Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Florian Boor]

Hi,

GIT User account schrieb:
> +		kernel_ver = bb.data.getVar("OLDEST_KERNEL", d, 1)
> +		if kernel_ver is None:
> +			qemu = "qemu-%s  -s 1048576" % target_arch
> +		else:
> +			qemu = "qemu-%s  -s 1048576 -r %s" % (target_arch, kernel_ver)

are you sure this is a good idea?
OLDEST_KERNEL defaults to "2.4.0", so this change is likely to break locale
generation for every machine that doesn't have OLDEST_KERNEL set.

Greetings

Florian

-- 
The dream of yesterday                  Florian Boor
is the hope of today                    Tel: +49 271-771091-15
and the reality of tomorrow.            Fax: +49 271-771091-19
[Robert Hutchings Goddard, 1904]        florian.boor@kernelconcepts.de
                                        http://www.kernelconcepts.de/en

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 16:15:33 Florian Boor wrote:
> Hi,
>
> GIT User account schrieb:
> > +		kernel_ver = bb.data.getVar("OLDEST_KERNEL", d, 1)
> > +		if kernel_ver is None:
> > +			qemu = "qemu-%s  -s 1048576" % target_arch
> > +		else:
> > +			qemu = "qemu-%s  -s 1048576 -r %s" % (target_arch, kernel_ver)
>
> are you sure this is a good idea?
> OLDEST_KERNEL defaults to "2.4.0", so this change is likely to break locale
> generation for every machine that doesn't have OLDEST_KERNEL set.

I've not heard yet of this really breaking something. And frankly I don't see 
any reason for that. If it actually does - feel free to shout and scream of 
course, we'll think how to handle it better.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 17:16 +0400, Roman I Khimov wrote:
> On Monday 18 May 2009 16:15:33 Florian Boor wrote:
> > Hi,
> >
> > GIT User account schrieb:
> > > +		kernel_ver = bb.data.getVar("OLDEST_KERNEL", d, 1)
> > > +		if kernel_ver is None:
> > > +			qemu = "qemu-%s  -s 1048576" % target_arch
> > > +		else:
> > > +			qemu = "qemu-%s  -s 1048576 -r %s" % (target_arch, kernel_ver)
> >
> > are you sure this is a good idea?
> > OLDEST_KERNEL defaults to "2.4.0", so this change is likely to break locale
> > generation for every machine that doesn't have OLDEST_KERNEL set.
> 
> I've not heard yet of this really breaking something. And frankly I don't see 
> any reason for that. If it actually does - feel free to shout and scream of 
> course, we'll think how to handle it better.

I guess it'd break on targets where OLDEST_KERNEL isn't set (and hence
defaults to 2.4.0) but you need a kernel newer than 2.4.0 for correct
operation.  Presumably all ARM EABI targets fall into that latter
category, since the TLS syscalls weren't in 2.4.x, and there might be
some others where glibc's default setting for --enable-kernel is
something newer than 2.4.0.

All in all I'd be happier with this change if it enforced a minimum
kernel version of 2.6.16 (i.e. what it used to be) so that there is no
possibility of versions going backwards for any target.

Incidentally, if there is a default value for OLDEST_KERNEL in
bitbake.conf then checking the retrieved value against None seems fairly
futile since that will never be true.

p.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 17:41:29 Phil Blundell wrote:
> On Mon, 2009-05-18 at 17:16 +0400, Roman I Khimov wrote:
> > On Monday 18 May 2009 16:15:33 Florian Boor wrote:
> > > Hi,
> > >
> > > GIT User account schrieb:
> > > > +		kernel_ver = bb.data.getVar("OLDEST_KERNEL", d, 1)
> > > > +		if kernel_ver is None:
> > > > +			qemu = "qemu-%s  -s 1048576" % target_arch
> > > > +		else:
> > > > +			qemu = "qemu-%s  -s 1048576 -r %s" % (target_arch, kernel_ver)
> > >
> > > are you sure this is a good idea?
> > > OLDEST_KERNEL defaults to "2.4.0", so this change is likely to break
> > > locale generation for every machine that doesn't have OLDEST_KERNEL
> > > set.
> >
> > I've not heard yet of this really breaking something. And frankly I don't
> > see any reason for that. If it actually does - feel free to shout and
> > scream of course, we'll think how to handle it better.
>
> I guess it'd break on targets where OLDEST_KERNEL isn't set (and hence
> defaults to 2.4.0) but you need a kernel newer than 2.4.0 for correct
> operation.  Presumably all ARM EABI targets fall into that latter
> category, since the TLS syscalls weren't in 2.4.x, and there might be
> some others where glibc's default setting for --enable-kernel is
> something newer than 2.4.0.

If any target device requires kernel version not less then $SMTH, it must set 
it via OLDEST_KERNEL, isn't it? So if target device doesn't specify any 
OLDEST_KERNEL, it should be OK with 2.4.0. And then if glibc is configured for 
2.4.0 it either runs on host's 2.6.smth or not (presumably runs fine), 
regardless of what it thinks about the kernel version. Or am I missing 
something?

> Incidentally, if there is a default value for OLDEST_KERNEL in
> bitbake.conf then checking the retrieved value against None seems fairly
> futile since that will never be true.

Well, things happen. We can change it to

if kernel_ver is None:
	bb.fatal("Your config is completely broken")
...

of course.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 18:12 +0400, Roman I Khimov wrote:
> If any target device requires kernel version not less then $SMTH, it must set 
> it via OLDEST_KERNEL, isn't it?

No, this has never previously been a constraint.  ${OLDEST_KERNEL}
exists as a hint to say "there is no need to include support for kernels
older than this one", but it doesn't necessarily work the other way
around: there is no guarantee that ${OLDEST_KERNEL} is actually a
sufficient version for correct operation.

If OLDEST_KERNEL were to be interpreted more widely, as you propose, it
would not make much sense to have a global default of 2.4.0 since there
are all kinds of targets that won't run with such an old version.

p.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 18:40:03 Phil Blundell wrote:
> On Mon, 2009-05-18 at 18:12 +0400, Roman I Khimov wrote:
> > If any target device requires kernel version not less then $SMTH, it must
> > set it via OLDEST_KERNEL, isn't it?
>
> No, this has never previously been a constraint.  ${OLDEST_KERNEL}
> exists as a hint to say "there is no need to include support for kernels
> older than this one", but it doesn't necessarily work the other way
> around: there is no guarantee that ${OLDEST_KERNEL} is actually a
> sufficient version for correct operation.

That seems strange to me. 

I've tried locale generation for qemuarm that doesn't set any OLDEST_KERNEL 
and it really is broken with

FATAL: kernel too old

The problem is, glibc configuration started with "--enable-kernel=2.4.0" gets 
reconfigured to 2.6.14:

checking installed Linux kernel header files... 2.0.10 or later
configure: WARNING: minimum kernel version reset to 2.6.14
checking for kernel header at least 2.6.14... ok

and the reason is in ports/sysdeps/unix/sysv/linux/arm/eabi/configure.in.

ARM EABI has highest requirements, m68k wants at least 2.6.10, then 
sparc/sparc64 wants 2.4.21 and then hppa and ppc/ppc64 wants 2.4.19. Anything 
else is OK with 2.4.0.

Now, I think that this should be fixed on target level. OLDEST_KERNEL should 
be something sane. Global 2.4.0 from bitbake.conf is good enough for most 
targets, but if we know that none of ARM EABI works with kernel versions prior 
to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that targets. Or higher, 
of course, if target needs/wants that. If that's OK with all, I'm doing a 
patch.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> Now, I think that this should be fixed on target level. OLDEST_KERNEL should 
> be something sane. Global 2.4.0 from bitbake.conf is good enough for most 
> targets, but if we know that none of ARM EABI works with kernel versions prior 
> to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that targets. Or higher, 
> of course, if target needs/wants that. If that's OK with all, I'm doing a 
> patch.

