backtrace() not working on ARMv7a

[Filippo ARCIDIACONO]

On 9/12/2013 6:51 PM, Rajendra Dendukuri wrote:
> Filippo,
>
> That really helps!
great!
> Now a minor variation of the program does not throw a full backtrace and stops short after showing the c library call. Again on x86 host machine it correctly shows a deep backtrace.
I got a full backtrace also with this test.

root at arcidiaf:/home/filippo# gcc -o crash_test_1 -g -O0 -fexceptions 
-finstrument-functions crash_tst_1.c
root at arcidiaf:/home/filippo# ./crash_test_1

Trying to construct invalid buffer!
SIGSEGV Handler!
Got Backtrace Size=10
0x00008854
0x0000894c
0x76e62c24
0x76e8c890
0x76e862d0
0x76e837cc
0x76e835a8
0x00008a6c
0x00008b18
0x76ead7bc

To fix this you need some patches that are on my list to be submitted to 
the mailing list.
I'm going to submit at least the patch you need.


>
> Attached is the file. I am calling "func1()" which crashes inside snprintf library call implementation.
>
> ARMv7a# ./crash_test
>
> Trying to construct invalid buffer!
> SIGSEGV Handler!
> Got Backtrace Size=3
> 0x000087bc
> 0x0000887c
> 0xb6f02420
> ARMv7a #
>
> I think we need to add unwind-tables or -finstrument-functions for uClibc as well. I am using buildroot for the toolchain. Any ideas?
yes, as I said before.
>
> /Thanks
you are welcome.

regards,
Filippo
>
> -----Original Message-----
> From: Filippo ARCIDIACONO [mailto:filippo.arcidiacono at st.com]
> Sent: Thursday, September 12, 2013 10:14 AM
> To: Rajendra Dendukuri
> Cc: SHREYAS JOSHI; uclibc at uclibc.org
> Subject: Re: backtrace() not working on ARMv7a
>
> On 9/12/2013 2:49 PM, Rajendra Dendukuri wrote:
>> Thanks Shreyas for the tip. But it is not working even with "-fexceptions" or "-fasynchronous-unwind-tables" options.
>>
>> Forum,
>> Anyone faced this issue before?
> Yes. I got the same behavior using your test.
> Doing some googling , see
> http://stackoverflow.com/questions/9229702/gcc-return-address-of-calling-function-in-arm-architecture
> on ARM arch the gcc __builtin_return_address() function does not work
> beyond the current function|||| (only __builtin_return_address(0) works),
> this explain because we don't obtain a full backtrace.
> Using -finstrument-function compiler option (see gcc documentation at
> http://gcc.gnu.org/onlinedocs/gcc-4.7.3/gcc/Code-Gen-Options.html#Code-Gen-Options)
> I got a full backtrace:
>
> root at arcidiaf:/home/filippo# gcc -o crash_test -O0 -fexceptions
> -finstrument-functions crash_tst.c
> root at arcidiaf:/home/filippo# ./crash_test
> Trying to access NULL pointer!
> SIGSEGV Handler!
> Got Backtrace Size=6
> 0x00008854
> 0x0000894c
> 0x76ef1c24
> 0x000089b4
> 0x00008b18
> 0x76f3c7bc
>
>
>
> ||
>> /Thanks
>>
>> From: SHREYAS JOSHI [mailto:dexterous.me82 at yahoo.com]
>> Sent: Thursday, September 12, 2013 2:33 AM
>> To: Rajendra Dendukuri; uclibc at uclibc.org
>> Subject: Re: backtrace() not working on ARMv7a
>>
>> Try -fexceptions.
>>
>> For further details, check the following link.
>>
>> http://www.tune2wizard.com/backtrace/
>>
>>
>> Thanks & regards,
>> Shreyas Joshi
>>
>>
>> ________________________________
>> From: Rajendra Dendukuri <rajendra at broadcom.com<mailto:rajendra at broadcom.com>>
>> To: "uclibc at uclibc.org<mailto:uclibc at uclibc.org>" <uclibc at uclibc.org<mailto:uclibc at uclibc.org>>
>> Sent: Wednesday, September 11, 2013 10:01 PM
>> Subject: backtrace() not working on ARMv7a
>>
>> Hi uclibc list,
>>
>> I am trying to generate a call stack trace using backtrace() API which is part of libubacktrace. The idea is to catch a SIGSEGV signal and call backtrace() to print all the address values and compare the obtained addresses from objdump -D of executable. Attached is a simplified version of the application. Following is the output seen on ARMv7a platform with buildroot cross-toolchain (gcc- 4.6.3, uClibc-0.9.33.2, binutis-2.21.1).
>>
>> To compile the application:
>> ${CROSS_COMPILE}gcc -o crash_test -O0 -funwind-tables -rdynamic crash_test.c
>>
>> ARMv7a# ./crash_test
>> Trying to access NULL pointer!
>> SIGSEGV Handler!
>> Got Backtrace Size=2
>> 0x00008724
>> 0x000087c4
>> ARMv7a#
>>
>> The above function call trace is only that of the signal hander (print_back_trace, sigsegv_handler) and does not show the entire call stack for the function where program crashed (print_back_trace, sigsegv_handler, func2, main).
>>
>> On x86 host same program shows a deeper function call trace right up to the main function and even beyond:
>> x86-RHEL5-host$./crash_test
>> Trying to access NULL pointer!
>> SIGSEGV Handler!
>> Got Backtrace Size=7
>> 0x00400939
>> 0x004009ab
>> 0x30930302f0
>> 0x004009d3
>> 0x00400a7b
>> 0x309301d994
>> 0x00400889
>> x86-RHEL5-host$
>>
>> Can someone please comment on what is going wrong. Any special flags that need to be used while compiling/linking the program.
>>
>> /Thanks
>>
>> _______________________________________________
>> uClibc mailing list
>> uClibc at uclibc.org<mailto:uClibc at uclibc.org>
>> http://lists.busybox.net/mailman/listinfo/uclibc
>> _______________________________________________
>> uClibc mailing list
>> uClibc at uclibc.org
>> http://lists.busybox.net/mailman/listinfo/uclibc
>>
>>
>