Glibc-ports status? Creating glibc-ports tarball for 2.13?

Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Carlos O'Donell]

Joseph,

Thank you for creating the glibc-2.13 tag an branch (3 weeks ago) for
glibc-ports!

libc-ports,

Sorry for the second email. Content is identical.

Status?
=======

What is the status of all the machines in glibc-ports with respect to
the glibc-2.13 tag in libc-ports?

Alpha - Richard?
ARM - Joseph/Phil?
HPPA - Carlos (myself): Does not build, missing patches.
M68K - AS?
MIPS - Joseph/AJ?
Power - Ryan?

I would like to know the status of the machines such that a public
2.13.N announcement sets the appropriate expectation for users of
these tarballs.

Note that there has currently been no announcement on libc-announce
about this release. My plan is to make a release announcement at some
point soon.

Tarball?
========

I was asked to create a glibc-ports 2.13 tarball. I see two courses of
action (in truth we could do both):

(a) Create a tarball from the current glibc-2.13 tag. Even if your
machine doesn't build against the upstream glibc-2.13 tarball, it's a
common place for packagers to start.

(b) Ask glibc-ports maintainers to chekin any lingering patches.
Create a new "glibc-2.13.1" release with some cherry-picks, and create
a tarball from *that* tag name for both glibc and glibc-ports.

The intent is to move towards a process where the glibc-ports tarballs
work correctly with glibc tarballs (of the same version) without
additional patches. It feels like this would make life simpler for the
packagers.

Is that a worthy or even useful goal for the community?

Otherwise the process is mostly mechanical, offering only a common
tarball for patching.

Comments?

Cheers,
Carlos.
-- 
Carlos O'Donell
Mentor Graphics / CodeSourcery
carlos@codesourcery.com
+1 (650) 331-3385 x716

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Andreas Jaeger]

On Monday, February 14, 2011 02:38:37 PM Carlos O'Donell wrote:
> Joseph,
> 
> Thank you for creating the glibc-2.13 tag an branch (3 weeks ago) for
> glibc-ports!
> 
> libc-ports,
> 
> Sorry for the second email. Content is identical.
> 
> Status?
> =======
> 
> What is the status of all the machines in glibc-ports with respect to
> the glibc-2.13 tag in libc-ports?
> 
> Alpha - Richard?
> ARM - Joseph/Phil?
> HPPA - Carlos (myself): Does not build, missing patches.
> M68K - AS?
> MIPS - Joseph/AJ?

No idea - I'm not involved with MIPS anymore.

Andreas
-- 
 Andreas Jaeger, Program Manager openSUSE,  aj@{novell.com,opensuse.org}
  Twitter: jaegerandi | Identica: jaegerandi
   SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
    Maxfeldstr. 5, 90409 Nürnberg, Germany
     GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Ryan Arnold]

On Mon, Feb 14, 2011 at 7:38 AM, Carlos O'Donell
<carlos@codesourcery.com> wrote:
> Joseph,
>
> Thank you for creating the glibc-2.13 tag an branch (3 weeks ago) for
> glibc-ports!
>
> libc-ports,
>
> Sorry for the second email. Content is identical.
>
> Status?
> =======
>
> What is the status of all the machines in glibc-ports with respect to
> the glibc-2.13 tag in libc-ports?
>
> Alpha - Richard?
> ARM - Joseph/Phil?
> HPPA - Carlos (myself): Does not build, missing patches.
> M68K - AS?
> MIPS - Joseph/AJ?
> Power - Ryan?
>
> I would like to know the status of the machines such that a public
> 2.13.N announcement sets the appropriate expectation for users of
> these tarballs.

I'll try a 4** build for the powerpc port sometime in the next few days.

Ryan

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Matt Turner]

On Mon, Feb 14, 2011 at 1:38 PM, Carlos O'Donell
<carlos@codesourcery.com> wrote:
> What is the status of all the machines in glibc-ports with respect to
> the glibc-2.13 tag in libc-ports?
>
> Alpha - Richard?

Needs a couple header-updating patches, but otherwise only trivial
changes. I'd be nice to get these into glibc-2.13, but it won't really
hurt anything if they don't get in.

It sure would be nice if we could avoid patching nearly identical
headers in nearly identical ways for each architecture...

