[Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Anthony Liguori]

The QEMU team is pleased to announce the availability of the 0.10.1 
release.  This is a stable release of the 0.10 tree and contains only 
bug fixes since the 0.10.0.  The attached Changelog includes all of the 
changes that have went in since 0.10.0.

It can be downloaded from Savannah at:

 http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz

It may take a couple hours for the Savannah mirrors to update so if you 
get an error, try again later.

  - virtio-net: allow masking of notifications on empty queue (Alex 
Williamson)
  - e1000: fix rx descriptor low threshold logic (Alex Willaimson)
  - x86 tcg: add NULL checks to lsl instruction (Jan Kiszka)
  - kvm vga: fix screen corruption with -std-vga and Windows (Avi Kivity)
  - kvm vga: fix screen corruption with Ubuntu installations (Glauber Costa)
  - virtio-net: check right return size on sg list (Alex Williamson)
  - Make qemu_announce_self handle holes (live migration after hotplug)
    (Marcelo Tosatti)
  - Revert r6804-r6808 (qcow2 allocation info).  This series of changes 
added
    a high cost to startup for large qcow2 images (Anthony Liguori)
  - qemu-img: fix help message (Aurelien Jarno)
  - Fix build for non-default installs of SDL (Anthony Liguori)
  - Fix race condition in env->interrupt_request.  When using TCG and a 
dynticks
    host timer, this condition could cause TCG to get stuck in an infinite
    loop (Aurelien Jarno)
  - Fix reading encrypted hard disk passwords during early startup (Jan 
Kiszka)
  - Fix encrypted disk reporting in 'info block' (Jan Kiszka)
  - Fix console size with tiny displays (MusicPal) (Jan Kiszka)
  - Improve error handling in bdrv_open2 (Jan Kiszka)
  - Avoid leaking data in mux'ed character devices (Jan Kiszka)
  - Fix initial character device reset (no banner in monitor) (Jan Kiszka)
  - Fix cpuid KVM crash on i386 host (Lubomir Rintel)
  - Fix SLES10sp2 installation by adding ISTAT1 register to LSI SCSI 
emulation
    (Ryan Harper)

Regards,

Anthony Liguori

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Rob Landley]

On Saturday 21 March 2009 18:25:23 Anthony Liguori wrote:
> The QEMU team is pleased to announce the availability of the 0.10.1
> release.  This is a stable release of the 0.10 tree and contains only
> bug fixes since the 0.10.0.  The attached Changelog includes all of the
> changes that have went in since 0.10.0.
>
> It can be downloaded from Savannah at:
>
>  http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz

This doesn't include a fix powerpc -nographic (either by reverting to the 
previous openbios-ppc or by upgrading to current OpenBios svn):
  http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg01070.html

Nor does it include the fix for sh4 -append:
  http://lists.gnu.org/archive/html/qemu-devel/2009-02/msg00727.html

Rob

P.S.  Sorry I didn't get to respond to your friday message.  My laptop's 
Ubuntu 8.10 decided to eat itself over the weekend (I hate selinux) and I 
just finished reinstalling last night...

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Anthony Liguori]

Rob Landley wrote:
> On Saturday 21 March 2009 18:25:23 Anthony Liguori wrote:
>   
>> The QEMU team is pleased to announce the availability of the 0.10.1
>> release.  This is a stable release of the 0.10 tree and contains only
>> bug fixes since the 0.10.0.  The attached Changelog includes all of the
>> changes that have went in since 0.10.0.
>>
>> It can be downloaded from Savannah at:
>>
>>  http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
>>     
>
> This doesn't include a fix powerpc -nographic (either by reverting to the 
> previous openbios-ppc or by upgrading to current OpenBios svn):
>   http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg01070.html
>
> Nor does it include the fix for sh4 -append:
>   http://lists.gnu.org/archive/html/qemu-devel/2009-02/msg00727.html
>   

Part of my criteria for committing to stable is pretty heavy testing of 
the actual patch.  For areas I don't maintain, I'm not really setup to 
do the level of testing that would be required for stable.

I'd suggest that the original maintainers who commit the patches to 
trunk should be the ones to determine whether a patch should also go in 
stable.

Regards,

Anthony Liguori


> Rob
>
> P.S.  Sorry I didn't get to respond to your friday message.  My laptop's 
> Ubuntu 8.10 decided to eat itself over the weekend (I hate selinux) and I 
> just finished reinstalling last night...
>   

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Felipe Contreras]

On Sun, Mar 22, 2009 at 2:25 AM, Anthony Liguori <anthony@codemonkey.ws> wrote:
> The QEMU team is pleased to announce the availability of the 0.10.1 release.
>  This is a stable release of the 0.10 tree and contains only bug fixes since
> the 0.10.0.  The attached Changelog includes all of the changes that have
> went in since 0.10.0.
>
> It can be downloaded from Savannah at:
>
> http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
>
> It may take a couple hours for the Savannah mirrors to update so if you get
> an error, try again later.

