[Mesa-dev] Initial amdgpu driver release

Zhou, Jammy Jammy.Zhou at amd.com
Fri Apr 24 00:43:11 PDT 2015


Hi Alex,

For the core driver patch:

+config DRM_AMDGPU
+	tristate "AMD GPU"
+	depends on DRM && PCI
+	select FB_CFB_FILLRECT
+	select FB_CFB_COPYAREA
+	select FB_CFB_IMAGEBLIT
+	select FW_LOADER
+        select DRM_KMS_HELPER
+	select DRM_KMS_FB_HELPER
+        select DRM_TTM
+	select POWER_SUPPLY
+	select HWMON
+	select BACKLIGHT_CLASS_DEVICE
+	select DRM_AMD_GNB_BUS
+	select INTERVAL_TREE

I think DRM_AMD_GNB_BUS is not used, we can probably remove it now.

+/* TODO: Here are things that needs to be done :
+ *	- surface allocator & initializer : (bit like scratch reg) should
+ *	  initialize HDP_ stuff on RS600, R600, R700 hw, well anythings
+ *	  related to surface
+ *	- WB : write back stuff (do it bit like scratch reg things)
+ *	- Vblank : look at Jesse's rework and what we should do
+ *	- r600/r700: gart & cp
+ *	- cs : clean cs ioctl use bitmap & things like that.
+ *	- power management stuff
+ *	- Barrier in gart code
+ *	- Unmappabled vram ?
+ *	- TESTING, TESTING, TESTING
+ */
+
+/* Initialization path:
+ *  We expect that acceleration initialization might fail for various
+ *  reasons even thought we work hard to make it works on most
+ *  configurations. In order to still have a working userspace in such
+ *  situation the init path must succeed up to the memory controller
+ *  initialization point. Failure before this point are considered as
+ *  fatal error. Here is the init callchain :
+ *      amdgpu_device_init  perform common structure, mutex initialization
+ *      asic_init           setup the GPU memory layout and perform all
+ *                          one time initialization (failure in this
+ *                          function are considered fatal)
+ *      asic_startup        setup the GPU acceleration, in order to
+ *                          follow guideline the first thing this
+ *                          function should do is setting the GPU
+ *                          memory controller (only MC setup failure
+ *                          are considered as fatal)
+ */
+
These should be outdated, and I think they can be removed now.

For the uapi header patch:

+#define AMDGPU_TILING_MACRO				0x1
+#define AMDGPU_TILING_MICRO				0x2
+#define AMDGPU_TILING_SWAP_16BIT			0x4
+#define AMDGPU_TILING_R600_NO_SCANOUT			AMDGPU_TILING_SWAP_16BIT
+#define AMDGPU_TILING_SWAP_32BIT			0x8
+/* this object requires a surface when mapped - i.e. front buffer */
+#define AMDGPU_TILING_SURFACE				0x10
+#define AMDGPU_TILING_MICRO_SQUARE			0x20
+#define AMDGPU_TILING_EG_BANKW_SHIFT			8
+#define AMDGPU_TILING_EG_BANKW_MASK			0xf
+#define AMDGPU_TILING_EG_BANKH_SHIFT			12
+#define AMDGPU_TILING_EG_BANKH_MASK			0xf
+#define AMDGPU_TILING_EG_MACRO_TILE_ASPECT_SHIFT	16
+#define AMDGPU_TILING_EG_MACRO_TILE_ASPECT_MASK	0xf
+#define AMDGPU_TILING_EG_TILE_SPLIT_SHIFT		24
+#define AMDGPU_TILING_EG_TILE_SPLIT_MASK		0xf
+#define AMDGPU_TILING_EG_STENCIL_TILE_SPLIT_SHIFT	28
+#define AMDGPU_TILING_EG_STENCIL_TILE_SPLIT_MASK	0xf
......
+#define SI_TILE_MODE_COLOR_LINEAR_ALIGNED	8
+#define SI_TILE_MODE_COLOR_1D			13
+#define SI_TILE_MODE_COLOR_1D_SCANOUT		9
+#define SI_TILE_MODE_COLOR_2D_8BPP		14
+#define SI_TILE_MODE_COLOR_2D_16BPP		15
+#define SI_TILE_MODE_COLOR_2D_32BPP		16
+#define SI_TILE_MODE_COLOR_2D_64BPP		17
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_16BPP	11
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_32BPP	12
+#define SI_TILE_MODE_DEPTH_STENCIL_1D		4
+#define SI_TILE_MODE_DEPTH_STENCIL_2D		0
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_2AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_4AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_8AA	2
+
+#define CIK_TILE_MODE_DEPTH_STENCIL_1D		5

It looks these definitions are not used by libdrm_amdgpu anymore (and even by the kernel driver). Maybe we can remove the unused definitions, and move the used ones to amdgpu.h instead. Besides, I think we'd better remove 'R600/EG/SI/CIK' from the naming.

Other than the comments above, the kernel series are Acked-by: Jammy Zhou <Jammy.Zhou at amd.com>

Regards,
Jammy

-----Original Message-----
From: dri-devel [mailto:dri-devel-bounces at lists.freedesktop.org] On Behalf Of Alex Deucher
Sent: Tuesday, April 21, 2015 6:34 AM
To: Maling list - DRI developers; mesa-dev at lists.freedesktop.org
Cc: Deucher, Alexander
Subject: Initial amdgpu driver release

I'm pleased to announce the initial release of the new amdgpu driver.
This is a partial replacement for the radeon driver for newer AMD asics.  A number of components are still shared.  Here is a comparison of the radeon and amdgpu stacks:

1. radeon stack
kernel driver: radeon.ko
libdrm: libdrm_radeon
mesa: radeon, r200, r300, r600, radeonsi
ddx: xf86-video-ati

2. amdgpu stack
kernel driver: amdgpu.ko
libdrm: libdrm_amdgpu
mesa: radeonsi
ddx: xf86-video-amdgpu

Older asics will continue to be supported by the radeon stack; new asics will be supported by the amdgpu stack.  CI (Sea Islands) asics have support in both driver stacks, but this is purely for testing purposes.  CI parts are officially supported in the radeon stack.
Support for CI on the amdgpu stack is determined by a config option in the kernel.  CI support is not enabled by default for amdgpu.

Most of our focus has been on Carrizo support, so there are some gaps in the dGPU support for Tonga and Iceland, notably power management.
Those gaps will be filled in eventually.

Also included in this code base are full register headers for just about every block on the asics.

Barring the gaps mentioned above, the driver stack is functionally on par with radeon including:
- OpenGL 3.3 support using the radeonsi mesa driver
- Video decode support using UVD
- Video encode support using VCE

The code can be found in the amdgpu branches of the following git trees.
xf86-video-amdgpu:
http://cgit.freedesktop.org/~agd5f/xf86-video-amdgpu/log/?h=amdgpu
libdrm:
http://cgit.freedesktop.org/~agd5f/drm/log/?h=amdgpu
kernel:
http://cgit.freedesktop.org/~agd5f/linux/log/?h=amdgpu
mesa:
http://cgit.freedesktop.org/~mareko/mesa/log/?h=amdgpu

To test the new driver stack you will need to specify a device section in your xorg.conf with the driver set to amdgpu rather than radeon.

Please review!

Thanks,

The AMD Linux Driver Team
_______________________________________________
dri-devel mailing list
dri-devel at lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel


More information about the mesa-dev mailing list