glibc 2.5+

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

glibc 2.5+

Davide Cittaro-2
Hi all, I've found that glibc 2.5+ cannot be compiled on my alpha, as reported on gentoo bugzilla by another person.


I have a couple of questions, then:
1- it is safe to use '-nptlonly' flag?
2- when the patched glibc will be available in portage (comment 2 of that bug has been posted  one week ago)?

d

/*
Davide Cittaro
HPC and Bioinformatics Systems @ Informatics Core

IFOM - Istituto FIRC di Oncologia Molecolare
via adamello, 16
20139 Milano
Italy

tel.: +39(02)574303007
*/


Reply | Threaded
Open this post in threaded view
|

Re: glibc 2.5+

Bryan Østergaard
On Mon, May 28, 2007 at 04:01:15PM +0200, Davide Cittaro wrote:
> Hi all, I've found that glibc 2.5+ cannot be compiled on my alpha, as  
> reported on gentoo bugzilla by another person.
>
> http://bugs.gentoo.org/show_bug.cgi?id=179353
>
> I have a couple of questions, then:
> 1- it is safe to use '-nptlonly' flag?
Not really. In the past we've found Linuxthreads to be unstable and
nptlonly is much preferred for that reason (we had programs randomly
crashing with Linuxthreads and no problems with NPTL).

> 2- when the patched glibc will be available in portage (comment 2 of  
> that bug has been posted  one week ago)?
As soon as glibc upstream fixes the problem. Meanwhile you have two
different options.

1. Upgrade to binutils >=2.17.50.0.15 and risk possible unknown bugs
from doing so.
2. Ignoring glibc-2.5 updates until upstream fixes the issue.

My recommendation would be to wait and not try to work around the issue
yourself.

Regards,
Bryan Østergaard
--
[hidden email] mailing list

Reply | Threaded
Open this post in threaded view
|

Re: glibc 2.5+

Davide Cittaro-2

On May 28, 2007, at 5:06 PM, Bryan Østergaard wrote:

On Mon, May 28, 2007 at 04:01:15PM +0200, Davide Cittaro wrote:
Hi all, I've found that glibc 2.5+ cannot be compiled on my alpha, as  
reported on gentoo bugzilla by another person.


I have a couple of questions, then:
1- it is safe to use '-nptlonly' flag?
Not really. In the past we've found Linuxthreads to be unstable and
nptlonly is much preferred for that reason (we had programs randomly
crashing with Linuxthreads and no problems with NPTL).


After 1 hour of building I've discovered that glibc doesn't compile even without nptlonly flag

2- when the patched glibc will be available in portage (comment 2 of  
that bug has been posted  one week ago)?
As soon as glibc upstream fixes the problem. Meanwhile you have two
different options.

1. Upgrade to binutils >=2.17.50.0.15 and risk possible unknown bugs
from doing so.
2. Ignoring glibc-2.5 updates until upstream fixes the issue.

My recommendation would be to wait and not try to work around the issue
yourself.

Unfortunately I have strict times and this general system update has been scheduled for this week. Since glibc-2.5-r2 is marked stable on alpha I thought I wouldn't run into such issues... :-(
If I upgrade binutils can I use both nptl* flags?

d

/*
Davide Cittaro
HPC and Bioinformatics Systems @ Informatics Core

IFOM - Istituto FIRC di Oncologia Molecolare
via adamello, 16
20139 Milano
Italy

tel.: +39(02)574303007
*/


Reply | Threaded
Open this post in threaded view
|

Re: glibc 2.5+

Bryan Østergaard
On Mon, May 28, 2007 at 05:11:56PM +0200, Davide Cittaro wrote:

>
> On May 28, 2007, at 5:06 PM, Bryan Østergaard wrote:
>
> >On Mon, May 28, 2007 at 04:01:15PM +0200, Davide Cittaro wrote:
> >>Hi all, I've found that glibc 2.5+ cannot be compiled on my alpha, as
> >>reported on gentoo bugzilla by another person.
> >>
> >>http://bugs.gentoo.org/show_bug.cgi?id=179353
> >>
> >>I have a couple of questions, then:
> >>1- it is safe to use '-nptlonly' flag?
> >Not really. In the past we've found Linuxthreads to be unstable and
> >nptlonly is much preferred for that reason (we had programs randomly
> >crashing with Linuxthreads and no problems with NPTL).
> >
>
> After 1 hour of building I've discovered that glibc doesn't compile  
> even without nptlonly flag
>
> >>2- when the patched glibc will be available in portage (comment 2 of
> >>that bug has been posted  one week ago)?
> >As soon as glibc upstream fixes the problem. Meanwhile you have two
> >different options.
> >
> >1. Upgrade to binutils >=2.17.50.0.15 and risk possible unknown bugs
> >from doing so.
> >2. Ignoring glibc-2.5 updates until upstream fixes the issue.
> >
> >My recommendation would be to wait and not try to work around the  
> >issue
> >yourself.
>
> Unfortunately I have strict times and this general system update has  
> been scheduled for this week. Since glibc-2.5-r2 is marked stable on  
> alpha I thought I wouldn't run into such issues... :-(
> If I upgrade binutils can I use both nptl* flags?
>
I'd still recommend putting this particular update off tbh. I would have
stabled binutils-2.17.50.16 already if I was sure it wouldn't cause
other problems. As it is we really haven't tested it well enough to
stable it yet and you can't downgrade binutils again if it turns out to
be a bad idea running .2.17.50.*.

That said, if you really really want to go through with this then
2.17.50.16 does solve the glibc compilation problems with nptlonly.

Regards,
Bryan Østergaard
--
[hidden email] mailing list

Reply | Threaded
Open this post in threaded view
|

Re: glibc 2.5+

Davide Cittaro-2
Hi again
On May 28, 2007, at 5:49 PM, Bryan Østergaard wrote:

That said, if you really really want to go through with this then

2.17.50.16 does solve the glibc compilation problems with nptlonly.



Hey, it seems that binutils 2.17.50.16 and glibc-2.5-r3 are pretty stable on this machine  (ES40)!
Ok, I know it is risky :-) BTW many thanks for this hint.

d

/*
Davide Cittaro
HPC and Bioinformatics Systems @ Informatics Core

IFOM - Istituto FIRC di Oncologia Molecolare
via adamello, 16
20139 Milano
Italy

tel.: +39(02)574303007
*/