warning may remain for a prolonged period on the bare iron.
|
|
|
|
* (actually till the machine is powered off, or until the problem is gone)
|
|
|
|
* So we just stop listening for the WARNING MCH and prevent 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) processor.
|
|
|
|
*/
|
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;
|
|
|
|
/*
|
|
|
|
* Use single machine clear, as we cannot handle smp right now
|
|
|
|
*/
|
|
|
|
__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
|
|
|
}
|
|
|
|
#endif
|
2005-06-26 01:55:30 +04:00
|
|
|
|
|
|
|
if (mcck.kill_task) {
|
|
|
|
local_irq_enable();
|
|
|
|
printk(KERN_EMERG "mcck: Terminating task because of machine "
|
|
|
|
"malfunction (code 0x%016llx).\n", mcck.mcck_code);
|
|
|
|
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
|
|
|
|
|
|
|
/*
|
|
|
|
* returns 0 if all registers could be validated
|
|
|
|
* returns 1 otherwise
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
s390_revalidate_registers(struct mci *mci)
|
|
|
|
{
|
|
|
|
int kill_task;
|
|
|
|
u64 tmpclock;
|
|
|
|
u64 zero;
|
|
|
|
void *fpt_save_area, *fpt_creg_save_area;
|
|
|
|
|
|
|
|
kill_task = 0;
|
|
|
|
zero = 0;
|
|
|
|
/* General purpose registers */
|
|
|
|
if (!mci->gr)
|
|
|
|
/*
|
|
|
|
* General purpose registers couldn't be restored and have
|
|
|
|
* unknown contents. Process needs to be terminated.
|
|
|
|
*/
|
|
|
|
kill_task = 1;
|
|
|
|
|
|
|
|
/* Revalidate floating point registers */
|
|
|
|
if (!mci->fp)
|
|
|
|
/*
|
|
|
|
* Floating point registers can't be restored and
|
|
|
|
* therefore the process needs to be terminated.
|
|
|
|
*/
|
|
|
|
kill_task = 1;
|
|
|
|
|
2006-01-06 11:19:28 +03:00
|
|
|
#ifndef CONFIG_64BIT
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" ld 0,0(%0)\n"
|
|
|
|
" ld 2,8(%0)\n"
|
|
|
|
" ld 4,16(%0)\n"
|
|
|
|
" ld 6,24(%0)"
|
|
|
|
: : "a" (&S390_lowcore.floating_pt_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
#endif
|
|
|
|
|
|
|
|
if (MACHINE_HAS_IEEE) {
|
2006-01-06 11:19:28 +03:00
|
|
|
#ifdef CONFIG_64BIT
|
2005-06-26 01:55:30 +04:00
|
|
|
fpt_save_area = &S390_lowcore.floating_pt_save_area;
|
|
|
|
fpt_creg_save_area = &S390_lowcore.fpt_creg_save_area;
|
|
|
|
#else
|
|
|
|
fpt_save_area = (void *) S390_lowcore.extended_save_area_addr;
|
|
|
|
fpt_creg_save_area = fpt_save_area+128;
|
|
|
|
#endif
|
|
|
|
/* Floating point control register */
|
|
|
|
if (!mci->fc) {
|
|
|
|
/*
|
|
|
|
* Floating point control register can't be restored.
|
|
|
|
* Task will be terminated.
