Problem with futex call on 8xx

Frank Svendsbøe frank.svendsboe at gmail.com
Sat Sep 26 22:52:29 EST 2009


I'll answer myself, since I just found the solution to the problem -
with help from 'kos_tom'
at #uclibc on freenode.

The problem is caused by a missing 'CONFIG_FUTEX=y' in the defconfig
for our target.

The system we're developing is based on the Adder port by Scott Wood, which is
also missing this, so a patch should be committed for this and
possible other 8xx targets
as well.

Best regards,
Frank

On Sat, Sep 26, 2009 at 2:00 PM, Frank Svendsbøe
<frank.svendsboe at gmail.com> wrote:
> Hi
> I'm having a problem with ~100% CPU load on MPC8xx when calling
> pthread_cond_wait.
> Running strace, I get:
>
> futex(0x10040d5c, FUTEX_WAIT, 1, NULL)  = -1 ENOSYS (Function not implemented)
>
> .. and this call is inifinitely repeating, which explains the high load.
>
> I'm running Linux 2.6.26-rc2 (not newer due to the slowdown problem
> introduced by commit
> 8d30c14cab30d405a05f2aaceda1e9ad57800f36, as pointed out by Rex Feany
> this week), and
> glibc v2.6 (part of ELDK 4.2).
>
> Is this a problem due to an error in the futex implementation in
> glibc, or a kernel problem?
>
> Best regards,
> Frank
>


More information about the Linuxppc-dev mailing list