I'm still not convinced this is either practical or desirable.  The
kernel versions you mentioned are the minimum required for the
particular version of glibc you were looking at, but different libcs or
different versions of the same libc will have different minimum kernel
requirements even for the same target.  (As an obvious example, the m68k
port has been supported in glibc since its inception and clearly didn't
require a 2.6.18 kernel to start with.)

I think it would be better to leave the semantics of OLDEST_KERNEL alone
and just teach qemu to use either the same __LINUX_KERNEL_VERSION as
glibc's configure script has settled on, or (perhaps better) just the
newest version that it supports for the given target.

p.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Florian Boor]

Hi,

Roman I Khimov schrieb:

> I've not heard yet of this really breaking something. And frankly I don't see 
> any reason for that. If it actually does - feel free to shout and scream of 
> course, we'll think how to handle it better.

I assume it breaks for everyone using qemu 0.9.x. I see it failing with this
message now:

NOTE: Running task 635 of 6760 (ID: 194,
/disk/d2/dev/recipes/glibc/glibc_2.6.1.bb, do_package)
NOTE: preparing tree for binary locale generation
NOTE: generating locale de_DE (UTF-8)
qemu-arm version 0.9.1, Copyright (c) 2003-2008 Fabrice Bellard
usage: qemu-arm [options] program [arguments...]
Linux CPU emulator (compiled for arm emulation)
[...]

This happens building Angstrom for both armv4 and armv5 targets and even with
the latest OLDEST_KERNEL changes Koen made some hours ago.

Greetings

Florian

-- 
The dream of yesterday                  Florian Boor
is the hope of today                    Tel: +49 271-771091-15
and the reality of tomorrow.            Fax: +49 271-771091-19
[Robert Hutchings Goddard, 1904]        florian.boor@kernelconcepts.de
                                        http://www.kernelconcepts.de/en

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 19:54 +0100, Phil Blundell wrote:
> On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> > Now, I think that this should be fixed on target level. OLDEST_KERNEL should 
> > be something sane. Global 2.4.0 from bitbake.conf is good enough for most 
> > targets, but if we know that none of ARM EABI works with kernel versions prior 
> > to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that targets. Or higher, 
> > of course, if target needs/wants that. If that's OK with all, I'm doing a 
> > patch.
> 
> I'm still not convinced this is either practical or desirable.  The
> kernel versions you mentioned are the minimum required for the
> particular version of glibc you were looking at, but different libcs or
> different versions of the same libc will have different minimum kernel
> requirements even for the same target.  (As an obvious example, the m68k
> port has been supported in glibc since its inception and clearly didn't
> require a 2.6.18 kernel to start with.)

The other thing to say about OLDEST_KERNEL is that it was intended to be
controlled by the DISTRO, not by the MACHINE.  In other words, it's a
policy control ("kernel x.xx is the earliest one we care about; it's
okay to omit the compatibility bits for older ones") and not a target
attribute ("kernel x.xx is the minimum required for correct
functionality").  We don't currently have a variable with the latter
semantics and nor am I very convinced that it would be a good idea to
introduce one.

p.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 22:54:53 Phil Blundell wrote:
> On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> > Now, I think that this should be fixed on target level. OLDEST_KERNEL
> > should be something sane. Global 2.4.0 from bitbake.conf is good enough
> > for most targets, but if we know that none of ARM EABI works with kernel
> > versions prior to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that
> > targets. Or higher, of course, if target needs/wants that. If that's OK
> > with all, I'm doing a patch.
>
> ...
>
> I think it would be better to leave the semantics of OLDEST_KERNEL alone
> and just teach qemu to use either the same __LINUX_KERNEL_VERSION as
> glibc's configure script has settled on, or (perhaps better) just the
> newest version that it supports for the given target.

Well, from my point of view, any static version is bad, because it's magic 
thrown in. When I've stumbled upon this, I've scratched my head a bit on why 
the hell is it 2.6.16, not 2.6.10, not 2.6.20, not anything else but 2.6.16. 
Sure enough, it's just something that worked some time ago for someone and was 
hardcoded then.

Then, I think that OLDEST_KERNEL should be something sane, since this is what 
we try to configure glibc with. It properly warns that there is something 
wrong on our side if we try to configure glibc for 2.4.0 kernel with ARM EABI 
target, so why not fix it? And then, if there are updates to glibc or anything 
else, it's target-specific and the target should adjust its OLDEST_KERNEL 
appropriately.

Of course, we can take version from __LINUX_KERNEL_VERSION, but that's more 
work and it doesn't fix misconfiguration we have when glibc has to decide that 
it knows better its minimal requirements for target.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 23:23:32 Florian Boor wrote:
> Hi,
>
> Roman I Khimov schrieb:
> > I've not heard yet of this really breaking something. And frankly I don't
> > see any reason for that. If it actually does - feel free to shout and
> > scream of course, we'll think how to handle it better.
>
> I assume it breaks for everyone using qemu 0.9.x. I see it failing with
> this message now:
>
> NOTE: Running task 635 of 6760 (ID: 194,
> /disk/d2/dev/recipes/glibc/glibc_2.6.1.bb, do_package)
> NOTE: preparing tree for binary locale generation
> NOTE: generating locale de_DE (UTF-8)
> qemu-arm version 0.9.1, Copyright (c) 2003-2008 Fabrice Bellard
> usage: qemu-arm [options] program [arguments...]
> Linux CPU emulator (compiled for arm emulation)
> [...]
>
> This happens building Angstrom for both armv4 and armv5 targets and even
> with the latest OLDEST_KERNEL changes Koen made some hours ago.

Hmm. Did it say anything else? In my case it's been failing with "FATAL: 
kernel too old". But that was host qemu (0.10.3).

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 23:23:32 Florian Boor wrote:
> Hi,
>
> Roman I Khimov schrieb:
> > I've not heard yet of this really breaking something. And frankly I don't
> > see any reason for that. If it actually does - feel free to shout and
> > scream of course, we'll think how to handle it better.
>
> I assume it breaks for everyone using qemu 0.9.x. I see it failing with
> this message now:
>
> NOTE: Running task 635 of 6760 (ID: 194,
> /disk/d2/dev/recipes/glibc/glibc_2.6.1.bb, do_package)
> NOTE: preparing tree for binary locale generation
> NOTE: generating locale de_DE (UTF-8)
> qemu-arm version 0.9.1, Copyright (c) 2003-2008 Fabrice Bellard
> usage: qemu-arm [options] program [arguments...]
> Linux CPU emulator (compiled for arm emulation)
> [...]
>
> This happens building Angstrom for both armv4 and armv5 targets and even
> with the latest OLDEST_KERNEL changes Koen made some hours ago.

Does 0.9.1 support "-E ..." parameter? That's the only reason for such 
behavior I could think of currently.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 23:23:32 Florian Boor wrote:
> Hi,
>
> Roman I Khimov schrieb:
> > I've not heard yet of this really breaking something. And frankly I don't
> > see any reason for that. If it actually does - feel free to shout and
> > scream of course, we'll think how to handle it better.
>
> I assume it breaks for everyone using qemu 0.9.x. I see it failing with
> this message now:

OK, this is a problem in second patch from series, "-E" parameter was added 
recently, dated Jan 30 2009 in qemu git. Bad, bad, bad.

How to fix that? Check QEMU version?