|
|
|
|
*/
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile("lfpc 0(%0)" : : "a" (&zero), "m" (zero));
|
2005-06-26 01:55:30 +04:00
|
|
|
kill_task = 1;
|
|
|
|
|
2006-09-28 18:56:43 +04:00
|
|
|
} else
|
|
|
|
asm volatile("lfpc 0(%0)" : : "a" (fpt_creg_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" ld 0,0(%0)\n"
|
|
|
|
" ld 1,8(%0)\n"
|
|
|
|
" ld 2,16(%0)\n"
|
|
|
|
" ld 3,24(%0)\n"
|
|
|
|
" ld 4,32(%0)\n"
|
|
|
|
" ld 5,40(%0)\n"
|
|
|
|
" ld 6,48(%0)\n"
|
|
|
|
" ld 7,56(%0)\n"
|
|
|
|
" ld 8,64(%0)\n"
|
|
|
|
" ld 9,72(%0)\n"
|
|
|
|
" ld 10,80(%0)\n"
|
|
|
|
" ld 11,88(%0)\n"
|
|
|
|
" ld 12,96(%0)\n"
|
|
|
|
" ld 13,104(%0)\n"
|
|
|
|
" ld 14,112(%0)\n"
|
|
|
|
" ld 15,120(%0)\n"
|
|
|
|
: : "a" (fpt_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Revalidate access registers */
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" lam 0,15,0(%0)"
|
|
|
|
: : "a" (&S390_lowcore.access_regs_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
if (!mci->ar)
|
|
|
|
/*
|
|
|
|
* Access registers have unknown contents.
|
|
|
|
* Terminating task.
|
|
|
|
*/
|
|
|
|
kill_task = 1;
|
|
|
|
|
|
|
|
/* Revalidate control registers */
|
|
|
|
if (!mci->cr)
|
|
|
|
/*
|
|
|
|
* Control registers have unknown contents.
|
|
|
|
* Can't recover and therefore stopping machine.
|
|
|
|
*/
|
|
|
|
s390_handle_damage("invalid control registers.");
|
|
|
|
else
|
2006-01-06 11:19:28 +03:00
|
|
|
#ifdef CONFIG_64BIT
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" lctlg 0,15,0(%0)"
|
|
|
|
: : "a" (&S390_lowcore.cregs_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
#else
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" lctl 0,15,0(%0)"
|
|
|
|
: : "a" (&S390_lowcore.cregs_save_area));
|
2005-06-26 01:55:30 +04:00
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We don't even try to revalidate the TOD register, since we simply
|
|
|
|
* can't write something sensible into that register.
|
|
|
|
*/
|
|
|
|
|
2006-01-06 11:19:28 +03:00
|
|
|
#ifdef CONFIG_64BIT
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* See if we can revalidate the TOD programmable register with its
|
|
|
|
* old contents (should be zero) otherwise set it to zero.
|
|
|
|
*/
|
|
|
|
if (!mci->pr)
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" sr 0,0\n"
|
|
|
|
" sckpf"
|
|
|
|
: : : "0", "cc");
|
2005-06-26 01:55:30 +04:00
|
|
|
else
|
|
|
|
asm volatile(
|
2006-09-28 18:56:43 +04:00
|
|
|
" l 0,0(%0)\n"
|
|
|
|
" sckpf"
|
|
|
|
: : "a" (&S390_lowcore.tod_progreg_save_area)
|
|
|
|
: "0", "cc");
|
2005-06-26 01:55:30 +04:00
|
|
|
#endif
|
|
|
|
|
|
|
|
/* Revalidate clock comparator register */
|
2006-09-28 18:56:43 +04:00
|
|
|
asm volatile(
|
|
|
|
" stck 0(%1)\n"
|
|
|
|
" sckc 0(%1)"
|
|
|
|
: "=m" (tmpclock) : "a" (&(tmpclock)) : "cc", "memory");
|
2005-06-26 01:55:30 +04:00
|
|
|
|
|
|
|
/* Check if old PSW is valid */
|
|
|
|
if (!mci->wp)
|
|
|
|
/*
|
|
|
|
* Can't tell if we come from user or kernel mode
|
|
|
|
* -> stopping machine.
|
|
|
|
*/
|
|
|
|
s390_handle_damage("old psw invalid.");
|
|
|
|
|
|
|
|
if (!mci->ms || !mci->pm || !mci->ia)
|
|
|
|
kill_task = 1;
|
|
|
|
|
|
|
|
return kill_task;
|
|
|
|
}
|
|
|
|
|
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
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
/*
|
|
|
|
* machine check handler.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
s390_do_machine_check(struct pt_regs *regs)
|
|
|
|
{
|
2006-04-28 05:40:23 +04:00
|
|
|
static DEFINE_SPINLOCK(ipd_lock);
|
|
|
|
static unsigned long long last_ipd;
|
|
|
|
static int ipd_count;
|
|
|
|
unsigned long long tmp;
|
2005-06-26 01:55:30 +04:00
|
|
|
struct mci *mci;
|
|
|
|
struct mcck_struct *mcck;
|
|
|
|
int umode;
|
|
|
|
|
2006-07-03 11:25:00 +04:00
|
|
|
lockdep_off();
|
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
mci = (struct mci *) &S390_lowcore.mcck_interruption_code;
|
|
|
|
mcck = &__get_cpu_var(cpu_mcck);
|
|
|
|
umode = user_mode(regs);
|
|
|
|
|
|
|
|
if (mci->sd)
|
|
|
|
/* System damage -> stopping machine */
|
|
|
|
s390_handle_damage("received system damage machine check.");
|
|
|
|
|
|
|
|
if (mci->pd) {
|
|
|
|
if (mci->b) {
|
|
|
|
/* Processing backup -> verify if we can survive this */
|
|
|
|
u64 z_mcic, o_mcic, t_mcic;
|
2006-01-06 11:19:28 +03:00
|
|
|
#ifdef CONFIG_64BIT
|
2005-06-26 01:55:30 +04:00
|
|
|
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);
|
|
|
|
#else
|
|
|
|
z_mcic = (1ULL<<63 | 1ULL<<59 | 1ULL<<57 | 1ULL<<50 |
|
|
|
|
1ULL<<29);
|
|
|
|
o_mcic = (1ULL<<43 | 1ULL<<42 | 1ULL<<41 | 1ULL<<40 |
|
|
|
|
1ULL<<36 | 1ULL<<35 | 1ULL<<34 | 1ULL<<32 |
|
|
|
|
1ULL<<30 | 1ULL<<20 | 1ULL<<17 | 1ULL<<16);
|
|
|
|
#endif
|
|
|
|
t_mcic = *(u64 *)mci;
|
|
|
|
|
|
|
|
if (((t_mcic & z_mcic) != 0) ||
|
|
|
|
((t_mcic & o_mcic) != o_mcic)) {
|
|
|
|
s390_handle_damage("processing backup machine "
|
|
|
|
"check with damage.");
|
|
|
|
}
|
2006-04-28 05:40:23 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Nullifying exigent condition, therefore we might
|
|
|
|
* retry this instruction.
|
|
|
|
*/
|
|
|
|
|
|
|
|
spin_lock(&ipd_lock);
|
|
|
|
|
|
|
|
tmp = get_clock();
|
|
|
|
|
|
|
|
if (((tmp - last_ipd) >> 12) < MAX_IPD_TIME)
|
|
|
|
ipd_count++;
|
|
|
|
else
|
|
|
|
ipd_count = 1;
|
|
|
|
|
|
|
|
last_ipd = tmp;
|
|
|
|
|
|
|
|
if (ipd_count == MAX_IPD_COUNT)
|
|
|
|
s390_handle_damage("too many ipd retries.");
|
|
|
|
|
|
|
|
spin_unlock(&ipd_lock);
|
2005-06-26 01:55:30 +04:00
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* Processing damage -> stopping machine */
|
|
|
|
s390_handle_damage("received instruction processing "
|
|
|
|
"damage machine check.");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (s390_revalidate_registers(mci)) {
|
|
|
|
if (umode) {
|
|
|
|
/*
|
|
|
|
* Couldn't restore all register contents while in
|
|
|
|
* user mode -> mark task for termination.
|
|
|
|
*/
|
|
|
|
mcck->kill_task = 1;
|
|
|
|
mcck->mcck_code = *(unsigned long long *) mci;
|
|
|
|
set_thread_flag(TIF_MCCK_PENDING);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
/*
|
|
|
|
* Couldn't restore all register contents while in
|
|
|
|
* kernel mode -> stopping machine.
