Re: [issues] arm gdb backtrace for read syscall

[Matthieu CASTET <matthieu.castet@xxxxxxxxxx>]

Carlos O'Donell a écrit :
> On Fri, Oct 17, 2008 at 12:28 PM, Matthieu CASTET
> <matthieu.castet@xxxxxxxxxx> wrote:
>> I am using eglibc-2.7, and the gdb backtrace doesn't work when a thread
>> is block on a read syscall. [1]
>> Can't we manage to make the read entry code more gdb friendly ?
>> [1]
>> bt
>> #0  0x4003383c in read ()
>>   from /opt/arm-eglibc/arm-none-linux-gnueabi/libc/lib/libpthread.so.0
>> #1  0x400330ac in __pthread_enable_asynccancel () at cancellation.c:42
>> #2  0x00000000 in ?? ()
> 
> What you mean by more gdb friendly?
I mean it avoid to distrub gdb backtrace algorithm.

I don't what arm gdb use for backtrace (frame pointer, stack scan, ...).


> Could you describe the behaviour you would like?
I would like that when a process is block on the read system call, I can
get the complete backtrace ie getting caller of the read syscall.
This is the case for example for sigwaitinfo syscall.


Thanks.

Matthieu