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Monday 18 May 2009 23:37:09 Phil Blundell wrote:
> On Mon, 2009-05-18 at 19:54 +0100, Phil Blundell wrote:
> > On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> > > Now, I think that this should be fixed on target level. OLDEST_KERNEL
> > > should be something sane. Global 2.4.0 from bitbake.conf is good enough
> > > for most targets, but if we know that none of ARM EABI works with
> > > kernel versions prior to 2.6.14, we should set OLDEST_KERNEL to 2.6.14
> > > for that targets. Or higher, of course, if target needs/wants that. If
> > > that's OK with all, I'm doing a patch.
> >
> > I'm still not convinced this is either practical or desirable.  The
> > kernel versions you mentioned are the minimum required for the
> > particular version of glibc you were looking at, but different libcs or
> > different versions of the same libc will have different minimum kernel
> > requirements even for the same target.  (As an obvious example, the m68k
> > port has been supported in glibc since its inception and clearly didn't
> > require a 2.6.18 kernel to start with.)
>
> The other thing to say about OLDEST_KERNEL is that it was intended to be
> controlled by the DISTRO, not by the MACHINE.  In other words, it's a
> policy control ("kernel x.xx is the earliest one we care about; it's
> okay to omit the compatibility bits for older ones") and not a target
> attribute ("kernel x.xx is the minimum required for correct
> functionality").  We don't currently have a variable with the latter
> semantics and nor am I very convinced that it would be a good idea to
> introduce one.

We can add soft 'sane' assignments to machines and then distros can override 
that to whatever they want if they want to, or not?

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 23:37 +0400, Roman I Khimov wrote:
> On Monday 18 May 2009 22:54:53 Phil Blundell wrote:
> > I think it would be better to leave the semantics of OLDEST_KERNEL alone
> > and just teach qemu to use either the same __LINUX_KERNEL_VERSION as
> > glibc's configure script has settled on, or (perhaps better) just the
> > newest version that it supports for the given target.
> 
> Well, from my point of view, any static version is bad, because it's magic 
> thrown in. When I've stumbled upon this, I've scratched my head a bit on why 
> the hell is it 2.6.16, not 2.6.10, not 2.6.20, not anything else but 2.6.16. 
> Sure enough, it's just something that worked some time ago for someone and was 
> hardcoded then.

Agreed, a static version is clearly wrong.

> Of course, we can take version from __LINUX_KERNEL_VERSION, but that's more 
> work and it doesn't fix misconfiguration we have when glibc has to decide that 
> it knows better its minimal requirements for target.

I'm not sure what you're saying here.  __LINUX_KERNEL_VERSION (from
config.h) is precisely the version that glibc is compiled to expect (and
what dl-sysdep checks against).  So, if you run qemu as that same kernel
version, it is guaranteed to always be compatible with the glibc build
in question.

p.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Mon, 2009-05-18 at 23:54 +0400, Roman I Khimov wrote:
> On Monday 18 May 2009 23:37:09 Phil Blundell wrote:
> > The other thing to say about OLDEST_KERNEL is that it was intended to be
> > controlled by the DISTRO, not by the MACHINE.  In other words, it's a
> > policy control ("kernel x.xx is the earliest one we care about; it's
> > okay to omit the compatibility bits for older ones") and not a target
> > attribute ("kernel x.xx is the minimum required for correct
> > functionality").  We don't currently have a variable with the latter
> > semantics and nor am I very convinced that it would be a good idea to
> > introduce one.
> 
> We can add soft 'sane' assignments to machines and then distros can override 
> that to whatever they want if they want to, or not?

No, the MACHINE doesn't know what version of libc is going to be
selected and hence what kernel version will be needed.  Even if it did,
this plan would lose if the DISTRO selected an older kernel than the
MACHINE requires, which in turn would make it difficult in practice for
DISTROs to do anything useful with this variable.

What is your objection to just selecting the highest kernel version that
qemu supports for the target architecture?  That seems like an easy fix
and one that makes this whole problem go away without requiring either
any extra variables or any semantic changes to existing variables.

p.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Tuesday 19 May 2009 00:00:35 Phil Blundell wrote:
> On Mon, 2009-05-18 at 23:37 +0400, Roman I Khimov wrote:
> > Of course, we can take version from __LINUX_KERNEL_VERSION, but that's
> > more work and it doesn't fix misconfiguration we have when glibc has to
> > decide that it knows better its minimal requirements for target.
>
> I'm not sure what you're saying here.  __LINUX_KERNEL_VERSION (from
> config.h) is precisely the version that glibc is compiled to expect (and
> what dl-sysdep checks against).  So, if you run qemu as that same kernel
> version, it is guaranteed to always be compatible with the glibc build
> in question.

Yep, sure. What's bugging me is that we try to configure glibc for wrong 
kernel in first place. And if we switch to proper configuration, then there 
will be no need to dig into config.h to get __LINUX_KERNEL_VERSION (is its 
format reliable enough across all versions of glibc?).

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Tom Rini]

On Mon, May 18, 2009 at 11:51:06PM +0400, Roman I Khimov wrote:
> On Monday 18 May 2009 23:23:32 Florian Boor wrote:
> > Hi,
> >
> > Roman I Khimov schrieb:
> > > I've not heard yet of this really breaking something. And frankly I don't
> > > see any reason for that. If it actually does - feel free to shout and
> > > scream of course, we'll think how to handle it better.
> >
> > I assume it breaks for everyone using qemu 0.9.x. I see it failing with
> > this message now:
> 
> OK, this is a problem in second patch from series, "-E" parameter was added 
> recently, dated Jan 30 2009 in qemu git. Bad, bad, bad.
> 
> How to fix that? Check QEMU version?

Use the 0.10.3 recipe and build libc for ppc, arm OABI, arm EABI and
lets drop 0.9.x crap already, in dev

-- 
Tom Rini

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Florian Boor]

Hi,

Roman I Khimov schrieb:
> Does 0.9.1 support "-E ..." parameter? That's the only reason for such 
> behavior I could think of currently.

Obviously not... looks like commit 6065fa491c009118ae282ae933215649cccfcd24
introduced this. It works if I revert the qemu command line changes.

Greetings

Florian

-- 
The dream of yesterday                  Florian Boor
is the hope of today                    Tel: +49 271-771091-15
and the reality of tomorrow.            Fax: +49 271-771091-19
[Robert Hutchings Goddard, 1904]        florian.boor@kernelconcepts.de
                                        http://www.kernelconcepts.de/en

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[GNUtoo]

On Mon, 2009-05-18 at 14:41 +0100, Phil Blundell wrote:
> n targets where OLDEST_KERNEL isn't set (and hence
> defaults to 2.4.0) but you need a kernel newer than 2.4.0 for correct
> operation.  Presumably all ARM EABI targets fall into that latter
> category, since the TLS syscalls weren't in 2.4.x, and there might be
> some others where glibc's default setting for --enable-kernel is
> something newer than 2.4.0.
> 
> All in all I'd be happier with this change if it enforced a minimum
> kernel version of 2.6.16 (i.e. what it used to be) so that there is no
> possibility of versions going backwards for any target.
> 
> Incidentally, if there is a default value for OLDEST_KERNEL in
> bitbake.conf then checking the retrieved value against None seems
> fairly
> futile since that will never be true.
> 
> p.
> 
I don't mind if the default is 2.6.16 but is there a howto for someone
that has a 2.4(.19-rmk7) kernel ?
Denis.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Peter Chubb]

>>>>> "Phil" == Phil Blundell <philb@gnu.org> writes:

Phil> On Mon, 2009-05-18 at 23:54 +0400, Roman I Khimov wrote:

Phil> What is your objection to just selecting the highest kernel
Phil> version that qemu supports for the target architecture?  That
Phil> seems like an easy fix and one that makes this whole problem go
Phil> away without requiring either any extra variables or any
Phil> semantic changes to existing variables.


