2008-04-18 01:33:30 Pthread on BF54X/uClinux

Meihui Fan (CHINA)

Message: 54450

It seems link pthread into a program to run on BF54X/uClinux will make the process go into sleeping state.

The simple testing code is:

$ cat foo.c

#include <stdio.h>

#include <stdlib.h>

#include <unistd.h>

#include <pthread.h>

int main(void) { for (;;) printf("foo\n"); return 0; }

The toolchain version is:

$ bfin-linux-uclibc-gcc-4.1.2 -v

Using built-in specs.

Target: bfin-linux-uclibc

Configured with: /home/mhfan/devel/hhbf/uClinux-hg/toolchain/gcc-4.1/configure --build=i686-pc-linux-gnu --host=i686-pc-linux-gnu --target=bfin-linux-uclibc --prefix=/usr/local/toolchain-hhbf --enable-threads=posix --enable-shared --with-sysroot=/usr/local/toolchain-hhbf/bfin-linux-uclibc/runtime --disable-libstdcxx-pch --enable-languages=c,c++ --disable-symvers --disable-libssp --disable-libffi --enable-clocale=generic --enable-version-specific-runtime-libs --enable-__cxa_atexit

Thread model: posix

gcc version 4.1.2 (ADI svn)

Build with different flag:

$ bfin-linux-uclibc-gcc-4.1.2 foo.c -o foo

$ bfin-linux-uclibc-gcc-4.1.2 -pthread foo.c -o foo.pthread

Run the program on the target in background:

BF54X # ./foo > /dev/null &

BF54X # ./foo.pthread > /dev/null &

BF54X # ps

PID USER VSZ STAT COMMAND

1 0 560 S init

2 0 0 SW< [kthreadd]

3 0 0 SW< [ksoftirqd/0]

4 0 0 SW< [events/0]

5 0 0 SW< [khelper]

6 0 0 SW< [kblockd/0]

7 0 0 SW< [kmmcd]

8 0 0 SW [pdflush]

9 0 0 SW [pdflush]

10 0 0 SW< [kswapd0]

11 0 0 SW< [aio/0]

12 0 0 SW< [mtdblockd]

21 0 0 SW< [mmcqd]

23 0 668 R -/bin/msh

29 0 440 R ./foo

30 0 516 R ./foo.pthread

31 0 560 R ps

Nothing bad so far.

BF54X # sleep 30; ps

PID USER VSZ STAT COMMAND

1 0 560 S init

2 0 0 SW< [kthreadd]

3 0 0 SW< [ksoftirqd/0]

4 0 0 SW< [events/0]

5 0 0 SW< [khelper]

6 0 0 SW< [kblockd/0]

7 0 0 SW< [kmmcd]

8 0 0 SW [pdflush]

9 0 0 SW [pdflush]

10 0 0 SW< [kswapd0]

11 0 0 SW< [aio/0]

12 0 0 SW< [mtdblockd]

21 0 0 SW< [mmcqd]

23 0 668 R -/bin/msh

29 0 440 R ./foo

30 0 516 S ./foo.pthread

33 0 560 R ps

Here we get a problem: why foo.pthread goes into sleeping state now???

Is this a toolchain/kernel BUG?

QuoteReplyEditDelete

2008-04-18 13:11:28 Re: Pthread on BF54X/uClinux

Robin Getz (UNITED STATES)

Message: 54492 Meihui:

I talked to Mike, and he has been able to replicate on 537 - so I don't know why we have not noticed before.

We are digging into it, and will get back to you.

-Robin

QuoteReplyEditDelete

2008-04-24 13:54:39 Re: Pthread on BF54X/uClinux

SÃrgio Trofino (BRAZIL)

Message: 54881 Hi, Robin

I am using the BF537-STAMP, which run an image of uClinux_svn compiled with toolchain_svn.

Using the version of the svn or Release2008 I have problems in implementing the code attached.

It seems to me that the problem is with the function "mutex_unlock" because with the toolchain Release2008 accused of a mistake in libpthread.so Null Pointer + 0x8b42, which will be noticed the "mutex_unlock."

If find a solution, I kind enough to inform me also.

Regards

Sérgio Trofino

cond_signal.c

QuoteReplyEditDelete

2008-04-24 14:41:11 Re: Pthread on BF54X/uClinux

Robin Getz (UNITED STATES)

Message: 54886 SÃrgio:

Yeah, Mike is working on it - and has reproduced it, and is figuring it out.

-Robin

QuoteReplyEditDelete

2008-04-25 18:28:52 Re: Pthread on BF54X/uClinux