Mailing List Archive

Will there be any consequences for people that is already using gcc 3.4?


On Sat, 11 Sep 2004 13:43:57 -0500, Daniel Goller <morfic@gentoo.org> wrote:
> GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
> profiles anymore.
> There are no critical packages that dont compile with it.
> If anyone sees a reason to not unmask it, let me know, otherwise we
> would like to unamsk it soon.
>
> --
> gentoo-dev@gentoo.org mailing list
>
>



--
Ioannis Aslanidis
<deathwing00[at]forums.gentoo.org> 0xC2539DA3
<aioannis[at]tinet.org> 0xF202D067
<dwcommander[at]users.sourceforge.net>

Hellenic Gentoo GNU/Linux project manager (http://hellenicgentoo.sf.net)
FIRECOPS++ project manager (http://firecops.sf.net)

Computer Engineering student at Universitat Rovira i Virgili

--
gentoo-dev@gentoo.org mailing list

On Sat, 2004-09-11 at 14:43, Daniel Goller wrote:
> GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
> profiles anymore.
> There are no critical packages that dont compile with it.
> If anyone sees a reason to not unmask it, let me know, otherwise we
> would like to unamsk it soon.

Ideally I'd like to see us get all arches that can use gcc-3.4 on it and
in stable for the next release cycle. So this ~x86 unmask in profiles
has my blessing. Works best when used in tandem with >=binutils-2.15
(feature wise)

Users that experience failures of gcc/binutils/libc compiling itself
should send any bugs toolchain@g.o
Usually this is only limited the users with buggy athlon{,-xp} hardware
that ever seem to have problems.

Programs that fail to compile with 3.4.x that previously compiled with
3.3.4 should be routed to gcc-porting@g.o and package maintainers
respectively.

>
>
> --
> gentoo-dev@gentoo.org mailing list
--
Ned Ludd <solar@gentoo.org>
Gentoo (hardened,security,infrastructure,embedded,toolchain) Developer

your impact would be to no longer need a special profile
if it works now, it will work then



Ioannis Aslanidis wrote:

>Will there be any consequences for people that is already using gcc 3.4?
>
>
>On Sat, 11 Sep 2004 13:43:57 -0500, Daniel Goller <morfic@gentoo.org> wrote:
>
>
>>GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
>>profiles anymore.
>>There are no critical packages that dont compile with it.
>>If anyone sees a reason to not unmask it, let me know, otherwise we
>>would like to unamsk it soon.
>>
>>--
>>gentoo-dev@gentoo.org mailing list
>>
>>
>>
>>
>
>
>
>
>

--
gentoo-dev@gentoo.org mailing list

and blessed be my work

Ned Ludd wrote:

>On Sat, 2004-09-11 at 14:43, Daniel Goller wrote:
>
>
>>GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
>>profiles anymore.
>>There are no critical packages that dont compile with it.
>>If anyone sees a reason to not unmask it, let me know, otherwise we
>>would like to unamsk it soon.
>>
>>
>
>Ideally I'd like to see us get all arches that can use gcc-3.4 on it and
>in stable for the next release cycle. So this ~x86 unmask in profiles
>has my blessing. Works best when used in tandem with >=binutils-2.15
>(feature wise)
>
>Users that experience failures of gcc/binutils/libc compiling itself
>should send any bugs toolchain@g.o
>Usually this is only limited the users with buggy athlon{,-xp} hardware
>that ever seem to have problems.
>
>Programs that fail to compile with 3.4.x that previously compiled with
>3.3.4 should be routed to gcc-porting@g.o and package maintainers
>respectively.
>
>
>
>>--
>>gentoo-dev@gentoo.org mailing list
>>
>>

--
gentoo-dev@gentoo.org mailing list

El Sábado, 11 de Septiembre de 2004 20:43, Daniel Goller escribió:
> GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
> profiles anymore.
> There are no critical packages that dont compile with it.
> If anyone sees a reason to not unmask it, let me know, otherwise we
> would like to unamsk it soon.

IMHO, GCC 3.4 shouldn't go ~x86 until openoffice.org can be built with it.
Just my 2 cents.