That doesn't make sense for our case.  We're using openembedded to
generate a rootfs for an embedded virtual environment whose Linux
kernel is currently at 2.6.10.  Glibcs built for later kernels will
not run on this (because of the changes for handling
thread-local-storage introduced in the 2.6.11--2.6.12 timeframe)

--
Dr Peter Chubb  http://www.gelato.unsw.edu.au  peterc AT gelato.unsw.edu.au
http://www.ertos.nicta.com.au           ERTOS within National ICT Australia

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Tom Rini]

On Mon, May 18, 2009 at 08:37:09PM +0100, Phil Blundell wrote:
> On Mon, 2009-05-18 at 19:54 +0100, Phil Blundell wrote:
> > On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> > > Now, I think that this should be fixed on target level. OLDEST_KERNEL should 
> > > be something sane. Global 2.4.0 from bitbake.conf is good enough for most 
> > > targets, but if we know that none of ARM EABI works with kernel versions prior 
> > > to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that targets. Or higher, 
> > > of course, if target needs/wants that. If that's OK with all, I'm doing a 
> > > patch.
> > 
> > I'm still not convinced this is either practical or desirable.  The
> > kernel versions you mentioned are the minimum required for the
> > particular version of glibc you were looking at, but different libcs or
> > different versions of the same libc will have different minimum kernel
> > requirements even for the same target.  (As an obvious example, the m68k
> > port has been supported in glibc since its inception and clearly didn't
> > require a 2.6.18 kernel to start with.)
> 
> The other thing to say about OLDEST_KERNEL is that it was intended to be
> controlled by the DISTRO, not by the MACHINE.  In other words, it's a
> policy control ("kernel x.xx is the earliest one we care about; it's
> okay to omit the compatibility bits for older ones") and not a target
> attribute ("kernel x.xx is the minimum required for correct
> functionality").  We don't currently have a variable with the latter
> semantics and nor am I very convinced that it would be a good idea to
> introduce one.

I've re-read most of the thread, and I think what we need to do is:

- Acknowledge we've not set OLDEST_KERNEL in a sane way for a while.
  Except in some _machines_ we just use the 2.4.0 default, and glibc
  corrects it up.

- Acknowledge this is a GLIBC / EGLIBC only variable and move it over,
  with weak assignments to sane values in glibc.inc / eglibc.inc.  We
  might have to have some other possibilities commented out explaining
  what they do.  For example, arm-linux (oabi), non-nptl x86, etc.

-- 
Tom Rini

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Tom Rini]

On Mon, May 18, 2009 at 08:44:17PM -0700, Tom Rini wrote:
> On Mon, May 18, 2009 at 08:37:09PM +0100, Phil Blundell wrote:
> > On Mon, 2009-05-18 at 19:54 +0100, Phil Blundell wrote:
> > > On Mon, 2009-05-18 at 22:42 +0400, Roman I Khimov wrote:
> > > > Now, I think that this should be fixed on target level. OLDEST_KERNEL should 
> > > > be something sane. Global 2.4.0 from bitbake.conf is good enough for most 
> > > > targets, but if we know that none of ARM EABI works with kernel versions prior 
> > > > to 2.6.14, we should set OLDEST_KERNEL to 2.6.14 for that targets. Or higher, 
> > > > of course, if target needs/wants that. If that's OK with all, I'm doing a 
> > > > patch.
> > > 
> > > I'm still not convinced this is either practical or desirable.  The
> > > kernel versions you mentioned are the minimum required for the
> > > particular version of glibc you were looking at, but different libcs or
> > > different versions of the same libc will have different minimum kernel
> > > requirements even for the same target.  (As an obvious example, the m68k
> > > port has been supported in glibc since its inception and clearly didn't
> > > require a 2.6.18 kernel to start with.)
> > 
> > The other thing to say about OLDEST_KERNEL is that it was intended to be
> > controlled by the DISTRO, not by the MACHINE.  In other words, it's a
> > policy control ("kernel x.xx is the earliest one we care about; it's
> > okay to omit the compatibility bits for older ones") and not a target
> > attribute ("kernel x.xx is the minimum required for correct
> > functionality").  We don't currently have a variable with the latter
> > semantics and nor am I very convinced that it would be a good idea to
> > introduce one.
> 
> I've re-read most of the thread, and I think what we need to do is:
> 
> - Acknowledge we've not set OLDEST_KERNEL in a sane way for a while.
>   Except in some _machines_ we just use the 2.4.0 default, and glibc
>   corrects it up.
> 
> - Acknowledge this is a GLIBC / EGLIBC only variable and move it over,
>   with weak assignments to sane values in glibc.inc / eglibc.inc.  We
>   might have to have some other possibilities commented out explaining
>   what they do.  For example, arm-linux (oabi), non-nptl x86, etc.

Oh! And since this is glibc/eglibc only, we can pass qemu the min
version the libc was compiled for.

-- 
Tom Rini

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Tue, 2009-05-19 at 09:10 +1000, Peter Chubb wrote:
> >>>>> "Phil" == Phil Blundell <philb@gnu.org> writes:
> 
> Phil> On Mon, 2009-05-18 at 23:54 +0400, Roman I Khimov wrote:
> 
> Phil> What is your objection to just selecting the highest kernel
> Phil> version that qemu supports for the target architecture?  That
> Phil> seems like an easy fix and one that makes this whole problem go
> Phil> away without requiring either any extra variables or any
> Phil> semantic changes to existing variables.
> 
> 
> That doesn't make sense for our case.  We're using openembedded to
> generate a rootfs for an embedded virtual environment whose Linux
> kernel is currently at 2.6.10.  Glibcs built for later kernels will
> not run on this (because of the changes for handling
> thread-local-storage introduced in the 2.6.11--2.6.12 timeframe)

I think you misunderstood what I was proposing.  Glibc would continue to
be configured just as it is now: the only difference would be that
_qemu_ would be set to emulate the newest kernel version that it
supports, rather than a fixed version as previously.

p.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Tuesday 19 May 2009 11:38:51 Phil Blundell wrote:
> I think you misunderstood what I was proposing.  Glibc would continue to
> be configured just as it is now: the only difference would be that
> _qemu_ would be set to emulate the newest kernel version that it
> supports, rather than a fixed version as previously.

As I can tell from a quick glance, there is no special handling for different 
kernel versions in QEMU. It translates syscalls from target application to 
host kernel and if we tell it to fake kernel version reported to application 
it does so and nothing more (linux-user/syscall.c in qemu). So I don't think 
there is any newest kernel version supported by QEMU for any target, it just 
doesn't care.

Re: [oe-commits] Roman I Khimov : (e) glibc-package: fix kernel version passed to qemu

[Phil Blundell]

On Tue, 2009-05-19 at 12:55 +0400, Roman I Khimov wrote:
> As I can tell from a quick glance, there is no special handling for different 
> kernel versions in QEMU. It translates syscalls from target application to 
> host kernel and if we tell it to fake kernel version reported to application 
> it does so and nothing more (linux-user/syscall.c in qemu). So I don't think 
> there is any newest kernel version supported by QEMU for any target, it just 
> doesn't care.

Oh right.  Well, in that case, it's very easy: we can just tell qemu to
report itself as 9999.9999.9999 or something.  Or have a NEWEST_KERNEL
variable that's set (globally in bitbake.conf or somesuch) to the newest
kernel that might conceivably be used.

p.

[PATCH] (e)glibc.inc: add minimal sane kernel versions

[Roman I Khimov]

(e)glibc is configured for OLDEST_KERNEL, certain target architectures have
specific requirements for kernel versions, if that are not satisfied (e)glibc
reconfigures itself with what it think more appropriate and warns user that
there is misconfiguration on his side.

