[blfs-dev] i686 lame fails to build for me

[Ken Moffat]

I haven't built i686 for ages - my last time was while testing 7.4,
for 7.5 there was nothing which I felt it would be useful for me to
test on that architecture. But all the recent LFS changes made me
decide to try it again. The machine is an AMD A4, and LFS itself
built ok.

But lame, even with
sed -i -e 's/fast-math/& -msse2/' configure
fails to build :

xmm_quantize_sub.c: In function 'init_xrpow_core_sse':
xmm_quantize_sub.c:65:18: warning: SSE vector return without SSE
enabled changes the ABI [-Wpsabi]
const __m128 vec_fabs_mask =
_mm_loadu_ps(&fabs_mask._float[0]);
^
In file included from xmm_quantize_sub.c:37:0:
/usr/lib/gcc/i686-pc-linux-gnu/4.9.0/include/xmmintrin.h:929:1:
error: inlining failed in call to always_inline '_mm_loadu_ps':
target specific option mismatch
_mm_loadu_ps (float const *__P)
^

That first warning made me wonder if the CPU was a bit lacking in
sse* support (ISTR some of the earlier athlon64 processors did not
have a full complement of sse{,2,3}. But /proc/cpuinfo shows -

vendor_id : AuthenticAMD
cpu family : 21
model : 16
model name : AMD A4-5300 APU with Radeon(tm) HD Graphics
...
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr
pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx
mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc nonstop_tsc
extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1
sse4_2 popcnt aes xsave avx f16c lahf_lm cmp_legacy svm extapic
cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs xop skinit
wdt lwp fma4 tce nodeid_msr tbm topoext perfctr_core perfctr_nb arat
cpb hw_pstate npt lbrv svm_lock nrip_save tsc_scale vmcb_clean
flushbyasid decodeassists pausefilter pfthreshold bmi1

which I think is all the information that matters, and it does
claim to support sse, sse2, sse3, and even sse4_2.

For the moment I've commented lame out, but I'm sure that
something, possibly ffmpeg, will want lame. Any other ideas about
fixing the gcc-4.9 problem on 32-bit ?

ĸen

[Ken Moffat]

Post by Ken Moffat
I haven't built i686 for ages - my last time was while testing 7.4,
for 7.5 there was nothing which I felt it would be useful for me to
test on that architecture. But all the recent LFS changes made me
decide to try it again. The machine is an AMD A4, and LFS itself
built ok.
But lame, even with
sed -i -e 's/fast-math/& -msse2/' configure

I also get a similar failure in gst-plugins-base, with
sed -i -e 's/-O2/& -msse2/' configure

make[3]: Entering directory
'/scratch/working/gst-plugins-base-0.10.36/gst/subparse'
CC libgstsubparse_la-gstssaparse.lo
CC libgstaudioresample_la-speex_resampler_float.lo
In file included from resample.c:134:0,
from speex_resampler_float.c:26:
resample_sse.h: In function ‘interpolate_product_single’:
resample_sse.h:62:10: warning: SSE vector return without SSE enabled
changes the ABI [-Wpsabi]
__m128 sum = _mm_setzero_ps();
^
In file included from resample_sse.h:38:0,
from resample.c:134,
from speex_resampler_float.c:26:
resample_sse.h: In function ‘inner_product_single’:
/usr/lib/gcc/i686-pc-linux-gnu/4.9.0/include/xmmintrin.h:115:1:
error: inlining failed in call to always_inline ‘_mm_setzero_ps’:
target specific option mismatch
_mm_setzero_ps (void)
^

etc.

This time, I'm stopping the build - firefox without gst-plugins-base
will not be useful to me. Again, any suggestions will be gratefully
received.

ĸen

[Pierre Labastie]

Post by Ken Moffat
I haven't built i686 for ages - my last time was while testing 7.4,
for 7.5 there was nothing which I felt it would be useful for me to
test on that architecture. But all the recent LFS changes made me
decide to try it again. The machine is an AMD A4, and LFS itself
built ok.
But lame, even with
sed -i -e 's/fast-math/& -msse2/' configure

[...]

Post by Ken Moffat
For the moment I've commented lame out, but I'm sure that
something, possibly ffmpeg, will want lame. Any other ideas about
fixing the gcc-4.9 problem on 32-bit ?
ĸen

Have you tried Fernando's solution?
(http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027263.html).
That is for gst-plugins-base, but I guess it may be adapted to lame.

Pierre

[Pierre Labastie]

Post by Pierre Labastie

Post by Ken Moffat
I haven't built i686 for ages - my last time was while testing 7.4,
for 7.5 there was nothing which I felt it would be useful for me to
test on that architecture. But all the recent LFS changes made me
decide to try it again. The machine is an AMD A4, and LFS itself
built ok.
But lame, even with
sed -i -e 's/fast-math/& -msse2/' configure

[...]

