linux_dsm_epyc7002/drivers/gpu/drm/radeon/radeon_gem.c

828 lines
20 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 19:42:42 +07:00
/*
* 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
*/
#include <linux/pci.h>
#include <drm/drm_debugfs.h>
#include <drm/drm_device.h>
#include <drm/drm_file.h>
#include <drm/radeon_drm.h>
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 19:42:42 +07:00
#include "radeon.h"
void radeon_gem_object_free(struct drm_gem_object *gobj)
{
struct radeon_bo *robj = gem_to_radeon_bo(gobj);
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 19:42:42 +07:00
if (robj) {
radeon_mn_unregister(robj);
radeon_bo_unref(&robj);
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 19:42:42 +07:00
}
}
int radeon_gem_object_create(struct radeon_device *rdev, unsigned long size,
int alignment, int initial_domain,
u32 flags, bool kernel,
struct drm_gem_object **obj)
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 19:42:42 +07:00
{
struct radeon_bo *robj;
unsigned long max_size;
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 19:42:42 +07:00
int r;
*obj = NULL;
/* At least align on page size */
if (alignment < PAGE_SIZE) {
alignment = PAGE_SIZE;
}
/* Maximum bo size is the unpinned gtt size since we use the gtt to
* handle vram to system pool migrations.
*/
max_size = rdev->mc.gtt_size - rdev->gart_pin_size;
if (size > max_size) {
DRM_DEBUG("Allocation size %ldMb bigger than %ldMb limit\n",
size >> 20, max_size >> 20);
return -ENOMEM;
}
retry:
r = radeon_bo_create(rdev, size, alignment, kernel, initial_domain,
flags, NULL, NULL, &robj);
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 19:42:42 +07:00
if (r) {
if (r != -ERESTARTSYS) {
if (initial_domain == RADEON_GEM_DOMAIN_VRAM) {
initial_domain |= RADEON_GEM_DOMAIN_GTT;
goto retry;
}
DRM_ERROR("Failed to allocate GEM object (%ld, %d, %u, %d)\n",
size, initial_domain, alignment, r);
}
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 19:42:42 +07:00
return r;
}
*obj = &robj->tbo.base;
robj->pid = task_pid_nr(current);
mutex_lock(&rdev->gem.mutex);
list_add_tail(&robj->list, &rdev->gem.objects);
mutex_unlock(&rdev->gem.mutex);
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 19:42:42 +07:00
return 0;
}
static int radeon_gem_set_domain(struct drm_gem_object *gobj,
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 19:42:42 +07:00
uint32_t rdomain, uint32_t wdomain)
{
struct radeon_bo *robj;
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 19:42:42 +07:00
uint32_t domain;
long r;
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 19:42:42 +07:00
/* FIXME: reeimplement */
robj = gem_to_radeon_bo(gobj);
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 19:42:42 +07:00
/* work out where to validate the buffer to */
domain = wdomain;
if (!domain) {
domain = rdomain;
}
if (!domain) {
/* Do nothings */
pr_warn("Set domain without domain !\n");
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 19:42:42 +07:00
return 0;
}
if (domain == RADEON_GEM_DOMAIN_CPU) {
/* Asking for cpu access wait for object idle */
r = dma_resv_wait_timeout_rcu(robj->tbo.base.resv, true, true, 30 * HZ);
if (!r)
r = -EBUSY;
if (r < 0 && r != -EINTR) {
pr_err("Failed to wait for object: %li\n", r);
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 19:42:42 +07:00
return r;
}
}
if (domain == RADEON_GEM_DOMAIN_VRAM && robj->prime_shared_count) {
/* A BO that is associated with a dma-buf cannot be sensibly migrated to VRAM */
return -EINVAL;
}
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 19:42:42 +07:00
return 0;
}
int radeon_gem_init(struct radeon_device *rdev)
{
INIT_LIST_HEAD(&rdev->gem.objects);
return 0;
}
void radeon_gem_fini(struct radeon_device *rdev)
{
radeon_bo_force_delete(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 19:42:42 +07:00
}
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
/*
* Call from drm_gem_handle_create which appear in both new and open ioctl
* case.
