linux-kernel - Re: [PATCH v3 4/5] drm/sched: Use drm sched lockdep map for submit

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <ZruxG8YaHqV1CAgS@DUT025-TGLU.fm.intel.com>
Date: Tue, 13 Aug 2024 19:16:43 +0000
From: Matthew Brost <matthew.brost@...el.com>
To: "Ghimiray, Himal Prasad" <himal.prasad.ghimiray@...el.com>
CC: <intel-xe@...ts.freedesktop.org>, <dri-devel@...ts.freedesktop.org>,
	<linux-kernel@...r.kernel.org>, <tj@...nel.org>, <jiangshanlai@...il.com>,
	<christian.koenig@....com>, <ltuikov89@...il.com>, <daniel@...ll.ch>
Subject: Re: [PATCH v3 4/5] drm/sched: Use drm sched lockdep map for submit_wq

On Mon, Aug 12, 2024 at 11:05:31AM +0530, Ghimiray, Himal Prasad wrote:
> 
> 
> On 10-08-2024 03:58, Matthew Brost wrote:
> > Avoid leaking a lockdep map on each drm sched creation and destruction
> > by using a single lockdep map for all drm sched allocated submit_wq.
> > 
> > v2:
> >   - Use alloc_ordered_workqueue_lockdep_map (Tejun)
> > 
> > Cc: Luben Tuikov <ltuikov89@...il.com>
> > Cc: Christian König <christian.koenig@....com>
> > Signed-off-by: Matthew Brost <matthew.brost@...el.com>
> > ---
> >   drivers/gpu/drm/scheduler/sched_main.c | 11 +++++++++++
> >   1 file changed, 11 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
> > index ab53ab486fe6..cf79d348cb32 100644
> > --- a/drivers/gpu/drm/scheduler/sched_main.c
> > +++ b/drivers/gpu/drm/scheduler/sched_main.c
> > @@ -87,6 +87,12 @@
> >   #define CREATE_TRACE_POINTS
> >   #include "gpu_scheduler_trace.h"
> > +#ifdef CONFIG_LOCKDEP
> > +static struct lockdep_map drm_sched_lockdep_map = {
> > +	.name = "drm_sched_lockdep_map"
> > +};
> 
> 
> will it be better to use STATIC_LOCKDEP_MAP_INIT ? Initializing key here
> instead of while registering the class ?
> 

Most existing design patterns in the kernel define static lockdep class
this way so I think this is fine. But honestly don't really have an
opinion here.

Matt

> 
> > +#endif
> > +
> >   #define to_drm_sched_job(sched_job)		\
> >   		container_of((sched_job), struct drm_sched_job, queue_node)
> > @@ -1272,7 +1278,12 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
> >   		sched->submit_wq = submit_wq;
> >   		sched->own_submit_wq = false;
> >   	} else {
> > +#ifdef CONFIG_LOCKDEP
> > +		sched->submit_wq = alloc_ordered_workqueue_lockdep_map(name, 0,
> > +								       &drm_sched_lockdep_map);
> > +#else
> >   		sched->submit_wq = alloc_ordered_workqueue(name, 0);
> > +#endif
> >   		if (!sched->submit_wq)
> >   			return -ENOMEM;