# HG changeset patch
# User john.le...@sun.com
# Date 1231990064 28800
# Node ID 629c101c9ec11f3eb5cb56eb9548c96c33c8daf6
# Parent 0f488fb716b1ab0a1379509b8b3594f32f0ea980
Solaris least privilege support
On Solaris dom0, virtd runs as a privilege barrier: all libvirt
connections are routed throug
# HG changeset patch
# User john.le...@sun.com
# Date 1231982481 28800
# Node ID 082e0f7d5de236e69bea177c8d4c4204350144c1
# Parent 1a16d4b76232845091c1f800569ec462860cb1fb
Improve xend_get error message
Signed-off-by: John Levon
diff --git a/src/xend_internal.c b/src/xend_internal.c
--- a/src/x
# HG changeset patch
# User john.le...@sun.com
# Date 1231985011 28800
# Node ID 2766ee91dd5ea8e99dac27ce730af0dc46a1d107
# Parent 082e0f7d5de236e69bea177c8d4c4204350144c1
Avoid passing NULL to printf %s specifier
This is non-portable.
Signed-off-by: John Levon
diff --git a/src/internal.h b/sr
On Wed, Jan 14, 2009 at 10:38:02PM +, Daniel P. Berrange wrote:
> > # virsh start domu-defefe
> > 12:09:30.581: error : Domain not found: xenUnifiedDomainLookupByName
> > libvir: Xen error : Domain not found: xenUnifiedDomainLookupByName
> > libvir: Xen error : Domain not found: xenUnifiedDom
On Mon, Jan 12, 2009 at 04:34:25PM +, Richard W.M. Jones wrote:
> We discovered over the new year that the XDR code in glibc, which was
> derived from some original Sun code, doesn't have an unambiguously
> free license[1]. Furthermore the rpcgen in glibc is unmaintained and
> produces such ba
On Wed, Jan 14, 2009 at 08:39:02PM +0100, Jim Meyering wrote:
> "Daniel P. Berrange" wrote:
> > This patch ensures all public API methods only have a single exit
> > path, to make mutex unlocking simpler.
> >
> > remote_internal.c | 1256
> > +++---
On Wed, Jan 14, 2009 at 08:03:49PM +, John Levon wrote:
>
> We do:
>
> 700 /* Not found. */
> 701 xenUnifiedError (conn, VIR_ERR_NO_DOMAIN, __FUNCTION__);
>
> But:
>
> # virsh start domu-defefe
> 12:09:30.581: error : Domain not found: xenUnifiedDomainLookupByName
> libvir: Xen e
On Wed, Jan 14, 2009 at 08:27:36PM +0100, Jim Meyering wrote:
> John Levon wrote:
> > On Wed, Jan 14, 2009 at 10:53:51AM -0800, john.le...@sun.com wrote:
> >
> >> Fix activeDomainList handling
> >>
> >> It must be NULLed out after free()ing.
> >
> > This one is incomplete - I didn't find the bug s
On Wed, Jan 14, 2009 at 10:54:18AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User john.le...@sun.com
> # Date 1231958879 28800
> # Node ID 672cdb87a32fb0b275bb7cdc9abb9c2d506ec1b0
> # Parent 46dc909bda5b20d11f08d1516f21c05f27b6648f
> Fix xs_unwatch() to correctly pass the token
>
Howdy all!
