Patchwork [PULL] pci,msi,virtio

login
register
mail settings
Submitter Alex Williamson
Date July 23, 2012, 5:05 p.m.
Message ID <1343063125.2229.269.camel@bling.home>
Download mbox | patch
Permalink /patch/172714/
State New
Headers show

Comments

Alex Williamson - July 23, 2012, 5:05 p.m.
On Mon, 2012-07-23 at 11:39 -0500, Anthony Liguori wrote:
> Andreas Färber <afaerber@suse.de> writes:
> 
> > Am 19.07.2012 17:15, schrieb Michael S. Tsirkin:
> >> The following changes since commit 80aa796bf38b7ef21daa42673b4711510c450d8a:
> >> 
> >>   pci_bridge_dev: fix error path in pci_bridge_dev_initfn() (2012-06-11 22:55:13 +0300)
> >> 
> >> are available in the git repository at:
> >> 
> >>   git://git.kernel.org/pub/scm/virt/kvm/mst/qemu.git tags/for_anthony
> >> 
> >> for you to fetch changes up to 932d4a42afa28829fadf3cbfbb0507cc09aafd8b:
> >> 
> >>   msi/msix: added API to set MSI message address and data (2012-07-19 17:56:42 +0300)
> >> 
> >> ----------------------------------------------------------------
> >> pci,msi,virtio
> >> 
> >> This pull includes preparation patches mostly by Jan and Alex
> >> that should help merge device assignment down the road.
> >> And there's a new API needed for emulating POWER firmware.
> >> 
> >> So no new functionality and some unused APIs but it looks like
> >> merging will help people make progress.
> >> 
> >> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> >
> > Usually, PULLs are expected to carry the individual patches as replies.
> >
> > But more important, did something go wrong with rebasing before sending
> > out the PULL? June 11 is more than a month ago. And if I try to rebase
> > my pci_host branch on your "pci" branch it tries to replay loads of
> > really old post-1.1 commits (e.g., my "Pass PowerPCCPU to...") and
> > fails... am I doing something wrong? "for_anthony" tag and "pci" branch
> > seem to match in date at least.
> 
> 
> It's a tag, not a branch.
> 
> I had to add an explicit remote to pull tags in.  I'm not sure if
> there's a better way to do it in git.

Looks like it merged ok for me, but you'll need this to complete the
PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.
Anthony, it'd be great if we could merge and fix this in your tree since
MST is out of the office.  Thanks,

Alex

commit 34e37573f1a983dab43673d664eb500bc87a46d4
Author: Alex Williamson <alex.williamson@redhat.com>
Date:   Mon Jul 23 10:59:43 2012 -0600

    esp: Update PCIUnregisterFunc function
    
    Function prototype changed in f90c2bcd
    
    Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Anthony Liguori - July 23, 2012, 6:16 p.m.
On 07/23/2012 12:05 PM, Alex Williamson wrote:
> On Mon, 2012-07-23 at 11:39 -0500, Anthony Liguori wrote:
>> Andreas Färber<afaerber@suse.de>  writes:
>>
>>> Am 19.07.2012 17:15, schrieb Michael S. Tsirkin:
>>>> The following changes since commit 80aa796bf38b7ef21daa42673b4711510c450d8a:
>>>>
>>>>    pci_bridge_dev: fix error path in pci_bridge_dev_initfn() (2012-06-11 22:55:13 +0300)
>>>>
>>>> are available in the git repository at:
>>>>
>>>>    git://git.kernel.org/pub/scm/virt/kvm/mst/qemu.git tags/for_anthony
>>>>
>>>> for you to fetch changes up to 932d4a42afa28829fadf3cbfbb0507cc09aafd8b:
>>>>
>>>>    msi/msix: added API to set MSI message address and data (2012-07-19 17:56:42 +0300)
>>>>
>>>> ----------------------------------------------------------------
>>>> pci,msi,virtio
>>>>
>>>> This pull includes preparation patches mostly by Jan and Alex
>>>> that should help merge device assignment down the road.
>>>> And there's a new API needed for emulating POWER firmware.
>>>>
>>>> So no new functionality and some unused APIs but it looks like
>>>> merging will help people make progress.
>>>>
>>>> Signed-off-by: Michael S. Tsirkin<mst@redhat.com>
>>>
>>> Usually, PULLs are expected to carry the individual patches as replies.
>>>
>>> But more important, did something go wrong with rebasing before sending
>>> out the PULL? June 11 is more than a month ago. And if I try to rebase
>>> my pci_host branch on your "pci" branch it tries to replay loads of
>>> really old post-1.1 commits (e.g., my "Pass PowerPCCPU to...") and
>>> fails... am I doing something wrong? "for_anthony" tag and "pci" branch
>>> seem to match in date at least.
>>
>>
>> It's a tag, not a branch.
>>
>> I had to add an explicit remote to pull tags in.  I'm not sure if
>> there's a better way to do it in git.
>
> Looks like it merged ok for me, but you'll need this to complete the
> PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.
> Anthony, it'd be great if we could merge and fix this in your tree since
> MST is out of the office.  Thanks,
>
> Alex

I can't fix it without rewriting git history or breaking bisection which isn't 
something I'm really willing to do.

