Re: [PATCH v3 3/3] drm/vram: Implement lazy unmapping for GEM VRAM buffers

From: Thomas Zimmermann <tzimmermann@suse.de>
To: Gerd Hoffmann <kraxel@redhat.com>
Cc: feng.tang@intel.com, dave@stgolabs.net, rong.a.chen@intel.com,
	airlied@linux.ie, dri-devel@lists.freedesktop.org,
	maxime.ripard@bootlin.com, ying.huang@intel.com, sean@poorly.run
Subject: Re: [PATCH v3 3/3] drm/vram: Implement lazy unmapping for GEM VRAM buffers
Date: Fri, 6 Sep 2019 12:37:47 +0200	[thread overview]
Message-ID: <b08a47cf-8e2f-6632-7704-94c8ec4421d4@suse.de> (raw)
In-Reply-To: <20190906093905.umz4wkwn4lukg2jg@sirius.home.kraxel.org>

[-- Attachment #1.1.1: Type: text/plain, Size: 992 bytes --]

Hi

Am 06.09.19 um 11:39 schrieb Gerd Hoffmann:
>> +void drm_gem_vram_bo_driver_move_notify(struct ttm_buffer_object *bo,
>> +					bool evict,
>> +					struct ttm_mem_reg *new_mem)
>> +{
> [ ... ]
>> +	if (!kmap->virtual)
>> +		return;
>> +	ttm_bo_kunmap(kmap);
>> +	kmap->virtual = NULL;
>> +}
> 
> I think ttm_buffer_object_destroy() needs "if (kmap->virtual)
> ttm_bo_kunmap()" too, due to the lazy unmap you can land there
> with an active mapping.

Hmm, from my understanding, that final call to move_notify() is done by
ttm_bo_cleanup_memtype_use(), which is called from within ttm_bo_put().

If we want to add a final kunmap, There's also the release_notify()
callback, which is probably the right place to put it.

Best regards
Thomas

> 
> cheers,
>   Gerd
> 

-- 
Thomas Zimmermann
Graphics Driver Developer
SUSE Linux GmbH, Maxfeldstrasse 5, 90409 Nuernberg, Germany
GF: Felix Imendörffer, Mary Higgins, Sri Rasiah
HRB 21284 (AG Nürnberg)

[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel