Re: Questions on patch set: x86/fault: #PF improvements, mostly related to USER bit

[Luming Yu]

try again.

On Thu, Aug 5, 2021 at 9:28 AM Yu, Luming <luming.yu@xxxxxxxxx> wrote:
>
> Cc’ gregkh
>
>
>
> It would be nice if we can see the patch set available in 5.10-xx stable tree.
>
> But Jing’s finding on Linus tree about the commit seemed to indicate the commit id
>
> Of the patch and the commit  log and patch could not lead to the  revision of arch/x86/mm/fault.c
>
> By the patch of the commit itself. So it is hard for Jing to do a clean back port with the inconsistent git log.
>
>
>
> BR
>
> Luming
>
>
>
> From: Han, Jing1 <jing1.han@xxxxxxxxx>
> Sent: Wednesday, August 4, 2021 8:44 PM
> To: Lutomirski, Andy <luto@xxxxxxxxxx>
> Cc: Yu, Luming <luming.yu@xxxxxxxxx>
> Subject: Questions on patch set: x86/fault: #PF improvements, mostly related to USER bit
>
>
>
> Hi Luto,
>
>
>
> When trying to backport your patch from upstream to 5.10.18: https://lore.kernel.org/lkml/cover.1612924255.git.luto@xxxxxxxxxx/
>
> I find something wrong with this commit: 2cc624b0a7e68ba8957b18600181f7d5b0f3e1b6 x86/fault: Split the OOPS code out from no_context()
>
> The code after this commit in upstream does not align with the code showed in the patch.
>
> To be exactly, the function name is “no_context” in the patch, while it is “kernelmode_fixup_or_oops” in upstream code.
>
> Could you please check what is the problem?
>
> (Another confusion is, there are 14 patches total in the link above, but 13 patches in upstream code.)
>
>
>
> I do the backport it because it is preparation for CET, I will backport CET to 5.10.18 when it is upstreamed as there is customer request.
>
> Will you provide the backport patch to 5.10 stable branch?
>
>
>
> Thank you very much,
>
> Jing
>
>