*/
int radeon_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_priv)
{
struct radeon_bo *rbo = gem_to_radeon_bo(obj);
struct radeon_device *rdev = rbo->rdev;
struct radeon_fpriv *fpriv = file_priv->driver_priv;
struct radeon_vm *vm = &fpriv->vm;
struct radeon_bo_va *bo_va;
int r;
if ((rdev->family < CHIP_CAYMAN) ||
(!rdev->accel_working)) {
return 0;
}
r = radeon_bo_reserve(rbo, false);
if (r) {
return r;
}
bo_va = radeon_vm_bo_find(vm, rbo);
if (!bo_va) {
bo_va = radeon_vm_bo_add(rdev, vm, rbo);
} else {
++bo_va->ref_count;
}
radeon_bo_unreserve(rbo);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
return 0;
}
void radeon_gem_object_close(struct drm_gem_object *obj,
struct drm_file *file_priv)
{
struct radeon_bo *rbo = gem_to_radeon_bo(obj);
struct radeon_device *rdev = rbo->rdev;
struct radeon_fpriv *fpriv = file_priv->driver_priv;
struct radeon_vm *vm = &fpriv->vm;
struct radeon_bo_va *bo_va;
int r;
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
if ((rdev->family < CHIP_CAYMAN) ||
(!rdev->accel_working)) {
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
return;
}
r = radeon_bo_reserve(rbo, true);
if (r) {
dev_err(rdev->dev, "leaking bo va because "
"we fail to reserve bo (%d)\n", r);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
return;
}
bo_va = radeon_vm_bo_find(vm, rbo);
if (bo_va) {
if (--bo_va->ref_count == 0) {
radeon_vm_bo_rmv(rdev, bo_va);
}
}
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
radeon_bo_unreserve(rbo);
}
static int radeon_gem_handle_lockup(struct radeon_device *rdev, int r)
{
if (r == -EDEADLK) {
r = radeon_gpu_reset(rdev);
if (!r)
r = -EAGAIN;
}
return r;
}
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 19:42:42 +07:00
/*
* GEM ioctls.
*/
int radeon_gem_info_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct radeon_device *rdev = dev->dev_private;
struct drm_radeon_gem_info *args = data;
struct ttm_mem_type_manager *man;
man = &rdev->mman.bdev.man[TTM_PL_VRAM];
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 19:42:42 +07:00
args->vram_size = (u64)man->size << PAGE_SHIFT;
args->vram_visible = rdev->mc.visible_vram_size;
args->vram_visible -= rdev->vram_pin_size;
args->gart_size = rdev->mc.gtt_size;
args->gart_size -= rdev->gart_pin_size;
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 19:42:42 +07:00
return 0;
}
int radeon_gem_pread_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
/* TODO: implement */
DRM_ERROR("unimplemented %s\n", __func__);
return -ENOSYS;
}
int radeon_gem_pwrite_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
/* TODO: implement */
DRM_ERROR("unimplemented %s\n", __func__);
return -ENOSYS;
}
int radeon_gem_create_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct radeon_device *rdev = dev->dev_private;
struct drm_radeon_gem_create *args = data;
struct drm_gem_object *gobj;
uint32_t handle;
int r;
down_read(&rdev->exclusive_lock);
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 19:42:42 +07:00
/* create a gem object to contain this object in */
args->size = roundup(args->size, PAGE_SIZE);
r = radeon_gem_object_create(rdev, args->size, args->alignment,
args->initial_domain, args->flags,
false, &gobj);
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 19:42:42 +07:00
if (r) {
up_read(&rdev->exclusive_lock);
r = radeon_gem_handle_lockup(rdev, r);
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 19:42:42 +07:00
return r;
}
r = drm_gem_handle_create(filp, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_put_unlocked(gobj);
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 19:42:42 +07:00
if (r) {
up_read(&rdev->exclusive_lock);
r = radeon_gem_handle_lockup(rdev, r);
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 19:42:42 +07:00
return r;
}
args->handle = handle;
up_read(&rdev->exclusive_lock);
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 19:42:42 +07:00
return 0;
}
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
int radeon_gem_userptr_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct ttm_operation_ctx ctx = { true, false };
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
struct radeon_device *rdev = dev->dev_private;
struct drm_radeon_gem_userptr *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *bo;
uint32_t handle;
int r;