When does MST get back?  If it's not for a while, I can merge it all manually 
with a fixup.

Regards,

Anthony Liguori

>
> commit 34e37573f1a983dab43673d664eb500bc87a46d4
> Author: Alex Williamson<alex.williamson@redhat.com>
> Date:   Mon Jul 23 10:59:43 2012 -0600
>
>      esp: Update PCIUnregisterFunc function
>
>      Function prototype changed in f90c2bcd
>
>      Signed-off-by: Alex Williamson<alex.williamson@redhat.com>
>
> diff --git a/hw/esp.c b/hw/esp.c
> index c6422ad..a011347 100644
> --- a/hw/esp.c
> +++ b/hw/esp.c
> @@ -1153,13 +1153,11 @@ static int esp_pci_scsi_init(PCIDevice *dev)
>       return 0;
>   }
>
> -static int esp_pci_scsi_uninit(PCIDevice *d)
> +static void esp_pci_scsi_uninit(PCIDevice *d)
>   {
>       PCIESPState *pci = DO_UPCAST(PCIESPState, dev, d);
>
>       memory_region_destroy(&pci->io);
> -
> -    return 0;
>   }
>
>   static void esp_pci_class_init(ObjectClass *klass, void *data)
>
>
>
>
Alex Williamson - July 23, 2012, 6:29 p.m.
On Mon, 2012-07-23 at 13:16 -0500, Anthony Liguori wrote:
> On 07/23/2012 12:05 PM, Alex Williamson wrote:
> > On Mon, 2012-07-23 at 11:39 -0500, Anthony Liguori wrote:
> >> Andreas Färber<afaerber@suse.de>  writes:
> >>
> >>> Am 19.07.2012 17:15, schrieb Michael S. Tsirkin:
> >>>> The following changes since commit 80aa796bf38b7ef21daa42673b4711510c450d8a:
> >>>>
> >>>>    pci_bridge_dev: fix error path in pci_bridge_dev_initfn() (2012-06-11 22:55:13 +0300)
> >>>>
> >>>> are available in the git repository at:
> >>>>
> >>>>    git://git.kernel.org/pub/scm/virt/kvm/mst/qemu.git tags/for_anthony
> >>>>
> >>>> for you to fetch changes up to 932d4a42afa28829fadf3cbfbb0507cc09aafd8b:
> >>>>
> >>>>    msi/msix: added API to set MSI message address and data (2012-07-19 17:56:42 +0300)
> >>>>
> >>>> ----------------------------------------------------------------
> >>>> pci,msi,virtio
> >>>>
> >>>> This pull includes preparation patches mostly by Jan and Alex
> >>>> that should help merge device assignment down the road.
> >>>> And there's a new API needed for emulating POWER firmware.
> >>>>
> >>>> So no new functionality and some unused APIs but it looks like
> >>>> merging will help people make progress.
> >>>>
> >>>> Signed-off-by: Michael S. Tsirkin<mst@redhat.com>
> >>>
> >>> Usually, PULLs are expected to carry the individual patches as replies.
> >>>
> >>> But more important, did something go wrong with rebasing before sending
> >>> out the PULL? June 11 is more than a month ago. And if I try to rebase
> >>> my pci_host branch on your "pci" branch it tries to replay loads of
> >>> really old post-1.1 commits (e.g., my "Pass PowerPCCPU to...") and
> >>> fails... am I doing something wrong? "for_anthony" tag and "pci" branch
> >>> seem to match in date at least.
> >>
> >>
> >> It's a tag, not a branch.
> >>
> >> I had to add an explicit remote to pull tags in.  I'm not sure if
> >> there's a better way to do it in git.
> >
> > Looks like it merged ok for me, but you'll need this to complete the
> > PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.
> > Anthony, it'd be great if we could merge and fix this in your tree since
> > MST is out of the office.  Thanks,
> >
> > Alex
> 
> I can't fix it without rewriting git history or breaking bisection which isn't 
> something I'm really willing to do.
> 
> When does MST get back?  If it's not for a while, I can merge it all manually 
> with a fixup.

