Skip to contentSkip to conversation

no backtrace after "printk/nmi: generic solution for safe printk in NMI"

linux-next, x86_64 Sergey Senozhatsky 1 participant 3 messages Feb 25, 2016 – Feb 26, 2016

lore
Share & tools
lore
#0 lore raw

Hello Petr,

seem that commit b927968830676373caf4241e80d8b447133f84b2
Author: Petr Mladek <[email protected]>
Date: Thu Feb 25 13:00:35 2016 +1100

printk/nmi: generic solution for safe printk in NMI

printk() takes some locks and could not be used a safe way in NMI context.

The chance of a deadlock is real especially when printing stacks from all CPUs. This particular problem has been addressed on x86 by the commit a9edc8809328 ("x86/nmi: Perform a safe NMI stack trace on all CPUs").

The patchset brings two big advantages. First, it makes the NMI backtraces safe on all architectures for free. Second, it makes all NMI messages almost safe on all architectures (the temporary buffer is limited. We still should keep the number of messages in NMI context at minimum).
[..]

makes my x86_64 boxen unhappy, I see no CPU backtraces and no panic messages on HARDLOCKUPs (CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE=1).

does it work for you?

	-ss
#1 lore raw
Quoted 22 lines · Sergey Senozhatsky

Hello Petr,

seem that commit b927968830676373caf4241e80d8b447133f84b2
Author: Petr Mladek <[email protected]>
Date: Thu Feb 25 13:00:35 2016 +1100

printk/nmi: generic solution for safe printk in NMI

printk() takes some locks and could not be used a safe way in NMI context.

The chance of a deadlock is real especially when printing stacks from all CPUs. This particular problem has been addressed on x86 by the commit a9edc8809328 ("x86/nmi: Perform a safe NMI stack trace on all CPUs").

The patchset brings two big advantages. First, it makes the NMI backtraces safe on all architectures for free. Second, it makes all NMI messages almost safe on all architectures (the temporary buffer is limited. We still should keep the number of messages in NMI context at minimum).
[..]

Quoted 4 lines

makes my x86_64 boxen unhappy, I see no CPU backtraces and no panic messages on HARDLOCKUPs (CONFIG_BOOTPARAM_HARDLOCKUP_PANIC_VALUE=1).

does it work for you?

just for note,

I reverted (next-20160225) the following commits:

bdcdb76696402e1a0821db6bcccb2b4de4049f27
7ad17383fa428d1c89d2be95d00806f33df69dfd
054a26c149a5add6d27d154f2025ee2605d9f25d
c2dbc8fd415835a71b0dee0f3f3a353d116f8731
3fd57f62ac2a8de6b402bf5f6aeb3a7d1e3f11a3
3ff0033c1e878b8bd8fce6efcd1b432b24f1d321

-- still no HARDLOCKUP panic backtraces on a console.

reverting of

b927968830676373caf4241e80d8b447133f84b2 (plus MIPS config conflict resolution)

-- fixed the problem for me. I've tested several times.

the sample code that I use for testing is very simple (well.. but it does the job):

---

       u64 s, e;
       unsigned long flags;

       local_irq_save(flags);
       s = local_clock() >> 31UL;
       pr_err(">>>>>>>>>>>>>>>>>START_THE_TEST\n");
       while (1) {
               e = local_clock() >> 31UL;
               if (e - s > %%YOUR_WATCHDOG_THRESHOLD%%)
                       goto out;
       }
out:
       pr_err(">>>>>>>>>>>>>>>>>END_THE_TEST\n");
       local_irq_restore(flags);

---

most likely I'll be able to reply only tomorrow (in case if there will be any questions).

	-ss
#2 lore raw

On (02/25/16 19:31), Sergey Senozhatsky wrote:
[..]

Quoted 1 line

most likely I'll be able to reply only tomorrow (in case if there will be any questions).

solely for the record

lkml.kernel.org/r/[email protected]

	-ss