linux-kernel - Re: KASAN: use-after-free Read in io_uring

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>] [day] [month] [year] [list]

Date:   Thu, 30 Jul 2020 20:53:10 -0600
From:   Jens Axboe <axboe@...nel.dk>
To:     Hillf Danton <hdanton@...a.com>
Cc:     syzbot <syzbot+9d46305e76057f30c74e@...kaller.appspotmail.com>,
        io-uring@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        linux-kernel@...r.kernel.org, syzkaller-bugs@...glegroups.com,
        viro@...iv.linux.org.uk, Markus Elfring <Markus.Elfring@....de>
Subject: Re: KASAN: use-after-free Read in io_uring_setup (2)

On 7/30/20 8:28 PM, Hillf Danton wrote:
> 
> On Thu, 30 Jul 2020 20:07:59 -0600 Jens Axboe wrote:
>> On 7/30/20 7:45 PM, Hillf Danton wrote:
>>>
>>> Add the missing percpu_ref_get when creating ctx.
>>>
> [...]
>> The error path doesn't care, the issue is only after fd install. Hence
> 
> Yes you are right.
> 
>> we don't need to grab a reference, just make sure we don't touch the ctx
>> after fd install.
> 
> This is a cure, not a generic one as it maybe a potpit for anyone adding
> changes here since on. But that's quite unlikely as this is a way one-off
> path.
> 
>> Since you saw this one, you must have also seen my
>> patch. Why not comment on that instead?
> 
> You know, it is unusually hard to add anything in your field, and I hit the
> send button after staring at the screen for two minutes, given a different
> approach.

The patch was sent out 7h ago. My suggestion would be to at least see
what other people may have commented or posted on the topic first, instead
of just ignoring it point blank and sending something else out.

A good way to start a discussion would be to reply to my email in this
very thread, with why you think an alternate solution might be better.
Or point out of there are errors in it. Just ignoring what else has been
posted just comes off as rude, to be honest.

You've got patches in for io_uring in the past, and I'd surely like to
see that continue. But working together is helping each other out, not
working in a vacuum, pretending not to see what else is being discussed
or posted.

-- 
Jens Axboe