So in working with libvirt-qpid, I've come to realize that it imposes a
new difficulty in doing migrations. Traditionally you would connect
out to each individual libvirt instance from a central location and be
able to pass the destination connection pointer to the virDomainMigrate
f
On Wed, Jan 14, 2009 at 10:53:51AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User john.le...@sun.com
> # Date 1231958879 28800
> # Node ID 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631
> # Parent 097f5b4497d7ecc5cce14d10559244313d26ce46
> Fix activeDomainList handling
>
> It must be
On Wed, Jan 14, 2009 at 10:54:02AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User john.le...@sun.com
> # Date 1231958879 28800
> # Node ID 46dc909bda5b20d11f08d1516f21c05f27b6648f
> # Parent 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631
> Fix ref-counting for Xen driver event registra
On Thu, Jan 15, 2009 at 07:32:25AM +1100, James Morris wrote:
> On Wed, 14 Jan 2009, Daniel J Walsh wrote:
>
> > I think labeling can be done to allow the access to directories, and
> > files. So libvirt could go in an label a file/directory in such a way
> > that the running qemu_t:s0.c10 can re
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Itamar Heim wrote:
>> From: libvir-list-boun...@redhat.com [mailto:libvir-list-
>> boun...@redhat.com] On Behalf Of Daniel J Walsh
>> I think labeling can be done to allow the access to directories, and
>> files. So libvirt could go in an label a file
> From: libvir-list-boun...@redhat.com [mailto:libvir-list-
> boun...@redhat.com] On Behalf Of Daniel J Walsh
> I think labeling can be done to allow the access to directories, and
> files. So libvirt could go in an label a file/directory in such a way
> that the running qemu_t:s0.c10 can read or
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
James Morris wrote:
> On Wed, 14 Jan 2009, Daniel J Walsh wrote:
>
>> I think labeling can be done to allow the access to directories, and
>> files. So libvirt could go in an label a file/directory in such a way
>> that the running qemu_t:s0.c10 can
Jim Meyering wrote:
> "Daniel P. Berrange" wrote:
>> This patch ensures all public API methods only have a single exit
>> path, to make mutex unlocking simpler.
>>
>> remote_internal.c | 1256
>> +++---
>
> Hi Dan,
>
> I got about halfway through i
On Wed, 14 Jan 2009, Daniel J Walsh wrote:
> I think labeling can be done to allow the access to directories, and
> files. So libvirt could go in an label a file/directory in such a way
> that the running qemu_t:s0.c10 can read or read/write the file/directory.
>
> Same with the ability to creat
We do:
700 /* Not found. */
701 xenUnifiedError (conn, VIR_ERR_NO_DOMAIN, __FUNCTION__);
But:
# virsh start domu-defefe
12:09:30.581: error : Domain not found: xenUnifiedDomainLookupByName
libvir: Xen error : Domain not found: xenUnifiedDomainLookupByName
libvir: Xen error : Domain n
"Daniel P. Berrange" wrote:
> This patch ensures all public API methods only have a single exit
> path, to make mutex unlocking simpler.
>
> remote_internal.c | 1256
> +++---
Hi Dan,
I got about halfway through it and spotted something odd:
> di
John Levon wrote:
> On Wed, Jan 14, 2009 at 10:53:51AM -0800, john.le...@sun.com wrote:
>
>> Fix activeDomainList handling
>>
>> It must be NULLed out after free()ing.
>
> This one is incomplete - I didn't find the bug since I have to use
> 0...@0.so with LIBVIRT_DEBUG. Annoyingly, Solaris printf
john.le...@sun.com wrote:
> Fix ref-counting for Xen driver event registration
> diff --git a/src/xen_unified.c b/src/xen_unified.c
> --- a/src/xen_unified.c
> +++ b/src/xen_unified.c
> @@ -1359,15 +1359,21 @@ xenUnifiedDomainEventRegister (virConnec
> void *opaque,
On Wed, Jan 14, 2009 at 10:53:51AM -0800, john.le...@sun.com wrote:
> Fix activeDomainList handling
>
> It must be NULLed out after free()ing.
This one is incomplete - I didn't find the bug since I have to use
0...@0.so with LIBVIRT_DEBUG. Annoyingly, Solaris printf requires %s
parameters to not
john.le...@sun.com wrote:
> Fix activeDomainList handling
In your eventual ChangeLog entry, please
mention xsWatchList, too.
> It must be NULLed out after free()ing.
