Re: [RFCv2 2/3] iomap: Change uptodate variable name to state

From: Christoph Hellwig <hch@infradead.org>
To: Matthew Wilcox <willy@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>,
	"Ritesh Harjani (IBM)" <ritesh.list@gmail.com>,
	linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	Aravinda Herle <araherle@in.ibm.com>
Subject: Re: [RFCv2 2/3] iomap: Change uptodate variable name to state
Date: Tue, 31 Jan 2023 07:07:01 -0800	[thread overview]
Message-ID: <Y9kulWxXxcYye09a@infradead.org> (raw)
In-Reply-To: <Y9hDu8hVBa3qJTNw@casper.infradead.org>

On Mon, Jan 30, 2023 at 10:24:59PM +0000, Matthew Wilcox wrote:
> > a single bitmap of undefined length, then change the declaration and
> > the structure size calculation away from using array notation and
> > instead just use pointers to the individual bitmap regions within
> > the allocated region.
> 
> Hard to stomach that solution when the bitmap is usually 2 bytes long
> (in Ritesh's case).  Let's see a version of this patchset with
> accessors before rendering judgement.

Yes.  I think what we need is proper helpers that are self-documenting
for every bitmap update as I already suggsted last round.  That keeps
the efficient allocation, and keeps all the users self-documenting.
It just adds a bit of boilerplate for all these helpers, but that
should be worth having the clarity and performance benefits.