netdev - Re: [RFC PATCH net-next 2/6] net: ethtool: record custom RSS contexts in the IDR

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

Date:   Wed, 5 Apr 2023 10:34:07 +0100
From:   Edward Cree <ecree.xilinx@...il.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     edward.cree@....com, linux-net-drivers@....com,
        davem@...emloft.net, pabeni@...hat.com, edumazet@...gle.com,
        netdev@...r.kernel.org, habetsm.xilinx@...il.com,
        sudheer.mogilappagari@...el.com
Subject: Re: [RFC PATCH net-next 2/6] net: ethtool: record custom RSS contexts
 in the IDR

On 05/04/2023 00:40, Jakub Kicinski wrote:
> On Tue, 4 Apr 2023 12:49:18 +0100 Edward Cree wrote:
>>> GFP_USER? Do you mean it for accounting? GFP_KERNEL_ACCOUNT?  
>>
>> It's an allocation triggerable by userland; I was under the
>>  impression that those were supposed to use GFP_USER for some
>>  reason; the rss_config alloc further up this function does.
> 
> That's what I thought, too, and that it implies memory accounting.
> But then someone from MM told me that that's not the case, and
> that GFP_USER is supposed to be mmap()able. Or maybe the latter
> part I got from the kdoc in gfp_types.h.
> 
> I think we should make sure the memory is accounted.

Okay.  Presumably this doesn't apply to `rss_config` because it's
 short-lived (freed on the way out of the function)?
(In which case I think `rss_config` should just be GFP_KERNEL;
 we don't try to map it to userspace, we just copy_{from,to}_user
 between it and the ioctl data.  But that's unrelated cleanup
 which we can worry about later, if at all.)