[PATCH v2] drm/xe/uapi: Uniform async vs sync handling
Matthew Brost
matthew.brost at intel.com
Wed Dec 13 17:59:20 UTC 2023
On Wed, Dec 13, 2023 at 06:36:58PM +0100, Thomas Hellström wrote:
>
> On 12/13/23 15:13, Souza, Jose wrote:
> > On Tue, 2023-12-12 at 20:35 -0800, Matthew Brost wrote:
> > > Remove concept of async vs sync VM bind queues, rather make async vs
> > > sync a per IOCTL choice. Since this is per IOCTL, it makes sense to have
> > > a singular flag IOCTL rather than per VM bind op flag too. Add
> > > DRM_XE_SYNCS_FLAG_WAIT_FOR_OP which is an input sync flag to support
> > > this. Support this new flag for both the VM bind IOCTL only with a path
> > > to support this for execs too.
> > >
> > > v2: Add cookie, move sync wait outside of any locks.
> > >
> > > Cc: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > > Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> > > Cc: Francois Dugast <francois.dugast at intel.com>
> > > Signed-off-by: Matthew Brost <matthew.brost at intel.com>
> > > ---
> > > drivers/gpu/drm/xe/xe_exec.c | 12 ++-
> > > drivers/gpu/drm/xe/xe_exec_queue.c | 7 +-
> > > drivers/gpu/drm/xe/xe_exec_queue_types.h | 2 -
> > > drivers/gpu/drm/xe/xe_vm.c | 116 +++++++++++------------
> > > drivers/gpu/drm/xe/xe_vm_types.h | 13 ++-
> > > include/uapi/drm/xe_drm.h | 64 ++++++++-----
> > > 6 files changed, 109 insertions(+), 105 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/xe/xe_exec.c b/drivers/gpu/drm/xe/xe_exec.c
> > > index ba92e5619da3..8a1530dab65f 100644
> > > --- a/drivers/gpu/drm/xe/xe_exec.c
> > > +++ b/drivers/gpu/drm/xe/xe_exec.c
> > > @@ -104,7 +104,7 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > struct xe_device *xe = to_xe_device(dev);
> > > struct xe_file *xef = to_xe_file(file);
> > > struct drm_xe_exec *args = data;
> > > - struct drm_xe_sync __user *syncs_user = u64_to_user_ptr(args->syncs);
> > > + struct drm_xe_sync __user *syncs_user = u64_to_user_ptr(args->syncs.syncs);
> > > u64 __user *addresses_user = u64_to_user_ptr(args->address);
> > > struct xe_exec_queue *q;
> > > struct xe_sync_entry *syncs = NULL;
> > > @@ -120,7 +120,9 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > int err = 0;
> > > if (XE_IOCTL_DBG(xe, args->extensions) ||
> > > - XE_IOCTL_DBG(xe, args->pad[0] || args->pad[1] || args->pad[2]) ||
> > > + XE_IOCTL_DBG(xe, args->pad || args->pad2[0] || args->pad2[1] || args->pad2[2]) ||
> > > + XE_IOCTL_DBG(xe, args->syncs.flags) ||
> > > + XE_IOCTL_DBG(xe, args->syncs.cookie) ||
> > > XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1]))
> > > return -EINVAL;
> > > @@ -140,8 +142,8 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > goto err_exec_queue;
> > > }
> > > - if (args->num_syncs) {
> > > - syncs = kcalloc(args->num_syncs, sizeof(*syncs), GFP_KERNEL);
> > > + if (args->syncs.num_syncs) {
> > > + syncs = kcalloc(args->syncs.num_syncs, sizeof(*syncs), GFP_KERNEL);
> > > if (!syncs) {
> > > err = -ENOMEM;
> > > goto err_exec_queue;
> > > @@ -150,7 +152,7 @@ int xe_exec_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > vm = q->vm;
> > > - for (i = 0; i < args->num_syncs; i++) {
> > > + for (i = 0; i < args->syncs.num_syncs; i++) {
> > > err = xe_sync_entry_parse(xe, xef, &syncs[num_syncs++],
> > > &syncs_user[i], SYNC_PARSE_FLAG_EXEC |
> > > (xe_vm_in_lr_mode(vm) ?
