lkml.org 
[lkml]   [2009]   [Sep]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH v3 2/5] kmemleak: add clear command support
From
Date
On Fri, 2009-09-04 at 17:44 -0700, Luis R. Rodriguez wrote:
> /*
> + * We use grey instead of black to ensure we can do future
> + * scans on the same objects. If we did not do future scans
> + * these black objects could potentially contain references to
> + * newly allocated objects in the future and we'd end up with
> + * false positives.
> + */
> +static void kmemleak_clear(void)
> +{
> + struct kmemleak_object *object;
> + unsigned long flags;
> +
> + stop_scan_thread();
> +
> + rcu_read_lock();
> + list_for_each_entry_rcu(object, &object_list, object_list) {
> + spin_lock_irqsave(&object->lock, flags);
> + if ((object->flags & OBJECT_REPORTED) &&
> + unreferenced_object(object))
> + object->min_count = -1;
> + spin_unlock_irqrestore(&object->lock, flags);
> + }
> + rcu_read_unlock();
> +
> + start_scan_thread();
> +}

Do we need to stop and start the scanning thread here? When starting it,
it will trigger a memory scan automatically. I don't think we want this
as a side-effect, so I dropped these lines from your patch.

Also you set min_count to -1 here which means black object, so a
subsequent patch corrects it. I'll set min_count to 0 here in case
anyone bisects over it.

--
Catalin



\
 
 \ /
  Last update: 2009-09-08 18:15    [W:0.126 / U:0.412 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site