You forgot this one:
http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg00315.html

---------- Forwarded message ----------
From: Paul Brook <paul@codesourcery.com>
Date: Tue, Mar 17, 2009 at 2:54 PM
Subject: Re: [Qemu-devel] [6735] The _exit syscall is used for both
thread  termination in NPTL applications ,
To: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>


> >>> The _exit syscall is used for both thread termination in NPTL
> >>> applications, and process termination in legacy applications.  Try to
> >>> guess which we want based on the presence of multiple threads.
> >>>
> >>> Also implement locking when modifying the CPU list.
> >>
> >> This fixes a regression in arm-linux-user,

I'm not so sure this is really a regression. Multithreaded applications never
worked.

> >> it should be pushed into the stable branch, right?
> >
> > If it fixes something *and* if there is no risk of regression, yes. But
> > I am not the one able to decide. Paul Brook is the person to ask.

There's always a risk of a regression.
I suspect this change is probably fairly safe though.

Paul


-- 
Felipe Contreras

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Rob Landley]

On Tuesday 24 March 2009 16:20:14 Anthony Liguori wrote:
> Rob Landley wrote:
> > On Saturday 21 March 2009 18:25:23 Anthony Liguori wrote:
> >> The QEMU team is pleased to announce the availability of the 0.10.1
> >> release.  This is a stable release of the 0.10 tree and contains only
> >> bug fixes since the 0.10.0.  The attached Changelog includes all of the
> >> changes that have went in since 0.10.0.
> >>
> >> It can be downloaded from Savannah at:
> >>
> >>  http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
> >
> > This doesn't include a fix powerpc -nographic (either by reverting to the
> > previous openbios-ppc or by upgrading to current OpenBios svn):
> >   http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg01070.html
> >
> > Nor does it include the fix for sh4 -append:
> >   http://lists.gnu.org/archive/html/qemu-devel/2009-02/msg00727.html
>
> Part of my criteria for committing to stable is pretty heavy testing of
> the actual patch.  For areas I don't maintain, I'm not really setup to
> do the level of testing that would be required for stable.

By the way, if you want a quick and easy way to do testing, my Firmware Linux 
project builds bootable system images for arm, mips, ppc, sh4, x86, and 
x86-64, targeted at qemu.  You can download prebuilt binary tarballs if you 
don't want to run the build on your local machine, and it includes boot 
scripts for qemu in the system image tarballs.

I'm in the process of putting together the 0.9.6 release, there's an -rc up at 
http://impactlinux.com/fwl/downloads/snapshots/2009-03-30/

You can download the appropriate system image tarball from there (or the older 
0.9.5 release versions from 
http://impactlinux.com/fwl/downloads/binaries/system-image/ ) extract it 
and "./run-emulator.sh".  You should get kernel boot messages ending in a 
shell prompt inside the emulator.  (It's got a complete native toolchain in 
there so you can even compile stuff, the virtual network's set up to 
masquerade out so you can wget stuff, etc.)

In the case of the above two bugs, system-image.powerpc doesn't boot under 
qemu 0.10.1 without the openbios-ppc fix, and system-image-sh4 doesn't boot 
under qemu 0.10.1 without the sh4 fix.  With said fixes, they both boot fine.  
There's a locally reproducible test with complete source code.  (It even 
builds its own cross compilers.)

FWL also builds a sparc image that doesn't work due to a uClibc bug, and an 
m68k image that doesn't boot because qemu doesn't actually support m68k, just 
coldfire.  My goals for the 1.0 release are to build images for every target 
qemu supports.  I was recently surprised to build a 440 image and find that 
qemu's bamboo board is kvm only. :P

Anyway, more details on http://impactlinux.com/fwl

I'm in the process of putting together the 0.9.6 release, there's an -rc up at 
http://impactlinux.com/fwl/downloads/snapshots/2009-03-30/

> I'd suggest that the original maintainers who commit the patches to
> trunk should be the ones to determine whether a patch should also go in
> stable.

I'm all for it, but in the case of sh4 they already submitted the patch here 
(that's the post I'm linking to) and it never got merged even into 0.10.0 or 
0.10.1.  I see it's in now.

(On the ppc front, I suspect Blue Swirl can take care of himself. :)

Rob
-- 
GPLv3 is to GPLv2 what Attack of the Clones is to The Empire Strikes Back.