> > > diff --git a/drivers/gpu/drm/xe/xe_exec_queue.c b/drivers/gpu/drm/xe/xe_exec_queue.c
> > > index eeb9605dd45f..a25d67971fdd 100644
> > > --- a/drivers/gpu/drm/xe/xe_exec_queue.c
> > > +++ b/drivers/gpu/drm/xe/xe_exec_queue.c
> > > @@ -625,10 +625,7 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data,
> > > if (XE_IOCTL_DBG(xe, eci[0].gt_id >= xe->info.gt_count))
> > > return -EINVAL;
> > > - if (eci[0].engine_class >= DRM_XE_ENGINE_CLASS_VM_BIND_ASYNC) {
> > > - bool sync = eci[0].engine_class ==
> > > - DRM_XE_ENGINE_CLASS_VM_BIND_SYNC;
> > > -
> > > + if (eci[0].engine_class == DRM_XE_ENGINE_CLASS_VM_BIND) {
> > > for_each_gt(gt, xe, id) {
> > > struct xe_exec_queue *new;
> > > @@ -654,8 +651,6 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data,
> > > args->width, hwe,
> > > EXEC_QUEUE_FLAG_PERSISTENT |
> > > EXEC_QUEUE_FLAG_VM |
> > > - (sync ? 0 :
> > > - EXEC_QUEUE_FLAG_VM_ASYNC) |
> > > (id ?
> > > EXEC_QUEUE_FLAG_BIND_ENGINE_CHILD :
> > > 0));
> > > diff --git a/drivers/gpu/drm/xe/xe_exec_queue_types.h b/drivers/gpu/drm/xe/xe_exec_queue_types.h
> > > index c7aefa1c8c31..0f7f6cded4a3 100644
> > > --- a/drivers/gpu/drm/xe/xe_exec_queue_types.h
> > > +++ b/drivers/gpu/drm/xe/xe_exec_queue_types.h
> > > @@ -84,8 +84,6 @@ struct xe_exec_queue {
> > > #define EXEC_QUEUE_FLAG_VM BIT(4)
> > > /* child of VM queue for multi-tile VM jobs */
> > > #define EXEC_QUEUE_FLAG_BIND_ENGINE_CHILD BIT(5)
> > > -/* VM jobs for this queue are asynchronous */
> > > -#define EXEC_QUEUE_FLAG_VM_ASYNC BIT(6)
> > > /**
> > > * @flags: flags for this exec queue, should statically setup aside from ban
> > > diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> > > index 2f3df9ee67c9..ab38685d2daf 100644
> > > --- a/drivers/gpu/drm/xe/xe_vm.c
> > > +++ b/drivers/gpu/drm/xe/xe_vm.c
> > > @@ -1343,9 +1343,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags)
> > > struct xe_gt *gt = tile->primary_gt;
> > > struct xe_vm *migrate_vm;
> > > struct xe_exec_queue *q;
> > > - u32 create_flags = EXEC_QUEUE_FLAG_VM |
> > > - ((flags & XE_VM_FLAG_ASYNC_DEFAULT) ?
> > > - EXEC_QUEUE_FLAG_VM_ASYNC : 0);
> > > + u32 create_flags = EXEC_QUEUE_FLAG_VM;
> > > if (!vm->pt_root[id])
> > > continue;
> > > @@ -1712,12 +1710,6 @@ xe_vm_bind_vma(struct xe_vma *vma, struct xe_exec_queue *q,
> > > return ERR_PTR(err);
> > > }
> > > -static bool xe_vm_sync_mode(struct xe_vm *vm, struct xe_exec_queue *q)
> > > -{
> > > - return q ? !(q->flags & EXEC_QUEUE_FLAG_VM_ASYNC) :
> > > - !(vm->flags & XE_VM_FLAG_ASYNC_DEFAULT);
> > > -}
> > > -
> > > static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma,
> > > struct xe_exec_queue *q, struct xe_sync_entry *syncs,
> > > u32 num_syncs, bool immediate, bool first_op,
> > > @@ -1747,8 +1739,6 @@ static int __xe_vm_bind(struct xe_vm *vm, struct xe_vma *vma,
> > > if (last_op)
> > > xe_exec_queue_last_fence_set(wait_exec_queue, vm, fence);
> > > - if (last_op && xe_vm_sync_mode(vm, q))
> > > - dma_fence_wait(fence, true);
> > > dma_fence_put(fence);
> > > return 0;
> > > @@ -1791,8 +1781,6 @@ static int xe_vm_unbind(struct xe_vm *vm, struct xe_vma *vma,
> > > xe_vma_destroy(vma, fence);
> > > if (last_op)
> > > xe_exec_queue_last_fence_set(wait_exec_queue, vm, fence);
> > > - if (last_op && xe_vm_sync_mode(vm, q))
> > > - dma_fence_wait(fence, true);