Matt

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Carlos O'Donell]

On 2/14/2011 9:51 AM, Matt Turner wrote:
> On Mon, Feb 14, 2011 at 1:38 PM, Carlos O'Donell
> <carlos@codesourcery.com> wrote:
>> What is the status of all the machines in glibc-ports with respect to
>> the glibc-2.13 tag in libc-ports?
>>
>> Alpha - Richard?
> 
> Needs a couple header-updating patches, but otherwise only trivial
> changes. I'd be nice to get these into glibc-2.13, but it won't really
> hurt anything if they don't get in.

We could get them into glibc-2.13.1, how much time do you need?

> It sure would be nice if we could avoid patching nearly identical
> headers in nearly identical ways for each architecture...

Yes, this is quite lame. Guy Martin was working on a patch for glibc
core, but this was rejected by Ulrich. It was my hope that Guy would
instead rework the patch for libc-ports, allowing all the ports to
share similar headers.

See:
http://sourceware.org/ml/libc-alpha/2010-08/msg00030.html

Cheers,
Carlos.
-- 
Carlos O'Donell
Mentor Graphics / CodeSourcery
carlos@codesourcery.com
+1 (650) 331-3385 x716

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Andreas Schwab]

Carlos O'Donell <carlos@codesourcery.com> writes:

> M68K - AS?

Works fine modulo BFD bug.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Joseph S. Myers]

On Mon, 14 Feb 2011, Carlos O'Donell wrote:

> Status?
> =======
> 
> What is the status of all the machines in glibc-ports with respect to
> the glibc-2.13 tag in libc-ports?
> 
> Alpha - Richard?
> ARM - Joseph/Phil?
> HPPA - Carlos (myself): Does not build, missing patches.
> M68K - AS?
> MIPS - Joseph/AJ?
> Power - Ryan?

As far as I know, ARM, M68K, MIPS and Power are up to date with all 
relevant libc changes.

> (a) Create a tarball from the current glibc-2.13 tag. Even if your
> machine doesn't build against the upstream glibc-2.13 tarball, it's a
> common place for packagers to start.
> 
> (b) Ask glibc-ports maintainers to chekin any lingering patches.
> Create a new "glibc-2.13.1" release with some cherry-picks, and create
> a tarball from *that* tag name for both glibc and glibc-ports.

Yes, that seems logical.  Sometimes libc release branches get new symbols 
in bits/ headers cherry-picked to them, in which case the corresponding 
ports changes should also be cherry-picked to those branches.  And of 
course ports changes to bring ports up to date with libc 2.13 will need to 
go on master and 2.13 branch in any case.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Guy Martin]

> > It sure would be nice if we could avoid patching nearly identical
> > headers in nearly identical ways for each architecture...
> 
> Yes, this is quite lame. Guy Martin was working on a patch for glibc
> core, but this was rejected by Ulrich. It was my hope that Guy would
> instead rework the patch for libc-ports, allowing all the ports to
> share similar headers.

I actually tried this approach. Altho my patch wasn't the best, Joseph
made it clear that a -port only patch isn't feasible.

His comment :
"Putting a file directly in sysdeps/unix/sysv/linux/bits/socket.h in
ports also seems rather fragile; it causes the installed headers to
depend on whether the ports add-on is used or not, and for the sake of
multilib configurations such as a Power Architecture compiler
supporting both hard-float (libc) and soft-float (ports) libraries it's
desirable to avoid such differences within the same target architecture
even when it has both libc and ports variants.  As such, I don't think
it makes sense to try to change this without libc buy-in (the same
issue of different O_* values affects SPARC, it's not just a ports
issue)."
http://sourceware.org/ml/libc-ports/2010-08/msg00027.html


I've then sent a patch following his recommandations but it was
dismissed :
http://sourceware.org/ml/libc-alpha/2010-08/msg00082.html


Regards,
  Guy

Re: Glibc-ports status? Creating glibc-ports tarball for 2.13?

[Roland McGrath]

I hadn't noticed this go by before, so I'm glad Joseph rejected it.  I'd
like to make a stronger blanket statement.  No ports add-on should ever
override a sysdeps file in the main tree.  That is, it must always be a
no-op to include the ports tree as an add-on when configuring for a
non-ports architecture.  

If you want to add new shared files, then the way to do that is to put them
in a new or different sysdeps subdirectory, adding Implies files as needed.
Note that doing this would break the current situation where each port can
be an independent add-on distributed separately.  Up until now, they all
live in one repository/tree together only for convenience.  If there is a
real need to add a new file shared across ports, we should give the issue
more careful thought before deciding how to handle it.


Thanks,
Roland