linux-kernel - Re: [PATCH v3 1/4] riscv: implement user_access

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: <518bfbeb-6ae9-4b56-b55e-ffa4b9438188@tenstorrent.com>
Date: Wed, 19 Mar 2025 15:24:29 +1100
From: Cyril Bur <cyrilbur@...storrent.com>
To: Alexandre Ghiti <alex@...ti.fr>, palmer@...belt.com,
 aou@...s.berkeley.edu, paul.walmsley@...ive.com, charlie@...osinc.com,
 jrtc27@...c27.com, ben.dooks@...ethink.co.uk
Cc: linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org,
 jszhang@...nel.org
Subject: Re: [PATCH v3 1/4] riscv: implement user_access_begin() and families



On 15/3/2025 12:28 am, Alexandre Ghiti wrote:
> Hi Cyril,
> 
> 
> There is a bunch of checkpatch errors to fix, see https:// 
> gist.github.com/linux-riscv-bot/98f23fd1b04d6da7c23c6cb18245a158
> 
> Why isn't there an implementation for unsafe_copy_from_user()? Let's 
> take the following example:
> 
> user_access_begin()
> unsafe_copy_from_user()
> unsafe_get_user() <==== This one will fail since unsafe_copy_from_user() 
> -> raw_copy_from_user() -> __asm_vector_usercopy() which enables and 
> disables the SUM bit.
> user_access_end()
> 
> Another thing is that with this patch, we lose the vectorized user 
> access functions, can you fix that too?

I've just been looking at doing that. I think teasing out an 'unsafe' 
style vectorised version and shoehorning it into this patch will be too 
much.

Would it be possible for me to do this as a separate series? I'm 
motivated to have this happen - we have a vector unit, we'll definitely 
want to use it.

I will add that a 'standard' copy_to_user() would still find its way to 
calling raw_copy_to_user() and so could be vectorised if all other 
criteria are met.

Thanks,

Cyril

> 
> Thanks,
> 
> Alex
> 
> 
>