> > > dma_fence_put(fence);
> > > return 0;
> > > @@ -1800,7 +1788,6 @@ static int xe_vm_unbind(struct xe_vm *vm, struct xe_vma *vma,
> > > #define ALL_DRM_XE_VM_CREATE_FLAGS (DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE | \
> > > DRM_XE_VM_CREATE_FLAG_LR_MODE | \
> > > - DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT | \
> > > DRM_XE_VM_CREATE_FLAG_FAULT_MODE)
> > > int xe_vm_create_ioctl(struct drm_device *dev, void *data,
> > > @@ -1854,8 +1841,6 @@ int xe_vm_create_ioctl(struct drm_device *dev, void *data,
> > > flags |= XE_VM_FLAG_SCRATCH_PAGE;
> > > if (args->flags & DRM_XE_VM_CREATE_FLAG_LR_MODE)
> > > flags |= XE_VM_FLAG_LR_MODE;
> > > - if (args->flags & DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT)
> > > - flags |= XE_VM_FLAG_ASYNC_DEFAULT;
> > > if (args->flags & DRM_XE_VM_CREATE_FLAG_FAULT_MODE)
> > > flags |= XE_VM_FLAG_FAULT_MODE;
> > > @@ -2263,8 +2248,7 @@ static int xe_vma_op_commit(struct xe_vm *vm, struct xe_vma_op *op)
> > > static int vm_bind_ioctl_ops_parse(struct xe_vm *vm, struct xe_exec_queue *q,
> > > struct drm_gpuva_ops *ops,
> > > struct xe_sync_entry *syncs, u32 num_syncs,
> > > - struct list_head *ops_list, bool last,
> > > - bool async)
> > > + struct list_head *ops_list, bool last)
> > > {
> > > struct xe_vma_op *last_op = NULL;
> > > struct drm_gpuva_op *__op;
> > > @@ -2696,16 +2680,16 @@ static int vm_bind_ioctl_ops_execute(struct xe_vm *vm,
> > > #ifdef TEST_VM_ASYNC_OPS_ERROR
> > > #define SUPPORTED_FLAGS \
> > > - (FORCE_ASYNC_OP_ERROR | DRM_XE_VM_BIND_FLAG_ASYNC | \
> > > - DRM_XE_VM_BIND_FLAG_READONLY | DRM_XE_VM_BIND_FLAG_IMMEDIATE | \
> > > - DRM_XE_VM_BIND_FLAG_NULL | 0xffff)
> > > + (FORCE_ASYNC_OP_ERROR | DRM_XE_VM_BIND_FLAG_READONLY | \
> > > + DRM_XE_VM_BIND_FLAG_IMMEDIATE | DRM_XE_VM_BIND_FLAG_NULL | 0xffff)
> > > #else
> > > #define SUPPORTED_FLAGS \
> > > - (DRM_XE_VM_BIND_FLAG_ASYNC | DRM_XE_VM_BIND_FLAG_READONLY | \
> > > + (DRM_XE_VM_BIND_FLAG_READONLY | \
> > > DRM_XE_VM_BIND_FLAG_IMMEDIATE | DRM_XE_VM_BIND_FLAG_NULL | \
> > > 0xffff)
> > > #endif
> > > #define XE_64K_PAGE_MASK 0xffffull
> > > +#define ALL_DRM_XE_SYNCS_FLAGS (DRM_XE_SYNCS_FLAG_WAIT_FOR_OP)
> > > #define MAX_BINDS 512 /* FIXME: Picking random upper limit */
> > > @@ -2717,7 +2701,7 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe,
> > > int err;
> > > int i;
> > > - if (XE_IOCTL_DBG(xe, args->pad || args->pad2) ||
> > > + if (XE_IOCTL_DBG(xe, args->pad) ||
> > > XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1]))
> > > return -EINVAL;
> > > @@ -2745,6 +2729,14 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe,
> > > *bind_ops = &args->bind;
> > > }
> > > + *async = !(args->syncs.flags & DRM_XE_SYNCS_FLAG_WAIT_FOR_OP);
> > > +
> > > + if (XE_IOCTL_DBG(xe, args->syncs.flags & ~ALL_DRM_XE_SYNCS_FLAGS) ||
> > > + XE_IOCTL_DBG(xe, !*async && args->syncs.num_syncs)) {
> > > + err = -EINVAL;
> > > + goto free_bind_ops;
> > > + }
> > > +
> > > for (i = 0; i < args->num_binds; ++i) {
> > > u64 range = (*bind_ops)[i].range;
> > > u64 addr = (*bind_ops)[i].addr;
> > > @@ -2775,18 +2767,6 @@ static int vm_bind_ioctl_check_args(struct xe_device *xe,
> > > goto free_bind_ops;
> > > }
> > > - if (i == 0) {
> > > - *async = !!(flags & DRM_XE_VM_BIND_FLAG_ASYNC);
> > > - if (XE_IOCTL_DBG(xe, !*async && args->num_syncs)) {
