2022-12-02 18:40:57

by Reinette Chatre

[permalink] [raw]
Subject: [PATCH 0/3] dmaengine: idxd: Error path fixes

Dear Maintainers,

I have been using the IDXD driver to experiment with the upcoming core
changes in support of IMS ([1], [2], [3]). As part of this work I
happened to exercise the error paths within IDXD and encountered
a few issues that are addressed in this series. These changes are
independent from IMS and just aims to make the IDXD driver more
robust against errors.

It is not clear to me if these are appropriate for stable so I am
not including the stable team. Please let me know if you think
otherwise and I can add the necessary Cc. With the refactoring
through the history of the driver I was not able to
identify a Fixes: candidate for all. Patch #3 does look to be a
potentially complicated backport.

Your feedback is greatly appreciated.

Reinette

[1] https://lore.kernel.org/lkml/[email protected]
[2] https://lore.kernel.org/lkml/[email protected]
[3] https://lore.kernel.org/lkml/[email protected]

Reinette Chatre (3):
dmaengine: idxd: Let probe fail when workqueue cannot be enabled
dmaengine: idxd: Prevent use after free on completion memory
dmaengine: idxd: Do not call DMX TX callbacks during workqueue disable

drivers/dma/idxd/device.c | 15 ++++++++++++---
1 file changed, 12 insertions(+), 3 deletions(-)


base-commit: a4412fdd49dc011bcc2c0d81ac4cab7457092650
--
2.34.1


2022-12-02 22:11:02

by Fenghua Yu

[permalink] [raw]
Subject: RE: [PATCH 0/3] dmaengine: idxd: Error path fixes

Hi, Reinette,

> It is not clear to me if these are appropriate for stable so I am not including the
> stable team. Please let me know if you think otherwise and I can add the
> necessary Cc. With the refactoring through the history of the driver I was not
> able to identify a Fixes: candidate for all. Patch #3 does look to be a potentially
> complicated backport.

It would be better to add Cc: stable because they cause series issues.

In patch 3, you may add Fixes: 403a2e236538. This is in 5.16. 5.17 and later support
SVA. We don't need to backport to too early versions.

Thanks.

-Fenghua

2022-12-02 22:44:25

by Reinette Chatre

[permalink] [raw]
Subject: Re: [PATCH 0/3] dmaengine: idxd: Error path fixes

Hi Fenghua,

On 12/2/2022 1:32 PM, Yu, Fenghua wrote:
> Hi, Reinette,
>
>> It is not clear to me if these are appropriate for stable so I am not including the
>> stable team. Please let me know if you think otherwise and I can add the
>> necessary Cc. With the refactoring through the history of the driver I was not
>> able to identify a Fixes: candidate for all. Patch #3 does look to be a potentially
>> complicated backport.
>
> It would be better to add Cc: stable because they cause series issues.

Thank you. Will do.

>
> In patch 3, you may add Fixes: 403a2e236538. This is in 5.16. 5.17 and later support
> SVA. We don't need to backport to too early versions.

ok. Thank you very much.

Reinette