[PATCH 0/5] xen/vpci: miscellaneous fixes

Roger Pau Monne posted 5 patches 2 months, 2 weeks ago
Patches applied successfully (tree, apply log)
git fetch https://gitlab.com/xen-project/patchew/xen tags/patchew/20250814160358.95543-1-roger.pau@citrix.com
xen/drivers/vpci/header.c | 186 ++++++++++++++++----------------------
xen/include/xen/vpci.h    |   5 +
2 files changed, 82 insertions(+), 109 deletions(-)
[PATCH 0/5] xen/vpci: miscellaneous fixes
Posted by Roger Pau Monne 2 months, 2 weeks ago
Hello,

The following series started with the report in:

https://lore.kernel.org/xen-devel/dbc003a2-9202-46ec-bf87-2829d8a63d53@citrix.com/

And ended up doing a bit of cleanup on the way.  I think the overall
result is nicer, I never really liked the deferral of the setting of the
command and ROM BAR registers, as it makes the logic much harder to
follow.  I've done some testing, but I will probably do some
specific testing to ensure the error paths work as expected - as I don't
have a system that trigger those.  Posting it ahead because some patches
are fairly trivial, and to get feedback on the approach.

Thanks, Roger.

Roger Pau Monne (5):
  xen/vpci: purge BAR rangeset contents before use
  xen/vpci: make BAR mapping more resilient for the hardware domain
  xen/vpci: simplify handling of memory decoding and ROM enable writes
  vpci/msix: move MSI-X hole punching as a result of memory decoding
    enable
  xen/vpci: only check BAR validity once

 xen/drivers/vpci/header.c | 186 ++++++++++++++++----------------------
 xen/include/xen/vpci.h    |   5 +
 2 files changed, 82 insertions(+), 109 deletions(-)

-- 
2.49.0
Re: [PATCH 0/5] xen/vpci: miscellaneous fixes
Posted by Chen, Jiqian 1 month, 2 weeks ago
On 2025/8/15 00:03, Roger Pau Monne wrote:
> Hello,
> 
> The following series started with the report in:
> 
> https://lore.kernel.org/xen-devel/dbc003a2-9202-46ec-bf87-2829d8a63d53@citrix.com/
> 
> And ended up doing a bit of cleanup on the way.  I think the overall
> result is nicer, I never really liked the deferral of the setting of the
> command and ROM BAR registers, as it makes the logic much harder to
> follow.  I've done some testing, but I will probably do some
> specific testing to ensure the error paths work as expected - as I don't
> have a system that trigger those. 
I have applied this series to my local environment and run some normal graphics apps without finding errors currently.
If you need to do some specific tests, I am happy to help testing.

> Posting it ahead because some patches are fairly trivial, and to get feedback on the approach.
> 
> Thanks, Roger.
> 
> Roger Pau Monne (5):
>   xen/vpci: purge BAR rangeset contents before use
>   xen/vpci: make BAR mapping more resilient for the hardware domain
>   xen/vpci: simplify handling of memory decoding and ROM enable writes
>   vpci/msix: move MSI-X hole punching as a result of memory decoding
>     enable
>   xen/vpci: only check BAR validity once
> 
>  xen/drivers/vpci/header.c | 186 ++++++++++++++++----------------------
>  xen/include/xen/vpci.h    |   5 +
>  2 files changed, 82 insertions(+), 109 deletions(-)
> 

-- 
Best regards,
Jiqian Chen.