Add minimal sane kernel versions for architectures requiring that.
---
 conf/distro/include/eglibc.inc |    9 +++++++++
 conf/distro/include/glibc.inc  |    9 +++++++++
 2 files changed, 18 insertions(+), 0 deletions(-)

diff --git a/conf/distro/include/eglibc.inc b/conf/distro/include/eglibc.inc
index 0d2087e..7901949 100644
--- a/conf/distro/include/eglibc.inc
+++ b/conf/distro/include/eglibc.inc
@@ -7,3 +7,12 @@ PREFERRED_PROVIDER_virtual/libintl ?= "eglibc"
 PREFERRED_PROVIDER_virtual/libc = "eglibc"
 PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "eglibc-initial"
 PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-for-gcc = "eglibc"
+
+# These are minimal sane kernel versions for targets that woudln't work with
+# default 2.4.0
+
+# ARM OABI is fine with 2.4.0, EABI is not, so we override on "-gnueabi"
+# TARGET_OS here, not on architecture
+OLDEST_KERNEL_linux-gnueabi ?= "2.6.16"
+OLDEST_KERNEL_sparc64 ?= "2.4.21"
+OLDEST_KERNEL_ppc64 ?= "2.4.21"
diff --git a/conf/distro/include/glibc.inc b/conf/distro/include/glibc.inc
index f9140ff..25f98ce 100644
--- a/conf/distro/include/glibc.inc
+++ b/conf/distro/include/glibc.inc
@@ -7,3 +7,12 @@ PREFERRED_PROVIDER_virtual/libintl ?= "glibc"
 PREFERRED_PROVIDER_virtual/libc = "glibc"
 PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "glibc-initial"
 PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-for-gcc = "glibc"
+
+# These are minimal sane kernel versions for targets that woudln't work with
+# default 2.4.0
+
+# ARM OABI is fine with 2.4.0, EABI is not, so we override on "-gnueabi"
+# TARGET_OS here, not on architecture
+OLDEST_KERNEL_linux-gnueabi ?= "2.6.14"
+OLDEST_KERNEL_sparc64 ?= "2.4.21"
+OLDEST_KERNEL_ppc64 ?= "2.4.21"
-- 
1.6.3.1


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Roman I Khimov]

On Tuesday 19 May 2009 00:17:07 Tom Rini wrote:
> On Mon, May 18, 2009 at 11:51:06PM +0400, Roman I Khimov wrote:
> > On Monday 18 May 2009 23:23:32 Florian Boor wrote:
> > > Hi,
> > >
> > > Roman I Khimov schrieb:
> > > > I've not heard yet of this really breaking something. And frankly I
> > > > don't see any reason for that. If it actually does - feel free to
> > > > shout and scream of course, we'll think how to handle it better.
> > >
> > > I assume it breaks for everyone using qemu 0.9.x. I see it failing with
> > > this message now:
> >
> > OK, this is a problem in second patch from series, "-E" parameter was
> > added recently, dated Jan 30 2009 in qemu git. Bad, bad, bad.
> >
> > How to fix that? Check QEMU version?
>
> Use the 0.10.3 recipe and build libc for ppc, arm OABI, arm EABI and
> lets drop 0.9.x crap already, in dev

Ported OE QEMU patches to 0.10.3, apart from two:

revert_arm_tcg.patch.gz
qemu-n800-support.patch

Just don't know why the first was needed and the second requires some real 
effort to update to 0.10.3. Could be interesting to hear something about 
those.

Re: [oe-commits] Roman I Khimov : (e)glibc-package: fix kernel version passed to qemu

[Marcin Juszkiewicz]

Dnia wtorek, 19 maja 2009 o 17:24:16 Roman I Khimov napisał(a):

> Ported OE QEMU patches to 0.10.3, apart from two:
>
> revert_arm_tcg.patch.gz
> qemu-n800-support.patch
>
> Just don't know why the first was needed and the second requires some
> real effort to update to 0.10.3. Could be interesting to hear
> something about those.

I would drop both and test do binary locales works for ARM targets (as 
this is main use of qemu-native). n800 support is merged into qemu, 
revert_arm_tcg.patch removed qemu new code generator for ARM.


Regards, 
-- 
JID:      hrw@jabber.org
Website:  http://marcin.juszkiewicz.com.pl/
LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz

[PATCH] qemu 0.10.3: port OE's patches, make preferable version

[Roman I Khimov]

Fixes binary locale generation broken by
6065fa491c009118ae282ae933215649cccfcd24.
---
 recipes/qemu/qemu-0.10.3/06_exit_segfault.patch    |   45 ++++++++++++++++
 recipes/qemu/qemu-0.10.3/11_signal_sigaction.patch |   21 ++++++++
 recipes/qemu/qemu-0.10.3/22_net_tuntap_stall.patch |   13 +++++
 recipes/qemu/qemu-0.10.3/31_syscalls.patch         |   22 ++++++++
 recipes/qemu/qemu-0.10.3/32_syscall_sysctl.patch   |   55 ++++++++++++++++++++
 recipes/qemu/qemu-0.10.3/52_ne2000_return.patch    |   17 ++++++
 recipes/qemu/qemu-0.10.3/61_safe_64bit_int.patch   |   27 ++++++++++
 recipes/qemu/qemu-0.10.3/63_sparc_build.patch      |   18 ++++++
 .../qemu/qemu-0.10.3/64_ppc_asm_constraints.patch  |   18 ++++++
 recipes/qemu/qemu-0.10.3/66_tls_ld.patch           |   55 ++++++++++++++++++++
 recipes/qemu/qemu-0.10.3/91-oh-sdl-cursor.patch    |   18 ++++++
 .../qemu/qemu-0.10.3/fix_baum_c_compilation.patch  |   30 +++++++++++
 recipes/qemu/qemu-0.10.3/fix_segfault.patch        |   37 +++++++++++++
 recipes/qemu/qemu-0.10.3/no-strip.patch            |   26 +++++++++
 recipes/qemu/qemu_0.10.3.bb                        |   17 +++++-
 15 files changed, 416 insertions(+), 3 deletions(-)
 create mode 100644 recipes/qemu/qemu-0.10.3/06_exit_segfault.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/11_signal_sigaction.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/22_net_tuntap_stall.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/31_syscalls.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/32_syscall_sysctl.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/52_ne2000_return.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/61_safe_64bit_int.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/63_sparc_build.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/64_ppc_asm_constraints.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/66_tls_ld.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/91-oh-sdl-cursor.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/fix_baum_c_compilation.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/fix_segfault.patch
 create mode 100644 recipes/qemu/qemu-0.10.3/no-strip.patch