drm/radeon: untag user pointers in radeon_gem_userptr_ioctl This patch is a part of a series that extends kernel ABI to allow to pass tagged user pointers (with the top byte set to something else other than 0x00) as syscall arguments. In radeon_gem_userptr_ioctl() an MMU notifier is set up with a (tagged) userspace pointer. The untagged address should be used so that MMU notifiers for the untagged address get correctly matched up with the right BO. This funcation also calls radeon_ttm_tt_pin_userptr(), which uses provided user pointers for vma lookups, which can only by done with untagged pointers. This patch untags user pointers in radeon_gem_userptr_ioctl(). Link: http://lkml.kernel.org/r/c856babeb67195b35603b8d5ba386a2819cec5ff.1563904656.git.andreyknvl@google.com Signed-off-by: Andrey Konovalov <andreyknvl@google.com> Reviewed-by: Khalid Aziz <khalid.aziz@oracle.com> Reviewed-by: Kees Cook <keescook@chromium.org> Suggested-by: Felix Kuehling <Felix.Kuehling@amd.com> Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Eric Auger <eric.auger@redhat.com> Cc: Jens Wiklander <jens.wiklander@linaro.org> Cc: Mauro Carvalho Chehab <mchehab+samsung@kernel.org> Cc: Mike Rapoport <rppt@linux.ibm.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Will Deacon <will@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-09-26 06:48:51 +07:00
args->addr = untagged_addr(args->addr);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
if (offset_in_page(args->addr | args->size))
return -EINVAL;
/* reject unknown flag values */
if (args->flags & ~(RADEON_GEM_USERPTR_READONLY |
RADEON_GEM_USERPTR_ANONONLY | RADEON_GEM_USERPTR_VALIDATE |
RADEON_GEM_USERPTR_REGISTER))
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
return -EINVAL;
if (args->flags & RADEON_GEM_USERPTR_READONLY) {
/* readonly pages not tested on older hardware */
if (rdev->family < CHIP_R600)
return -EINVAL;
} else if (!(args->flags & RADEON_GEM_USERPTR_ANONONLY) ||
!(args->flags & RADEON_GEM_USERPTR_REGISTER)) {
/* if we want to write to it we must require anonymous
memory and install a MMU notifier */
return -EACCES;
}
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
down_read(&rdev->exclusive_lock);
/* create a gem object to contain this object in */
r = radeon_gem_object_create(rdev, args->size, 0,
RADEON_GEM_DOMAIN_CPU, 0,
false, &gobj);
if (r)
goto handle_lockup;
bo = gem_to_radeon_bo(gobj);
r = radeon_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
if (r)
goto release_object;
if (args->flags & RADEON_GEM_USERPTR_REGISTER) {
r = radeon_mn_register(bo, args->addr);
if (r)
goto release_object;
}
if (args->flags & RADEON_GEM_USERPTR_VALIDATE) {
down_read(&current->mm->mmap_sem);
r = radeon_bo_reserve(bo, true);
if (r) {
up_read(&current->mm->mmap_sem);
goto release_object;
}
radeon_ttm_placement_from_domain(bo, RADEON_GEM_DOMAIN_GTT);
r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
radeon_bo_unreserve(bo);
up_read(&current->mm->mmap_sem);
if (r)
goto release_object;
}
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
r = drm_gem_handle_create(filp, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_put_unlocked(gobj);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
if (r)
goto handle_lockup;
args->handle = handle;
up_read(&rdev->exclusive_lock);
return 0;
release_object:
drm_gem_object_put_unlocked(gobj);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
handle_lockup:
up_read(&rdev->exclusive_lock);
r = radeon_gem_handle_lockup(rdev, r);
return r;
}
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 19:42:42 +07:00
int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
/* transition the BO to a domain -
* just validate the BO into a certain domain */
struct radeon_device *rdev = dev->dev_private;
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 19:42:42 +07:00
struct drm_radeon_gem_set_domain *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *robj;
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 19:42:42 +07:00
int r;
/* for now if someone requests domain CPU -
* just make sure the buffer is finished with */
down_read(&rdev->exclusive_lock);
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 19:42:42 +07:00
/* just do a BO wait for now */
gobj = drm_gem_object_lookup(filp, args->handle);
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 19:42:42 +07:00
if (gobj == NULL) {
up_read(&rdev->exclusive_lock);
return -ENOENT;
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 19:42:42 +07:00
}
robj = gem_to_radeon_bo(gobj);
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 19:42:42 +07:00
r = radeon_gem_set_domain(gobj, args->read_domains, args->write_domain);
