Re: [PATCH 6/6] vfs: change inode times to use struct timespec64

[Deepa Dinamani]

Al,

Are you ok with this approach to changing vfs timestamps?

Kees mentioned that he wants to merge a patch to pstore that changes
it to use timespec64 internally for 4.17:
https://lkml.org/lkml/2018/5/13/3

I'm not sure how we usually merge such flag day patches. Should this
be targeted for 4.17 or 4.18? The above might or might not be a
problem based on when this series is merged.

If you are ok with this approach, I could post a v2 with a couple of
requested fix-ups.

-Deepa

On Fri, May 11, 2018 at 11:44 PM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> On Fri, May 11, 2018 at 9:59 PM, Deepa Dinamani <deepa.kernel@xxxxxxxxx> wrote:
>> diff --git a/fs/pstore/inode.c b/fs/pstore/inode.c
>> index 5fcb845b9fec..fb681d302bb3 100644
>> --- a/fs/pstore/inode.c
>> +++ b/fs/pstore/inode.c
>> @@ -392,7 +392,7 @@ int pstore_mkfile(struct dentry *root, struct pstore_record *record)
>>         inode->i_private = private;
>>
>>         if (record->time.tv_sec)
>> -               inode->i_mtime = inode->i_ctime = record->time;
>> +               inode->i_mtime = inode->i_ctime = timespec_to_timespec64(record->time);
>>
>>         d_add(dentry, inode);
>
> I'm fine to just convert pstore internally to timespec64 right now. Is
> it correct to say that I should use timespec64_to_timespec() here
> until this flag day patch? And I'd need to do this as well, yes?
>
> fs/pstore/platform.c:     record->time =
> ns_to_timespec64(ktime_get_real_fast_ns());
>
> Thanks!
>
> -Kees
>
> --
> Kees Cook
> Pixel Security