Re: [PATCH] iov_iter: optimise iter type checking

From: Pavel Begunkov <asml.silence@gmail.com>
To: David Laight <David.Laight@ACULAB.COM>,
	'Al Viro' <viro@zeniv.linux.org.uk>
Cc: "linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
	Jens Axboe <axboe@kernel.dk>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] iov_iter: optimise iter type checking
Date: Tue, 12 Jan 2021 16:04:51 +0000	[thread overview]
Message-ID: <67d33206-9208-dd36-6063-be76150af9df@gmail.com> (raw)
In-Reply-To: <e125521b90d5405898c21c0e896c3525@AcuMS.aculab.com>

On 11/01/2021 09:35, David Laight wrote:
> From: Pavel Begunkov
>> Sent: 09 January 2021 22:11
> ....
>>> Does any code actually look at the fields as a pair?
>>> Would it even be better to use separate bytes?
>>> Even growing the on-stack structure by a word won't really matter.
>>
>> u8 type, rw;
>>
>> That won't bloat the struct. I like the idea. If used together compilers
>> can treat it as u16.
>>
>> btw there is a 4B hole just after for x64.
> 
> I've just had a quick look at the sources.
> (Nothing was powered up earlier.)
> 
> AFAICT nothing needs the RW flag to be in the same word
> as the type.
> If you think about it, the call site is specific to read/write.
> The only places iov_iter_rw() is used in inside helper functions
> to save the direction being passed from the caller.
> 
> I hope the comment about bit 1 being BVEC_FLAG_NO_REF is old.
> I can't find any references to that flag.

Yep, long dead.

-- 
Pavel Begunkov