Next week, so (unfortunately for me) probably not worth you messing with
it.  Thanks,

Alex


> > commit 34e37573f1a983dab43673d664eb500bc87a46d4
> > Author: Alex Williamson<alex.williamson@redhat.com>
> > Date:   Mon Jul 23 10:59:43 2012 -0600
> >
> >      esp: Update PCIUnregisterFunc function
> >
> >      Function prototype changed in f90c2bcd
> >
> >      Signed-off-by: Alex Williamson<alex.williamson@redhat.com>
> >
> > diff --git a/hw/esp.c b/hw/esp.c
> > index c6422ad..a011347 100644
> > --- a/hw/esp.c
> > +++ b/hw/esp.c
> > @@ -1153,13 +1153,11 @@ static int esp_pci_scsi_init(PCIDevice *dev)
> >       return 0;
> >   }
> >
> > -static int esp_pci_scsi_uninit(PCIDevice *d)
> > +static void esp_pci_scsi_uninit(PCIDevice *d)
> >   {
> >       PCIESPState *pci = DO_UPCAST(PCIESPState, dev, d);
> >
> >       memory_region_destroy(&pci->io);
> > -
> > -    return 0;
> >   }
> >
> >   static void esp_pci_class_init(ObjectClass *klass, void *data)
> >
> >
> >
> >
> 
>
Andreas Färber - July 24, 2012, 11:32 a.m.
Am 23.07.2012 20:29, schrieb Alex Williamson:
> On Mon, 2012-07-23 at 13:16 -0500, Anthony Liguori wrote:
>> On 07/23/2012 12:05 PM, Alex Williamson wrote:
>>> Looks like it merged ok for me, but you'll need this to complete the
>>> PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.
>>> Anthony, it'd be great if we could merge and fix this in your tree since
>>> MST is out of the office.
>>
>> I can't fix it without rewriting git history or breaking bisection which isn't 
>> something I'm really willing to do.
>>
>> When does MST get back?  If it's not for a while, I can merge it all manually 
>> with a fixup.
> 
> Next week, so (unfortunately for me) probably not worth you messing with
> it.

So Anthony, how urgent is your/Wanpeng's i440fx rework and through whose
queue is that supposed to go? Can you wait for mst's return? Or do you
want to handle it yourself and have mst rebase on it while at it?

My pci-host branch [1] is updated with mst's requested spapr_pci.c code
movement, so I just need to know whether to rebase on some other branch
and when to re-post. Soft Freeze is in three weeks iirc.

Regards,
Andreas

[1] http://repo.or.cz/w/qemu/afaerber.git/shortlog/refs/heads/pci-host
Alexey Kardashevskiy - July 25, 2012, 1:56 a.m.
On 24/07/12 21:32, Andreas Färber wrote:
> Am 23.07.2012 20:29, schrieb Alex Williamson:
>> On Mon, 2012-07-23 at 13:16 -0500, Anthony Liguori wrote:
>>> On 07/23/2012 12:05 PM, Alex Williamson wrote:
>>>> Looks like it merged ok for me, but you'll need this to complete the
>>>> PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.
>>>> Anthony, it'd be great if we could merge and fix this in your tree since
>>>> MST is out of the office.
>>>
>>> I can't fix it without rewriting git history or breaking bisection which isn't
>>> something I'm really willing to do.
>>>
>>> When does MST get back?  If it's not for a while, I can merge it all manually
>>> with a fixup.
>>
>> Next week, so (unfortunately for me) probably not worth you messing with
>> it.
>
> So Anthony, how urgent is your/Wanpeng's i440fx rework and through whose
> queue is that supposed to go? Can you wait for mst's return? Or do you
> want to handle it yourself and have mst rebase on it while at it?
>
> My pci-host branch [1] is updated with mst's requested spapr_pci.c code
> movement, so I just need to know whether to rebase on some other branch
> and when to re-post. Soft Freeze is in three weeks iirc.


A bit curious - where is spapr_pci.c going to be moved to?


>
> Regards,
> Andreas
>
> [1] http://repo.or.cz/w/qemu/afaerber.git/shortlog/refs/heads/pci-host
>
Andreas Färber - July 25, 2012, 9:16 a.m.
Am 25.07.2012 03:56, schrieb Alexey Kardashevskiy:
> On 24/07/12 21:32, Andreas Färber wrote:
>> My pci-host branch [1] is updated with mst's requested spapr_pci.c code
>> movement, so I just need to know whether to rebase on some other branch
>> and when to re-post. Soft Freeze is in three weeks iirc.
> 
> 
> A bit curious - where is spapr_pci.c going to be moved to?

