[Intel-gfx] [RFC-v2 11/26] drm/i915/pxp: Func to send hardware session termination
Huang, Sean Z
sean.z.huang at intel.com
Sat Nov 21 00:35:25 UTC 2020
Implement the functions to allow PXP to send a GPU command, in
order to terminate the hardware session, so hardware can recycle
this session slot for the next usage.
Signed-off-by: Huang, Sean Z <sean.z.huang at intel.com>
---
drivers/gpu/drm/i915/pxp/intel_pxp_sm.c | 288 ++++++++++++++++++++++++
drivers/gpu/drm/i915/pxp/intel_pxp_sm.h | 16 ++
2 files changed, 304 insertions(+)
diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_sm.c b/drivers/gpu/drm/i915/pxp/intel_pxp_sm.c
index b69f5f8bf238..f41e50911688 100644
--- a/drivers/gpu/drm/i915/pxp/intel_pxp_sm.c
+++ b/drivers/gpu/drm/i915/pxp/intel_pxp_sm.c
@@ -3,13 +3,163 @@
* Copyright(c) 2020, Intel Corporation. All rights reserved.
*/
+#include "gt/intel_gpu_commands.h"
+#include "gt/intel_gt.h"
#include "gt/intel_context.h"
+#include "gt/intel_gt_buffer_pool.h"
#include "gt/intel_engine_pm.h"
#include "intel_pxp.h"
#include "intel_pxp_sm.h"
#include "intel_pxp_context.h"
+static struct i915_vma *pxp_get_batch(struct drm_i915_private *i915,
+ struct intel_context *ce,
+ struct intel_gt_buffer_pool_node *pool,
+ u32 *cmd_buf, int cmd_size_in_dw)
+{
+ struct i915_vma *batch = ERR_PTR(-EINVAL);
+ u32 *cmd;
+
+ if (!ce || !ce->engine || !cmd_buf)
+ return ERR_PTR(-EINVAL);
+
+ if (cmd_size_in_dw * 4 > PAGE_SIZE) {
+ drm_err(&i915->drm, "Failed to %s, invalid cmd_size_id_dw=[%d]\n",
+ __func__, cmd_size_in_dw);
+ return ERR_PTR(-EINVAL);
+ }
+
+ cmd = i915_gem_object_pin_map(pool->obj, I915_MAP_FORCE_WC);
+ if (IS_ERR(cmd)) {
+ drm_err(&i915->drm, "Failed to i915_gem_object_pin_map()\n");
+ return ERR_PTR(-EINVAL);
+ }
+
+ memcpy(cmd, cmd_buf, cmd_size_in_dw * 4);
+
+ if (drm_debug_enabled(DRM_UT_DRIVER)) {
+ print_hex_dump(KERN_DEBUG, "cmd binaries:",
+ DUMP_PREFIX_OFFSET, 4, 4, cmd, cmd_size_in_dw * 4, true);
+ }
+
+ i915_gem_object_unpin_map(pool->obj);
+
+ batch = i915_vma_instance(pool->obj, ce->vm, NULL);
+ if (IS_ERR(batch)) {
+ drm_err(&i915->drm, "Failed to i915_vma_instance()\n");
+ return batch;
+ }
+
+ return batch;
+}
+
+static int pxp_submit_cmd(struct drm_i915_private *i915, u32 *cmd, int cmd_size_in_dw)
+{
+ int err = -EINVAL;
+ struct i915_vma *batch;
+ struct i915_request *rq;
+ struct intel_context *ce = NULL;
+ bool is_engine_pm_get = false;
+ bool is_batch_vma_pin = false;
+ bool is_skip_req_on_err = false;
+ bool is_engine_get_pool = false;
+ struct intel_gt_buffer_pool_node *pool = NULL;
+ struct intel_gt *gt = NULL;
+
+ if (!i915 || !HAS_ENGINE(&i915->gt, VCS0) ||
+ !i915->gt.engine[VCS0]->kernel_context) {
+ err = -EINVAL;
+ goto end;
+ }
+
+ if (!cmd || (cmd_size_in_dw * 4) > PAGE_SIZE) {
+ drm_err(&i915->drm, "Failed to %s bad params\n", __func__);
+ return -EINVAL;
+ }
+
+ gt = &i915->gt;
+ ce = i915->gt.engine[VCS0]->kernel_context;
+
+ intel_engine_pm_get(ce->engine);
+ is_engine_pm_get = true;
+
+ pool = intel_gt_get_buffer_pool(gt, PAGE_SIZE);
+ if (IS_ERR(pool)) {
+ drm_err(&i915->drm, "Failed to intel_engine_get_pool()\n");
+ goto end;
+ }
+ is_engine_get_pool = true;
+
+ batch = pxp_get_batch(i915, ce, pool, cmd, cmd_size_in_dw);
+ if (IS_ERR(batch)) {