drm_gem_object_put_unlocked(gobj);
up_read(&rdev->exclusive_lock);
r = radeon_gem_handle_lockup(robj->rdev, r);
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 19:42:42 +07:00
return r;
}
int radeon_mode_dumb_mmap(struct drm_file *filp,
struct drm_device *dev,
uint32_t handle, uint64_t *offset_p)
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 19:42:42 +07:00
{
struct drm_gem_object *gobj;
struct radeon_bo *robj;
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 19:42:42 +07:00
gobj = drm_gem_object_lookup(filp, handle);
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 19:42:42 +07:00
if (gobj == NULL) {
return -ENOENT;
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 19:42:42 +07:00
}
robj = gem_to_radeon_bo(gobj);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
if (radeon_ttm_tt_has_userptr(robj->tbo.ttm)) {
drm_gem_object_put_unlocked(gobj);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
return -EPERM;
}
*offset_p = radeon_bo_mmap_offset(robj);
drm_gem_object_put_unlocked(gobj);
return 0;
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 19:42:42 +07:00
}
int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_mmap *args = data;
return radeon_mode_dumb_mmap(filp, dev, args->handle, &args->addr_ptr);
}
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 19:42:42 +07:00
int radeon_gem_busy_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_busy *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *robj;
int r;
uint32_t cur_placement = 0;
gobj = drm_gem_object_lookup(filp, args->handle);
if (gobj == NULL) {
return -ENOENT;
}
robj = gem_to_radeon_bo(gobj);
r = dma_resv_test_signaled_rcu(robj->tbo.base.resv, true);
if (r == 0)
r = -EBUSY;
else
r = 0;
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 04:07:29 +07:00
cur_placement = READ_ONCE(robj->tbo.mem.mem_type);
args->domain = radeon_mem_type_to_domain(cur_placement);
drm_gem_object_put_unlocked(gobj);
return r;
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 19:42:42 +07:00
}
int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct radeon_device *rdev = dev->dev_private;
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 19:42:42 +07:00
struct drm_radeon_gem_wait_idle *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *robj;
int r = 0;
uint32_t cur_placement = 0;
long ret;
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 19:42:42 +07:00
gobj = drm_gem_object_lookup(filp, args->handle);
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 19:42:42 +07:00
if (gobj == NULL) {
return -ENOENT;
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 19:42:42 +07:00
}
robj = gem_to_radeon_bo(gobj);
ret = dma_resv_wait_timeout_rcu(robj->tbo.base.resv, true, true, 30 * HZ);
if (ret == 0)
r = -EBUSY;
else if (ret < 0)
r = ret;
/* Flush HDP cache via MMIO if necessary */
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-24 04:07:29 +07:00
cur_placement = READ_ONCE(robj->tbo.mem.mem_type);
if (rdev->asic->mmio_hdp_flush &&
radeon_mem_type_to_domain(cur_placement) == RADEON_GEM_DOMAIN_VRAM)
robj->rdev->asic->mmio_hdp_flush(rdev);
drm_gem_object_put_unlocked(gobj);
r = radeon_gem_handle_lockup(rdev, r);
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 19:42:42 +07:00
return r;
}
int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_set_tiling *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *robj;
int r = 0;
DRM_DEBUG("%d \n", args->handle);
gobj = drm_gem_object_lookup(filp, args->handle);
if (gobj == NULL)
return -ENOENT;
robj = gem_to_radeon_bo(gobj);
r = radeon_bo_set_tiling_flags(robj, args->tiling_flags, args->pitch);
drm_gem_object_put_unlocked(gobj);
return r;
}
int radeon_gem_get_tiling_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_get_tiling *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *rbo;
int r = 0;
DRM_DEBUG("\n");
gobj = drm_gem_object_lookup(filp, args->handle);
if (gobj == NULL)
return -ENOENT;
rbo = gem_to_radeon_bo(gobj);
r = radeon_bo_reserve(rbo, false);
if (unlikely(r != 0))
goto out;
radeon_bo_get_tiling_flags(rbo, &args->tiling_flags, &args->pitch);
radeon_bo_unreserve(rbo);
out:
drm_gem_object_put_unlocked(gobj);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
return r;
}
/**
* radeon_gem_va_update_vm -update the bo_va in its VM
*
* @rdev: radeon_device pointer
* @bo_va: bo_va to update
*
* Update the bo_va directly after setting it's address. Errors are not
* vital here, so they are not reported back to userspace.
