[PATCH] tools/mm/slabinfo: Fix trace disabling logic

[PATCH] tools/mm/slabinfo: Fix trace disabling logic

[Kaushlendra Kumar]

When tracing is disabled (!tracing) but the slab has tracing enabled
(s->trace), the code should disable tracing by writing 0, not enable
it by writing 1. Fix the inverted logic to correctly turn off tracing.

Signed-off-by: Kaushlendra Kumar <kaushlendra.kumar@intel.com>
---
 tools/mm/slabinfo.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/mm/slabinfo.c b/tools/mm/slabinfo.c
index 80cdbd3db82d..7c51d283504d 100644
--- a/tools/mm/slabinfo.c
+++ b/tools/mm/slabinfo.c
@@ -798,7 +798,7 @@ static void slab_debug(struct slabinfo *s)
 			fprintf(stderr, "%s can only enable trace for one slab at a time\n", s->name);
 	}
 	if (!tracing && s->trace)
-		set_obj(s, "trace", 1);
+		set_obj(s, "trace", 0);
 }
 
 static void totals(void)
-- 
2.34.1

Re: [PATCH] tools/mm/slabinfo: Fix trace disabling logic

[Harry Yoo]

On Mon, Dec 08, 2025 at 02:12:55PM +0530, Kaushlendra Kumar wrote:
> When tracing is disabled (!tracing) but the slab has tracing enabled
> (s->trace), the code should disable tracing by writing 0, not enable
> it by writing 1. Fix the inverted logic to correctly turn off tracing.
> 
> Signed-off-by: Kaushlendra Kumar <kaushlendra.kumar@intel.com>
> ---

After commit 060807f841ac9 ("mm, slub: make remaining slub_debug related
attributes read-only"), enabling or disabling debug option at runtime is
not supported anymore. slabinfo's -d and -dA flags are not supposed work on
recent kernels.

As it turns out enabling or disabling debug features at runtime is not a
good idea anyway, could you please working on dropping the feature instead?

>  tools/mm/slabinfo.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tools/mm/slabinfo.c b/tools/mm/slabinfo.c
> index 80cdbd3db82d..7c51d283504d 100644
> --- a/tools/mm/slabinfo.c
> +++ b/tools/mm/slabinfo.c
> @@ -798,7 +798,7 @@ static void slab_debug(struct slabinfo *s)
>  			fprintf(stderr, "%s can only enable trace for one slab at a time\n", s->name);
>  	}
>  	if (!tracing && s->trace)
> -		set_obj(s, "trace", 1);
> +		set_obj(s, "trace", 0);
>  }
>  
>  static void totals(void)
> -- 
> 2.34.1

-- 
Cheers,
Harry / Hyeonggon

RE: [PATCH] tools/mm/slabinfo: Fix trace disabling logic

[Kumar, Kaushlendra]

On Mon, Dec 08, 2025, Harry Hyeonggon Yoo wrote:
> On Mon, Dec 08, 2025 at 02:12:55PM +0530, Kaushlendra Kumar wrote:
>> When tracing is disabled (!tracing) but the slab has tracing enabled 
>> (s->trace), the code should disable tracing by writing 0, not enable 
>> it by writing 1. Fix the inverted logic to correctly turn off tracing.
>>
>> Signed-off-by: Kaushlendra Kumar <kaushlendra.kumar@intel.com>
>> ---
> 
> After commit 060807f841ac9 ("mm, slub: make remaining slub_debug related 
> attributes read-only"), enabling or disabling debug option at runtime is 
> not supported anymore. slabinfo's -d and -dA flags are not supposed work 
> on recent kernels.
> 
> As it turns out enabling or disabling debug features at runtime is not a 
> good idea anyway, could you please working on dropping the feature instead?

Thanks for the context. You're right-fixing inverted logic in code 
that's been non-functional since 060807f841ac9 doesn't add value.

I'll submit a cleanup series removing the obsolete 
runtime debug functionality:
- Remove -d/-dA options and related usage text
- Drop slab_debug() and set_obj() for debug attributes


-Kaushlendra