+ drm_err(&i915->drm, "Failed to pxp_get_batch()\n");
+ goto end;
+ }
+
+ err = i915_vma_pin(batch, 0, 0, PIN_USER);
+ if (err) {
+ drm_err(&i915->drm, "Failed to i915_vma_pin()\n");
+ goto end;
+ }
+ is_batch_vma_pin = true;
+
+ rq = intel_context_create_request(ce);
+ if (IS_ERR(rq)) {
+ drm_err(&i915->drm, "Failed to intel_context_create_request()\n");
+ goto end;
+ }
+ is_skip_req_on_err = true;
+
+ err = intel_gt_buffer_pool_mark_active(pool, rq);
+ if (err) {
+ drm_err(&i915->drm, "Failed to intel_engine_pool_mark_active()\n");
+ goto end;
+ }
+
+ i915_vma_lock(batch);
+ err = i915_request_await_object(rq, batch->obj, false);
+ if (!err)
+ err = i915_vma_move_to_active(batch, rq, 0);
+ i915_vma_unlock(batch);
+ if (err) {
+ drm_err(&i915->drm, "Failed to i915_request_await_object()\n");
+ goto end;
+ }
+
+ if (ce->engine->emit_init_breadcrumb) {
+ err = ce->engine->emit_init_breadcrumb(rq);
+ if (err) {
+ drm_err(&i915->drm, "Failed to emit_init_breadcrumb()\n");
+ goto end;
+ }
+ }
+
+ err = ce->engine->emit_bb_start(rq, batch->node.start,
+ batch->node.size, 0);
+ if (err) {
+ drm_err(&i915->drm, "Failed to emit_bb_start()\n");
+ goto end;
+ }
+
+ i915_request_add(rq);
+
+end:
+ if (unlikely(err) && is_skip_req_on_err)
+ i915_request_set_error_once(rq, err);
+
+ if (is_batch_vma_pin)
+ i915_vma_unpin(batch);
+
+ if (is_engine_get_pool)
+ intel_gt_buffer_pool_put(pool);
+
+ if (is_engine_pm_get)
+ intel_engine_pm_put(ce->engine);
+
+ return err;
+}
+
static int pxp_sm_reg_read(struct drm_i915_private *i915, u32 offset, u32 *regval)
{
intel_wakeref_t wakeref;
@@ -514,6 +664,144 @@ int pxp_sm_mark_protected_session_in_play(struct drm_i915_private *i915, int ses
return -EINVAL;
}
+static int add_pxp_prolog(struct drm_i915_private *i915, u32 *cmd, int session_type,
+ int session_index)
+{
+ u32 increased_size_in_dw = 0;
+ u32 *cmd_prolog = cmd;
+ const int cmd_prolog_size_in_dw = 10;
+
+ if (!cmd)
+ return cmd_prolog_size_in_dw;
+
+ /* MFX_WAIT - stall until prior PXP and MFX/HCP/HUC objects are cmopleted */
+ *cmd_prolog++ = (MFX_WAIT | MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG |
+ MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG);
+
+ /* MI_FLUSH_DW - pxp off */
+ *cmd_prolog++ = MI_FLUSH_DW; /* DW0 */
+ *cmd_prolog++ = 0; /* DW1 */
+ *cmd_prolog++ = 0; /* DW2 */
+
+ /* MI_SET_APPID */
+ if (session_type == SESSION_TYPE_TYPE1) {
+ if (session_index >= MAX_TYPE1_SESSIONS) {
+ drm_err(&i915->drm, "Failed to %s invalid session_index\n", __func__);
+ goto end;
+ }
+
+ *cmd_prolog++ = (MI_SET_APPID | MI_SET_APPID_TYPE1_APP |
+ MI_SET_APPID_SESSION_ID(session_index));
+ } else {
+ if (session_index >= MAX_TYPE0_SESSIONS) {
+ drm_err(&i915->drm, "Failed to %s invalid session_index\n", __func__);
+ goto end;
+ }
+
+ *cmd_prolog++ = (MI_SET_APPID | MI_SET_APPID_SESSION_ID(session_index));
+ }
+
+ /* MFX_WAIT */
+ *cmd_prolog++ = (MFX_WAIT | MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG |
+ MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG);
+
+ /* MI_FLUSH_DW - pxp on */
+ *cmd_prolog++ = (MI_FLUSH_DW | MI_FLUSH_DW_DW0_PROTECTED_MEMORY_ENABLE); /* DW0 */
+ *cmd_prolog++ = 0; /* DW1 */
+ *cmd_prolog++ = 0; /* DW2 */
+
+ /* MFX_WAIT */
+ *cmd_prolog++ = (MFX_WAIT | MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG |
+ MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG);
+
+ increased_size_in_dw = (cmd_prolog - cmd);
+end:
+ return increased_size_in_dw;
+}
+
+static int add_pxp_epilog(u32 *cmd)
+{
+ u32 increased_size_in_dw = 0;
+ u32 *cmd_epilog = cmd;
+ const int cmd_epilog_size_in_dw = 5;
+
+ if (!cmd)
+ return cmd_epilog_size_in_dw;
+
+ /* MI_FLUSH_DW - pxp off */
+ *cmd_epilog++ = MI_FLUSH_DW; /* DW0 */
+ *cmd_epilog++ = 0; /* DW1 */
+ *cmd_epilog++ = 0; /* DW2 */
+
+ /* MFX_WAIT - stall until prior PXP and MFX/HCP/HUC objects are cmopleted */
+ *cmd_epilog++ = (MFX_WAIT | MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG |
+ MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG);
+
+ /* MI_BATCH_BUFFER_END */
+ *cmd_epilog++ = MI_BATCH_BUFFER_END;
+
+ increased_size_in_dw = (cmd_epilog - cmd);
+ return increased_size_in_dw;
+}
+
+static int add_pxp_inline_termination(u32 *cmd)
+{
+ u32 increased_size_in_dw = 0;
+ u32 *cmd_termin = cmd;
+ const int cmd_termin_size_in_dw = 2;
+
+ if (!cmd)
+ return cmd_termin_size_in_dw;
+
+ /* CRYPTO_KEY_EXCHANGE - session inline termination */
+ *cmd_termin++ = CRYPTO_KEY_EXCHANGE; /* DW0 */
+ *cmd_termin++ = 0; /* DW1 */
+
+ increased_size_in_dw = (cmd_termin - cmd);
+ return increased_size_in_dw;
+}
+
+static int issue_hw_terminate_for_session(struct drm_i915_private *i915, int session_type,
+ int session_index)
+{
+ u32 *cmd = NULL;
+ u32 *cmd_ptr = NULL;
+ int cmd_size_in_dw = 0;
+ int ret;
+
+ if (!i915)
+ return -EINVAL;
+
+ cmd_size_in_dw += add_pxp_prolog(i915, NULL, session_type, session_index);
+ cmd_size_in_dw += add_pxp_inline_termination(NULL);
+ cmd_size_in_dw += add_pxp_epilog(NULL);
+
+ cmd = kzalloc(cmd_size_in_dw * 4, GFP_KERNEL);
+ if (!cmd)
+ return -ENOMEM;
+
+ cmd_ptr = cmd;
+ cmd_ptr += add_pxp_prolog(i915, cmd_ptr, session_type, session_index);
+ cmd_ptr += add_pxp_inline_termination(cmd_ptr);
+ cmd_ptr += add_pxp_epilog(cmd_ptr);
+
+ if (cmd_size_in_dw != (cmd_ptr - cmd)) {
+ ret = -EINVAL;
+ drm_err(&i915->drm, "Failed to %s\n", __func__);
+ goto end;
+ }
+
+ ret = pxp_submit_cmd(i915, cmd, cmd_size_in_dw);
+ if (ret) {
+ drm_err(&i915->drm, "Failed to pxp_submit_cmd()\n");
+ goto end;
+ }
+
+end:
+ kfree(cmd);
+ return ret;
+}
+
int pxp_sm_set_kcr_init_reg(struct drm_i915_private *i915)
{
int ret;
diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_sm.h b/drivers/gpu/drm/i915/pxp/intel_pxp_sm.h
index 90f7d74cacdf..1b3173bca281 100644
--- a/drivers/gpu/drm/i915/pxp/intel_pxp_sm.h
+++ b/drivers/gpu/drm/i915/pxp/intel_pxp_sm.h
@@ -23,6 +23,22 @@
#define SESSION_TYPE_MASK BIT(7)
#define SESSION_ID_MASK (BIT(7) - 1)
+/* PXP GPU command definitions */
+
+/* MI_SET_APPID */
+#define MI_SET_APPID_TYPE1_APP BIT(7)
+#define MI_SET_APPID_SESSION_ID(x) ((x) << 0)
+
+/* MI_FLUSH_DW */
+#define MI_FLUSH_DW_DW0_PROTECTED_MEMORY_ENABLE BIT(22)
+
+/* MI_WAIT */
+#define MFX_WAIT_DW0_PXP_SYNC_CONTROL_FLAG BIT(9)
+#define MFX_WAIT_DW0_MFX_SYNC_CONTROL_FLAG BIT(8)
+
+/* CRYPTO_KEY_EXCHANGE */
+#define CRYPTO_KEY_EXCHANGE ((0x3 << 29) | (0x01609 << 16))
+
enum pxp_session_types {
SESSION_TYPE_TYPE0 = 0,
SESSION_TYPE_TYPE1 = 1,
--
2.17.1
More information about the Intel-gfx
mailing list