Re: eglibc / FP / Movidis

[Shane McDonald <mcdonald.shane@gmail.com>]

Some late follow-up on this issue...

On Thu, May 6, 2010 at 10:28 PM, Shane McDonald
<mcdonald.shane@gmail.com> wrote:
> On Thu, May 6, 2010 at 8:53 PM, Clint Adams <schizo@debian.org> wrote:
>> On Thu, May 06, 2010 at 05:05:37PM -0600, Shane McDonald wrote:
>>> I tried building eglibc_2.10.2-7 on my FPU-less mips machine with
>>> a kernel patched for the identified FPU emulator bug.  Although I still
>>> see some unexpected failures, the floating point failures you reported
>>> seem to be resolved with this kernel patch.
>>
>> That's good news.  Is the SMP issue going to keep the patch out of
>> mainline?
>
> Nope -- there's just some cleanup / coding style issues remaining,
> I believe.

The patch has been accepted into the linux-mips.org repository,
both on the HEAD and on the stable versions since 2.6.16.  However,
it didn't make it into Linus' 2.6.34-rc7, nor Greg KH's 2.6.32.13 or 2.6.33.4.

Linus hasn't pulled the l-m.o tree since the patch went in, so I don't
know whether it will show up in -rc8 (if there is one) or the final release.

>>> Encountered regressions that don't match expected failures:
>>> tst-cancel4.out, Error 1
>>> tst-cancel5.out, Error 1
>>> tst-mmap-eofsync.out, Error 1
>>> tst-mmap-fflushsync.out, Error 1
>>
>> If you feel like rebuilding, you could peek into those .out files
>> to see if anything is obvious.
>
> Will do.  I wasn't thinking, and my pbuilder cleaned up after itself.
> I'm rebuilding now in a sid chroot so that those files will stay around.
> I'll let you know what I find.

In my re-build, I only got three of the four failures I previously got:

    Encountered regressions that don't match expected failures:
    tst-cancel4.out, Error 1
    tst-cancel5.out, Error 1
    tst-mmap-fflushsync.out, Error 1

I couldn't see anything obvious from the .out files.  Here are their contents.

tst-cancel4.out:

in-time cancel test of 'read' successful
in-time cancel test of 'readv' successful
in-time cancel test of 'select' successful
tf_pselect: pselect returns with 0 (Success)

tst-cancel5.out:

in-time cancel test of 'read' successful
in-time cancel test of 'readv' successful
in-time cancel test of 'select' successful
tf_pselect: pselect returns with 0 (Success)

tst-mmap-fflushsync.out

feof = 0, ferror = 0 immediately after fgets
wrote more to file
second time read "", expected "Line the second


I hope this information helps you.

Shane