*/
static void radeon_gem_va_update_vm(struct radeon_device *rdev,
struct radeon_bo_va *bo_va)
{
struct ttm_validate_buffer tv, *entry;
struct radeon_bo_list *vm_bos;
struct ww_acquire_ctx ticket;
struct list_head list;
unsigned domain;
int r;
INIT_LIST_HEAD(&list);
tv.bo = &bo_va->bo->tbo;
tv.num_shared = 1;
list_add(&tv.head, &list);
vm_bos = radeon_vm_get_bos(rdev, bo_va->vm, &list);
if (!vm_bos)
return;
r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL);
if (r)
goto error_free;
list_for_each_entry(entry, &list, head) {
domain = radeon_mem_type_to_domain(entry->bo->mem.mem_type);
/* if anything is swapped out don't swap it in here,
just abort and wait for the next CS */
if (domain == RADEON_GEM_DOMAIN_CPU)
goto error_unreserve;
}
mutex_lock(&bo_va->vm->mutex);
r = radeon_vm_clear_freed(rdev, bo_va->vm);
if (r)
goto error_unlock;
if (bo_va->it.start)
r = radeon_vm_bo_update(rdev, bo_va, &bo_va->bo->tbo.mem);
error_unlock:
mutex_unlock(&bo_va->vm->mutex);
error_unreserve:
ttm_eu_backoff_reservation(&ticket, &list);
error_free:
kvfree(vm_bos);
if (r && r != -ERESTARTSYS)
DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
}
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
int radeon_gem_va_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_va *args = data;
struct drm_gem_object *gobj;
struct radeon_device *rdev = dev->dev_private;
struct radeon_fpriv *fpriv = filp->driver_priv;
struct radeon_bo *rbo;
struct radeon_bo_va *bo_va;
u32 invalid_flags;
int r = 0;
if (!rdev->vm_manager.enabled) {
args->operation = RADEON_VA_RESULT_ERROR;
return -ENOTTY;
}
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
/* !! DONT REMOVE !!
* We don't support vm_id yet, to be sure we don't have have broken
* userspace, reject anyone trying to use non 0 value thus moving
* forward we can use those fields without breaking existant userspace
*/
if (args->vm_id) {
args->operation = RADEON_VA_RESULT_ERROR;
return -EINVAL;
}
if (args->offset < RADEON_VA_RESERVED_SIZE) {
dev_err(&dev->pdev->dev,
"offset 0x%lX is in reserved area 0x%X\n",
(unsigned long)args->offset,
RADEON_VA_RESERVED_SIZE);
args->operation = RADEON_VA_RESULT_ERROR;
return -EINVAL;
}
/* don't remove, we need to enforce userspace to set the snooped flag
* otherwise we will endup with broken userspace and we won't be able
* to enable this feature without adding new interface
*/
invalid_flags = RADEON_VM_PAGE_VALID | RADEON_VM_PAGE_SYSTEM;
if ((args->flags & invalid_flags)) {
dev_err(&dev->pdev->dev, "invalid flags 0x%08X vs 0x%08X\n",
args->flags, invalid_flags);
args->operation = RADEON_VA_RESULT_ERROR;
return -EINVAL;
}
switch (args->operation) {
case RADEON_VA_MAP:
case RADEON_VA_UNMAP:
break;
default:
dev_err(&dev->pdev->dev, "unsupported operation %d\n",
args->operation);
args->operation = RADEON_VA_RESULT_ERROR;
return -EINVAL;
}
gobj = drm_gem_object_lookup(filp, args->handle);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
if (gobj == NULL) {
args->operation = RADEON_VA_RESULT_ERROR;
return -ENOENT;
}
rbo = gem_to_radeon_bo(gobj);
r = radeon_bo_reserve(rbo, false);
if (r) {
args->operation = RADEON_VA_RESULT_ERROR;
drm_gem_object_put_unlocked(gobj);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
return r;
}
bo_va = radeon_vm_bo_find(&fpriv->vm, rbo);
if (!bo_va) {
args->operation = RADEON_VA_RESULT_ERROR;
radeon_bo_unreserve(rbo);
drm_gem_object_put_unlocked(gobj);
return -ENOENT;
}
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
switch (args->operation) {
case RADEON_VA_MAP:
if (bo_va->it.start) {
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
args->operation = RADEON_VA_RESULT_VA_EXIST;
args->offset = bo_va->it.start * RADEON_GPU_PAGE_SIZE;
radeon_bo_unreserve(rbo);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
goto out;
}
r = radeon_vm_bo_set_addr(rdev, bo_va, args->offset, args->flags);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
