openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/drivers/gpu/drm/radeon/radeon_asic.h

436 lines
15 KiB
C
Raw Normal View History

drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
/*
* Copyright 2008 Advanced Micro Devices, Inc.
* Copyright 2008 Red Hat Inc.
* Copyright 2009 Jerome Glisse.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Dave Airlie
* Alex Deucher
* Jerome Glisse
*/
#ifndef __RADEON_ASIC_H__
#define __RADEON_ASIC_H__
/*
* common functions
*/
void radeon_legacy_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock);
void radeon_legacy_set_clock_gating(struct radeon_device *rdev, int enable);
void radeon_atom_set_engine_clock(struct radeon_device *rdev, uint32_t eng_clock);
void radeon_atom_set_memory_clock(struct radeon_device *rdev, uint32_t mem_clock);
void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable);
/*
* r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280
*/
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
int r100_init(struct radeon_device *rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg);
void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
void r100_errata(struct radeon_device *rdev);
void r100_vram_info(struct radeon_device *rdev);
int r100_gpu_reset(struct radeon_device *rdev);
int r100_mc_init(struct radeon_device *rdev);
void r100_mc_fini(struct radeon_device *rdev);
int r100_wb_init(struct radeon_device *rdev);
void r100_wb_fini(struct radeon_device *rdev);
int r100_gart_enable(struct radeon_device *rdev);
void r100_pci_gart_disable(struct radeon_device *rdev);
void r100_pci_gart_tlb_flush(struct radeon_device *rdev);
int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr);
int r100_cp_init(struct radeon_device *rdev, unsigned ring_size);
void r100_cp_fini(struct radeon_device *rdev);
void r100_cp_disable(struct radeon_device *rdev);
void r100_ring_start(struct radeon_device *rdev);
int r100_irq_set(struct radeon_device *rdev);
int r100_irq_process(struct radeon_device *rdev);
void r100_fence_ring_emit(struct radeon_device *rdev,
struct radeon_fence *fence);
int r100_cs_parse(struct radeon_cs_parser *p);
void r100_pll_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
uint32_t r100_pll_rreg(struct radeon_device *rdev, uint32_t reg);
int r100_copy_blit(struct radeon_device *rdev,
uint64_t src_offset,
uint64_t dst_offset,
unsigned num_pages,
struct radeon_fence *fence);
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
int r100_set_surface_reg(struct radeon_device *rdev, int reg,
uint32_t tiling_flags, uint32_t pitch,
uint32_t offset, uint32_t obj_size);
int r100_clear_surface_reg(struct radeon_device *rdev, int reg);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
static struct radeon_asic r100_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r100_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &r100_errata,
.vram_info = &r100_vram_info,
.gpu_reset = &r100_gpu_reset,
.mc_init = &r100_mc_init,
.mc_fini = &r100_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &r100_gart_enable,
.gart_disable = &r100_pci_gart_disable,
.gart_tlb_flush = &r100_pci_gart_tlb_flush,
.gart_set_page = &r100_pci_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r100_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r100_fence_ring_emit,
.cs_parse = &r100_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = NULL,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_legacy_set_engine_clock,
.set_memory_clock = NULL,
.set_pcie_lanes = NULL,
.set_clock_gating = &radeon_legacy_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* r300,r350,rv350,rv380
*/
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
int r300_init(struct radeon_device *rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
void r300_errata(struct radeon_device *rdev);
void r300_vram_info(struct radeon_device *rdev);
int r300_gpu_reset(struct radeon_device *rdev);
int r300_mc_init(struct radeon_device *rdev);
void r300_mc_fini(struct radeon_device *rdev);
