[PATCH] glibc_2.9: disable parallel make

[PATCH] glibc_2.9: disable parallel make

[Steffen Sledz]

Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
(e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
with various errors. :(

Signed-off-by: Steffen Sledz <sledz@dresearch.de>
---
 recipes/glibc/glibc_2.9.bb |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/recipes/glibc/glibc_2.9.bb b/recipes/glibc/glibc_2.9.bb
index cddae98..428af03 100644
--- a/recipes/glibc/glibc_2.9.bb
+++ b/recipes/glibc/glibc_2.9.bb
@@ -75,6 +75,8 @@ SRC_URI_append_sh4 = " file://no-z-defs.patch;patch=1"
 #powerpc patches to add support for soft-float
 SRC_URI_append_powerpc= " file://powerpc-sqrt-hack.diff;patch=1"
 
+PARALLEL_MAKE = " "
+
 S = "${WORKDIR}/glibc-${PV}"
 B = "${WORKDIR}/build-${TARGET_SYS}"
 
-- 
1.6.4.2

Re: [PATCH] glibc_2.9: disable parallel make

[Stanislav Brabec]

Steffen Sledz wrote:
> Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
> (e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
> with various errors. :(

It would make bootstrap longer on multi-CPU machines. :-(

I never seen such failure with PARALLEL_MAKE = "-j 8" and
BB_NUMBER_THREADS = "8", 8x Xeon, at least with glibc-2.9-r35.2 and not
before 4249269b.

Does it happen in the same directory every time? Is there any common
error or missing library?

-- 
Stanislav Brabec
http://www.penguin.cz/~utx/zaurus

Re: [PATCH] glibc_2.9: disable parallel make

[Steffen Sledz]

Stanislav Brabec wrote:
> Steffen Sledz wrote:
>> Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
>> (e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
>> with various errors. :(
> 
> It would make bootstrap longer on multi-CPU machines. :-(

Yes, i know. But i found no other solution for it. :(

> I never seen such failure with PARALLEL_MAKE = "-j 8" and
> BB_NUMBER_THREADS = "8", 8x Xeon, at least with glibc-2.9-r35.2 and not
> before 4249269b.
> 
> Does it happen in the same directory every time? Is there any common
> error or missing library?

Sorry, no.

See thread
<http://thread.gmane.org/gmane.comp.handhelds.openembedded/26781>

Steffen

Re: [PATCH] glibc_2.9: disable parallel make

[Steffen Sledz]

Steffen Sledz wrote:
> Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
> (e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
> with various errors. :(
> 
> Signed-off-by: Steffen Sledz <sledz@dresearch.de>
> ...

BTW: It is the first time i proposed a patch this way and i expected
to see it in the patchwork queue but it doesn't. :(

Steffen

Re: [PATCH] glibc_2.9: disable parallel make

[Khem Raj]

On Tue, Oct 27, 2009 at 1:23 AM, Steffen Sledz <sledz@dresearch.de> wrote:
> Steffen Sledz wrote:
>> Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
>> (e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
>> with various errors. :(

I would suggest to refrain from applying this patch as it affects
people who do not suffer from this problem
and compile time for glibc is quite huge that everyone will feel the
pinch. I would still urge to
debug the real problem and fix it. May be there is something wrong
with the OS on the box or even in the
parallel build mechanism of glibc.  Or if there is a way to say 'only
disable parallel build when build machine is
"AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1" would
be ok too.

Thanks

-Khem

>>
>> Signed-off-by: Steffen Sledz <sledz@dresearch.de>
>> ...
>
> BTW: It is the first time i proposed a patch this way and i expected
> to see it in the patchwork queue but it doesn't. :(
>
> Steffen
>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH] glibc_2.9: disable parallel make

[Steffen Sledz]

Khem Raj wrote:
>>> Parallel make for glibc_2.9.bb fails on different multi-cpu build hosts
>>> (e.g. AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1)
>>> with various errors. :(
> 
> I would suggest to refrain from applying this patch as it affects
> people who do not suffer from this problem
> and compile time for glibc is quite huge that everyone will feel the
> pinch. I would still urge to
> debug the real problem and fix it. May be there is something wrong
> with the OS on the box or even in the
> parallel build mechanism of glibc.  Or if there is a way to say 'only
> disable parallel build when build machine is
> "AMD Phenom(tm) 9150e Quad-Core Processor running openSUSE 11.1" would
> be ok too.

I agree. As long as nobody else has this problem i'll not push this patch.

>> BTW: It is the first time i proposed a patch this way and i expected
>> to see it in the patchwork queue but it doesn't. :(

Any idea why this ptach suggestion is not visible in the patchwork queue?

Steffen