Regards,
Alberto

--
/* Alberto García Hierro (Skyhusker) */

If you're bothered, check out OpenOffice 1.9.52. That builds fine. Don't
forget, every GCC 3.x for new x hasn't compiled OpenOffice when it went
to ~x86. 3.3 hit and 3.2 was needed. 3.2 hit and it didn't compile at
all. 3.1 was pre-OOo I think. We're not doing anything out of the ordinary.

However, I'd be inclined to suggest that perhaps a 1.9.52 OOo ebuild,
package.mask'd and -* keyworded, should perhaps be committed for testing
purposes. Any objections? If not, I'll make the tarball and ebuild - it
works here but my ebuild is filthy. Also when 2.0 is released the 64-bit
fixes will be merged too.

Alberto Garcia Hierro wrote:
> El Sábado, 11 de Septiembre de 2004 20:43, Daniel Goller escribió:
>
>>GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
>>profiles anymore.
>>There are no critical packages that dont compile with it.
>>If anyone sees a reason to not unmask it, let me know, otherwise we
>>would like to unamsk it soon.
>
>
> IMHO, GCC 3.4 shouldn't go ~x86 until openoffice.org can be built with it.
> Just my 2 cents.
>
> Regards,
> Alberto
>


--
gentoo-dev@gentoo.org mailing list

Alberto Garcia Hierro wrote:
> El Sábado, 11 de Septiembre de 2004 20:43, Daniel Goller escribió:
> > GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
> > profiles anymore.
>
> IMHO, GCC 3.4 shouldn't go ~x86 until openoffice.org can be built with it.

And sun-j2sdk...

Regards,
Norberto

--
gentoo-dev@gentoo.org mailing list

Daniel Goller wrote:

> GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
profiles anymore.
> There are no critical packages that dont compile with it.
> If anyone sees a reason to not unmask it, let me know, otherwise we
would like to unamsk it soon.


Thoughts/comments on bug #57602?
http://bugs.gentoo.org/show_bug.cgi?id=57602

It seems to affect many/most using xorg-x11 with -march=pentium4.
There's a pretty simple fix, but no one from the X11 herd seems to have
taken an interest yet. I'll update the bug ASAP with some more test
results. I welcome any input from others who have compiled xorg-x11 on a
P4 with gcc-3.4.

~ Patrick Dawson



--
gentoo-dev@gentoo.org mailing list

On Sunday 12 September 2004 07:49 pm, Patrick Dawson wrote:
> It seems to affect many/most using xorg-x11 with -march=pentium4.

it's sse2 code, not pentium4
-mike

--
gentoo-dev@gentoo.org mailing list

added -mno-sse2 conditionally to xorg CFLAGS with gcc 3.4 till this is
sorted out, since we have pentium4, pentium-m and amd64 users with
problems and all three of those cpus have sse2



Mike Frysinger wrote:

>On Sunday 12 September 2004 07:49 pm, Patrick Dawson wrote:
>
>
>>It seems to affect many/most using xorg-x11 with -march=pentium4.
>>
>>
>
>it's sse2 code, not pentium4
>-mike
>
>--
>gentoo-dev@gentoo.org mailing list
>
>
>
>

--
gentoo-dev@gentoo.org mailing list

On Monday 13 September 2004 12:33 am, Daniel Goller wrote:
> added -mno-sse2 conditionally to xorg CFLAGS with gcc 3.4 till this is
> sorted out, since we have pentium4, pentium-m and amd64 users with
> problems and all three of those cpus have sse2

xorg working fine w/ amd64 here... reading the comments, it seems that this is
a nVidia-binary-related bug, not really so much an xorg bug. Perhaps somehow
make it conditional on usage of such drivers?

>
> Mike Frysinger wrote:
> >On Sunday 12 September 2004 07:49 pm, Patrick Dawson wrote:
> >>It seems to affect many/most using xorg-x11 with -march=pentium4.
> >
> >it's sse2 code, not pentium4
> >-mike
--
Luke-Jr
Developer, Utopios
http://utopios.org/

i admit i dont know how to check for those, if i would it was a none
issue, from what i read i agree that it is binary driver issue, i talked
with seemant about it, but don't want to work on his ebuild more than
adding the line spanky suggested from 56702
this will keep things sane till he comes up with a permanent fix

or more likely till nvidia comes out with newer drivers , everyone jumps
on those and problem goes magically away

binaries are always fun



Luke-Jr wrote:

