linux-kernel - Re: [PATCH] workqueue.c: Change workqueue to accept variable length name

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-next>] [day] [month] [year] [list]

Message-ID: <ZYXNMEXBnuumL3D6@mac.lan>
Date: Sat, 23 Dec 2023 02:53:52 +0900
From: Tejun Heo <tj@...nel.org>
To: Audra Mitchell <aubaker@...hat.com>
Cc: Audra Mitchell <audra@...hat.com>, linux-kernel@...r.kernel.org,
	raquini@...hat.com, jiangshanlai@...il.com,
	hirokazu.yamauchi.hk@...achi.com,
	Laurence Oberman <loberman@...hat.com>,
	Donald Douwsma <ddouwsma@...hat.com>
Subject: Re: [PATCH] workqueue.c: Change workqueue to accept variable length
 name

Hello,

On Fri, Dec 22, 2023 at 10:35:03AM -0500, Audra Mitchell wrote:
> We have one concrete example from a Hitachi block device driver (notice the
> 47a1/47a2 gets
> cut off with the workqueue name):
> 
> Device                              Workqueue Name (24char zero terminated)
> /dev/sd0279b080047a1   xfs-blockgc/sd0279b0800
> /dev/sd0279b080047a2   xfs-blockgc/sd0279b0800

I see, so it's a combination of somewhat lengthy device names and then xfs
adding a prefix to them. Neither is particularly long but the combination
is.

> I can also imagine this issue being present with nvme devices, but the
> request came from Hitachi.
> I believe it would be up to the device driver to determine if the name can
> be shortened and I've
> included Hitachi requester on this email thread.
> 
> Alternatively, we could increase the size of the WQ_NAME_LEN, but it seems
> highly likely we are
> going to butt against the static size again in the future. We previously
> had variable length names
> and it seems (to me) to be the best long term path forward.

Can we just bump the length to 32 and trigger a warning if the requested
name overruns? I want to provide some pressure to limit the length of the
name so that it doesn't get too long over time. If folks bump into it and
can't find a different way to deal with it, we can get bring back the
subject.

Thanks.

-- 
tejun