Re: [PATCH net-next v22 00/14] Replace page_frag with page_frag_cache for sk_page_frag()

From: Yunsheng Lin
Date: Sat Oct 19 2024 - 04:27:23 EST


On 10/19/2024 1:39 AM, Alexander Duyck wrote:


So I still think this set should be split in half in order to make
this easier to review. The ones I have provided a review-by for so far
seem fine to me. I really think if you just submitted that batch first
we can get that landed and let them stew in the kernel for a bit to
make sure we didn't miss anything there.

It makes sense to me too that it might be better to get those submitted
to get more testing if there is no more comment about it.

I am guessing they should be targetting net-next tree to get more
testing as all the callers of page_frag API seem to be in the
networking, right?

Hi, David, Jakub & Paolo
It would be good if those patches are just cherry-picked from this
patchset as those patches with 'Reviewed-by' tag seem to be applying
cleanly. Or any better suggestion here?


As far as the others there is a bunch there for me to try and chew
through. A bunch of that is stuff not related necessarily to my
version of the page frag stuff that I did so merging the two is a bit
less obvious to me. The one thing I am wondering about is the behavior
for why we are pulling apart the logic with this "commit" approach
that is deferring the offset update which seems to have to happen
unless we need to abort.

Let's discuss that in patch 7.


My review time is going to be limited for the next several weeks. As
such I will likely not be able to get to a review until Friday or
Saturday each week so sending out updates faster than that will not
get you any additional reviews from me.

Thanks for the time reviewing and reminding about the above.
It makes sense to have more time to have more reviewing as long as we
are making productive progress.


Thanks,

- Alex