break;
case RADEON_VA_UNMAP:
r = radeon_vm_bo_set_addr(rdev, bo_va, 0, 0);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
break;
default:
break;
}
if (!r)
radeon_gem_va_update_vm(rdev, bo_va);
drm/radeon: GPU virtual memory support v22 Virtual address space are per drm client (opener of /dev/drm). Client are in charge of virtual address space, they need to map bo into it by calling DRM_RADEON_GEM_VA ioctl. First 16M of virtual address space is reserved by the kernel. Once using 2 level page table we should be able to have a small vram memory footprint for each pt (there would be one pt for all gart, one for all vram and then one first level for each virtual address space). Plan include using the sub allocator for a common vm page table area and using memcpy to copy vm page table in & out. Or use a gart object and copy things in & out using dma. v2: agd5f fixes: - Add vram base offset for vram pages. The GPU physical address of a vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete cards and the physical bus address of the stolen memory on integrated chips. - VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1 v3: agd5f: - integrate with the semaphore/multi-ring stuff v4: - rebase on top ttm dma & multi-ring stuff - userspace is now in charge of the address space - no more specific cs vm ioctl, instead cs ioctl has a new chunk v5: - properly handle mem == NULL case from move_notify callback - fix the vm cleanup path v6: - fix update of page table to only happen on valid mem placement v7: - add tlb flush for each vm context - add flags to define mapping property (readable, writeable, snooped) - make ring id implicit from ib->fence->ring, up to each asic callback to then do ring specific scheduling if vm ib scheduling function v8: - add query for ib limit and kernel reserved virtual space - rename vm->size to max_pfn (maximum number of page) - update gem_va ioctl to also allow unmap operation - bump kernel version to allow userspace to query for vm support v9: - rebuild page table only when bind and incrementaly depending on bo referenced by cs and that have been moved - allow virtual address space to grow - use sa allocator for vram page table - return invalid when querying vm limit on non cayman GPU - dump vm fault register on lockup v10: agd5f: - Move the vm schedule_ib callback to a standalone function, remove the callback and use the existing ib_execute callback for VM IBs. v11: - rebase on top of lastest Linus v12: agd5f: - remove spurious backslash - set IB vm_id to 0 in radeon_ib_get() v13: agd5f: - fix handling of RADEON_CHUNK_ID_FLAGS v14: - fix va destruction - fix suspend resume - forbid bo to have several different va in same vm v15: - rebase v16: - cleanup left over of vm init/fini v17: agd5f: - cs checker v18: agd5f: - reworks the CS ioctl to better support multiple rings and VM. Rather than adding a new chunk id for VM, just re-use the IB chunk id and add a new flags for VM mode. Also define additional dwords for the flags chunk id to define the what ring we want to use (gfx, compute, uvd, etc.) and the priority. v19: - fix cs fini in weird case of no ib - semi working flush fix for ni - rebase on top of sa allocator changes v20: agd5f: - further CS ioctl cleanups from Christian's comments v21: agd5f: - integrate CS checker improvements v22: agd5f: - final cleanups for release, only allow VM CS on cayman Signed-off-by: Jerome Glisse <jglisse@redhat.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-01-06 10:11:05 +07:00
args->operation = RADEON_VA_RESULT_OK;
if (r) {
args->operation = RADEON_VA_RESULT_ERROR;
}
out:
drm_gem_object_put_unlocked(gobj);
return r;
}
int radeon_gem_op_ioctl(struct drm_device *dev, void *data,
struct drm_file *filp)
{
struct drm_radeon_gem_op *args = data;
struct drm_gem_object *gobj;
struct radeon_bo *robj;
int r;
gobj = drm_gem_object_lookup(filp, args->handle);
if (gobj == NULL) {
