On 23.2.2010 10:00, Andreas Jaeger wrote:

The current kernel in openSUSE Factory (rc7 based) does not boot on my laptop, it fails very early in the boot sequence.

Updating to kernel head instead works for me. I propose to submit a new kernel to openSUSE:Factory asap.

Current kernel-xen fails to build on Factory :( /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:179:26: error: variably modified 'tx_skbs' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:180:26: error: variably modified 'rx_skbs' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:184:27: error: variably modified 'grant_tx_ref' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:186:27: error: variably modified 'grant_rx_ref' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:193:29: error: variably modified 'rx_pfn_array' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:194:32: error: variably modified 'rx_mcl' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront/netfront.h:195:27: error: variably modified 'rx_mmu' at file scope (https://build.opensuse.org/package/live_build_log?arch=x86_64&package=kernel-xen&project=Kernel%3AHEAD&repository=openSUSE_Factory). Michal -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

On Tuesday 23 February 2010 10:38:23 Jan Beulich wrote:

Has this been pointed out to the gcc guys? That code hasn't changed for a very long time...

Let's ask the Richard and Michael, Andreas

Jan

...

...

...

Michal Marek 23.02.10 10:16 >>>

On 23.2.2010 10:00, Andreas Jaeger wrote:

...

The current kernel in openSUSE Factory (rc7 based) does not boot on my laptop, it fails very early in the boot sequence.

Updating to kernel head instead works for me. I propose to submit a new kernel to openSUSE:Factory asap.

Current kernel-xen fails to build on Factory :( /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:179:26: error: variably modified 'tx_skbs' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:180:26: error: variably modified 'rx_skbs' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:184:27: error: variably modified 'grant_tx_ref' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:186:27: error: variably modified 'grant_rx_ref' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:193:29: error: variably modified 'rx_pfn_array' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:194:32: error: variably modified 'rx_mcl' at file scope /usr/src/packages/BUILD/kernel-xen-2.6.33/linux-2.6.32/drivers/xen/netfront /netfront.h:195:27: error: variably modified 'rx_mmu' at file scope (https://build.opensuse.org/package/live_build_log?arch=x86_64&package=kern el-xen&project=Kernel%3AHEAD&repository=openSUSE_Factory).

Michal

-- 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

On 02/23/2010 04:00 AM, Andreas Jaeger wrote:

The current kernel in openSUSE Factory (rc7 based) does not boot on my laptop, it fails very early in the boot sequence.

Updating to kernel head instead works for me. I propose to submit a new kernel to openSUSE:Factory asap.

FYI: I see Kernel panic: Fatal exception in interrupt - and the rest is scrolled out of the screen and therefore not readable.

This is on my Thinkpad x86-64 laptop with kernel-desktop,

Ideally I'd like to see Kernel:HEAD pushed to Factory every time we have a successful build across all arches and flavors. Right now I'm not sure how do automate that. -Jeff -- Jeff Mahoney SUSE Labs -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

On Tue, 23 Feb 2010, Jeff Mahoney wrote:

Another user reported a bug on this: https://bugzilla.novell.com/show_bug.cgi?id=582222

He blames the gcc 4.5 switch but I haven't tracked it down yet.

I can't help wondering why we adopt new compilers before they even carved their release branch upstream. Are we in the for the extra pain? Doing as we do is certainly beneficial for GCC, but I have a hard time seeing the benefit for openSUSE. By the way, is ftp://ftp.suse.com/pub/projects/kernel/kotd/HEAD/x86_64/ being behind by six days related to this issue, or is it something else? Would be nice to have that one more up-to-date. Gerald -- Dr. Gerald Pfeifer gp@novell.com | SUSE Linux Products GmbH Director Product Management | HRB 16746 (AG Nuremberg) SUSE Linux Enterprise, openSUSE, Appliances | GF Markus Rex -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 02/23/2010 03:32 PM, Jeff Mahoney wrote:

On 02/23/2010 11:07 AM, Jeff Mahoney wrote:

...

On 02/23/2010 04:00 AM, Andreas Jaeger wrote:

...

The current kernel in openSUSE Factory (rc7 based) does not boot on my laptop, it fails very early in the boot sequence.

...

...

Updating to kernel head instead works for me. I propose to submit a new kernel to openSUSE:Factory asap.

...

...

FYI: I see Kernel panic: Fatal exception in interrupt - and the rest is scrolled out of the screen and therefore not readable.

...

...

This is on my Thinkpad x86-64 laptop with kernel-desktop,

...

Another user reported a bug on this: https://bugzilla.novell.com/show_bug.cgi?id=582222

...

He blames the gcc 4.5 switch but I haven't tracked it down yet.

Sure enough, identical source built with gcc 4.4 boots but not when built with gcc 4.5.

I'll try to track down why.