diff --git a/recipes/qemu/qemu-0.10.3/06_exit_segfault.patch b/recipes/qemu/qemu-0.10.3/06_exit_segfault.patch
new file mode 100644
index 0000000..bc02d31
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/06_exit_segfault.patch
@@ -0,0 +1,45 @@
+#DPATCHLEVEL=0
+---
+# linux-user/main.c |    8 ++++----
+# 1 file changed, 4 insertions(+), 4 deletions(-)
+#
+Index: linux-user/main.c
+===================================================================
+--- linux-user/main.c.orig	2008-04-24 20:15:46.000000000 +0100
++++ linux-user/main.c	2008-04-24 20:15:53.000000000 +0100
+@@ -765,7 +765,7 @@
+         default:
+             printf ("Unhandled trap: 0x%x\n", trapnr);
+             cpu_dump_state(env, stderr, fprintf, 0);
+-            exit (1);
++            _exit (1);
+         }
+         process_pending_signals (env);
+     }
+@@ -1697,7 +1697,7 @@
+         default:
+             printf ("Unhandled trap: 0x%x\n", trapnr);
+             cpu_dump_state(env, stderr, fprintf, 0);
+-            exit (1);
++            _exit (1);
+         }
+         process_pending_signals (env);
+     }
+@@ -2026,7 +2026,7 @@
+                 for(item = cpu_log_items; item->mask != 0; item++) {
+                     printf("%-10s %s\n", item->name, item->help);
+                 }
+-                exit(1);
++                _exit(1);
+             }
+             cpu_set_log(mask);
+         } else if (!strcmp(r, "s")) {
+@@ -2045,7 +2045,7 @@
+             if (qemu_host_page_size == 0 ||
+                 (qemu_host_page_size & (qemu_host_page_size - 1)) != 0) {
+                 fprintf(stderr, "page size must be a power of two\n");
+-                exit(1);
++                _exit(1);
+             }
+         } else if (!strcmp(r, "g")) {
+             gdbstub_port = atoi(argv[optind++]);
diff --git a/recipes/qemu/qemu-0.10.3/11_signal_sigaction.patch b/recipes/qemu/qemu-0.10.3/11_signal_sigaction.patch
new file mode 100644
index 0000000..cd56541
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/11_signal_sigaction.patch
@@ -0,0 +1,21 @@
+#DPATCHLEVEL=0
+---
+# linux-user/signal.c |    5 +++++
+# 1 file changed, 5 insertions(+)
+#
+Index: linux-user/signal.c
+===================================================================
+--- linux-user/signal.c.orig	2008-04-24 20:15:55.000000000 +0100
++++ linux-user/signal.c	2008-04-24 20:15:57.000000000 +0100
+@@ -512,6 +512,11 @@
+ 
+     if (sig < 1 || sig > TARGET_NSIG || sig == SIGKILL || sig == SIGSTOP)
+         return -EINVAL;
++
++    /* no point doing the stuff as those are not allowed for sigaction */
++    if ((sig == TARGET_SIGKILL) || (sig == TARGET_SIGSTOP))
++        return -EINVAL;
++
+     k = &sigact_table[sig - 1];
+ #if defined(DEBUG_SIGNAL)
+     fprintf(stderr, "sigaction sig=%d act=0x%08x, oact=0x%08x\n",
diff --git a/recipes/qemu/qemu-0.10.3/22_net_tuntap_stall.patch b/recipes/qemu/qemu-0.10.3/22_net_tuntap_stall.patch
new file mode 100644
index 0000000..993960b
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/22_net_tuntap_stall.patch
@@ -0,0 +1,13 @@
+Index: qemu-0.10.3/net.c
+===================================================================
+--- qemu-0.10.3.orig/net.c	2009-05-19 17:46:54.650870195 +0400
++++ qemu-0.10.3/net.c	2009-05-19 17:47:05.612854521 +0400
+@@ -955,7 +955,7 @@
+         return -1;
+     }
+     memset(&ifr, 0, sizeof(ifr));
+-    ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
++    ifr.ifr_flags = IFF_TAP | IFF_NO_PI | IFF_ONE_QUEUE;
+     if (ifname[0] != '\0')
+         pstrcpy(ifr.ifr_name, IFNAMSIZ, ifname);
+     else
diff --git a/recipes/qemu/qemu-0.10.3/31_syscalls.patch b/recipes/qemu/qemu-0.10.3/31_syscalls.patch
new file mode 100644
index 0000000..20a1499
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/31_syscalls.patch
@@ -0,0 +1,22 @@
+Index: linux-user/syscall.c
+===================================================================
+--- linux-user/syscall.c.orig	2008-04-24 20:15:46.000000000 +0100
++++ linux-user/syscall.c	2008-04-24 20:15:59.000000000 +0100
+@@ -250,6 +250,7 @@
+ extern int setresgid(gid_t, gid_t, gid_t);
+ extern int getresgid(gid_t *, gid_t *, gid_t *);
+ extern int setgroups(int, gid_t *);
++extern int uselib(const char*);
+ 
+ #define ERRNO_TABLE_SIZE 1200
+ 
+@@ -4041,7 +4042,8 @@
+ #endif
+ #ifdef TARGET_NR_uselib
+     case TARGET_NR_uselib:
+-        goto unimplemented;
++        ret = get_errno(uselib(path((const char*)arg1)));
++        break;
+ #endif
+ #ifdef TARGET_NR_swapon
+     case TARGET_NR_swapon:
diff --git a/recipes/qemu/qemu-0.10.3/32_syscall_sysctl.patch b/recipes/qemu/qemu-0.10.3/32_syscall_sysctl.patch
new file mode 100644
index 0000000..2a670c9
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/32_syscall_sysctl.patch
@@ -0,0 +1,55 @@
+#DPATCHLEVEL=0
+---
+# linux-user/syscall.c |   32 +++++++++++++++++++++++++++++---
+# 1 file changed, 29 insertions(+), 3 deletions(-)
+#
+Index: linux-user/syscall.c
+===================================================================
+--- linux-user/syscall.c.orig	2009-05-19 17:50:28.000000000 +0400
++++ linux-user/syscall.c	2009-05-19 17:52:19.094103462 +0400
+@@ -55,6 +55,7 @@
+ #include <netinet/ip.h>
+ #include <netinet/tcp.h>
+ #include <qemu-common.h>
++#include <sys/sysctl.h>
+ #ifdef HAVE_GPROF
+ #include <sys/gmon.h>
+ #endif
+@@ -5193,9 +5194,34 @@
+         break;
+ #endif
+     case TARGET_NR__sysctl:
+-        /* We don't implement this, but ENOTDIR is always a safe
+-           return value. */
+-        ret = -TARGET_ENOTDIR;
++        {
++            struct __sysctl_args *args = (struct __sysctl_args *) arg1;
++            int *name_target, *name, nlen, *oldlenp, oldlen, newlen, i;
++            void *oldval, *newval;
++
++            name_target = (int *) tswapl((long) args->name);
++            nlen = tswapl(args->nlen);
++            oldval = (void *) tswapl((long) args->oldval);
++            oldlenp = (int *) tswapl((long) args->oldlenp);
++            oldlen = tswapl(*oldlenp);
++            newval = (void *) tswapl((long) args->newval);
++            newlen = tswapl(args->newlen);
++
++            name = alloca(nlen * sizeof (int));
++            for (i = 0; i < nlen; i++)
++                name[i] = tswapl(name_target[i]);
++
++            if (nlen == 2 && name[0] == CTL_KERN && name[1] == KERN_VERSION) {
++                ret = get_errno(
++                        sysctl(name, nlen, oldval, &oldlen, newval, newlen));
++                if (!is_error(ret)) {
++                    *oldlenp = tswapl(oldlen);
++                }
++            } else {
++                gemu_log("qemu: Unsupported sysctl name\n");
++                ret = -ENOSYS;
++            }
++        }
+         break;
+     case TARGET_NR_sched_setparam:
+         {
diff --git a/recipes/qemu/qemu-0.10.3/52_ne2000_return.patch b/recipes/qemu/qemu-0.10.3/52_ne2000_return.patch
new file mode 100644
index 0000000..e364bff
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/52_ne2000_return.patch
@@ -0,0 +1,17 @@
+---
+ hw/ne2000.c |    2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+Index: trunk/hw/ne2000.c
+===================================================================
+--- trunk.orig/hw/ne2000.c	2008-04-24 20:15:46.000000000 +0100
++++ trunk/hw/ne2000.c	2008-04-24 20:16:05.000000000 +0100
+@@ -217,7 +217,7 @@
+     NE2000State *s = opaque;
+ 
+     if (s->cmd & E8390_STOP)
+-        return 1;
++        return 0;
+     return !ne2000_buffer_full(s);
+ }
+ 
diff --git a/recipes/qemu/qemu-0.10.3/61_safe_64bit_int.patch b/recipes/qemu/qemu-0.10.3/61_safe_64bit_int.patch
new file mode 100644
index 0000000..80e7b22
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/61_safe_64bit_int.patch
@@ -0,0 +1,27 @@
+#DPATCHLEVEL=0
+---
+# dyngen-exec.h |    4 ++--
+# 1 file changed, 2 insertions(+), 2 deletions(-)
+#
+Index: dyngen-exec.h
+===================================================================
+--- dyngen-exec.h.orig	2009-05-01 21:02:44.000000000 +0400
++++ dyngen-exec.h	2009-05-19 17:56:28.164854448 +0400
+@@ -41,7 +41,7 @@
+ // Linux/Sparc64 defines uint64_t
+ #if !(defined (__sparc_v9__) && defined(__linux__)) && !(defined(__APPLE__) && defined(__x86_64__))
+ /* XXX may be done for all 64 bits targets ? */
+-#if defined (__x86_64__) || defined(__ia64) || defined(__s390x__) || defined(__alpha__) || defined(_ARCH_PPC64)
++#if defined (__x86_64__) || defined(__ia64) || defined(__s390x__) || defined(__alpha__) || defined(_ARCH_PPC64) || defined(__sparc__)
+ typedef unsigned long uint64_t;
+ #else
+ typedef unsigned long long uint64_t;
+@@ -58,7 +58,7 @@
+ typedef signed int int32_t;
+ // Linux/Sparc64 defines int64_t
+ #if !(defined (__sparc_v9__) && defined(__linux__)) && !(defined(__APPLE__) && defined(__x86_64__))
+-#if defined (__x86_64__) || defined(__ia64) || defined(__s390x__) || defined(__alpha__) || defined(_ARCH_PPC64)
++#if defined (__x86_64__) || defined(__ia64) || defined(__s390x__) || defined(__alpha__) || defined(_ARCH_PPC64) || defined(__sparc__)
+ typedef signed long int64_t;
+ #else
+ typedef signed long long int64_t;