Post by Ken Moffat
For the moment I've commented lame out, but I'm sure that
something, possibly ffmpeg, will want lame. Any other ideas about
fixing the gcc-4.9 problem on 32-bit ?
ĸen

Have you tried Fernando's solution?
(http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027263.html).
That is for gst-plugins-base, but I guess it may be adapted to lame.
Pierre

Actually, for lame, one possible solution is in this post:
http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027260.html

Pierre

[Fernando de Oliveira]

Post by Pierre Labastie

Post by Pierre Labastie

Post by Ken Moffat
I haven't built i686 for ages - my last time was while testing 7.4,
for 7.5 there was nothing which I felt it would be useful for me to
test on that architecture. But all the recent LFS changes made me
decide to try it again. The machine is an AMD A4, and LFS itself
built ok.
But lame, even with
sed -i -e 's/fast-math/& -msse2/' configure

[...]

Post by Ken Moffat
For the moment I've commented lame out, but I'm sure that
something, possibly ffmpeg, will want lame. Any other ideas about
fixing the gcc-4.9 problem on 32-bit ?
ĸen

Have you tried Fernando's solution?
(http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027263.html).
That is for gst-plugins-base, but I guess it may be adapted to lame.
Pierre

http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027260.html
Pierre

Please, read also the ticket about LibreOffice:

http://wiki.linuxfromscratch.org/blfs/ticket/5042

Since the version in the book, it cannot build with gcc-4.9.0 in i686.

This night I tried again, for the first time in the new udev system, and
it was the same as in the hybrid LFS.

So, we have already lame, gst-plugins-base (both versions) and LibreOffice.

Perhaps we could change this thread subject to "Some build problems with
gcc-4.9.0: lame, gst-plugins-base and LibreOffice" or something like that.

[Ken Moffat]

Post by Pierre Labastie
http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027260.html

Thanks. I'll be trying sed -i -e '/xmmintrin\.h/d' configure

ĸen

[Ken Moffat]

Post by Pierre Labastie
Have you tried Fernando's solution?
(http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027263.html).
That is for gst-plugins-base, but I guess it may be adapted to lame.
Pierre

Thanks for the pointers. I thought both of these issues had been
addressed by what is currently in the book. For gst-plugins-base I
had to apply fernando's patch manually (I think I sorted out the
wrapping caused by putting it in the body of the mail, but it did
not apply - the block seems to be a few lines earlier in my source,
and that was enough for it to be treated as "reversal - apply
anyway?" and then fail to apply.

I've rediffed it, will test sometime without the existing sed. Not
sure if that will be tonight. Will reply separately for lame.

ĸen

[Ken Moffat]

Post by Ken Moffat

Post by Pierre Labastie
Have you tried Fernando's solution?
(http://lists.linuxfromscratch.org/pipermail/blfs-dev/2014-April/027263.html).
That is for gst-plugins-base, but I guess it may be adapted to lame.
Pierre

Thanks for the pointers. I thought both of these issues had been
addressed by what is currently in the book. For gst-plugins-base I
had to apply fernando's patch manually (I think I sorted out the
wrapping caused by putting it in the body of the mail, but it did
not apply - the block seems to be a few lines earlier in my source,
and that was enough for it to be treated as "reversal - apply
anyway?" and then fail to apply.
I've rediffed it, will test sometime without the existing sed. Not
sure if that will be tonight. Will reply separately for lame.

Realised I could test these while leaving the box to run. They
both build (no idea if the runtime is acceptable, and I probably
won't be in a position to test that for several days.

I'll attach the rediffed patch.

Äžen

[Ken Moffat]

Post by Ken Moffat
Realised I could test these while leaving the box to run. They
both build (no idea if the runtime is acceptable, and I probably
won't be in a position to test that for several days.
I'll attach the rediffed patch.

Also needed for 1.2.4. The same patch applies, but with fuzz 1 and
an enormous offset which causes it to create configure.orig. I've
got a clean version, if wanted.

The 1.2.4 version seems ok (parole is working), for 0.10 I'm less
sure - almost every video at youtube which I try with firefox
appears to offer me a flash version, which does not play for me,
although one gave me video but no sound. I then tried opening a
local video file from firefox - the video was fine, but again no
sound. :-(

ĸen

[Fernando de Oliveira]

Post by Ken Moffat

Post by Ken Moffat
Realised I could test these while leaving the box to run. They
both build (no idea if the runtime is acceptable, and I probably
won't be in a position to test that for several days.
I'll attach the rediffed patch.

Also needed for 1.2.4. The same patch applies, but with fuzz 1 and
an enormous offset which causes it to create configure.orig. I've
got a clean version, if wanted.
The 1.2.4 version seems ok (parole is working), for 0.10 I'm less
sure - almost every video at youtube which I try with firefox
appears to offer me a flash version, which does not play for me,
although one gave me video but no sound. I then tried opening a
local video file from firefox - the video was fine, but again no
sound. :-(
ĸen

Thanks, ĸen. I don't know. It seems that we need this solution, instead
of the one in the book. Do you prefer the patches, instead of the sed
suggested by Bruce? Also, I preferred the solution by Pierre, because
reading his post, I thought it was better in terms of speed.