void r300_ring_start(struct radeon_device *rdev);
void r300_fence_ring_emit(struct radeon_device *rdev,
struct radeon_fence *fence);
int r300_cs_parse(struct radeon_cs_parser *p);
int r300_gart_enable(struct radeon_device *rdev);
void rv370_pcie_gart_disable(struct radeon_device *rdev);
void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev);
int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr);
uint32_t rv370_pcie_rreg(struct radeon_device *rdev, uint32_t reg);
void rv370_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes);
int r300_copy_dma(struct radeon_device *rdev,
uint64_t src_offset,
uint64_t dst_offset,
unsigned num_pages,
struct radeon_fence *fence);
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
static struct radeon_asic r300_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r300_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &r300_errata,
.vram_info = &r300_vram_info,
.gpu_reset = &r300_gpu_reset,
.mc_init = &r300_mc_init,
.mc_fini = &r300_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &r300_gart_enable,
.gart_disable = &r100_pci_gart_disable,
.gart_tlb_flush = &r100_pci_gart_tlb_flush,
.gart_set_page = &r100_pci_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r300_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
.cs_parse = &r300_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_legacy_set_engine_clock,
.set_memory_clock = NULL,
.set_pcie_lanes = &rv370_set_pcie_lanes,
.set_clock_gating = &radeon_legacy_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* r420,r423,rv410
*/
void r420_errata(struct radeon_device *rdev);
void r420_vram_info(struct radeon_device *rdev);
int r420_mc_init(struct radeon_device *rdev);
void r420_mc_fini(struct radeon_device *rdev);
static struct radeon_asic r420_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r300_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &r420_errata,
.vram_info = &r420_vram_info,
.gpu_reset = &r300_gpu_reset,
.mc_init = &r420_mc_init,
.mc_fini = &r420_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &r300_gart_enable,
.gart_disable = &rv370_pcie_gart_disable,
.gart_tlb_flush = &rv370_pcie_gart_tlb_flush,
.gart_set_page = &rv370_pcie_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r300_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
.cs_parse = &r300_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_atom_set_engine_clock,
.set_memory_clock = &radeon_atom_set_memory_clock,
.set_pcie_lanes = &rv370_set_pcie_lanes,
.set_clock_gating = &radeon_atom_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* rs400,rs480
*/
void rs400_errata(struct radeon_device *rdev);
void rs400_vram_info(struct radeon_device *rdev);
int rs400_mc_init(struct radeon_device *rdev);
void rs400_mc_fini(struct radeon_device *rdev);
int rs400_gart_enable(struct radeon_device *rdev);
void rs400_gart_disable(struct radeon_device *rdev);
void rs400_gart_tlb_flush(struct radeon_device *rdev);
int rs400_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr);
uint32_t rs400_mc_rreg(struct radeon_device *rdev, uint32_t reg);
void rs400_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
static struct radeon_asic rs400_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r300_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &rs400_errata,
.vram_info = &rs400_vram_info,
.gpu_reset = &r300_gpu_reset,
.mc_init = &rs400_mc_init,
.mc_fini = &rs400_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &rs400_gart_enable,
.gart_disable = &rs400_gart_disable,
.gart_tlb_flush = &rs400_gart_tlb_flush,
.gart_set_page = &rs400_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r300_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
.cs_parse = &r300_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_legacy_set_engine_clock,
.set_memory_clock = NULL,
.set_pcie_lanes = NULL,
.set_clock_gating = &radeon_legacy_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* rs600.
