Re: use memcpy_mcsafe() for copy_to_iter() (was: Re: [PATCH v3 0/9] Series short description)

[Dan Williams]

On Mon, May 14, 2018 at 11:49 PM, Ingo Molnar <mingo@xxxxxxxxxx> wrote:
>
> * Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
>
>> On Mon, May 14, 2018 at 12:26 AM, Ingo Molnar <mingo@xxxxxxxxxx> wrote:
>> >
>> > * Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
>> >
>> >> Ingo, Thomas, Al, any concerns with this series?
>> >
>> > Yeah, so:
>> >
>> >    "[PATCH v3 0/9] Series short description"
>> >
>> > ... isn't the catchiest of titles to capture my [all too easily distracted]
>> > attention! ;-)
>>
>> My bad! After that mistake it became a toss-up between more spam and
>> hoping the distraction would not throw you off.
>>
>> > I have marked it now for -tip processing. Linus was happy with this and acked the
>> > approach, right?
>>
>> I think "happy" is a strong word when it comes to x86 machine check
>> handling. My interpretation is that he and Andy acquiesced that this
>> is about the best we can do with dax+mce as things stand today.
>
> So, how would you like to go about this series?
>
> To help move it forward I applied the first 5 commits to tip:x86/dax, on a
> vanilla v4.17-rc5 base, did some minor edits to the changelogs, tested it
> superficially (I don't have DAX so this essentially means build tests) and
> pushed out the result.

Thanks for that. Technically speaking, you do have dax, but setting up
our unit tests is currently not friction free, so I would not expect
you to go through that effort. Hopefully we can revive 0day running
our unit tests one of these days.

> Barring some later generic-x86 regression (unlikely) this looks good to me - feel
> free to cross-pull that branch into your DAX/nvdimm tree.
>
> Or we could apply the remaining changes to -tip too - your call.

The remainder patches have developed a conflict with another topic
branch in the nvdimm tree, in particular "dax: introduce a
->copy_to_iter dax operation". I think the best course is for me to
rebase the remaining 4 on top of tip/x86/dax and carry the merge
conflict through the nvdimm tree.

> Thanks,
>
>         Ingo

Thanks!