I've done some analysis and posted the results to the bug report. I'd like to continue the discussion there. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.13 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/ iEYEARECAAYFAkuEVoYACgkQLPWxlyuTD7LevwCfUorGrIRa6Xu320gUb30iYTl8 wkAAoJ0ETS2eQ1shxY3CVhZxrPf20l9x =ak0f -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

Am Dienstag 23 Februar 2010 schrieb Gerald Pfeifer:

On Tue, 23 Feb 2010, Jeff Mahoney wrote:

...

Another user reported a bug on this: https://bugzilla.novell.com/show_bug.cgi?id=582222

He blames the gcc 4.5 switch but I haven't tracked it down yet.

I can't help wondering why we adopt new compilers before they even carved their release branch upstream. Are we in the for the extra pain? Doing as we do is certainly beneficial for GCC, but I have a hard time seeing the benefit for openSUSE.

We switch when gcc team thinks they are ready to release (very often gcc releases are delayed by politics) and the openSUSE release cycle fits. Having released Milestone2 was a good opportunity. Greetings, Stephan -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

On Wed, 24 Feb 2010, Michael Matz wrote:

...

I can't help wondering why we adopt new compilers before they even carved their release branch upstream. Are we in the for the extra pain? Without trying the compiler we won't know of any bugs in there. The extra pain would have been mostly the same had we switched later, as most problems were in the packages, not the compiler.

That would be the case if we were the only distro out there. If we only switch, and I am not suggesting this of course, to GCC 4.5 in two years, I bet that all projects still alive upstream will have been fixed. ;-)

And even the problems in the compiler wouldn't have been found by then if nobody tries it.

Absolutely. The openSUSE Build Service is great for this kind of stuff. When it gets tricky is when a broken kernel may either be due to the kernel being broken genuinely or by the compiler, and you don't know which is the case. (And, yes, I had to rescue boot my workstation and install an older kernel the other day.) Gerald -- Dr. Gerald Pfeifer gp@novell.com | SUSE Linux Products GmbH Director Product Management | HRB 16746 (AG Nuremberg) SUSE Linux Enterprise, openSUSE, Appliances | GF Markus Rex -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 02/24/2010 11:31 AM, Michael Matz wrote:

Hi,

On Wed, 24 Feb 2010, Gerald Pfeifer wrote:

...

...

And even the problems in the compiler wouldn't have been found by then if nobody tries it.

Absolutely. The openSUSE Build Service is great for this kind of stuff.

When it gets tricky is when a broken kernel may either be due to the kernel being broken genuinely or by the compiler, and you don't know which is the case.

It's not different if any other component on which you rely is broken for whatever reason. What specifically are you asking for? That your system please be not broken? Ever? Or something else?

Ciao, Michael. P.S: it's all a red-herring anyway, the non-booting kernel is in fact a kernel bug, not a GCC one, go figure.

Agreed. A warning would have been nice, though. I definitely wouldn't have tracked that one down to a missing align statement without your help. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.13 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/ iEYEARECAAYFAkuFVfQACgkQLPWxlyuTD7IdjACeMR4PsoQv2+PWAVz6PygHSDBT nGIAnAj5ozqRIQntpI6I2MYnSx5gE7ue =DH9/ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 02/24/2010 11:44 AM, Michael Matz wrote:

Hi,

On Wed, 24 Feb 2010, Jeff Mahoney wrote:

...

Agreed. A warning would have been nice, though. I definitely wouldn't have tracked that one down to a missing align statement without your help.

Any suggestion what the warning should say? We probably can cook something up, though I fear that it might trigger too often to be appreciated by users :)

How about: Warning: struct xyz declared with multiple alignment constraints, using the largest (#) The documentation can explain more. Warning: struct xyz declared with multiple alignment constraints. GCC will align structures using the largest alignment specified. The default alignment is 32 bytes. When smaller alignment constraints are specified with the __aligned__(<size>) attribute, this alignment is used. If there are mixed __aligned__ attributes, or if there is a declaration left using the default, this may cause issues with code that depends on the structures being aligned to one of the smaller values. What "normal" scenarios would cause this to trigger? If they're frequent then you might be right about it triggering too often. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.13 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/ iEYEARECAAYFAkuFWgsACgkQLPWxlyuTD7J2+gCePcsYnQn3SZ+J92XcO2V6d3vI P3IAoKiBSTgCjX+8M/PrNbEUXydezN52 =2IN4 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

Am Donnerstag 25 Februar 2010 schrieb Gerald Pfeifer:

On Wed, 24 Feb 2010, Michael Matz wrote:

...

It's not different if any other component on which you rely is broken for whatever reason. What specifically are you asking for? That your system please be not broken? Ever? Or something else?

We should have moved this one to -factory, I guess, but without context that's hard. My point is that there is a difference between using a .0 version even or using some snapshot, and especially for a critical tool like the system compiler the question is whether the benefit for openSUSE outweights the risk/pain/cost -- again, for openSUSE. You know that GCC is near and dear to my heart, and from the perspective I think it's cool.

Again: the benefit for openSUSE is the time we win by doing it now when there are tons of bugs anyway and not when we want to stabilize and then are hit much harder by unbootable systems all around. Greetings, Stephan -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org