Please, whatever you decide, submit the patches and fix the book, if you
do not mind.

And please, mark in the ticket that the gst-base, is partially fixed.
I'm the two versions of FF that you updated behind. No FF in the new dev
box, only Midori and the one from gnome.

I am doing something for the book and with a small headache, so would
prefer not doing it myself.

[Ken Moffat]

Post by Fernando de Oliveira

Post by Ken Moffat

Post by Ken Moffat
Realised I could test these while leaving the box to run. They
both build (no idea if the runtime is acceptable, and I probably
won't be in a position to test that for several days.
I'll attach the rediffed patch.

Also needed for 1.2.4. The same patch applies, but with fuzz 1 and
an enormous offset which causes it to create configure.orig. I've
got a clean version, if wanted.
The 1.2.4 version seems ok (parole is working), for 0.10 I'm less
sure - almost every video at youtube which I try with firefox
appears to offer me a flash version, which does not play for me,
although one gave me video but no sound. I then tried opening a
local video file from firefox - the video was fine, but again no
sound. :-(
ĸen

Thanks, ĸen. I don't know. It seems that we need this solution, instead
of the one in the book. Do you prefer the patches, instead of the sed
suggested by Bruce? Also, I preferred the solution by Pierre, because
reading his post, I thought it was better in terms of speed.

I think that Bruce suggested a sed for lame ? I'm running with
that, but have not tested it at runtime. At the moment that box is
having another attempt to backup my windoze system (realised I
needed to remove my recent large HDtracks.com downloads, and from
ntfs3g I only mount windows as r/o - talk about up/down/up/down, I'd
forgotten what a pain that is ;)

For the gst-base plugins, does "i686-gcc-4.9.0-1.patch" sound like
a decent name ?

Post by Fernando de Oliveira
Please, whatever you decide, submit the patches and fix the book, if you
do not mind.

OK.

Post by Fernando de Oliveira
And please, mark in the ticket that the gst-base, is partially fixed.
I'm the two versions of FF that you updated behind. No FF in the new dev
box, only Midori and the one from gnome.
I am doing something for the book and with a small headache, so would
prefer not doing it myself.

Fine, it won't be today, and perhaps not tomorrow. For headaches I
have no general advice, but best wishes.

ĸen

[Bruce Dubbs]

Post by Ken Moffat
For the gst-base plugins, does "i686-gcc-4.9.0-1.patch" sound like
a decent name ?

Use gcc-4.9.0-i686-1.patch

-- Bruce

[Fernando de Oliveira]

Post by Ken Moffat

Post by Fernando de Oliveira
Thanks, ĸen. I don't know. It seems that we need this solution, instead
of the one in the book. Do you prefer the patches, instead of the sed
suggested by Bruce? Also, I preferred the solution by Pierre, because
reading his post, I thought it was better in terms of speed.

Sorry. Apologies to Pierre. He was the one who suggested the sed's in
the book for gst-plugins-base.

Post by Ken Moffat
I think that Bruce suggested a sed for lame ?

Yes, but I was referring to the other ones, by Pierre.

Post by Ken Moffat
I'm running with
that, but have not tested it at runtime. At the moment that box is
having another attempt to backup my windoze system (realised I
needed to remove my recent large HDtracks.com downloads, and from
ntfs3g I only mount windows as r/o - talk about up/down/up/down, I'd
forgotten what a pain that is ;)
For the gst-base plugins, does "i686-gcc-4.9.0-1.patch" sound like
a decent name ?

For me, yes. Oh, no, by the book, should be your choice with some
replacements:


"packgnamewithpkgversion-i686_gcc_4_9_0-1.patch

Post by Ken Moffat

Post by Fernando de Oliveira
Please, whatever you decide, submit the patches and fix the book, if you
do not mind.

OK.

Post by Fernando de Oliveira
And please, mark in the ticket that the gst-base, is partially fixed.
I'm the two versions of FF that you updated behind. No FF in the new dev
box, only Midori and the one from gnome.
I am doing something for the book and with a small headache, so would
prefer not doing it myself.

Fine, it won't be today, and perhaps not tomorrow. For headaches I
have no general advice, but best wishes.

Thanks. A little better, after some paracetamol. I think I am starting
to have trouble with chocolates (eat it once or twice a week).

Post by Ken Moffat
ĸen

[Fernando de Oliveira]

Post by Fernando de Oliveira
"packgnamewithpkgversion-i686_gcc_4_9_0-1.patch

After Bruce's post:


packgnamewithpkgversion-gcc_4_9_0_i686-1.patch

[Ken Moffat]

Post by Fernando de Oliveira

Post by Fernando de Oliveira
"packgnamewithpkgversion-i686_gcc_4_9_0-1.patch

packgnamewithpkgversion-gcc_4_9_0_i686-1.patch

Yes, how could I forget that the internal parts should be
underscores ?

ĸen