> > > - err = -EINVAL;
> > > - goto free_bind_ops;
> > > - }
> > > - } else if (XE_IOCTL_DBG(xe, *async !=
> > > - !!(flags & DRM_XE_VM_BIND_FLAG_ASYNC))) {
> > > - err = -EINVAL;
> > > - goto free_bind_ops;
> > > - }
> > > -
> > > if (XE_IOCTL_DBG(xe, op > DRM_XE_VM_BIND_OP_PREFETCH) ||
> > > XE_IOCTL_DBG(xe, flags & ~SUPPORTED_FLAGS) ||
> > > XE_IOCTL_DBG(xe, obj && is_null) ||
> > > @@ -2854,12 +2834,25 @@ static int vm_bind_ioctl_signal_fences(struct xe_vm *vm,
> > > xe_exec_queue_last_fence_set(to_wait_exec_queue(vm, q), vm,
> > > fence);
> > > + dma_fence_put(fence);
> > > - if (xe_vm_sync_mode(vm, q)) {
> > > - long timeout = dma_fence_wait(fence, true);
> > > + return err;
> > > +}
> > > - if (timeout < 0)
> > > - err = -EINTR;
> > > +static int vm_bind_ioctl_sync_wait(struct xe_vm *vm,
> > > + struct dma_fence *fence,
> > > + u64 __user *cookie)
> > > +{
> > > + long timeout;
> > > + int err = 0;
> > > +
> > > + timeout = dma_fence_wait(fence, true);
> > > + if (timeout < 0) {
> > > + u64 value = 1;
> > > +
> > > + err = -ERESTARTSYS;
> > > + if (copy_to_user(cookie, &value, sizeof(value)))
> > > + err = -EFAULT;
> > > }
> > > dma_fence_put(fence);
> > > @@ -2875,6 +2868,7 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > struct drm_xe_sync __user *syncs_user;
> > > struct xe_bo **bos = NULL;
> > > struct drm_gpuva_ops **ops = NULL;
> > > + struct dma_fence *fence = NULL;
> > > struct xe_vm *vm;
> > > struct xe_exec_queue *q = NULL;
> > > u32 num_syncs;
> > > @@ -2889,7 +2883,7 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > if (err)
> > > return err;
> > > - if (XE_IOCTL_DBG(xe, args->pad || args->pad2) ||
> > > + if (XE_IOCTL_DBG(xe, args->pad) ||
> > > XE_IOCTL_DBG(xe, args->reserved[0] || args->reserved[1]))
> > > return -EINVAL;
> > > @@ -2904,12 +2898,6 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > err = -EINVAL;
> > > goto put_exec_queue;
> > > }
> > > -
> > > - if (XE_IOCTL_DBG(xe, args->num_binds && async !=
> > > - !!(q->flags & EXEC_QUEUE_FLAG_VM_ASYNC))) {
> > > - err = -EINVAL;
> > > - goto put_exec_queue;
> > > - }
> > > }
> > > vm = xe_vm_lookup(xef, args->vm_id);
> > > @@ -2918,14 +2906,6 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > goto put_exec_queue;
> > > }
> > > - if (!args->exec_queue_id) {
> > > - if (XE_IOCTL_DBG(xe, args->num_binds && async !=
> > > - !!(vm->flags & XE_VM_FLAG_ASYNC_DEFAULT))) {
> > > - err = -EINVAL;
> > > - goto put_vm;
> > > - }
> > > - }
> > > -
> > > err = down_write_killable(&vm->lock);
> > > if (err)
> > > goto put_vm;
> > > @@ -3015,16 +2995,16 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > }
> > > }
> > > - if (args->num_syncs) {
> > > - syncs = kcalloc(args->num_syncs, sizeof(*syncs), GFP_KERNEL);
> > > + if (args->syncs.num_syncs) {
> > > + syncs = kcalloc(args->syncs.num_syncs, sizeof(*syncs), GFP_KERNEL);
> > > if (!syncs) {
> > > err = -ENOMEM;
> > > goto put_obj;
> > > }
> > > }
> > > - syncs_user = u64_to_user_ptr(args->syncs);
> > > - for (num_syncs = 0; num_syncs < args->num_syncs; num_syncs++) {
> > > + syncs_user = u64_to_user_ptr(args->syncs.syncs);
> > > + for (num_syncs = 0; num_syncs < args->syncs.num_syncs; num_syncs++) {
> > > err = xe_sync_entry_parse(xe, xef, &syncs[num_syncs],
> > > &syncs_user[num_syncs],
> > > (xe_vm_in_lr_mode(vm) ?
