Hi Jens,
On Thu, May 19, 2022 at 05:33:01PM -0600, Jens Axboe wrote:
> On 5/19/22 5:25 PM, Jason A. Donenfeld wrote:
> > Hi Jens,
> >
> > On Fri, May 20, 2022 at 1:22 AM Jens Axboe <[email protected]> wrote:
> >> I can certainly do the write side too. To fix this regression, I just
> >> valued doing read_iter first and I'd hate to hold that up to do the
> >> write side too. I'll do the write side later today, but let's keep them
> >> separate.
> >
> > Excellent, thanks. I plan to queue these up all in a row.
>
> Built and tested v2, just sent it out. Note that it deviates from your
> proposal a bit since with that we lost the
>
> if (!len)
> break;
>
> check, which is kind of important if you ever want to be done :-)
Heh, noticed that too, thanks.
> I'll do the write_iter side, but as mentioned, I'd prefer to keep it
> separate from this patchset as this one fixes a real regression that we
> need to get backported too.
No problem. Because of all the flux in random.c lately, I've been
preparing a massive backports branch, 2 branches actually, so I'll make
sure this is in there. Backport concern aside, though, I'll look for
your write_iter patch today. Thanks a bunch for doing this.
Jason
On 5/19/22 5:39 PM, Jason A. Donenfeld wrote:
> Hi Jens,
>
> On Thu, May 19, 2022 at 05:33:01PM -0600, Jens Axboe wrote:
>> On 5/19/22 5:25 PM, Jason A. Donenfeld wrote:
>>> Hi Jens,
>>>
>>> On Fri, May 20, 2022 at 1:22 AM Jens Axboe <[email protected]> wrote:
>>>> I can certainly do the write side too. To fix this regression, I just
>>>> valued doing read_iter first and I'd hate to hold that up to do the
>>>> write side too. I'll do the write side later today, but let's keep them
>>>> separate.
>>>
>>> Excellent, thanks. I plan to queue these up all in a row.
>>
>> Built and tested v2, just sent it out. Note that it deviates from your
>> proposal a bit since with that we lost the
>>
>> if (!len)
>> break;
>>
>> check, which is kind of important if you ever want to be done :-)
>
> Heh, noticed that too, thanks.
:-)
>> I'll do the write_iter side, but as mentioned, I'd prefer to keep it
>> separate from this patchset as this one fixes a real regression that we
>> need to get backported too.
>
> No problem. Because of all the flux in random.c lately, I've been
> preparing a massive backports branch, 2 branches actually, so I'll make
> sure this is in there. Backport concern aside, though, I'll look for
> your write_iter patch today. Thanks a bunch for doing this.
Sounds great, thanks - write patch has been sent out too.
--
Jens Axboe