return -ENOENT;
}
robj = gem_to_radeon_bo(gobj);
drm/radeon: add userptr support v8 This patch adds an IOCTL for turning a pointer supplied by userspace into a buffer object. It imposes several restrictions upon the memory being mapped: 1. It must be page aligned (both start/end addresses, i.e ptr and size). 2. It must be normal system memory, not a pointer into another map of IO space (e.g. it must not be a GTT mmapping of another object). 3. The BO is mapped into GTT, so the maximum amount of memory mapped at all times is still the GTT limit. 4. The BO is only mapped readonly for now, so no write support. 5. List of backing pages is only acquired once, so they represent a snapshot of the first use. Exporting and sharing as well as mapping of buffer objects created by this function is forbidden and results in an -EPERM. v2: squash all previous changes into first public version v3: fix tabs, map readonly, don't use MM callback any more v4: set TTM_PAGE_FLAG_SG so that TTM never messes with the pages, pin/unpin pages on bind/unbind instead of populate/unpopulate v5: rebased on 3.17-wip, IOCTL renamed to userptr, reject any unknown flags, better handle READONLY flag, improve permission check v6: fix ptr cast warning, use set_page_dirty/mark_page_accessed on unpin v7: add warning about it's availability in the API definition v8: drop access_ok check, fix VM mapping bits Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v4) Reviewed-by: Jérôme Glisse <jglisse@redhat.com> (v4) Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2014-08-07 14:36:00 +07:00
r = -EPERM;
if (radeon_ttm_tt_has_userptr(robj->tbo.ttm))
goto out;
r = radeon_bo_reserve(robj, false);
if (unlikely(r))
goto out;
switch (args->op) {
case RADEON_GEM_OP_GET_INITIAL_DOMAIN:
args->value = robj->initial_domain;
break;
case RADEON_GEM_OP_SET_INITIAL_DOMAIN:
robj->initial_domain = args->value & (RADEON_GEM_DOMAIN_VRAM |
RADEON_GEM_DOMAIN_GTT |
RADEON_GEM_DOMAIN_CPU);
break;
default:
r = -EINVAL;
}
radeon_bo_unreserve(robj);
out:
drm_gem_object_put_unlocked(gobj);
return r;
}
int radeon_mode_dumb_create(struct drm_file *file_priv,
struct drm_device *dev,
struct drm_mode_create_dumb *args)
{
struct radeon_device *rdev = dev->dev_private;
struct drm_gem_object *gobj;
uint32_t handle;
int r;
args->pitch = radeon_align_pitch(rdev, args->width,
DIV_ROUND_UP(args->bpp, 8), 0);
args->size = args->pitch * args->height;
args->size = ALIGN(args->size, PAGE_SIZE);
r = radeon_gem_object_create(rdev, args->size, 0,
RADEON_GEM_DOMAIN_VRAM, 0,
false, &gobj);
if (r)
return -ENOMEM;
r = drm_gem_handle_create(file_priv, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_put_unlocked(gobj);
if (r) {
return r;
}
args->handle = handle;
return 0;
}
#if defined(CONFIG_DEBUG_FS)
static int radeon_debugfs_gem_info(struct seq_file *m, void *data)
{
struct drm_info_node *node = (struct drm_info_node *)m->private;
struct drm_device *dev = node->minor->dev;
struct radeon_device *rdev = dev->dev_private;
struct radeon_bo *rbo;
unsigned i = 0;
mutex_lock(&rdev->gem.mutex);
list_for_each_entry(rbo, &rdev->gem.objects, list) {
unsigned domain;
const char *placement;
domain = radeon_mem_type_to_domain(rbo->tbo.mem.mem_type);
switch (domain) {
case RADEON_GEM_DOMAIN_VRAM:
placement = "VRAM";
break;
case RADEON_GEM_DOMAIN_GTT:
placement = " GTT";
break;
case RADEON_GEM_DOMAIN_CPU:
default:
placement = " CPU";
break;
}
seq_printf(m, "bo[0x%08x] %8ldkB %8ldMB %s pid %8ld\n",
i, radeon_bo_size(rbo) >> 10, radeon_bo_size(rbo) >> 20,
placement, (unsigned long)rbo->pid);
i++;
}
mutex_unlock(&rdev->gem.mutex);
return 0;
}
static struct drm_info_list radeon_debugfs_gem_list[] = {
{"radeon_gem_info", &radeon_debugfs_gem_info, 0, NULL},
};
#endif
int radeon_gem_debugfs_init(struct radeon_device *rdev)
{
#if defined(CONFIG_DEBUG_FS)
return radeon_debugfs_add_files(rdev, radeon_debugfs_gem_list, 1);
#endif
return 0;
}