>
> Signed-off-by: John Levon
>
> diff --git a/src/xs_internal.c b/src/xs_internal.c
> --- a/src/xs_internal.c
> +++ b/src/xs_inte
# HG changeset patch
# User john.le...@sun.com
# Date 1231958879 28800
# Node ID 672cdb87a32fb0b275bb7cdc9abb9c2d506ec1b0
# Parent 46dc909bda5b20d11f08d1516f21c05f27b6648f
Fix xs_unwatch() to correctly pass the token
Signed-off-by: John Levon
diff --git a/src/xs_internal.c b/src/xs_internal.c
-
# HG changeset patch
# User john.le...@sun.com
# Date 1231958879 28800
# Node ID 46dc909bda5b20d11f08d1516f21c05f27b6648f
# Parent 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631
Fix ref-counting for Xen driver event registration
Signed-off-by: John Levon
diff --git a/src/xen_unified.c b/src/xen_unifi
# HG changeset patch
# User john.le...@sun.com
# Date 1231958879 28800
# Node ID 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631
# Parent 097f5b4497d7ecc5cce14d10559244313d26ce46
Fix activeDomainList handling
It must be NULLed out after free()ing.
Signed-off-by: John Levon
diff --git a/src/xs_interna
john.le...@sun.com wrote:
...
Hi John,
This looks fine, but there are a couple of nits.
> +#ifdef __sun
> +
> +static int
> +get_cpu_flags(virConnectPtr conn, const char **hvm, int *pae, int *longmode)
> +{
> +struct {
> +uint32_t r_eax, r_ebx, r_ecx, r_edx;
> +} regs;
> +
> +
On Wed, Jan 14, 2009 at 07:23:30AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231946129 28800
> # Node ID 9d8302140e58e20c7fb22b3242a52b06467b08e6
> # Parent b614f5780261464e8f10e48306e67d43d56931b0
> Remove references to non-existent files
>
> Some doc
On Wed, Jan 14, 2009 at 07:22:47AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231946129 28800
> # Node ID b614f5780261464e8f10e48306e67d43d56931b0
> # Parent 53c621a65055f4752abef748c0ad15cbb1d8013d
> Separate objdir build fixes
>
> Let devhelp build in
On Wed, Jan 14, 2009 at 07:20:31AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231946129 28800
> # Node ID c08132809fb6e060b1301e97954a1b0ababb384c
> # Parent 52dcd17a9fd4bf4a6171386f083d235eebd5aa23
> Fix devhelp build
>
> For devhelp to actually be cre
On Mon, Jan 12, 2009 at 09:19:02AM +, Miloslav Trma?? wrote:
> Hello,
> here's a version of the patch updated to apply against current CVS.
I've realized there's a slight problem with our idea to add a simple
/some/path
element to the volume '' section. Specifically when creating
a new im
On Wed, Jan 14, 2009 at 03:38:52PM +, Daniel P. Berrange wrote:
> > Implement capabilities for Solaris
> >
> > Use Solaris interfaces to derive the hypervisor capabilities.
>
> Previously your patch for this did a hypercall to fetch the
> capabilities straight from the hypervisor. Out of int
On Wed, Jan 14, 2009 at 07:20:12AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231940906 28800
> # Node ID a0d98d39955f4f304d318c7c780742ab929eb351
> # Parent ddfcba6b4181ab433dce345d66dc399a81c92a3b
> Implement capabilities for Solaris
>
> Use Solaris i
On Wed, Jan 14, 2009 at 07:20:47AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231946129 28800
> # Node ID 53c621a65055f4752abef748c0ad15cbb1d8013d
> # Parent bde7017c447cf36451194e64179b60e1c8d9f039
> Mark Xen PV CD-ROM devices as such
>
> Add a ':cdrom
On Wed, Jan 14, 2009 at 07:22:18AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User John Levon
> # Date 1231946129 28800
> # Node ID c7538cb7fab4e6f3aa6714fdd4126dcf2f0371b4
> # Parent cfe04caba5de61620fa6d1c2bff85cfd62255c1d
> Fix oversized stack allocation
>
> REMOTE_MESSAGE_MA
# HG changeset patch
# User John Levon
# Date 1231946129 28800
# Node ID 9d8302140e58e20c7fb22b3242a52b06467b08e6
# Parent b614f5780261464e8f10e48306e67d43d56931b0
Remove references to non-existent files
Some docs files don't exist.
Signed-off-by: John Levon
diff --git a/docs/Makefile.am b/do
# HG changeset patch
# User John Levon
# Date 1231946129 28800
# Node ID b614f5780261464e8f10e48306e67d43d56931b0
# Parent 53c621a65055f4752abef748c0ad15cbb1d8013d
Separate objdir build fixes
Let devhelp build in a separate objdir.