> > > @@ -3060,8 +3040,7 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > err = vm_bind_ioctl_ops_parse(vm, q, ops[i], syncs, num_syncs,
> > > &ops_list,
> > > - i == args->num_binds - 1,
> > > - async);
> > > + i == args->num_binds - 1);
> > > if (err)
> > > goto unwind_ops;
> > > }
> > > @@ -3077,7 +3056,10 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > xe_exec_queue_get(q);
> > > err = vm_bind_ioctl_ops_execute(vm, &ops_list);
> > > -
> > > + if (!err && !async) {
> > > + fence = xe_exec_queue_last_fence_get(to_wait_exec_queue(vm, q), vm);
> > > + dma_fence_get(fence);
> > > + }
> > > up_write(&vm->lock);
> > > if (q)
> > > @@ -3092,13 +3074,19 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > if (args->num_binds > 1)
> > > kfree(bind_ops);
> > > - return err;
> > > + return fence ? vm_bind_ioctl_sync_wait(vm, fence, u64_to_user_ptr(args->syncs.cookie)) :
> > > + err;
> > > unwind_ops:
> > > vm_bind_ioctl_ops_unwind(vm, ops, args->num_binds);
> > > free_syncs:
> > > - if (err == -ENODATA)
> > > + if (err == -ENODATA) {
> > > err = vm_bind_ioctl_signal_fences(vm, q, syncs, num_syncs);
> > > + if (!async) {
> > > + fence = xe_exec_queue_last_fence_get(to_wait_exec_queue(vm, q), vm);
> > > + dma_fence_get(fence);
> > > + }
> > > + }
> > > while (num_syncs--)
> > > xe_sync_entry_cleanup(&syncs[num_syncs]);
> > > @@ -3108,6 +3096,8 @@ int xe_vm_bind_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
> > > xe_bo_put(bos[i]);
> > > release_vm_lock:
> > > up_write(&vm->lock);
> > > + if (fence)
> > > + err = vm_bind_ioctl_sync_wait(vm, fence, u64_to_user_ptr(args->syncs.cookie));
> > > put_vm:
> > > xe_vm_put(vm);
> > > put_exec_queue:
> > > diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
> > > index 2e023596cb15..63e8a50b88e9 100644
> > > --- a/drivers/gpu/drm/xe/xe_vm_types.h
> > > +++ b/drivers/gpu/drm/xe/xe_vm_types.h
> > > @@ -138,13 +138,12 @@ struct xe_vm {
> > > */
> > > #define XE_VM_FLAG_64K BIT(0)
> > > #define XE_VM_FLAG_LR_MODE BIT(1)
> > > -#define XE_VM_FLAG_ASYNC_DEFAULT BIT(2)
> > > -#define XE_VM_FLAG_MIGRATION BIT(3)
> > > -#define XE_VM_FLAG_SCRATCH_PAGE BIT(4)
> > > -#define XE_VM_FLAG_FAULT_MODE BIT(5)
> > > -#define XE_VM_FLAG_BANNED BIT(6)
> > > -#define XE_VM_FLAG_TILE_ID(flags) FIELD_GET(GENMASK(8, 7), flags)
> > > -#define XE_VM_FLAG_SET_TILE_ID(tile) FIELD_PREP(GENMASK(8, 7), (tile)->id)
> > > +#define XE_VM_FLAG_MIGRATION BIT(2)
> > > +#define XE_VM_FLAG_SCRATCH_PAGE BIT(3)
> > > +#define XE_VM_FLAG_FAULT_MODE BIT(4)
> > > +#define XE_VM_FLAG_BANNED BIT(5)
> > > +#define XE_VM_FLAG_TILE_ID(flags) FIELD_GET(GENMASK(7, 6), flags)
> > > +#define XE_VM_FLAG_SET_TILE_ID(tile) FIELD_PREP(GENMASK(7, 6), (tile)->id)
> > > unsigned long flags;
> > > /** @composite_fence_ctx: context composite fence */
> > > diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h
> > > index 0895e4d2a981..d72e4441cc80 100644
> > > --- a/include/uapi/drm/xe_drm.h
> > > +++ b/include/uapi/drm/xe_drm.h
> > > @@ -140,8 +140,7 @@ struct drm_xe_engine_class_instance {
> > > * Kernel only classes (not actual hardware engine class). Used for
> > > * creating ordered queues of VM bind operations.
