linux-kernel - Re: [PATCH] slab: add flags in cache

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: <d1d729d1-2330-4710-801f-f007979abbcd@suse.cz>
Date: Fri, 24 Oct 2025 10:22:39 +0200
From: Vlastimil Babka <vbabka@...e.cz>
To: Kassey Li <kassey.li@....qualcomm.com>, akpm@...ux-foundation.org,
 cl@...two.org, rientjes@...gle.com, roman.gushchin@...ux.dev,
 harry.yoo@...cle.com, linux-mm@...ck.org
Cc: linux-kernel@...r.kernel.org
Subject: Re: [PATCH] slab: add flags in cache_show

On 10/24/25 05:06, Kassey Li wrote:
> 
> 在 10/17/2025 6:50 PM, Vlastimil Babka 写道:
>> On 10/17/25 08:48, Kassey Li wrote:
>>> Flags info is useful to check the slab type.
>>>
>>> for example, _SLAB_RECLAIM_ACCOUNT:
>>>
>>>      0x50100 _SLAB_PANIC _SLAB_RECLAIM_ACCOUNT _SLAB_CMPXCHG_DOUBLE
>>>
>>> Signed-off-by: Kassey Li <kassey.li@....qualcomm.com>
>> Many of the flags are represented by files in /sys/kernel/slab/<cache>/
>> If you miss some, we could add it there. Changing slabinfo output could
>> break some users and the raw hex value is not a stable representation of the
>> flags anyway.
> 
> may you review this draft change according your suggest  ? if ok, i will 
> share v2 change.

What I meant is that that are files for some specific flags, e.g. the
SLAB_RECLAIM_ACCOUNT you mentioned has a file "reclaim_account".
Other files could be added for other flags when it's useful. Printing a raw
number dumping the implementation details isn't great.

> diff --git a/mm/slub.c b/mm/slub.c
> index b1f15598fbfd..619f446d2251 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -9033,6 +9033,12 @@ static ssize_t slab_size_show(struct kmem_cache 
> *s, char *buf)
>   }
>   SLAB_ATTR_RO(slab_size);
> 
> +static ssize_t slab_flags_show(struct kmem_cache *s, char *buf)
> +{
> +       return sysfs_emit(buf, "%u\n", s->flags);
> +}
> +SLAB_ATTR_RO(slab_flags);
> +
>   static ssize_t align_show(struct kmem_cache *s, char *buf)
>   {
>          return sysfs_emit(buf, "%u\n", s->align);
> @@ -9480,6 +9486,7 @@ SLAB_ATTR(skip_kfence);
> 
>   static struct attribute *slab_attrs[] = {
>          &slab_size_attr.attr,
> +       &slab_flags_attr.attr,
>          &object_size_attr.attr,
>          &objs_per_slab_attr.attr,
>          &order_attr.attr,
> 
>>
>>> ---
>>>   mm/slab_common.c | 6 +++---
>>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/mm/slab_common.c b/mm/slab_common.c
>>> index 932d13ada36c..f43239211e69 100644
>>> --- a/mm/slab_common.c
>>> +++ b/mm/slab_common.c
>>> @@ -1083,7 +1083,7 @@ static void print_slabinfo_header(struct seq_file *m)
>>>   	 * without _too_ many complaints.
>>>   	 */
>>>   	seq_puts(m, "slabinfo - version: 2.1\n");
>>> -	seq_puts(m, "# name            <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab>");
>>> +	seq_puts(m, "# name            <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> <flags>");
>>>   	seq_puts(m, " : tunables <limit> <batchcount> <sharedfactor>");
>>>   	seq_puts(m, " : slabdata <active_slabs> <num_slabs> <sharedavail>");
>>>   	seq_putc(m, '\n');
>>> @@ -1112,9 +1112,9 @@ static void cache_show(struct kmem_cache *s, struct seq_file *m)
>>>   	memset(&sinfo, 0, sizeof(sinfo));
>>>   	get_slabinfo(s, &sinfo);
>>>   
>>> -	seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d",
>>> +	seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d 0x%-8x",
>>>   		   s->name, sinfo.active_objs, sinfo.num_objs, s->size,
>>> -		   sinfo.objects_per_slab, (1 << sinfo.cache_order));
>>> +		   sinfo.objects_per_slab, (1 << sinfo.cache_order), s->flags);
>>>   
>>>   	seq_printf(m, " : tunables %4u %4u %4u",
>>>   		   sinfo.limit, sinfo.batchcount, sinfo.shared);