2006-10-15 03:00:07

by Steve Youngs

[permalink] [raw]
Subject: Bad core files with 2.6.19-rc2

gdb doesn't like any core dump file generated while running
2.6.19-rc2. If I `kill -SIGSEGV $some_app_pid' and then...

gdb some_app core

I get...

warning: Couldn't find general-purpose registers in core file.
#0 0x00000000 in ?? ()

But if I gdb attach to a running process and then kill -SIGSEGV
the process, it produces a normal trace without problem.


--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| Te audire no possum. |
| Musa sapientum fixa est in aure. |
|----------------------------------<[email protected]>---|


2006-10-15 18:42:22

by Adrian Bunk

[permalink] [raw]
Subject: Re: Bad core files with 2.6.19-rc2

On Sun, Oct 15, 2006 at 12:53:41PM +1000, Steve Youngs wrote:

> gdb doesn't like any core dump file generated while running
> 2.6.19-rc2. If I `kill -SIGSEGV $some_app_pid' and then...
>
> gdb some_app core
>
> I get...
>
> warning: Couldn't find general-purpose registers in core file.
> #0 0x00000000 in ?? ()
>
> But if I gdb attach to a running process and then kill -SIGSEGV
> the process, it produces a normal trace without problem.

It seems this issue should be fixed in Linus' tree now.

Can you confirm it's fixed?

TIA
Adrian

--

"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed

2006-10-15 19:24:46

by Petr Vandrovec

[permalink] [raw]
Subject: Re: Bad core files with 2.6.19-rc2

Adrian Bunk wrote:
> On Sun, Oct 15, 2006 at 12:53:41PM +1000, Steve Youngs wrote:
>
>
>>gdb doesn't like any core dump file generated while running
>>2.6.19-rc2. If I `kill -SIGSEGV $some_app_pid' and then...
>>
>> gdb some_app core
>>
>>I get...
>>
>> warning: Couldn't find general-purpose registers in core file.
>> #0 0x00000000 in ?? ()
>>
>>But if I gdb attach to a running process and then kill -SIGSEGV
>>the process, it produces a normal trace without problem.
>
>
> It seems this issue should be fixed in Linus' tree now.
>
> Can you confirm it's fixed?

It should be fixed now (i.e. 26 minutes ago). Before my first fix core
dump contained only some phdrs and notes, nothing else. After first fix
core dump contained everything, but pieces after notes were shifted down
by couple of bytes. After second fix I believe that cores are written
out correctly - readelf still says that there is some zeroed note, but
it seems to me like feature - older core files have this as well:

$ readelf -a core
...
Notes at offset 0x00035608 with length 0x00000018:
Owner Data size Description
Linux 0x00000004 Unknown note type: (0x00000000)

Petr Vandrovec

2006-10-15 22:15:59

by Steve Youngs

[permalink] [raw]
Subject: Re: Bad core files with 2.6.19-rc2

* Petr Vandrovec <[email protected]> writes:

> Adrian Bunk wrote:
>> On Sun, Oct 15, 2006 at 12:53:41PM +1000, Steve Youngs wrote:
>>> warning: Couldn't find general-purpose registers in core file.
>>> #0 0x00000000 in ?? ()

>> It seems this issue should be fixed in Linus' tree now.
>>
>> Can you confirm it's fixed?

> It should be fixed now

Yes, it is. Thank you very much.

--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| Te audire no possum. |
| Musa sapientum fixa est in aure. |
|----------------------------------<[email protected]>---|