Code movement, not file movement. You can see the patches in the link
below. The above referred to code moved from patch 1 to the spapr_pci
patch - Ben was previously cc'ed on the cover letter but despite my
complaining no one added the sPAPR machine to MAINTAINERS, so the
individual patches don't show up on qemu-ppc.

Andreas

>> [1] http://repo.or.cz/w/qemu/afaerber.git/shortlog/refs/heads/pci-host
Michael S. Tsirkin - July 29, 2012, 1:58 p.m.
On Mon, Jul 23, 2012 at 11:05:25AM -0600, Alex Williamson wrote:
> On Mon, 2012-07-23 at 11:39 -0500, Anthony Liguori wrote:
> > Andreas Färber <afaerber@suse.de> writes:
> > 
> > > Am 19.07.2012 17:15, schrieb Michael S. Tsirkin:
> > >> The following changes since commit 80aa796bf38b7ef21daa42673b4711510c450d8a:
> > >> 
> > >>   pci_bridge_dev: fix error path in pci_bridge_dev_initfn() (2012-06-11 22:55:13 +0300)
> > >> 
> > >> are available in the git repository at:
> > >> 
> > >>   git://git.kernel.org/pub/scm/virt/kvm/mst/qemu.git tags/for_anthony
> > >> 
> > >> for you to fetch changes up to 932d4a42afa28829fadf3cbfbb0507cc09aafd8b:
> > >> 
> > >>   msi/msix: added API to set MSI message address and data (2012-07-19 17:56:42 +0300)
> > >> 
> > >> ----------------------------------------------------------------
> > >> pci,msi,virtio
> > >> 
> > >> This pull includes preparation patches mostly by Jan and Alex
> > >> that should help merge device assignment down the road.
> > >> And there's a new API needed for emulating POWER firmware.
> > >> 
> > >> So no new functionality and some unused APIs but it looks like
> > >> merging will help people make progress.
> > >> 
> > >> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > >
> > > Usually, PULLs are expected to carry the individual patches as replies.
> > >
> > > But more important, did something go wrong with rebasing before sending
> > > out the PULL? June 11 is more than a month ago. And if I try to rebase
> > > my pci_host branch on your "pci" branch it tries to replay loads of
> > > really old post-1.1 commits (e.g., my "Pass PowerPCCPU to...") and
> > > fails... am I doing something wrong? "for_anthony" tag and "pci" branch
> > > seem to match in date at least.
> > 
> > 
> > It's a tag, not a branch.
> > 
> > I had to add an explicit remote to pull tags in.  I'm not sure if
> > there's a better way to do it in git.
> 
> Looks like it merged ok for me, but you'll need this to complete the
> PCIUnregisterFunc conversion.  Not sure how this built in MST's tree.

I didn't rebase my tree, this one is newer.

> Anthony, it'd be great if we could merge and fix this in your tree since
> MST is out of the office.  Thanks,
> 
> Alex
> commit 34e37573f1a983dab43673d664eb500bc87a46d4
> Author: Alex Williamson <alex.williamson@redhat.com>
> Date:   Mon Jul 23 10:59:43 2012 -0600
> 
>     esp: Update PCIUnregisterFunc function
>     
>     Function prototype changed in f90c2bcd
>     
>     Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
> 
> diff --git a/hw/esp.c b/hw/esp.c
> index c6422ad..a011347 100644
> --- a/hw/esp.c
> +++ b/hw/esp.c
> @@ -1153,13 +1153,11 @@ static int esp_pci_scsi_init(PCIDevice *dev)
>      return 0;
>  }
>  
> -static int esp_pci_scsi_uninit(PCIDevice *d)
> +static void esp_pci_scsi_uninit(PCIDevice *d)
>  {
>      PCIESPState *pci = DO_UPCAST(PCIESPState, dev, d);
>  
>      memory_region_destroy(&pci->io);
> -
> -    return 0;
>  }
>  
>  static void esp_pci_class_init(ObjectClass *klass, void *data)
>

Patch

diff --git a/hw/esp.c b/hw/esp.c
index c6422ad..a011347 100644
--- a/hw/esp.c
+++ b/hw/esp.c
@@ -1153,13 +1153,11 @@  static int esp_pci_scsi_init(PCIDevice *dev)
     return 0;
 }
 
-static int esp_pci_scsi_uninit(PCIDevice *d)
+static void esp_pci_scsi_uninit(PCIDevice *d)
 {
     PCIESPState *pci = DO_UPCAST(PCIESPState, dev, d);
 
     memory_region_destroy(&pci->io);
-
-    return 0;
 }
 
 static void esp_pci_class_init(ObjectClass *klass, void *data)