*/
void rs600_errata(struct radeon_device *rdev);
void rs600_vram_info(struct radeon_device *rdev);
int rs600_mc_init(struct radeon_device *rdev);
void rs600_mc_fini(struct radeon_device *rdev);
int rs600_irq_set(struct radeon_device *rdev);
int rs600_gart_enable(struct radeon_device *rdev);
void rs600_gart_disable(struct radeon_device *rdev);
void rs600_gart_tlb_flush(struct radeon_device *rdev);
int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr);
uint32_t rs600_mc_rreg(struct radeon_device *rdev, uint32_t reg);
void rs600_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
static struct radeon_asic rs600_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r300_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &rs600_errata,
.vram_info = &rs600_vram_info,
.gpu_reset = &r300_gpu_reset,
.mc_init = &rs600_mc_init,
.mc_fini = &rs600_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &rs600_gart_enable,
.gart_disable = &rs600_gart_disable,
.gart_tlb_flush = &rs600_gart_tlb_flush,
.gart_set_page = &rs600_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r300_ring_start,
.irq_set = &rs600_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
.cs_parse = &r300_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_atom_set_engine_clock,
.set_memory_clock = &radeon_atom_set_memory_clock,
.set_pcie_lanes = NULL,
.set_clock_gating = &radeon_atom_set_clock_gating,
};
/*
* rs690,rs740
*/
void rs690_errata(struct radeon_device *rdev);
void rs690_vram_info(struct radeon_device *rdev);
int rs690_mc_init(struct radeon_device *rdev);
void rs690_mc_fini(struct radeon_device *rdev);
uint32_t rs690_mc_rreg(struct radeon_device *rdev, uint32_t reg);
void rs690_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
static struct radeon_asic rs690_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &r300_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &rs690_errata,
.vram_info = &rs690_vram_info,
.gpu_reset = &r300_gpu_reset,
.mc_init = &rs690_mc_init,
.mc_fini = &rs690_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &rs400_gart_enable,
.gart_disable = &rs400_gart_disable,
.gart_tlb_flush = &rs400_gart_tlb_flush,
.gart_set_page = &rs400_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &r300_ring_start,
.irq_set = &rs600_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
.cs_parse = &r300_cs_parse,
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r300_copy_dma,
.set_engine_clock = &radeon_atom_set_engine_clock,
.set_memory_clock = &radeon_atom_set_memory_clock,
.set_pcie_lanes = NULL,
.set_clock_gating = &radeon_atom_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* rv515
*/
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
int rv515_init(struct radeon_device *rdev);
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
void rv515_errata(struct radeon_device *rdev);
void rv515_vram_info(struct radeon_device *rdev);
int rv515_gpu_reset(struct radeon_device *rdev);
int rv515_mc_init(struct radeon_device *rdev);
void rv515_mc_fini(struct radeon_device *rdev);
uint32_t rv515_mc_rreg(struct radeon_device *rdev, uint32_t reg);
void rv515_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
void rv515_ring_start(struct radeon_device *rdev);
uint32_t rv515_pcie_rreg(struct radeon_device *rdev, uint32_t reg);
void rv515_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
static struct radeon_asic rv515_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &rv515_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &rv515_errata,
.vram_info = &rv515_vram_info,
.gpu_reset = &rv515_gpu_reset,
.mc_init = &rv515_mc_init,
.mc_fini = &rv515_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &r300_gart_enable,
.gart_disable = &rv370_pcie_gart_disable,
.gart_tlb_flush = &rv370_pcie_gart_tlb_flush,
.gart_set_page = &rv370_pcie_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &rv515_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.cs_parse = &r300_cs_parse,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_atom_set_engine_clock,
.set_memory_clock = &radeon_atom_set_memory_clock,
.set_pcie_lanes = &rv370_set_pcie_lanes,
.set_clock_gating = &radeon_atom_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* r520,rv530,rv560,rv570,r580
*/
void r520_errata(struct radeon_device *rdev);
void r520_vram_info(struct radeon_device *rdev);
int r520_mc_init(struct radeon_device *rdev);
void r520_mc_fini(struct radeon_device *rdev);
static struct radeon_asic r520_asic = {
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.init = &rv515_init,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.errata = &r520_errata,
.vram_info = &r520_vram_info,
.gpu_reset = &rv515_gpu_reset,
.mc_init = &r520_mc_init,
.mc_fini = &r520_mc_fini,
.wb_init = &r100_wb_init,
.wb_fini = &r100_wb_fini,
.gart_enable = &r300_gart_enable,
.gart_disable = &rv370_pcie_gart_disable,
.gart_tlb_flush = &rv370_pcie_gart_tlb_flush,
.gart_set_page = &rv370_pcie_gart_set_page,