Signed-off-by: John Levon
diff --git a/docs/devhelp/Makefile.
# HG changeset patch
# User John Levon
# Date 1231946129 28800
# Node ID c7538cb7fab4e6f3aa6714fdd4126dcf2f0371b4
# Parent cfe04caba5de61620fa6d1c2bff85cfd62255c1d
Fix oversized stack allocation
REMOTE_MESSAGE_MAX is a whopping 262Kb and shouldn't be allocated on the
stack.
Signed-off-by: Ryan
# HG changeset patch
# User John Levon
# Date 1231946128 28800
# Node ID dd17b3062611925baa2698ff5923579d0f2cd34e
# Parent a0d98d39955f4f304d318c7c780742ab929eb351
Introduce virt-console
Separate console handling out into a separate binary to allow management
of privileges.
Signed-off-by: John
# HG changeset patch
# User John Levon
# Date 1231946129 28800
# Node ID 53c621a65055f4752abef748c0ad15cbb1d8013d
# Parent bde7017c447cf36451194e64179b60e1c8d9f039
Mark Xen PV CD-ROM devices as such
Add a ':cdrom' marker for PV as well as HVM devices.
Signed-off-by: John Levon
diff --git a/sr
# HG changeset patch
# User John Levon
# Date 1231946129 28800
# Node ID c08132809fb6e060b1301e97954a1b0ababb384c
# Parent 52dcd17a9fd4bf4a6171386f083d235eebd5aa23
Fix devhelp build
For devhelp to actually be created during a build, we need to explicitly
depend on the generated files.
Signed-of
# HG changeset patch
# User John Levon
# Date 1231940906 28800
# Node ID a0d98d39955f4f304d318c7c780742ab929eb351
# Parent ddfcba6b4181ab433dce345d66dc399a81c92a3b
Implement capabilities for Solaris
Use Solaris interfaces to derive the hypervisor capabilities.
Signed-off-by: John Levon
diff -
"Daniel P. Berrange" wrote:
> On Tue, Jan 13, 2009 at 05:49:02PM +, Daniel P. Berrange wrote:
>> QEMU now has support for a sound card of type "ac97", so enable
>> that in the XML parser / qemu driver.
>>
>> Also remove some unused cruft relating to sound in Xen.
>>
>> domain_conf.c |3
On Sat, Dec 20, 2008 at 08:49:01AM -0800, john.le...@sun.com wrote:
> # HG changeset patch
> # User john.le...@sun.com
> # Date 1229789393 28800
> # Node ID a158bbc6df6fb8b15af875d08e73cfd3ca75f907
> # Parent 34706075e6c801cd9532490c11b20b64dfcf464b
> Improve compiler flag checking
>
> Some compi
Daniel P. Berrange wrote:
> On Tue, Jan 13, 2009 at 05:49:02PM +, Daniel P. Berrange wrote:
>> QEMU now has support for a sound card of type "ac97", so enable
>> that in the XML parser / qemu driver.
>>
>> Also remove some unused cruft relating to sound in Xen.
>>
>> domain_conf.c |3 ++-
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Daniel P. Berrange wrote:
> On Tue, Jan 13, 2009 at 05:18:46PM -0500, Daniel J Walsh wrote:
>> -BEGIN PGP SIGNED MESSAGE-
>> Hash: SHA1
>>
>> As I begin to work on the svirt lock down of the qemu process, I am
>> seeing a disturbing problem.
>>
"Daniel P. Berrange" wrote:
...
> No, they are all correct AFAIK. The *existing* code was buggy using
> the wrong macros in many places.
...
> You need to compare with the function context shown in the patch, rather
> than assume the original code was correct :-)
Yeah, "assuming" can cause troubl
On Wed, Jan 14, 2009 at 08:55:46AM +, Mark McLoughlin wrote:
> On Tue, 2009-01-13 at 20:49 +, Daniel P. Berrange wrote:
> > Currently when we shutdown the virtual networks are all shutdown too.