diff --git a/recipes/qemu/qemu-0.10.3/63_sparc_build.patch b/recipes/qemu/qemu-0.10.3/63_sparc_build.patch
new file mode 100644
index 0000000..097f55a
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/63_sparc_build.patch
@@ -0,0 +1,18 @@
+#DPATCHLEVEL=0
+---
+# sparc.ld |    2 +-
+# 1 file changed, 1 insertion(+), 1 deletion(-)
+#
+Index: sparc.ld
+===================================================================
+--- sparc.ld.orig	2008-04-24 20:15:46.000000000 +0100
++++ sparc.ld	2008-04-24 20:16:07.000000000 +0100
+@@ -6,7 +6,7 @@
+ SECTIONS
+ {
+   /* Read-only sections, merged into text segment: */
+-  . = 0x60000000 + SIZEOF_HEADERS;
++  . = 0x60000000 + 0x400;
+   .interp     : { *(.interp)    }
+   .hash          : { *(.hash)           }
+   .dynsym        : { *(.dynsym)         }
diff --git a/recipes/qemu/qemu-0.10.3/64_ppc_asm_constraints.patch b/recipes/qemu/qemu-0.10.3/64_ppc_asm_constraints.patch
new file mode 100644
index 0000000..7562daf
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/64_ppc_asm_constraints.patch
@@ -0,0 +1,18 @@
+#DPATCHLEVEL=1
+---
+# cpu-all.h |    2 +-
+# 1 file changed, 1 insertion(+), 1 deletion(-)
+#
+Index: qemu-0.10.3/cpu-all.h
+===================================================================
+--- qemu-0.10.3.orig/cpu-all.h	2009-05-01 21:02:44.000000000 +0400
++++ qemu-0.10.3/cpu-all.h	2009-05-19 17:59:28.357103179 +0400
+@@ -288,7 +288,7 @@
+ static inline void stl_le_p(void *ptr, int v)
+ {
+ #ifdef _ARCH_PPC
+-    __asm__ __volatile__ ("stwbrx %1,0,%2" : "=m" (*(uint32_t *)ptr) : "r" (v), "r" (ptr));
++     __asm__ __volatile__ ("stwbrx %0,0,%1" : : "r" (v), "r" (ptr) : "memory");
+ #else
+     uint8_t *p = ptr;
+     p[0] = v;
diff --git a/recipes/qemu/qemu-0.10.3/66_tls_ld.patch b/recipes/qemu/qemu-0.10.3/66_tls_ld.patch
new file mode 100644
index 0000000..cbd3f87
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/66_tls_ld.patch
@@ -0,0 +1,55 @@
+---
+ arm.ld  |    7 +++++++
+ i386.ld |    7 +++++++
+ 2 files changed, 14 insertions(+)
+
+Index: arm.ld
+===================================================================
+--- arm.ld.orig	2008-04-24 20:15:45.000000000 +0100
++++ arm.ld	2008-04-24 20:16:11.000000000 +0100
+@@ -26,6 +26,10 @@
+     { *(.rel.rodata) *(.rel.gnu.linkonce.r*) }
+   .rela.rodata   :
+     { *(.rela.rodata) *(.rela.gnu.linkonce.r*) }
++  .rel.tdata     : { *(.rel.tdata .rel.tdata.* .rel.gnu.linkonce.td.*) }
++  .rela.tdata    : { *(.rela.tdata .rela.tdata.* .rela.gnu.linkonce.td.*) }
++  .rel.tbss      : { *(.rel.tbss .rel.tbss.* .rel.gnu.linkonce.tb.*) }
++  .rela.tbss     : { *(.rela.tbss .rela.tbss.* .rela.gnu.linkonce.tb.*) }
+   .rel.got       : { *(.rel.got)		}
+   .rela.got      : { *(.rela.got)		}
+   .rel.ctors     : { *(.rel.ctors)	}
+@@ -58,6 +62,9 @@
+   .ARM.exidx   : { *(.ARM.exidx* .gnu.linkonce.armexidx.*) }
+    __exidx_end = .;
+   .reginfo : { *(.reginfo) }
++  /* Thread Local Storage sections  */
++  .tdata	  : { *(.tdata .tdata.* .gnu.linkonce.td.*) }
++  .tbss		  : { *(.tbss .tbss.* .gnu.linkonce.tb.*) *(.tcommon) }
+   /* Adjust the address for the data segment.  We want to adjust up to
+      the same address within the page on the next page up.  */
+   . = ALIGN(0x100000) + (. & (0x100000 - 1));
+Index: i386.ld
+===================================================================
+--- i386.ld.orig	2008-04-24 20:15:45.000000000 +0100
++++ i386.ld	2008-04-24 20:16:11.000000000 +0100
+@@ -28,6 +28,10 @@
+     { *(.rel.rodata) *(.rel.gnu.linkonce.r*) }
+   .rela.rodata   :
+     { *(.rela.rodata) *(.rela.gnu.linkonce.r*) }
++  .rel.tdata     : { *(.rel.tdata .rel.tdata.* .rel.gnu.linkonce.td.*) }
++  .rela.tdata    : { *(.rela.tdata .rela.tdata.* .rela.gnu.linkonce.td.*) }
++  .rel.tbss      : { *(.rel.tbss .rel.tbss.* .rel.gnu.linkonce.tb.*) }
++  .rela.tbss     : { *(.rela.tbss .rela.tbss.* .rela.gnu.linkonce.tb.*) }
+   .rel.got       : { *(.rel.got)		}
+   .rela.got      : { *(.rela.got)		}
+   .rel.ctors     : { *(.rel.ctors)	}
+@@ -53,6 +57,9 @@
+   _etext = .;
+   PROVIDE (etext = .);
+   .fini      : { *(.fini)    } =0x47ff041f
++  /* Thread Local Storage sections  */
++  .tdata	  : { *(.tdata .tdata.* .gnu.linkonce.td.*) }
++  .tbss		  : { *(.tbss .tbss.* .gnu.linkonce.tb.*) *(.tcommon) }
+   . = ALIGN(32 / 8);
+   PROVIDE (__preinit_array_start = .);
+   .preinit_array     : { *(.preinit_array) }
diff --git a/recipes/qemu/qemu-0.10.3/91-oh-sdl-cursor.patch b/recipes/qemu/qemu-0.10.3/91-oh-sdl-cursor.patch
new file mode 100644
index 0000000..b3d95a4
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/91-oh-sdl-cursor.patch
@@ -0,0 +1,18 @@
+=== modified file 'sdl.c'
+---
+ sdl.c |    2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+Index: sdl.c
+===================================================================
+--- sdl.c.orig	2008-04-24 20:15:45.000000000 +0100
++++ sdl.c	2008-04-24 20:16:12.000000000 +0100
+@@ -247,7 +247,7 @@
+ 
+     if (kbd_mouse_is_absolute()) {
+         SDL_ShowCursor(1);
+-        SDL_SetCursor(sdl_cursor_hidden);
++        /* SDL_SetCursor(sdl_cursor_hidden); */
+     } else {
+         SDL_ShowCursor(0);
+     }
diff --git a/recipes/qemu/qemu-0.10.3/fix_baum_c_compilation.patch b/recipes/qemu/qemu-0.10.3/fix_baum_c_compilation.patch
new file mode 100644
index 0000000..a2eb438
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/fix_baum_c_compilation.patch
@@ -0,0 +1,30 @@
+Index: trunk/hw/baum.c
+===================================================================
+--- trunk.orig/hw/baum.c	2008-11-29 05:12:40.418390606 +0300
++++ trunk/hw/baum.c	2008-11-29 05:13:11.498378426 +0300
+@@ -569,8 +569,10 @@ CharDriverState *chr_baum_init(void)
+     CharDriverState *chr;
+     brlapi_handle_t *handle;
+ #ifdef CONFIG_SDL
++#ifdef SDL_VIDEO_DRIVER_X11
+     SDL_SysWMinfo info;
+ #endif
++#endif
+     int tty;
+
+     baum = qemu_mallocz(sizeof(BaumDriverState));
+@@ -605,12 +607,14 @@ CharDriverState *chr_baum_init(void)
+     }
+
+ #ifdef CONFIG_SDL
++#ifdef SDL_VIDEO_DRIVER_X11
+     memset(&info, 0, sizeof(info));
+     SDL_VERSION(&info.version);
+     if (SDL_GetWMInfo(&info))
+         tty = info.info.x11.wmwindow;
+     else
+ #endif
++#endif
+         tty = BRLAPI_TTY_DEFAULT;
+
+     if (brlapi__enterTtyMode(handle, tty, NULL) == -1) {
diff --git a/recipes/qemu/qemu-0.10.3/fix_segfault.patch b/recipes/qemu/qemu-0.10.3/fix_segfault.patch
new file mode 100644
index 0000000..32ca3c0
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/fix_segfault.patch
@@ -0,0 +1,37 @@
+---
+ linux-user/syscall.c |   22 ----------------------
+ 1 file changed, 22 deletions(-)
+
+Index: qemu-0.10.3/linux-user/syscall.c
+===================================================================
+--- qemu-0.10.3.orig/linux-user/syscall.c	2009-05-19 18:18:04.000000000 +0400
++++ qemu-0.10.3/linux-user/syscall.c	2009-05-19 18:18:40.397103379 +0400
+@@ -6024,28 +6024,6 @@
+         goto unimplemented_nowarn;
+ #endif
+ 
+-#ifdef TARGET_NR_clock_gettime
+-    case TARGET_NR_clock_gettime:
+-    {
+-        struct timespec ts;
+-        ret = get_errno(clock_gettime(arg1, &ts));
+-        if (!is_error(ret)) {
+-            host_to_target_timespec(arg2, &ts);
+-        }
+-        break;
+-    }
+-#endif
+-#ifdef TARGET_NR_clock_getres
+-    case TARGET_NR_clock_getres:
+-    {
+-        struct timespec ts;
+-        ret = get_errno(clock_getres(arg1, &ts));
+-        if (!is_error(ret)) {
+-            host_to_target_timespec(arg2, &ts);
+-        }
+-        break;
+-    }
+-#endif
+ #ifdef TARGET_NR_clock_nanosleep
+     case TARGET_NR_clock_nanosleep:
+     {
diff --git a/recipes/qemu/qemu-0.10.3/no-strip.patch b/recipes/qemu/qemu-0.10.3/no-strip.patch
new file mode 100644
index 0000000..4813dd4
--- /dev/null
+++ b/recipes/qemu/qemu-0.10.3/no-strip.patch
@@ -0,0 +1,26 @@
+Index: trunk/Makefile
+===================================================================
+--- trunk.orig/Makefile	2008-04-24 20:15:37.000000000 +0100
++++ trunk/Makefile	2008-04-24 20:16:30.000000000 +0100
+@@ -196,7 +196,7 @@
+ install: all $(if $(BUILD_DOCS),install-doc)
+ 	mkdir -p "$(DESTDIR)$(bindir)"
+ ifneq ($(TOOLS),)
+-	$(INSTALL) -m 755 -s $(TOOLS) "$(DESTDIR)$(bindir)"
++	$(INSTALL) -m 755 $(TOOLS) "$(DESTDIR)$(bindir)"
+ endif
+ 	mkdir -p "$(DESTDIR)$(datadir)"
+ 	set -e; for x in bios.bin vgabios.bin vgabios-cirrus.bin ppc_rom.bin \
+Index: trunk/Makefile.target
+===================================================================
+--- trunk.orig/Makefile.target	2008-04-24 20:15:37.000000000 +0100
++++ trunk/Makefile.target	2008-04-24 20:16:30.000000000 +0100
+@@ -685,7 +685,7 @@
+ 
+ install: all
+ ifneq ($(PROGS),)
+-	$(INSTALL) -m 755 -s $(PROGS) "$(DESTDIR)$(bindir)"
++	$(INSTALL) -m 755 $(PROGS) "$(DESTDIR)$(bindir)"
+ endif
+ 
+ # Include automatically generated dependency files
diff --git a/recipes/qemu/qemu_0.10.3.bb b/recipes/qemu/qemu_0.10.3.bb
index d6ad1d7..890e878 100644
--- a/recipes/qemu/qemu_0.10.3.bb
+++ b/recipes/qemu/qemu_0.10.3.bb
@@ -1,14 +1,25 @@
 LICENSE = "GPL"
 DEPENDS = "zlib"
 
