From 721604a15b934f0a8d1909acb8017f029128be2f Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Thu, 5 Jan 2012 22:11:05 -0500 Subject: 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 Signed-off-by: Alex Deucher Signed-off-by: Dave Airlie --- drivers/gpu/drm/radeon/radeon_gem.c | 136 ++++++++++++++++++++++++++++++++++++ 1 file changed, 136 insertions(+) (limited to 'drivers/gpu/drm/radeon/radeon_gem.c') diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index ae321975283c..003eeec1b688 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -142,6 +142,44 @@ void radeon_gem_fini(struct radeon_device *rdev) radeon_bo_force_delete(rdev); } +/* + * 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) +{ + 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, *tmp; + + if (rdev->family < CHIP_CAYMAN) { + return; + } + + if (radeon_bo_reserve(rbo, false)) { + return; + } + list_for_each_entry_safe(bo_va, tmp, &rbo->va, bo_list) { + if (bo_va->vm == vm) { + /* remove from this vm address space */ + mutex_lock(&vm->mutex); + list_del(&bo_va->vm_list); + mutex_unlock(&vm->mutex); + list_del(&bo_va->bo_list); + kfree(bo_va); + } + } + radeon_bo_unreserve(rbo); +} + /* * GEM ioctls. @@ -354,6 +392,104 @@ out: return r; } +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; + + /* !! 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; + } + if (!(args->flags & RADEON_VM_PAGE_SNOOPED)) { + dev_err(&dev->pdev->dev, "only supported snooped mapping for now\n"); + 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(dev, filp, args->handle); + 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_unreference_unlocked(gobj); + return r; + } + switch (args->operation) { + case RADEON_VA_MAP: + bo_va = radeon_bo_va(rbo, &fpriv->vm); + if (bo_va) { + args->operation = RADEON_VA_RESULT_VA_EXIST; + args->offset = bo_va->soffset; + goto out; + } + r = radeon_vm_bo_add(rdev, &fpriv->vm, rbo, + args->offset, args->flags); + break; + case RADEON_VA_UNMAP: + r = radeon_vm_bo_rmv(rdev, &fpriv->vm, rbo); + break; + default: + break; + } + args->operation = RADEON_VA_RESULT_OK; + if (r) { + args->operation = RADEON_VA_RESULT_ERROR; + } +out: + radeon_bo_unreserve(rbo); + drm_gem_object_unreference_unlocked(gobj); + return r; +} + int radeon_mode_dumb_create(struct drm_file *file_priv, struct drm_device *dev, struct drm_mode_create_dumb *args) -- cgit v1.2.1