| Message ID | 1465312727-2211-12-git-send-email-zhi.a.wang@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Jun 07, 2016 at 11:18:47AM -0400, Zhi Wang wrote: > GVT workload scheduler needs special host LRC contexts, the so called > "shadow LRC context" to submit guest workload to host i915. During the > guest workload submission, workload scheduler fills the shadow LRC > context with the content of guest LRC context: engine context is copied > without changes, ring context is mostly owned by host i915. > > v7: > > - Move chart to a better place. (Joonas) > > v6: > > - Make GVT code as dead code when !CONFIG_DRM_I915_GVT. (Chris) > > v5: > - Only compile this feature when CONFIG_DRM_I915_GVT is enabled. (Tvrtko) > - Rebase the code into new repo. > - Add a comment about the ring buffer size. (Joonas) > > v2: > > Mostly based on Daniel's idea. Call the refactored core logic of GEM > context creation service and LRC context creation service to create the GVT > context. > > Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> > --- > drivers/gpu/drm/i915/i915_gem_context.c | 65 +++++++++++++++++++++++++++++++++ > 1 file changed, 65 insertions(+) > > diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c > index b0e82a1..2d8e22a 100644 > --- a/drivers/gpu/drm/i915/i915_gem_context.c > +++ b/drivers/gpu/drm/i915/i915_gem_context.c > @@ -343,6 +343,71 @@ i915_gem_create_context(struct drm_device *dev, > return ctx; > } > > +/** > + * i915_gem_create_gvt_context - create a GVT GEM context > + * @dev: drm device * > + * > + * This function is used to create a GVT specific GEM context. > + * > + * Returns: > + * pointer to i915_gem_context on success, error pointer if failed > + * > + */ > +/* GVT context usage flow: > + * > + * +-----------+ +-----------+ > + * | vGPU | | vGPU | > + * +-+-----^---+ +-+-----^---+ > + * | | | | > + * | | GVT-g | | GVT-g > + * vELSP write| | emulates vELSP write| | emulates > + * | | Execlist/CSB | | Execlist/CSB > + * | | Status | | Status > + * | | | | > + * +------v-----+-------------------------v-----+---------+ > + * | GVT Virtual Execlist Submission | > + * +------+-------------------------------+---------------+ > + * | | > + * | Per-VM/Ring Workoad Q | Per-VM/Ring Workload Q > + * +---------------------+--+ +------------------------+ > + * +---v--------+ ^ +---v--------+ > + * |GVT Workload|... | |GVT Workload|... > + * +------------+ | +------------+ > + * | > + * | Pick Workload from Q > + * +--------------------+---------------------------------+ > + * | GVT Workload Scheduler | > + * +--------------------+---------------------------------+ > + * | * Shadow guest LRC context > + * +------v------+ * Shadow guest ring buffer > + * | GVT Context | * Scan/Patch guest RB instructions > + * +------+------+ > + * | > + * v > + * Host i915 GEM Submission > + */ I presume you move this later. > +struct i915_gem_context * > +i915_gem_create_gvt_context(struct drm_device *dev) i915_gem_context_create_gvt No function declaration in the header, build incomplete. It is worth making sure that every patch in the series builds and doesn't introduce new (sparse) warnings. > +{ > + struct i915_gem_context *ctx; > + > + if (!IS_ENABLED(CONFIG_DRM_I915_GVT)) > + return ERR_PTR(-ENODEV); > + > + mutex_lock(&dev->struct_mutex); User facing? If so, this is the wrong type of mutex_lock. -Chris
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c index b0e82a1..2d8e22a 100644 --- a/drivers/gpu/drm/i915/i915_gem_context.c +++ b/drivers/gpu/drm/i915/i915_gem_context.c @@ -343,6 +343,71 @@ i915_gem_create_context(struct drm_device *dev, return ctx; } +/** + * i915_gem_create_gvt_context - create a GVT GEM context + * @dev: drm device * + * + * This function is used to create a GVT specific GEM context. + * + * Returns: + * pointer to i915_gem_context on success, error pointer if failed + * + */ +/* GVT context usage flow: + * + * +-----------+ +-----------+ + * | vGPU | | vGPU | + * +-+-----^---+ +-+-----^---+ + * | | | | + * | | GVT-g | | GVT-g + * vELSP write| | emulates vELSP write| | emulates + * | | Execlist/CSB | | Execlist/CSB + * | | Status | | Status + * | | | | + * +------v-----+-------------------------v-----+---------+ + * | GVT Virtual Execlist Submission | + * +------+-------------------------------+---------------+ + * | | + * | Per-VM/Ring Workoad Q | Per-VM/Ring Workload Q + * +---------------------+--+ +------------------------+ + * +---v--------+ ^ +---v--------+ + * |GVT Workload|... | |GVT Workload|... + * +------------+ | +------------+ + * | + * | Pick Workload from Q + * +--------------------+---------------------------------+ + * | GVT Workload Scheduler | + * +--------------------+---------------------------------+ + * | * Shadow guest LRC context + * +------v------+ * Shadow guest ring buffer + * | GVT Context | * Scan/Patch guest RB instructions + * +------+------+ + * | + * v + * Host i915 GEM Submission + */ +struct i915_gem_context * +i915_gem_create_gvt_context(struct drm_device *dev) +{ + struct i915_gem_context *ctx; + + if (!IS_ENABLED(CONFIG_DRM_I915_GVT)) + return ERR_PTR(-ENODEV); + + mutex_lock(&dev->struct_mutex); + + ctx = i915_gem_create_context(dev, NULL); + if (IS_ERR(ctx)) + goto out; + + ctx->enable_lrc_status_change_notification = true; + ctx->enable_lrc_single_submission = true; + ctx->lrc_ring_buffer_size = 512 * PAGE_SIZE; /* Max ring buffer size */ +out: + mutex_unlock(&dev->struct_mutex); + return ctx; +} + static void i915_gem_context_unpin(struct i915_gem_context *ctx, struct intel_engine_cs *engine) {
GVT workload scheduler needs special host LRC contexts, the so called "shadow LRC context" to submit guest workload to host i915. During the guest workload submission, workload scheduler fills the shadow LRC context with the content of guest LRC context: engine context is copied without changes, ring context is mostly owned by host i915. v7: - Move chart to a better place. (Joonas) v6: - Make GVT code as dead code when !CONFIG_DRM_I915_GVT. (Chris) v5: - Only compile this feature when CONFIG_DRM_I915_GVT is enabled. (Tvrtko) - Rebase the code into new repo. - Add a comment about the ring buffer size. (Joonas) v2: Mostly based on Daniel's idea. Call the refactored core logic of GEM context creation service and LRC context creation service to create the GVT context. Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> --- drivers/gpu/drm/i915/i915_gem_context.c | 65 +++++++++++++++++++++++++++++++++ 1 file changed, 65 insertions(+)