From: Randy Dunlap <randy.dunlap@oracle.com>
To: Eric Anholt <eric@anholt.net>
Cc: linux-kernel@vger.kernel.org, keithp@keithp.com
Subject: Re: [PATCH] drm: Add GEM ("graphics execution manager") to i915 driver.
Date: Fri, 1 Aug 2008 08:44:05 -0700 [thread overview]
Message-ID: <20080801084405.62d3c823.randy.dunlap@oracle.com> (raw)
In-Reply-To: <1217573919-7496-3-git-send-email-eric@anholt.net>
On Thu, 31 Jul 2008 23:58:39 -0700 Eric Anholt wrote:
> GEM allows the creation of persistent buffer objects accessible by the
> graphics device through new ioctls for managing execution of commands on the
> device. The userland API is almost entirely driver-specific to ensure that
> any driver building on this model can easily map the interface to individual
> driver requirements.
>
> GEM is used by the 2d driver for managing its internal state allocations and
> will be used for pixmap storage to reduce memory consumption and enable
> zero-copy GLX_EXT_texture_from_pixmap, and in the 3d driver is used to enable
> GL_EXT_framebuffer_object and GL_ARB_pixel_buffer_object.
"the 2d driver" ... "the 3d driver".
Just curious: Is there only one of each of these?
> diff --git a/drivers/gpu/drm/drm_agpsupport.c b/drivers/gpu/drm/drm_agpsupport.c
> index aefa5ac..2639be2 100644
> --- a/drivers/gpu/drm/drm_agpsupport.c
> +++ b/drivers/gpu/drm/drm_agpsupport.c
> @@ -33,6 +33,7 @@
>
> #include "drmP.h"
> #include <linux/module.h>
> +#include <asm/agp.h>
>
> #if __OS_HAS_AGP
>
> @@ -452,4 +453,52 @@ int drm_agp_unbind_memory(DRM_AGP_MEM * handle)
> return agp_unbind_memory(handle);
> }
>
> -#endif /* __OS_HAS_AGP */
> +/**
In the kernel source tree, "/**" means "beginning of kernel-doc notation",
so please don't use it when kernel-doc isn't being used.
(in multiple places/files)
> + * Binds a collection of pages into AGP memory at the given offset, returning
> + * the AGP memory structure containing them.
> + *
> + * No reference is held on the pages during this time -- it is up to the
> + * caller to handle that.
> + */
> +DRM_AGP_MEM *
> +drm_agp_bind_pages(struct drm_device *dev,
> + struct page **pages,
> + unsigned long num_pages,
> + uint32_t gtt_offset)
and the preferred function format is more like:
DRM_AGP_MEM *drm_agp_bind_page(struct drm_device *dev,
...
(many places)
> +{
> + DRM_AGP_MEM *mem;
> + int ret, i;
> +
> + DRM_DEBUG("\n");
> +
> + mem = drm_agp_allocate_memory(dev->agp->bridge, num_pages,
> + AGP_USER_MEMORY);
> + if (mem == NULL) {
> + DRM_ERROR("Failed to allocate memory for %ld pages\n",
> + num_pages);
> + return NULL;
> + }
> +
> + for (i = 0; i < num_pages; i++)
> + mem->memory[i] = phys_to_gart(page_to_phys(pages[i]));
> + mem->page_count = num_pages;
> +
> + mem->is_flushed = true;
> + ret = drm_agp_bind_memory(mem, gtt_offset / PAGE_SIZE);
> + if (ret != 0) {
> + DRM_ERROR("Failed to bind AGP memory: %d\n", ret);
> + agp_free_memory(mem);
> + return NULL;
> + }
> +
> + return mem;
> +}
> +EXPORT_SYMBOL(drm_agp_bind_pages);
> diff --git a/include/drm/drmP.h b/include/drm/drmP.h
> index 1c1b13e..a540db8 100644
> --- a/include/drm/drmP.h
> +++ b/include/drm/drmP.h
> @@ -104,6 +104,7 @@ struct drm_device;
> #define DRIVER_DMA_QUEUE 0x200
> #define DRIVER_FB_DMA 0x400
> #define DRIVER_IRQ_VBL2 0x800
> +#define DRIVER_GEM 0x1000
>
> /***********************************************************************/
> /** \name Begin the DRM... */
What uses/processes this comment notation, please?
> @@ -771,6 +838,22 @@ struct drm_device {
> spinlock_t drw_lock;
> struct idr drw_idr;
> /*@} */
> +
> + /** \name GEM information */
> + /*@{ */
and this one?
> + spinlock_t object_name_lock;
> + struct idr object_name_idr;
> + atomic_t object_count;
> + atomic_t object_memory;
> + atomic_t pin_count;
> + atomic_t pin_memory;
> + atomic_t gtt_count;
> + atomic_t gtt_memory;
> + uint32_t gtt_total;
> + uint32_t invalidate_domains; /* domains pending invalidation */
> + uint32_t flush_domains; /* domains pending flush */
> + /*@} */
> +
> };
---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/
next prev parent reply other threads:[~2008-08-01 15:49 UTC|newest]
Thread overview: 61+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-08-01 6:58 [PATCH] PCI: Add pci_read_base() API Eric Anholt
2008-08-01 6:58 ` [PATCH] Export shmem_file_setup and shmem_getpage for DRM-GEM Eric Anholt
2008-08-01 6:58 ` [PATCH] drm: Add GEM ("graphics execution manager") to i915 driver Eric Anholt
2008-08-01 15:44 ` Randy Dunlap [this message]
2008-08-01 18:11 ` Keith Packard
2008-08-06 1:11 ` Eric Anholt
2008-08-01 7:10 ` [PATCH] Export shmem_file_setup and shmem_getpage for DRM-GEM Eric Anholt
2008-08-01 10:57 ` Hugh Dickins
2008-08-01 18:06 ` Keith Packard
2008-08-01 20:50 ` Christoph Hellwig
2008-08-01 23:01 ` Keith Packard
2008-08-03 12:49 ` John Stoffel
2008-08-03 17:52 ` Keith Packard
2008-08-03 23:35 ` files/process scaling problem? (was: [PATCH] Export shmem_file_setup and shmem_getpage for DRM-GEM) Ingo Oeser
2008-08-04 0:19 ` Keith Packard
2008-08-04 8:19 ` Alan Cox
2008-08-04 13:51 ` Arjan van de Ven
2008-08-04 14:11 ` Alan Cox
2008-08-04 16:38 ` Arjan van de Ven
2008-08-04 16:58 ` Keith Packard
2008-08-04 21:46 ` Ingo Oeser
2008-08-04 22:20 ` Dave Airlie
2008-08-05 0:34 ` Keith Packard
2008-08-11 1:23 ` [PATCH] Export shmem_file_setup and shmem_getpage for DRM-GEM Christoph Hellwig
2008-08-11 3:03 ` [PATCH] Export shmem_file_setup " Keith Packard
2008-08-04 1:54 ` [PATCH] Export shmem_file_setup and shmem_getpage " Keith Packard
2008-08-04 9:02 ` Nick Piggin
2008-08-04 10:26 ` Keith Packard
2008-08-04 10:43 ` Nick Piggin
2008-08-04 11:45 ` Keith Packard
2008-08-04 17:09 ` Hugh Dickins
2008-08-04 17:25 ` Keith Packard
2008-08-04 18:39 ` Hugh Dickins
2008-08-04 19:20 ` Keith Packard
2008-08-04 19:55 ` Hugh Dickins
2008-08-04 21:37 ` Keith Packard
2008-08-05 2:25 ` John Stoffel
2008-08-05 4:28 ` Keith Packard
2008-08-06 16:20 ` Stephane Marchesin
2008-08-06 17:24 ` Arjan van de Ven
2008-08-06 17:32 ` Stephane Marchesin
2008-08-06 17:56 ` Keith Packard
2008-08-06 18:09 ` Stephane Marchesin
2008-08-06 21:22 ` Keith Packard
2008-08-07 2:16 ` Stephane Marchesin
2008-08-07 2:57 ` Keith Packard
2008-08-11 1:34 ` Christoph Hellwig
2008-08-05 4:28 ` Nick Piggin
2008-08-11 1:30 ` Christoph Hellwig
2008-08-04 21:58 ` Keith Packard
2008-08-04 22:22 ` Dave Airlie
2008-08-05 4:43 ` Nick Piggin
2008-08-05 5:19 ` Keith Packard
2008-08-07 0:45 ` Jesse Barnes
2008-08-19 1:17 ` Dave Airlie
2008-08-19 10:00 ` Nick Piggin
2008-08-19 16:46 ` Keith Packard
2008-08-19 18:50 ` Jesse Barnes
2008-08-21 13:42 ` Jerome Glisse
2008-08-21 16:15 ` Jesse Barnes
-- strict thread matches above, loose matches on Subject: below --
2008-08-26 19:43 [resubmit] Add GEM to i915 DRM driver Eric Anholt
2008-08-26 19:43 ` [PATCH] Export shmem_file_setup for DRM-GEM Eric Anholt
2008-08-26 19:43 ` [PATCH] Export kmap_atomic_pfn " Eric Anholt
2008-08-26 19:43 ` [PATCH] drm: Add GEM ("graphics execution manager") to i915 driver Eric Anholt
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080801084405.62d3c823.randy.dunlap@oracle.com \
--to=randy.dunlap@oracle.com \
--cc=eric@anholt.net \
--cc=keithp@keithp.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox