Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-11-16 17:12:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

x86, mce: Improve timeout error messages

Browse Source 
There are four different possible types of timeouts.  Distinguish
them in the logs to help debug them.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Link: http://lkml.kernel.org/r/0fa6d2653a54a01c48b43a3583caf950ea99606e.1419178397.git.luto@amacapital.net
Signed-off-by: Borislav Petkov <bp@suse.de>
This commit is contained in:
Andy Lutomirski 2014-12-21 08:18:25 -08:00 committed by Borislav Petkov
parent 97bf6af1f9
commit 6c80f87ed4
1 changed files with 11 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
arch/x86/kernel/cpu/mcheck/mce.c
View File

@ -311,7 +311,7 @@ static void wait_for_panic(void)
panic("Panicing machine check CPU died"); panic("Panicing machine check CPU died");
} }
static void mce_panic(char *msg, struct mce *final, char *exp) static void mce_panic(const char *msg, struct mce *final, char *exp)
{ {
int i, apei_err = 0; int i, apei_err = 0;
@ -735,7 +735,7 @@ static atomic_t mce_callin;
/* /*
* Check if a timeout waiting for other CPUs happened. * Check if a timeout waiting for other CPUs happened.
*/ */
static int mce_timed_out(u64 *t) static int mce_timed_out(u64 *t, const char *msg)
{ {
/* /*
* The others already did panic for some reason. * The others already did panic for some reason.
@ -750,8 +750,7 @@ static int mce_timed_out(u64 *t)
goto out; goto out;
if ((s64)*t < SPINUNIT) { if ((s64)*t < SPINUNIT) {
if (mca_cfg.tolerant <= 1) if (mca_cfg.tolerant <= 1)
mce_panic("Timeout synchronizing machine check over CPUs", mce_panic(msg, NULL, NULL);
NULL, NULL);
cpu_missing = 1; cpu_missing = 1;
return 1; return 1;
} }
@ -867,7 +866,8 @@ static int mce_start(int *no_way_out)
* Wait for everyone. * Wait for everyone.
*/ */
while (atomic_read(&mce_callin) != cpus) { while (atomic_read(&mce_callin) != cpus) {
if (mce_timed_out(&timeout)) { if (mce_timed_out(&timeout,
"Timeout: Not all CPUs entered broadcast exception handler")) {
atomic_set(&global_nwo, 0); atomic_set(&global_nwo, 0);
return -1; return -1;
} }
@ -892,7 +892,8 @@ static int mce_start(int *no_way_out)
* only seen by one CPU before cleared, avoiding duplicates. * only seen by one CPU before cleared, avoiding duplicates.
*/ */
while (atomic_read(&mce_executing) < order) { while (atomic_read(&mce_executing) < order) {
if (mce_timed_out(&timeout)) { if (mce_timed_out(&timeout,
"Timeout: Subject CPUs unable to finish machine check processing")) {
atomic_set(&global_nwo, 0); atomic_set(&global_nwo, 0);
return -1; return -1;
} }
@ -936,7 +937,8 @@ static int mce_end(int order)
* loops. * loops.
*/ */
while (atomic_read(&mce_executing) <= cpus) { while (atomic_read(&mce_executing) <= cpus) {
if (mce_timed_out(&timeout)) if (mce_timed_out(&timeout,
"Timeout: Monarch CPU unable to finish machine check processing"))
goto reset; goto reset;
ndelay(SPINUNIT); ndelay(SPINUNIT);
} }
@ -949,7 +951,8 @@ static int mce_end(int order)
* Subject: Wait for Monarch to finish. * Subject: Wait for Monarch to finish.
*/ */
while (atomic_read(&mce_executing) != 0) { while (atomic_read(&mce_executing) != 0) {
if (mce_timed_out(&timeout)) if (mce_timed_out(&timeout,
"Timeout: Monarch CPU did not finish machine check processing"))
goto reset; goto reset;
ndelay(SPINUNIT); ndelay(SPINUNIT);
} }