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Aurelien Jarno]

On Mon, Mar 30, 2009 at 07:25:03PM -0500, Rob Landley wrote:
> On Tuesday 24 March 2009 16:20:14 Anthony Liguori wrote:
> > Rob Landley wrote:
> > > On Saturday 21 March 2009 18:25:23 Anthony Liguori wrote:
> > >> The QEMU team is pleased to announce the availability of the 0.10.1
> > >> release.  This is a stable release of the 0.10 tree and contains only
> > >> bug fixes since the 0.10.0.  The attached Changelog includes all of the
> > >> changes that have went in since 0.10.0.
> > >>
> > >> It can be downloaded from Savannah at:
> > >>
> > >>  http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
> > >
> > > This doesn't include a fix powerpc -nographic (either by reverting to the
> > > previous openbios-ppc or by upgrading to current OpenBios svn):
> > >   http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg01070.html
> > >
> > > Nor does it include the fix for sh4 -append:
> > >   http://lists.gnu.org/archive/html/qemu-devel/2009-02/msg00727.html
> >
> > Part of my criteria for committing to stable is pretty heavy testing of
> > the actual patch.  For areas I don't maintain, I'm not really setup to
> > do the level of testing that would be required for stable.
> 
> By the way, if you want a quick and easy way to do testing, my Firmware Linux 
> project builds bootable system images for arm, mips, ppc, sh4, x86, and 
> x86-64, targeted at qemu.  You can download prebuilt binary tarballs if you 
> don't want to run the build on your local machine, and it includes boot 
> scripts for qemu in the system image tarballs.
> 
> I'm in the process of putting together the 0.9.6 release, there's an -rc up at 
> http://impactlinux.com/fwl/downloads/snapshots/2009-03-30/
> 
> You can download the appropriate system image tarball from there (or the older 
> 0.9.5 release versions from 
> http://impactlinux.com/fwl/downloads/binaries/system-image/ ) extract it 
> and "./run-emulator.sh".  You should get kernel boot messages ending in a 
> shell prompt inside the emulator.  (It's got a complete native toolchain in 
> there so you can even compile stuff, the virtual network's set up to 
> masquerade out so you can wget stuff, etc.)
> 
> In the case of the above two bugs, system-image.powerpc doesn't boot under 
> qemu 0.10.1 without the openbios-ppc fix, and system-image-sh4 doesn't boot 
> under qemu 0.10.1 without the sh4 fix.  With said fixes, they both boot fine.  
> There's a locally reproducible test with complete source code.  (It even 
> builds its own cross compilers.)
> 
> FWL also builds a sparc image that doesn't work due to a uClibc bug, and an 
> m68k image that doesn't boot because qemu doesn't actually support m68k, just 
> coldfire.  My goals for the 1.0 release are to build images for every target 
> qemu supports.  I was recently surprised to build a 440 image and find that 
> qemu's bamboo board is kvm only. :P
> 
> Anyway, more details on http://impactlinux.com/fwl
> 
> I'm in the process of putting together the 0.9.6 release, there's an -rc up at 
> http://impactlinux.com/fwl/downloads/snapshots/2009-03-30/
> 
> > I'd suggest that the original maintainers who commit the patches to
> > trunk should be the ones to determine whether a patch should also go in
> > stable.
> 
> I'm all for it, but in the case of sh4 they already submitted the patch here 
> (that's the post I'm linking to) and it never got merged even into 0.10.0 or 
> 0.10.1.  I see it's in now.
> 

The patch that you are linking is clearly not a candidate to stable, as
bugs have been found (like wrong kernel load address) during the
discussions that were still ongoing at the time of your requests, and it
was not even in trunk at this moment.

The patch that has been committed to trunk recently is different from
this one. Moreover I fail to see how it can be considered a regression
and not a new feature as it has never worked before. Note that 
qemu-system-sh4 works fine without this patch as long as you don't need
to pass a kernel argument.

Given that you keep insisting, I have committed a backported patch from
trunk. It is really new in trunk and so has not been really tested and
it may be reverted if we find any regression with it.

-- 
Aurelien Jarno	                        GPG: 1024D/F1BCDB73
aurelien@aurel32.net                 http://www.aurel32.net

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Felipe Contreras]

On Sun, Mar 29, 2009 at 2:08 PM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Sun, Mar 22, 2009 at 2:25 AM, Anthony Liguori <anthony@codemonkey.ws> wrote:
>> The QEMU team is pleased to announce the availability of the 0.10.1 release.
>>  This is a stable release of the 0.10 tree and contains only bug fixes since
>> the 0.10.0.  The attached Changelog includes all of the changes that have
>> went in since 0.10.0.
>>
>> It can be downloaded from Savannah at:
>>
>> http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
>>
>> It may take a couple hours for the Savannah mirrors to update so if you get
>> an error, try again later.
>
> You forgot this one:
> http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg00315.html

Ping?