> > This is less than useful if we're letting guest VMs hang around post
> > shutdown of libvirt
On Wed, Jan 14, 2009 at 01:23:28PM +0100, Jim Meyering wrote:
> "Daniel P. Berrange" wrote:
> > THis patch removes use of macros for accessing the privateDtaa
> > on a connection because they obscure data access making it harder
> > to visually validate correct thread locking
>
> However, some of
"Daniel P. Berrange" wrote:
> THis patch removes use of macros for accessing the privateDtaa
> on a connection because they obscure data access making it harder
> to visually validate correct thread locking
>
> remote_internal.c | 272
> ++
>
"Daniel P. Berrange" wrote:
> THis patch removes use of macros for accessing the privateDtaa
> on a connection because they obscure data access making it harder
> to visually validate correct thread locking
It's a trade-off, weighing loss of the magic-number check
against improved readability wrt
On Wed, Jan 14, 2009 at 5:58 AM, Daniel P. Berrange wrote:
> On Tue, Jan 13, 2009 at 03:00:17PM -0500, Sir Woody Hackswell wrote:
>
> It is desirable to stop things when the machine is shutting down.
>
> If we put this functionality in the stop() function of the initscript
> we cannot make this di
On Tue, Jan 13, 2009 at 05:49:02PM +, Daniel P. Berrange wrote:
> QEMU now has support for a sound card of type "ac97", so enable
> that in the XML parser / qemu driver.
>
> Also remove some unused cruft relating to sound in Xen.
>
> domain_conf.c |3 ++-
> domain_conf.h |1 +
>
"Daniel P. Berrange" wrote:
> On Wed, Jan 14, 2009 at 09:57:26AM +0100, Jim Meyering wrote:
>> Jim Meyering wrote:
>> > "Daniel P. Berrange" wrote:
>> >> On Tue, Jan 13, 2009 at 05:48:12PM +, Daniel P. Berrange wrote:
>> >>> This adds support for the domain events in the test driver. Code
>>
On Wed, Jan 14, 2009 at 09:57:26AM +0100, Jim Meyering wrote:
> Jim Meyering wrote:
> > "Daniel P. Berrange" wrote:
> >> On Tue, Jan 13, 2009 at 05:48:12PM +, Daniel P. Berrange wrote:
> >>> This adds support for the domain events in the test driver. Code
> >>> is following the same pattern a
On Wed, Jan 14, 2009 at 08:55:46AM +, Mark McLoughlin wrote:
> On Tue, 2009-01-13 at 20:49 +, Daniel P. Berrange wrote:
> > Currently when we shutdown the virtual networks are all shutdown too.
> > +errno = EINVAL;
> > +return -1;
> > +}
> > +
> > +memset(&ifr, 0, s
On Tue, Jan 13, 2009 at 03:00:17PM -0500, Sir Woody Hackswell wrote:
> Here is a small patch for libvirtd init and sysconfig script. Before
> killing libvirtd, we virsh shutdown any running domains. We also have
> a maximum time limit for shutdown (300 sec default), just in case the
> VM will not
On Tue, 2009-01-13 at 15:00 -0500, Sir Woody Hackswell wrote:
> Here is a small patch for libvirtd init and sysconfig script. Before
> killing libvirtd, we virsh shutdown any running domains. We also have
> a maximum time limit for shutdown (300 sec default), just in case the
> VM will not shut d
Jim Meyering wrote:
> "Daniel P. Berrange" wrote:
>> On Tue, Jan 13, 2009 at 05:48:12PM +, Daniel P. Berrange wrote:
>>> This adds support for the domain events in the test driver. Code
>>> is following the same pattern as the impl in the QEMU driver.
>>>
>>> test.c | 223
>>> +
On Tue, 2009-01-13 at 20:49 +, Daniel P. Berrange wrote:
> Currently when we shutdown the virtual networks are all shutdown too.
> This is less than useful if we're letting guest VMs hang around post
> shutdown of libvirtd, because it means we're tearing their network
> connection out from und
"Daniel P. Berrange" wrote:
> Tearing a guest's storage out from under its feet on libvirtd shutdown
> is just as bad as tearing out its network :-) This patch removes the
> code which shuts down storage pool when the daemon shuts down. So NFS
> mounts stay around, LVM VGs remain active, and iSCSI
62 matches
Mail list logo