|
|
|
|
*/
|
|
|
|
s390_handle_damage("unable to revalidate registers.");
|
|
|
|
}
|
|
|
|
|
2007-02-05 23:18:19 +03:00
|
|
|
if (mci->cd) {
|
|
|
|
/* Timing facility damage */
|
|
|
|
s390_handle_damage("TOD clock damaged");
|
|
|
|
}
|
|
|
|
|
|
|
|
if (mci->ed && mci->ec) {
|
|
|
|
/* External damage */
|
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_ETR_SYNC))
|
|
|
|
etr_sync_check();
|
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_ETR_SWITCH))
|
|
|
|
etr_switch_to_local();
|
2008-07-14 11:58:56 +04:00
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_STP_SYNC))
|
|
|
|
stp_sync_check();
|
|
|
|
if (S390_lowcore.external_damage_code & (1U << ED_STP_ISLAND))
|
|
|
|
stp_island_check();
|
2007-02-05 23:18:19 +03:00
|
|
|
}
|
|
|
|
|
2005-06-26 01:55:30 +04:00
|
|
|
if (mci->se)
|
|
|
|
/* Storage error uncorrected */
|
|
|
|
s390_handle_damage("received storage error uncorrected "
|
|
|
|
"machine check.");
|
|
|
|
|
|
|
|
if (mci->ke)
|
|
|
|
/* Storage key-error uncorrected */
|
|
|
|
s390_handle_damage("received storage key-error uncorrected "
|
|
|
|
"machine check.");
|
|
|
|
|
|
|
|
if (mci->ds && mci->fa)
|
|
|
|
/* Storage degradation */
|
|
|
|
s390_handle_damage("received storage degradation machine "
|
|
|
|
"check.");
|
|
|
|
|
|
|
|
if (mci->cp) {
|
|
|
|
/* Channel report word pending */
|
|
|
|
mcck->channel_report = 1;
|
|
|
|
set_thread_flag(TIF_MCCK_PENDING);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (mci->w) {
|
|
|
|
/* Warning pending */
|
|
|
|
mcck->warning = 1;
|
|
|
|
set_thread_flag(TIF_MCCK_PENDING);
|
|
|
|
}
|
2006-07-03 11:25:00 +04:00
|
|
|
lockdep_on();
|
2005-04-17 02:20:36 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* s390_init_machine_check
|
|
|
|
*
|
|
|
|
* initialize machine check handling
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
machine_check_init(void)
|
|
|
|
{
|
|
|
|
init_MUTEX_LOCKED(&m_sem);
|
2007-02-05 23:18:19 +03:00
|
|
|
ctl_set_bit(14, 25); /* enable external damage MCH */
|
2005-06-26 01:55:30 +04:00
|
|
|
ctl_set_bit(14, 27); /* enable system recovery MCH */
|
2005-04-17 02:20:36 +04:00
|
|
|
#ifdef CONFIG_MACHCHK_WARNING
|
|
|
|
ctl_set_bit(14, 24); /* enable warning MCH */
|
|
|
|
#endif
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Initialize the machine check handler really early to be able to
|
|
|
|
* catch all machine checks that happen during boot
|
|
|
|
*/
|
|
|
|
arch_initcall(machine_check_init);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Machine checks for the channel subsystem must be enabled
|
|
|
|
* after the channel subsystem is initialized
|
|
|
|
*/
|
|
|
|
static int __init
|
|
|
|
machine_check_crw_init (void)
|
|
|
|
{
|
2007-02-05 23:16:49 +03:00
|
|
|
struct task_struct *task;
|
|
|
|
|
|
|
|
task = kthread_run(s390_collect_crw_info, &m_sem, "kmcheck");
|
|
|
|
if (IS_ERR(task))
|
|
|
|
return PTR_ERR(task);
|
2005-04-17 02:20:36 +04:00
|
|
|
ctl_set_bit(14, 28); /* enable channel report MCH */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
device_initcall (machine_check_crw_init);
|