>On Monday 13 September 2004 12:33 am, Daniel Goller wrote:
>
>
>>added -mno-sse2 conditionally to xorg CFLAGS with gcc 3.4 till this is
>>sorted out, since we have pentium4, pentium-m and amd64 users with
>>problems and all three of those cpus have sse2
>>
>>
>
>xorg working fine w/ amd64 here... reading the comments, it seems that this is
>a nVidia-binary-related bug, not really so much an xorg bug. Perhaps somehow
>make it conditional on usage of such drivers?
>
>
>
>>Mike Frysinger wrote:
>>
>>
>>>On Sunday 12 September 2004 07:49 pm, Patrick Dawson wrote:
>>>
>>>
>>>>It seems to affect many/most using xorg-x11 with -march=pentium4.
>>>>
>>>>
>>>it's sse2 code, not pentium4
>>>-mike
>>>
>>>

--
gentoo-dev@gentoo.org mailing list

On Sun, 2004-09-12 at 21:30, Daniel Goller wrote:
> i admit i dont know how to check for those, if i would it was a none
> issue, from what i read i agree that it is binary driver issue, i talked
> with seemant about it, but don't want to work on his ebuild more than
> adding the line spanky suggested from 56702
> this will keep things sane till he comes up with a permanent fix
>
> or more likely till nvidia comes out with newer drivers , everyone jumps
> on those and problem goes magically away
>
> binaries are always fun
I'm planning on giving NVidia a call tomorrowish to get a feel for when
we can expect new binary drivers that are friendlier for sse2 + gcc-3.4
--
Seemant Kulleen
http://dev.gentoo.org/~seemant

Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x3458780E
Key fingerprint = 23A9 7CB5 9BBB 4F8D 549B 6593 EDA2 65D8 3458 780E

On Sunday 12 September 2004 02:24, Robert Moss wrote:
> If you're bothered, check out OpenOffice 1.9.52. That builds fine. Don't
> forget, every GCC 3.x for new x hasn't compiled OpenOffice when it went
> to ~x86. 3.3 hit and 3.2 was needed. 3.2 hit and it didn't compile at
> all. 3.1 was pre-OOo I think. We're not doing anything out of the ordinary.
>
> However, I'd be inclined to suggest that perhaps a 1.9.52 OOo ebuild,
> package.mask'd and -* keyworded, should perhaps be committed for testing
> purposes. Any objections? If not, I'll make the tarball and ebuild - it
> works here but my ebuild is filthy. Also when 2.0 is released the 64-bit
> fixes will be merged too.

If you could post them on bugzilla (not the tarball of course, but a link)
that would be nice. Assign them to openoffice@gentoo.org

Paul

--
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

On Monday 13 September 2004 05:51, Luke-Jr wrote:
> On Monday 13 September 2004 12:33 am, Daniel Goller wrote:
> > added -mno-sse2 conditionally to xorg CFLAGS with gcc 3.4 till this is
> > sorted out, since we have pentium4, pentium-m and amd64 users with
> > problems and all three of those cpus have sse2
>
> xorg working fine w/ amd64 here... reading the comments, it seems that this
> is a nVidia-binary-related bug, not really so much an xorg bug. Perhaps
> somehow make it conditional on usage of such drivers?

On my amd64 system (in i386 mode) I have a problem where I also have
reproduceable bugs that cause instant crashes for xorg-x11. A popular cause
is opening certain webpages in galeon (gtk based) while in konqueror it is
not a problem. I strongly suspect that something font-related is the cause.
Possibly in connection with sse2 and gcc-3.4.1. I'm trying to compile without
sse2 support now, and report whether that stops the bugs.

Paul

--
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

On Sunday 12 September 2004 16:49, Patrick Dawson wrote:
...
> Thoughts/comments on bug #57602?
> http://bugs.gentoo.org/show_bug.cgi?id=57602
>
> It seems to affect many/most using xorg-x11 with -march=pentium4.
> There's a pretty simple fix, but no one from the X11 herd seems to have
> taken an interest yet. I'll update the bug ASAP with some more test
> results. I welcome any input from others who have compiled xorg-x11 on a
> P4 with gcc-3.4.