> > > */
> > > -#define DRM_XE_ENGINE_CLASS_VM_BIND_ASYNC 5
> > > -#define DRM_XE_ENGINE_CLASS_VM_BIND_SYNC 6
> > > +#define DRM_XE_ENGINE_CLASS_VM_BIND 5
> > > /** @engine_class: engine class id */
> > > __u16 engine_class;
> > > /** @engine_instance: engine instance id */
> > > @@ -661,7 +660,6 @@ struct drm_xe_vm_create {
> > > * still enable recoverable pagefaults if supported by the device.
> > > */
> > > #define DRM_XE_VM_CREATE_FLAG_LR_MODE (1 << 1)
> > > -#define DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT (1 << 2)
> > > /*
> > > * DRM_XE_VM_CREATE_FLAG_FAULT_MODE requires also
> > > * DRM_XE_VM_CREATE_FLAG_LR_MODE. It allows memory to be allocated
> > > @@ -669,7 +667,7 @@ struct drm_xe_vm_create {
> > > * The xe driver internally uses recoverable pagefaults to implement
> > > * this.
> > > */
> > > -#define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 3)
> > > +#define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 2)
> > > /** @flags: Flags */
> > > __u32 flags;
> > > @@ -777,12 +775,11 @@ struct drm_xe_vm_bind_op {
> > > __u32 op;
> > > #define DRM_XE_VM_BIND_FLAG_READONLY (1 << 0)
> > > -#define DRM_XE_VM_BIND_FLAG_ASYNC (1 << 1)
> > > /*
> > > * Valid on a faulting VM only, do the MAP operation immediately rather
> > > * than deferring the MAP to the page fault handler.
> > > */
> > > -#define DRM_XE_VM_BIND_FLAG_IMMEDIATE (1 << 2)
> > > +#define DRM_XE_VM_BIND_FLAG_IMMEDIATE (1 << 1)
> > > /*
> > > * When the NULL flag is set, the page tables are setup with a special
> > > * bit which indicates writes are dropped and all reads return zero. In
> > > @@ -790,7 +787,7 @@ struct drm_xe_vm_bind_op {
> > > * operations, the BO handle MBZ, and the BO offset MBZ. This flag is
> > > * intended to implement VK sparse bindings.
> > > */
> > > -#define DRM_XE_VM_BIND_FLAG_NULL (1 << 3)
> > > +#define DRM_XE_VM_BIND_FLAG_NULL (1 << 2)
> > > /** @flags: Bind flags */
> > > __u32 flags;
> > > @@ -808,6 +805,35 @@ struct drm_xe_vm_bind_op {
> > > __u64 reserved[3];
> > > };
> > > +/**
> > > + * struct drm_xe_syncs - In / out syncs for IOCTLs.
> > > + */
> > > +struct drm_xe_syncs {
> > > + /** @num_syncs: amount of syncs to wait on */
> > > + __u32 num_syncs;
> > > +
> > > + /*
> > > + * Block in IOCTL until operation complete, num_syncs MBZ if set.
