libunwind-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Libunwind-devel] Need help: libunwind on ARMv5


From: Andrei Erofeev
Subject: Re: [Libunwind-devel] Need help: libunwind on ARMv5
Date: Thu, 27 Oct 2016 18:54:14 +0300

Hello,

 

A small update: I get a bit closer to the success with libunwind and ARMv5.

 

What I’ve done:

-          I built libunwind (the latest commit from the master branch).

-          I set unwind method to 3 on ARMv5 target board (I used environment variable to do that).

-          I got almost complete backtraces, the function where I cause the abort is missed from the backtraces for some reasons. Only that function is missed right now, all other seems at their places.

 

So the only issue I have at the moment is the missing “faulty” function in my backtraces. Maybe you can recommend a patch to fix that. I would very appreciate.

 

Thanks in advance,

Andrey

 

 

 

From: Andrei Erofeev [mailto:address@hidden
Sent: 27 October, 2016 2:28 PM
To: 'address@hidden' <address@hidden>
Subject: Need help: libunwind on ARMv5

 

Hello,

 

                First of all - Thanks you all for the cool product - libunwind!

 

                I have a question and hope that your answer helps me to solve a problem I stuck with.

 

I use a board with ARMv5 architecture (ARM926EJ-S). I have Linux OS and a C++ application running on that board type.  The problem is that I am not able to get backtraces in case of abort() and exceptions which cause abort(). If I use kill(getpid(0, SIGABRT) I get correct backtraces. The problem is well known. So I use libunwind and works if I execute the application on an ARMv6 board. But in case of ARMv5 it does not work for me. I cannot get the whole traces. I tried to change methods and it fails (I’ve attached the log).

 

                I use Libunwind version 1.1.

 

More info about CPU used on the board:

# cat /proc/cpuinfo

processor       : 0

model name      : ARM926EJ-S rev 5 (v5l)

BogoMIPS        : 198.76

Features        : swp half thumb fastmult edsp java

CPU implementer : 0x41

CPU architecture: 5TEJ

CPU variant     : 0x0

CPU part        : 0x926

CPU revision    : 5

Hardware        : Atmel AT91SAM (Device Tree)

 

I have a small app just to test backtrace functionality in case of abort()/throw. I’ve attached test1.c.

 

I do compilation using:

arm-buildroot-linux-gnueabi-g++ --std=c++11 -funwind-tables  -mapcs-frame -rdynamic test1.c  -lunwind -o test1

 

 

Thanks in advance,

Andrey

 


reply via email to

[Prev in Thread] Current Thread [Next in Thread]