linux-kernel - Re: [PATCH] netfs: Fix setting of BDP

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 for Android: free password hash cracker in your pocket

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

Date: Tue, 21 May 2024 14:09:16 -0600
From: Jens Axboe <axboe@...nel.dk>
To: David Howells <dhowells@...hat.com>
Cc: Steve French <stfrench@...rosoft.com>, Jeff Layton <jlayton@...nel.org>,
 Enzo Matsumiya <ematsumiya@...e.de>, Matthew Wilcox <willy@...radead.org>,
 Christian Brauner <brauner@...nel.org>, netfs@...ts.linux.dev,
 v9fs@...ts.linux.dev, linux-afs@...ts.infradead.org,
 linux-cifs@...r.kernel.org, linux-fsdevel@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH] netfs: Fix setting of BDP_ASYNC from iocb flags

On 5/21/24 2:05 PM, David Howells wrote:
> Jens Axboe <axboe@...nel.dk> wrote:
> 
>> On 5/21/24 9:54 AM, David Howells wrote:
>>> Jens Axboe <axboe@...nel.dk> wrote:
>>>
>>>> However, I'll note that BDP_ASYNC is horribly named, it should be
>>>> BDP_NOWAIT instead. But that's a separate thing, fix looks correct
>>>> as-is.
>>>
>>> I thought IOCB_NOWAIT was related to RWF_NOWAIT, but apparently not from the
>>> code.
>>
>> It is, something submitted with RWF_NOWAIT should have IOCB_NOWAIT set.
>> But RWF_NOWAIT isn't the sole user of IOCB_NOWAIT, and no assumptions
>> should be made about whether something is sync or async based on whether
>> or not RWF_NOWAIT is set. Those aren't related other than _some_ proper
>> async IO will have IOCB_NOWAIT set, and others will not.
> 
> Are you sure?  RWF_NOWAIT seems to set IOCB_NOIO.

As it should, no-wait should imply not blocking on other IO. This is
completely orthogonal to whether or not it's async or sync IO.

I have a distinct feeling we're talking past each other :-)

-- 
Jens Axboe