Re: fix buffer refcount races

[Christoph Hellwig <hch@lst.de>]

On Sun, Jan 12, 2025 at 09:08:46PM -0800, Darrick J. Wong wrote:
> > This might actually be able to trigger the first one, but otherwise
> > just means we're doing a pass through insert which will find it.
> > For pure lookups using xfs_buf_incore it could cause us to miss buffer
> > invalidation.  The fix for that is bigger and has bigger implications
> > because it not requires all b_hold increments to be done under d_lock.
> 
> Just to be clear, should this sentence say
> "...because it *now* requires"?

Yes.

> > This causes more contention, but as releasing the buffer always takes
> > the lock it can't be too horrible.  I also have a only minimally
> > tested series to switch it over to a lockref here:
> > 
> >     http://git.infradead.org/?p=users/hch/xfs.git;a=shortlog;h=refs/heads/xfs-buffer-locking
> 
> Will take a look; some of those patches look familiar. ;)

Well, the first batch after these fixes are the buffer cleanups I
reposted last week that you've mostly but not entirely reviewed.  The
reminder only really depends on them for changed context.