.cp_init = &r100_cp_init,
.cp_fini = &r100_cp_fini,
.cp_disable = &r100_cp_disable,
.ring_start = &rv515_ring_start,
.irq_set = &r100_irq_set,
.irq_process = &r100_irq_process,
.fence_ring_emit = &r300_fence_ring_emit,
drm/radeon: command stream checker for r3xx-r5xx hardware For security purpose we want to make sure the userspace process doesn't access memory beyond buffer it owns. To achieve this we need to check states the userspace program. For color buffer and zbuffer we check that the clipping register will discard access beyond buffers set as color or zbuffer. For vertex buffer we check that no vertex fetch will happen beyond buffer end. For texture we check various texture states (number of mipmap level, texture size, texture depth, ...) to compute the amount of memory the texture fetcher might access. The command stream checking impact the performances so far quick benchmark shows an average of 3% decrease in fps of various applications. It can be optimized a bit more by caching result of checking and thus avoid a full recheck if no states changed since last check. Note that this patch is still incomplete on checking side as it doesn't check 2d rendering states. Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-17 19:28:30 +08:00
.cs_parse = &r300_cs_parse,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
.copy_blit = &r100_copy_blit,
.copy_dma = &r300_copy_dma,
.copy = &r100_copy_blit,
.set_engine_clock = &radeon_atom_set_engine_clock,
.set_memory_clock = &radeon_atom_set_memory_clock,
.set_pcie_lanes = &rv370_set_pcie_lanes,
.set_clock_gating = &radeon_atom_set_clock_gating,
drm/radeon/kms: add initial colortiling support. This adds new set/get tiling interfaces where the pitch and macro/micro tiling enables can be set. Along with a flag to decide if this object should have a surface when mapped. The only thing we need to allocate with a mapped surface should be the frontbuffer. Note rotate scanout shouldn't require one, and back/depth shouldn't either, though mesa needs some fixes. It fixes the TTM interfaces along Thomas's suggestions, and I've tested the surface stealing code with two X servers and not seen any lockdep issues. I've stopped tiling the fbcon frontbuffer, as I don't see there being any advantage other than testing, I've left the testing commands in there, just flip the fb_tiled to true in radeon_fb.c Open: Can we integrate endian swapping in with this? Future features: texture tiling - need to relocate texture registers TXOFFSET* with tiling info. This also merges Michel's cleanup surfaces regs at init time patch even though it makes sense on its own, this patch really relies on it. Some PowerMac firmwares set up a tiling surface at the beginning of VRAM which messes us up otherwise. that patch is: Signed-off-by: Michel Dänzer <daenzer@vmware.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-24 07:48:08 +08:00
.set_surface_reg = r100_set_surface_reg,
.clear_surface_reg = r100_clear_surface_reg,
drm/radeon: introduce kernel modesetting for radeon hardware Add kernel modesetting support to radeon driver, use the ttm memory manager to manage memory and DRM/GEM to provide userspace API. In order to avoid backward compatibility issue and to allow clean design and code the radeon kernel modesetting use different code path than old radeon/drm driver. When kernel modesetting is enabled the IOCTL of radeon/drm driver are considered as invalid and an error message is printed in the log and they return failure. KMS enabled userspace will use new API to talk with the radeon/drm driver. The new API provide functions to create/destroy/share/mmap buffer object which are then managed by the kernel memory manager (here TTM). In order to submit command to the GPU the userspace provide a buffer holding the command stream, along this buffer userspace have to provide a list of buffer object used by the command stream. The kernel radeon driver will then place buffer in GPU accessible memory and will update command stream to reflect the position of the different buffers. The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory not owned by the process supplying the command stream. This part of the code is still incomplete and this why we propose that patch as a staging driver addition, future security might forbid current experimental userspace to run. This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX (radeon up to X1950). Works is underway to provide support for R6XX, R7XX and newer hardware (radeon from HD2XXX to HD4XXX). Authors: Jerome Glisse <jglisse@redhat.com> Dave Airlie <airlied@redhat.com> Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2009-06-05 20:42:42 +08:00
};
/*
* r600,rv610,rv630,rv620,rv635,rv670,rs780,rv770,rv730,rv710
*/
uint32_t r600_pciep_rreg(struct radeon_device *rdev, uint32_t reg);
void r600_pciep_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
#endif