Re: [RFC 2/2] iomap: Support subpage size dirty tracking to improve write performance

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Nov 02, 2022 at 10:35:38AM -0700, Darrick J. Wong wrote:
> Just so long as nobody imports that nonfeature of the bufferhead code
> where dirtying an already dirty bufferhead skips marking the folio dirty
> and writepage failures also fail to redirty the page.  That bit us hard
> recently and I strongly prefer not to repeat that.

Yes, that absolutely needs to be avoided.

> > We can always optimize by having a bit for the fairly common all dirty
> > case and only track and look at the array if that is no the case.
> 
> Yes, it would help to make the ranges in the bit array better defined
> than the semi-opencoded logic there is now.  (I'm whining specifically
> about the test_bit calls sprinkled around).

That is an absolutely requirement.  It was so obviosu to me that I
didn't bother to restate it after two others already pointed it out :)

> Once that's done it
> shouldn't be hard to add one more bit for the all-dirty state.  Though
> I'd want to see the numbers to prove that it saves us time anywhere.

We might be able to just use PG_private_2 in the page for now, but
maybe just adding a flags field to the iomap_page might be a better
idea, as the pageflags tend to have strange entanglements.



[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [NTFS 3]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [NTFS 3]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux