linux-kernel - Re: [PATCH 7/8] io_uring: cache io

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <fcf5df70-d709-4bec-b4ce-aa833d1d4da2@gmail.com>
Date: Tue, 4 Feb 2025 12:07:49 +0000
From: Pavel Begunkov <asml.silence@...il.com>
To: Max Kellermann <max.kellermann@...os.com>, axboe@...nel.dk,
 io-uring@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 7/8] io_uring: cache io_kiocb->flags in variable

On 1/29/25 19:11, Pavel Begunkov wrote:
> On 1/28/25 13:39, Max Kellermann wrote:
>> This eliminates several redundant reads, some of which probably cannot
>> be optimized away by the compiler.
> 
> Let's not, it hurts readability with no clear benefits. In most cases
> the compiler will be able to optimise it just where it matters, and
> in cold paths we're comparing the overhead of reading a cached variable
> with taking locks and doing indirect calls, and even then it'd likely
> need to be saved onto the stack and loaded back.
> 
> The only place where it might be worth it is io_issue_sqe(), and
> even then I'd doubt it.

Jens, I'd suggest to drop it out of the tree, for the reasons above.

-- 
Pavel Begunkov