some task not stopped during resume

some task not stopped during resume

[Johannes Berg]

[-- Attachment #1.1: Type: text/plain, Size: 777 bytes --]

Hi,

Oddly, I just got the following on 2.6.22-rc1:

[164571.322432] PM: Preparing system for mem sleep
[164571.883809] Stopping tasks ... done.
[...]
[164573.454973] PM: Entering mem sleep
[...]
[164574.580359] PM: Finishing wakeup.
[...]
[164578.238895] Restarting tasks ... <4> Strange, evolution not stopped
[164578.456389] done.

The interesting thing is that when I resumed, the gdb crash window came
up because evolution had just crashed (this is an option, otherwise
bug-buddy comes up.) But I'm not entirely sure how this could have
happened since it was definitely running while I suspended and must have
stopped at that point. Maybe some bad ptrace interaction? Though it
shouldn't have crashed when tasks were stopped? I'm confused.

johannes

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: some task not stopped during resume

[Rafael J. Wysocki]

Hi,

On Tuesday, 29 May 2007 14:42, Johannes Berg wrote:
> Hi,
> 
> Oddly, I just got the following on 2.6.22-rc1:
> 
> [164571.322432] PM: Preparing system for mem sleep
> [164571.883809] Stopping tasks ... done.
> [...]
> [164573.454973] PM: Entering mem sleep
> [...]
> [164574.580359] PM: Finishing wakeup.
> [...]
> [164578.238895] Restarting tasks ... <4> Strange, evolution not stopped
> [164578.456389] done.
> 
> The interesting thing is that when I resumed, the gdb crash window came
> up because evolution had just crashed (this is an option, otherwise
> bug-buddy comes up.) But I'm not entirely sure how this could have
> happened since it was definitely running while I suspended and must have
> stopped at that point. Maybe some bad ptrace interaction? Though it
> shouldn't have crashed when tasks were stopped? I'm confused.

The message seems to be a false-positive (this printk has been removed
recently, in 2.6.22-rc3, IIRC).

We don't freeze traced tasks the parents of which have been frozen.  So, in
fact, the evolution wasn't frozen, only the gdb that traced it, and during the
resume the gdb was thawed and allowed evolution to run, but it crashed.

Why it crashed is a different problem, actually, and that need not be related
to the freezer.  Still, if that turns out to be reproducible, I'll start to worry. ;-)

Greetings,
Rafael

Re: some task not stopped during resume

[Johannes Berg]

[-- Attachment #1.1: Type: text/plain, Size: 854 bytes --]

On Wed, 2007-05-30 at 15:50 +0200, Rafael J. Wysocki wrote:

> The message seems to be a false-positive (this printk has been removed
> recently, in 2.6.22-rc3, IIRC).
> 
> We don't freeze traced tasks the parents of which have been frozen.  So, in
> fact, the evolution wasn't frozen, only the gdb that traced it, and during the
> resume the gdb was thawed and allowed evolution to run, but it crashed.

Ah, that makes sense, thanks for the explanation.

> Why it crashed is a different problem, actually, and that need not be related
> to the freezer.  Still, if that turns out to be reproducible, I'll start to worry. ;-)

Well, I suppose it then crashed right before I suspended, gdb attached,
was frozen, and then I only saw that when I resumed. And evolution
really kills itself all the time without the freezer too ;)

johannes

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]