linux-kernel - Re: [PATCH 2/2] scsi:target:tcmu: reduce once copy by using uio ioctl

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

Message-ID: <YhyoMekPbsKjO7KG@kroah.com>
Date:   Mon, 28 Feb 2022 11:47:13 +0100
From:   Greg KH <gregkh@...uxfoundation.org>
To:     Xiaoguang Wang <xiaoguang.wang@...ux.alibaba.com>
Cc:     Bodo Stroesser <bostroesser@...il.com>,
        Guixin Liu <kanie@...ux.alibaba.com>,
        martin.petersen@...cle.com, linux-scsi@...r.kernel.org,
        target-devel@...r.kernel.org, linux-kernel@...r.kernel.org,
        xlpang@...ux.alibaba.com
Subject: Re: [PATCH 2/2] scsi:target:tcmu: reduce once copy by using uio ioctl

On Mon, Feb 28, 2022 at 04:52:52PM +0800, Xiaoguang Wang wrote:
> 
> hi Bodo,
> 
> > Liu,
> > 
> > generally I like ideas to speed up tcmu.
> > 
> > OTOH, since Andy Grover implemented tcmu based on uio device, we are
> > restricted to what uio offers. With today's knowledge I think we would
> > not use the uio device in tcmu again, but switching away from uio now
> > would break existing userspace SW.
> Yeah, it will have much work if deciding to switch away from uio.
> I came up with a hacky or crazy idea :) what about we create a new file
> in tcmu_open() by anon_inode_getfile_secure(), and export this fd by
> tcmu mail box, we can do ioctl() on this new file, then uio framework
> won't be touched...

No new ioctls please.  That is creating a new user/kernel api that you
must support for the next 20+ years.  Please do not do that.

thanks,

greg k-h