I'm adding Riku Voipio to the loop since I think he might be interested.

Again, arm-linux-user is broken on the stable branch since:
http://repo.or.cz/w/qemu.git?a=commit;h=a2f86d8

Cross-compilation with scratchbox doesn't work for anything that
checks for pthreads (using pthread_join), like GLib's configure
script.

It has been fixed by:
http://repo.or.cz/w/qemu.git?a=commit;h=c27647

-- 
Felipe Contreras

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Anthony Liguori]

Felipe Contreras wrote:
> On Sun, Mar 29, 2009 at 2:08 PM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>   
>> On Sun, Mar 22, 2009 at 2:25 AM, Anthony Liguori <anthony@codemonkey.ws> wrote:
>>     
>>> The QEMU team is pleased to announce the availability of the 0.10.1 release.
>>>  This is a stable release of the 0.10 tree and contains only bug fixes since
>>> the 0.10.0.  The attached Changelog includes all of the changes that have
>>> went in since 0.10.0.
>>>
>>> It can be downloaded from Savannah at:
>>>
>>> http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
>>>
>>> It may take a couple hours for the Savannah mirrors to update so if you get
>>> an error, try again later.
>>>       
>> You forgot this one:
>> http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg00315.html
>>     
>
> Ping?
>   

This is for Paul to decide.  I don't know linux-user emulation well 
enough to judge how much of an impact this patch exposes verses what it 
fixes.

I plan on cutting another stable release soon though so if it's going to 
go it, it should go in in the next day or two.

Regards,

Anthony Liguori

> I'm adding Riku Voipio to the loop since I think he might be interested.
>
> Again, arm-linux-user is broken on the stable branch since:
> http://repo.or.cz/w/qemu.git?a=commit;h=a2f86d8
>
> Cross-compilation with scratchbox doesn't work for anything that
> checks for pthreads (using pthread_join), like GLib's configure
> script.
>
> It has been fixed by:
> http://repo.or.cz/w/qemu.git?a=commit;h=c27647
>
>   

Re: [Qemu-devel] ANNOUNCE: Release 0.10.1 of QEMU

[Felipe Contreras]

On Tue, Mar 31, 2009 at 7:43 PM, Anthony Liguori <anthony@codemonkey.ws> wrote:
> Felipe Contreras wrote:
>>
>> On Sun, Mar 29, 2009 at 2:08 PM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>
>>>
>>> On Sun, Mar 22, 2009 at 2:25 AM, Anthony Liguori <anthony@codemonkey.ws>
>>> wrote:
>>>
>>>>
>>>> The QEMU team is pleased to announce the availability of the 0.10.1
>>>> release.
>>>>  This is a stable release of the 0.10 tree and contains only bug fixes
>>>> since
>>>> the 0.10.0.  The attached Changelog includes all of the changes that
>>>> have
>>>> went in since 0.10.0.
>>>>
>>>> It can be downloaded from Savannah at:
>>>>
>>>> http://savannah.nongnu.org/download/qemu/qemu-0.10.1.tar.gz
>>>>
>>>> It may take a couple hours for the Savannah mirrors to update so if you
>>>> get
>>>> an error, try again later.
>>>>
>>>
>>> You forgot this one:
>>> http://lists.gnu.org/archive/html/qemu-devel/2009-03/msg00315.html
>>>
>>
>> Ping?
>>
>
> This is for Paul to decide.  I don't know linux-user emulation well enough
> to judge how much of an impact this patch exposes verses what it fixes.
>
> I plan on cutting another stable release soon though so if it's going to go
> it, it should go in in the next day or two.

I'm attaching the message, again, where he says it's safe enough.

---------- Forwarded message ----------
From: Paul Brook <paul@codesourcery.com>
Date: Tue, Mar 17, 2009 at 2:54 PM
Subject: Re: [Qemu-devel] [6735] The _exit syscall is used for both
thread  termination in NPTL applications ,
To: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>


> >>> The _exit syscall is used for both thread termination in NPTL
> >>> applications, and process termination in legacy applications.  Try to
> >>> guess which we want based on the presence of multiple threads.
> >>>
> >>> Also implement locking when modifying the CPU list.
> >>
> >> This fixes a regression in arm-linux-user,

I'm not so sure this is really a regression. Multithreaded applications never
worked.

> >> it should be pushed into the stable branch, right?
> >
> > If it fixes something *and* if there is no risk of regression, yes. But
> > I am not the one able to decide. Paul Brook is the person to ask.

There's always a risk of a regression.
I suspect this change is probably fairly safe though.

Paul

-- 
Felipe Contreras