-# Need to port OE patches there
-DEFAULT_PREFERENCE = "-1"
-
 FILESPATH = "${FILE_DIRNAME}/qemu-${PV}"
 FILESDIR = "${WORKDIR}"
 
 SRC_URI = "\
     http://download.savannah.gnu.org/releases/qemu/qemu-${PV}.tar.gz \
+    file://06_exit_segfault.patch;patch=1;pnum=0 \
+    file://11_signal_sigaction.patch;patch=1;pnum=0 \
+    file://22_net_tuntap_stall.patch;patch=1 \
+    file://31_syscalls.patch;patch=1;pnum=0 \
+    file://32_syscall_sysctl.patch;patch=1;pnum=0 \
+    file://52_ne2000_return.patch;patch=1 \
+    file://61_safe_64bit_int.patch;patch=1;pnum=0 \
+    file://63_sparc_build.patch;patch=1;pnum=0 \
+    file://64_ppc_asm_constraints.patch;patch=1 \
+    file://66_tls_ld.patch;patch=1;pnum=0 \
+    file://91-oh-sdl-cursor.patch;patch=1;pnum=0 \
+    file://fix_segfault.patch;patch=1 \
+    file://no-strip.patch;patch=1 \
+    file://fix_baum_c_compilation.patch;patch=1 \
     "
 
 S = "${WORKDIR}/qemu-${PV}"
-- 
1.6.3.1