[v2] (v)pci: fixes related to memory decoding handling

[PATCH for-4.17 v2 0/5] (v)pci: fixes related to memory decoding handling

Posted by Roger Pau Monne 1 year, 6 months ago

Hello,

This patch series attempts to fix the regressions caused by 75cc460a1b
('xen/pci: detect when BARs are not suitably positioned') and the last
patch relaxes the check done when attempting to write to BARs with
memory decoding enabled.

I consider all of them bug fixes, albeit the last patch is not fixing a
regression (since vPCI code has always behaved this way).

Thanks, Roger.

Roger Pau Monne (5):
  vpci: don't assume that vpci per-device data exists unconditionally
  vpci/msix: remove from table list on detach
  vpci: introduce a local vpci_bar variable to modify_decoding()
  pci: do not disable memory decoding for devices
  vpci: refuse BAR writes only if the BAR is mapped

 xen/drivers/passthrough/pci.c | 69 -----------------------------------
 xen/drivers/vpci/header.c     | 52 ++++++++++++++++++--------
 xen/drivers/vpci/vpci.c       | 14 ++++---
 xen/include/xen/vpci.h        |  6 +++
 4 files changed, 52 insertions(+), 89 deletions(-)

-- 
2.37.3

Re: [PATCH for-4.17 v2 0/5] (v)pci: fixes related to memory decoding handling

Posted by Jan Beulich 1 year, 6 months ago

On 25.10.2022 16:44, Roger Pau Monne wrote:
> Hello,
> 
> This patch series attempts to fix the regressions caused by 75cc460a1b
> ('xen/pci: detect when BARs are not suitably positioned') and the last
> patch relaxes the check done when attempting to write to BARs with
> memory decoding enabled.
> 
> I consider all of them bug fixes, albeit the last patch is not fixing a
> regression (since vPCI code has always behaved this way).
> 
> Thanks, Roger.
> 
> Roger Pau Monne (5):
>   vpci: don't assume that vpci per-device data exists unconditionally
>   vpci/msix: remove from table list on detach
>   vpci: introduce a local vpci_bar variable to modify_decoding()
>   pci: do not disable memory decoding for devices
>   vpci: refuse BAR writes only if the BAR is mapped

Looks like you've lost Henry's release-ack that was given for all of v1
(i.e. only patch 2 isn't covered by it).

Jan

Re: [PATCH for-4.17 v2 0/5] (v)pci: fixes related to memory decoding handling

Posted by Roger Pau Monné 1 year, 6 months ago

On Tue, Oct 25, 2022 at 05:02:41PM +0200, Jan Beulich wrote:
> On 25.10.2022 16:44, Roger Pau Monne wrote:
> > Hello,
> > 
> > This patch series attempts to fix the regressions caused by 75cc460a1b
> > ('xen/pci: detect when BARs are not suitably positioned') and the last
> > patch relaxes the check done when attempting to write to BARs with
> > memory decoding enabled.
> > 
> > I consider all of them bug fixes, albeit the last patch is not fixing a
> > regression (since vPCI code has always behaved this way).
> > 
> > Thanks, Roger.
> > 
> > Roger Pau Monne (5):
> >   vpci: don't assume that vpci per-device data exists unconditionally
> >   vpci/msix: remove from table list on detach
> >   vpci: introduce a local vpci_bar variable to modify_decoding()
> >   pci: do not disable memory decoding for devices
> >   vpci: refuse BAR writes only if the BAR is mapped
> 
> Looks like you've lost Henry's release-ack that was given for all of v1
> (i.e. only patch 2 isn't covered by it).

I was worry about adding it again for the whole series (except patch
2), as I think release-acks are conditional to the time they are
given.  IOW: a release-ack given for a previous series sent maybe
weeks ago shouldn't be carried over because conditions to get changes
accepted might be tighter as we progress with the release.

I think Henry would be fine to reassess the suitability of the series
once it gets properly Acked.

Thanks, Roger.

RE: [PATCH for-4.17 v2 0/5] (v)pci: fixes related to memory decoding handling

Posted by Henry Wang 1 year, 6 months ago

Hi Roger and Jan,

> -----Original Message-----
> From: Roger Pau Monné <roger.pau@citrix.com>
> Subject: Re: [PATCH for-4.17 v2 0/5] (v)pci: fixes related to memory decoding
> handling
> 
> On Tue, Oct 25, 2022 at 05:02:41PM +0200, Jan Beulich wrote:
> > On 25.10.2022 16:44, Roger Pau Monne wrote:
> > > Hello,
> > >
> > > This patch series attempts to fix the regressions caused by 75cc460a1b
> > > ('xen/pci: detect when BARs are not suitably positioned') and the last
> > > patch relaxes the check done when attempting to write to BARs with
> > > memory decoding enabled.
> > >
> > > I consider all of them bug fixes, albeit the last patch is not fixing a
> > > regression (since vPCI code has always behaved this way).
> > >
> > > Thanks, Roger.
> > >
> > > Roger Pau Monne (5):
> > >   vpci: don't assume that vpci per-device data exists unconditionally
> > >   vpci/msix: remove from table list on detach
> > >   vpci: introduce a local vpci_bar variable to modify_decoding()
> > >   pci: do not disable memory decoding for devices
> > >   vpci: refuse BAR writes only if the BAR is mapped
> >
> > Looks like you've lost Henry's release-ack that was given for all of v1
> > (i.e. only patch 2 isn't covered by it).
> 
> I was worry about adding it again for the whole series (except patch
> 2), as I think release-acks are conditional to the time they are
> given.  IOW: a release-ack given for a previous series sent maybe
> weeks ago shouldn't be carried over because conditions to get changes
> accepted might be tighter as we progress with the release.

Thanks for being considerate! The release ack still holds.

> 
> I think Henry would be fine to reassess the suitability of the series
> once it gets properly Acked.

No problem, for the whole series (given that this is a bug fix series)

Release-acked-by: Henry Wang <Henry.Wang@arm.com>

But a reminder: patch #4 and #5 need review/ack from maintainers.

Kind regards,
Henry

> 
> Thanks, Roger.