> > > + */
> > > +#define DRM_XE_SYNCS_FLAG_WAIT_FOR_OP (1 << 0)
> > > + /** @flags: Sync flags */
> > > + __u32 flags;
> > > +
> > > + /**
> > > + * @cookie: pointer which is written with an non-zero value if IOCTL
> > > + * operation has been committed but the wait for completion
> > > + * (DRM_XE_SYNCS_FLAG_WAIT_FOR_OP is set) is interrupted. Value only
> > > + * valid when IOCTL returns -EINTR or -ERESTARTSYS.
> > > + */
> > > + __u64 cookie;
> > what is the usage of cookie?
> > from what I looked the xe_vm_bind_ioctl() will return -1 and errno will be set to -ERESTARTSYS in the case mentioned in the comment.
> >
> > if really needed would be nice if you could provide a example of how it should be used.
>
> It's opaque to user-space, except for clearing before first command
> submission.
>
> If the wait for operation complete hits a signal, then the IOCTL writes a 1
> to "cooke" and is then restarted either by the kernel itself or by
> user-space (Unless it was a CTRL-C) The KMD then reads the value of cookie.
> If it sees that it is 1, it skips performing the bind again, and goes
> directly to the wait part...
>
Ok, I missed this part that the KMD has responsible for checking the
cookie value. I thought it was up to the UMD to check the cookie on
restart and either issue the op again (num_binds != 0) or wait for
existing operations to be complete (num_binds == 0).
Easy enough to fix this.
> This is not very nice IMO, but is the price we pay for interruptible waits
> after the BIND has been commited. IMO it's better that user-space, at least
> in production code, implements sync binds using async binds with a follow-up
> wait for an out-(user) fence.
>
I think the argument is 3 IOCTLs (create sync, bind IOCTL, wait) vs. 1
(bind IOCTL).
> >
> > > +
> > > + /** @syncs: pointer to struct drm_xe_sync array */
> > > + __u64 syncs;
> > > +
> > > + /** @reserved: Reserved */
> > > + __u64 reserved[2];
> > > +};
> > > +
> > > struct drm_xe_vm_bind {
> > > /** @extensions: Pointer to the first extension struct, if any */
> > > __u64 extensions;
> > > @@ -839,14 +865,8 @@ struct drm_xe_vm_bind {
> > > __u64 vector_of_binds;
> > > };
> > > - /** @pad: MBZ */
> > > - __u32 pad2;
> > > -
> > > - /** @num_syncs: amount of syncs to wait on */
> > > - __u32 num_syncs;
> > > -
> > > - /** @syncs: pointer to struct drm_xe_sync array */
> > > - __u64 syncs;
> > > + /** @syncs: syncs for bind */
> > > + struct drm_xe_syncs syncs;
> > > /** @reserved: Reserved */
> > > __u64 reserved[2];
> > > @@ -975,14 +995,14 @@ struct drm_xe_exec {
> > > /** @extensions: Pointer to the first extension struct, if any */
> > > __u64 extensions;
> > > + /** @pad: MBZ */
> > > + __u32 pad;
> > > +
> > > /** @exec_queue_id: Exec queue ID for the batch buffer */
> > > __u32 exec_queue_id;
> > > - /** @num_syncs: Amount of struct drm_xe_sync in array. */
> > > - __u32 num_syncs;
> > > -
> > > - /** @syncs: Pointer to struct drm_xe_sync array. */
> > > - __u64 syncs;
> > > + /** @syncs: syncs for exec */
> > > + struct drm_xe_syncs syncs;
> > DRM_XE_SYNCS_FLAG_WAIT_FOR_OP is not implemented for DRM_IOCTL_XE_EXEC.
>
> That is intentional. The scheme above is awkward and IMO we shouldn't
> implement it for EXEC unless there is a specific need, that cannot be solved
> with waiting for an out-fence.
>
Yep intentionally implemented but bits are there to implemented if
needed. Implementation is trivial too.
Matt
> /Thomas
>
>
> >
> > > /**
> > > * @address: address of batch buffer if num_batch_buffer == 1 or an
> > > @@ -996,8 +1016,8 @@ struct drm_xe_exec {
> > > */
> > > __u16 num_batch_buffer;
> > > - /** @pad: MBZ */
> > > - __u16 pad[3];
> > > + /** @pad2: MBZ */
> > > + __u16 pad2[3];
> > > /** @reserved: Reserved */
> > > __u64 reserved[2];
More information about the Intel-xe
mailing list