linux-kernel - Re: [PATCH v3] mm: slub: move sysfs slab alloc/free interfaces to debugfs

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: <623f3165-4bce-8491-c9c4-8eac8404c21a@suse.cz>
Date:   Wed, 7 Apr 2021 12:30:53 +0200
From:   Vlastimil Babka <vbabka@...e.cz>
To:     Faiyaz Mohammed <faiyazm@...eaurora.org>, cl@...ux.com,
        penberg@...nel.org, rientjes@...gle.com, iamjoonsoo.kim@....com,
        akpm@...ux-foundation.org, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org
Cc:     vinmenon@...eaurora.org
Subject: Re: [PATCH v3] mm: slub: move sysfs slab alloc/free interfaces to
 debugfs

On 4/6/21 7:15 PM, Vlastimil Babka wrote:
> On 4/6/21 2:27 PM, Faiyaz Mohammed wrote:
>> alloc_calls and free_calls implementation in sysfs have two issues,
>> one is PAGE_SIZE limitiation of sysfs and other is it does not adhere
>> to "one value per file" rule.
>> 
>> To overcome this issues, move the alloc_calls and free_calls implemeation
>> to debugfs.
>> 
>> Signed-off-by: Faiyaz Mohammed <faiyazm@...eaurora.org>
> 
> Good direction, thanks. But I'm afraid we need a bit more:
> 
> - I don't see debugfs_remove() (or _recursive) used anywhere. When a cache is
> destroyed, do the dirs/files just linger in debugfs referencing removed
> kmem_cache objects?
> - There's a simple debugfs_create_dir(s->name, ...), for each cache while the
> sysfs variant handles merged caches with symlinks etc. For consistency, the
> hiearchy should look the same in debugfs as it does in sysfs.

Oh and one more suggestion. With full seq_file API (unlike sysfs) we should
really do the data gathering (to struct loc_track?) part just once per file
open, and cache it between individual reads.