Does anyone know if gcc 3.4.2 fixes these problems? I see there's a new
ebuild for it in portage...

--
Joshua J. Berry
Gentoo Linux -- GLSA Coordinator, Security Team
http://security.gentoo.org -- condordes@gentoo.org

On Monday 13 September 2004 03:20 pm, Paul de Vrieze wrote:
> A popular cause
> is opening certain webpages in galeon (gtk based) while in konqueror it is
> not a problem.

yeah i use pan to test it ... just open it up and click on some group names

also found clicking e-mails in kmail triggers it too
-mike

--
gentoo-dev@gentoo.org mailing list

On Monday 13 September 2004 03:54 pm, Joshua J. Berry wrote:
> Does anyone know if gcc 3.4.2 fixes these problems? I see there's a new
> ebuild for it in portage...

ive only tested with 3.4.1 but since i already have 3.4.2 on my system i'll
try rebuilding it now
-mike

--
gentoo-dev@gentoo.org mailing list

Arse, I will see if I can get round to fixing it. With luck it might not
be too bad!

Stefan

On Sun, 2004-09-12 at 08:25, Norberto Bensa wrote:
> Alberto Garcia Hierro wrote:
> > El Sábado, 11 de Septiembre de 2004 20:43, Daniel Goller escribió:
> > > GCC 3.4 should be ready to go ~x86 w/o being profile masked in x86
> > > profiles anymore.
> >
> > IMHO, GCC 3.4 shouldn't go ~x86 until openoffice.org can be built with it.
>
> And sun-j2sdk...
>
> Regards,
> Norberto
>
> --
> gentoo-dev@gentoo.org mailing list


--
gentoo-dev@gentoo.org mailing list

Stefan Jones wrote:
> Arse, I will see if I can get round to fixing it. With luck it might not
> be too bad!

Thanks Stefan.

> On Sun, 2004-09-12 at 08:25, Norberto Bensa wrote:
> > Alberto Garcia Hierro wrote:
> > > IMHO, GCC 3.4 shouldn't go ~x86 until openoffice.org can be built with
> > > it.
> >
> > And sun-j2sdk...

I really should try to compile it again now with gcc 3.4.2... I'll leave the
box doing that while I sleep.

Regards,
Norberto

--
gentoo-dev@gentoo.org mailing list

On Tuesday 14 September 2004 00:00, Mike Frysinger wrote:
> On Monday 13 September 2004 03:20 pm, Paul de Vrieze wrote:
> > A popular cause
> > is opening certain webpages in galeon (gtk based) while in konqueror
> > it is not a problem.
>
> yeah i use pan to test it ... just open it up and click on some group
> names
>
> also found clicking e-mails in kmail triggers it too
> -mike

Sometimes indeed. It's on an ati card (with the xorg drivers) so no closed
source involved.

Paul

--
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

On Monday 13 September 2004 06:02 pm, Mike Frysinger wrote:
> On Monday 13 September 2004 03:54 pm, Joshua J. Berry wrote:
> > Does anyone know if gcc 3.4.2 fixes these problems? I see there's a new
> > ebuild for it in portage...
>
> ive only tested with 3.4.1 but since i already have 3.4.2 on my system i'll
> try rebuilding it now

3.4.2 makes no difference
-mike

--
gentoo-dev@gentoo.org mailing list

On Tuesday 14 September 2004 00:00, Mike Frysinger wrote:
> On Monday 13 September 2004 03:20 pm, Paul de Vrieze wrote:
> > A popular cause
> > is opening certain webpages in galeon (gtk based) while in konqueror
> > it is not a problem.
>
> yeah i use pan to test it ... just open it up and click on some group
> names
>
> also found clicking e-mails in kmail triggers it too
> -mike

It seems that adding -fno-sse2 to the compile flags fixed it. This would
imply that it is definately a compiler bug.

Paul

--
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

On Wednesday 15 September 2004 04:21 am, Paul de Vrieze wrote:
> It seems that adding -fno-sse2 to the compile flags fixed it. This would
> imply that it is definately a compiler bug.

we knew that already :)
it's been added to the xorg ebuild already as a workaround
-mike

--
gentoo-dev@gentoo.org mailing list