[Intel-gfx] [PATCH 0/7] CT changes required for GuC submission

[Intel-gfx] [PATCH 0/7] CT changes required for GuC submission

[Matthew Brost]

As part of enabling GuC submission discussed in [1], [2], and [3] we
need optimize and update the CT code as this is now in the critical
path of submission. This series includes the patches to do that which is
the first 7 patches from [3]. The patches should have addressed all the
feedback in [3] and should be ready to merge once CI returns a we get a
few more RBs.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

[1] https://patchwork.freedesktop.org/series/89844/
[2] https://patchwork.freedesktop.org/series/91417/
[3] https://patchwork.freedesktop.org/series/91840/

John Harrison (1):
  drm/i915/guc: Module load failure test for CT buffer creation

Matthew Brost (6):
  drm/i915/guc: Relax CTB response timeout
  drm/i915/guc: Improve error message for unsolicited CT response
  drm/i915/guc: Increase size of CTB buffers
  drm/i915/guc: Add non blocking CTB send function
  drm/i915/guc: Add stall timer to non blocking CTB send function
  drm/i915/guc: Optimize CTB writes and reads

 .../gt/uc/abi/guc_communication_ctb_abi.h     |   3 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.h        |  11 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 253 +++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  14 +-
 4 files changed, 235 insertions(+), 46 deletions(-)

-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for CT changes required for GuC submission

[Patchwork]

== Series Details ==

Series: CT changes required for GuC submission
URL   : https://patchwork.freedesktop.org/series/91943/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
4de6bd659c01 drm/i915/guc: Relax CTB response timeout
9074da160af0 drm/i915/guc: Improve error message for unsolicited CT response
7ed4a3e0276a drm/i915/guc: Increase size of CTB buffers
3a1d54d48cfe drm/i915/guc: Add non blocking CTB send function
af75da893bfe drm/i915/guc: Add stall timer to non blocking CTB send function
609b9d9541a4 drm/i915/guc: Optimize CTB writes and reads
-:125: CHECK:PARENTHESIS_ALIGNMENT: Alignment should match open parenthesis
#125: FILE: drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c:542:
+		CT_ERROR(ct, "Corrupted descriptor head=%u tail=%u size=%u\n",
+			  ctb->desc->head, ctb->desc->tail, ctb->size);

total: 0 errors, 0 warnings, 1 checks, 188 lines checked
8dbeeecc18bf drm/i915/guc: Module load failure test for CT buffer creation
-:38: WARNING:FROM_SIGN_OFF_MISMATCH: From:/Signed-off-by: email address mismatch: 'From: John Harrison <John.C.Harrison@Intel.com>' != 'Signed-off-by: John Harrison <john.c.harrison@intel.com>'

total: 0 errors, 1 warnings, 0 checks, 20 lines checked


_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✗ Fi.CI.SPARSE: warning for CT changes required for GuC submission

[Patchwork]

== Series Details ==

Series: CT changes required for GuC submission
URL   : https://patchwork.freedesktop.org/series/91943/
State : warning

== Summary ==

$ dim sparse --fast origin/drm-tip
Sparse version: v0.6.2
Fast mode used, each commit won't be checked separately.
-
+drivers/gpu/drm/i915/display/intel_display.c:1893:21:    expected struct i915_vma *[assigned] vma
+drivers/gpu/drm/i915/display/intel_display.c:1893:21:    got void [noderef] __iomem *[assigned] iomem
+drivers/gpu/drm/i915/display/intel_display.c:1893:21: warning: incorrect type in assignment (different address spaces)
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:27:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:32:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:32:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:49:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:56:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_engine_stats.h:56:9: warning: trying to copy expression type 31
+drivers/gpu/drm/i915/gt/intel_reset.c:1396:5: warning: context imbalance in 'intel_gt_reset_trylock' - different lock contexts for basic block
+drivers/gpu/drm/i915/gt/intel_ring_submission.c:1207:24: warning: Using plain integer as NULL pointer
+drivers/gpu/drm/i915/i915_perf.c:1434:15: warning: memset with byte count of 16777216
+drivers/gpu/drm/i915/i915_perf.c:1488:15: warning: memset with byte count of 16777216
+./include/asm-generic/bitops/find.h:112:45: warning: shift count is negative (-262080)
+./include/asm-generic/bitops/find.h:32:31: warning: shift count is negative (-262080)
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read64' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_read8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'fwtable_write8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read64' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_read8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen11_fwtable_write8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read64' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_read8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen12_fwtable_write8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read64' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_read8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen6_write8' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write16' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write32' - different lock contexts for basic block
+./include/linux/spinlock.h:409:9: warning: context imbalance in 'gen8_write8' - different lock contexts for basic block


_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 1/7] drm/i915/guc: Relax CTB response timeout

[Matthew Brost]

In upcoming patch we will allow more CTB requests to be sent in
parallel to the GuC for processing, so we shouldn't assume any more
that GuC will always reply without 10ms.

Use bigger value hardcoded value of 1s instead.

v2: Add CONFIG_DRM_I915_GUC_CTB_TIMEOUT config option
v3:
 (Daniel Vetter)
  - Use hardcoded value of 1s rather than config option
v4:
 (Michal)
  - Use defines for timeout values

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 43409044528e..b86575b99537 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -474,14 +474,18 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	/*
 	 * Fast commands should complete in less than 10us, so sample quickly
 	 * up to that length of time, then switch to a slower sleep-wait loop.
-	 * No GuC command should ever take longer than 10ms.
+	 * No GuC command should ever take longer than 10ms but many GuC
+	 * commands can be inflight at time, so use a 1s timeout on the slower
+	 * sleep-wait loop.
 	 */
+#define GUC_CTB_RESPONSE_TIMEOUT_SHORT_MS 10
+#define GUC_CTB_RESPONSE_TIMEOUT_LONG_MS 1000
 #define done \
 	(FIELD_GET(GUC_HXG_MSG_0_ORIGIN, READ_ONCE(req->status)) == \
 	 GUC_HXG_ORIGIN_GUC)
-	err = wait_for_us(done, 10);
+	err = wait_for_us(done, GUC_CTB_RESPONSE_TIMEOUT_SHORT_MS);
 	if (err)
-		err = wait_for(done, 10);
+		err = wait_for(done, GUC_CTB_RESPONSE_TIMEOUT_LONG_MS);
 #undef done
 
 	if (unlikely(err))
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 2/7] drm/i915/guc: Improve error message for unsolicited CT response

[Matthew Brost]

Improve the error message when a unsolicited CT response is received by
printing fence that couldn't be found, the last fence, and all requests
with a response outstanding.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index b86575b99537..80db59b45c45 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -732,12 +732,16 @@ static int ct_handle_response(struct intel_guc_ct *ct, struct ct_incoming_msg *r
 		found = true;
 		break;
 	}
-	spin_unlock_irqrestore(&ct->requests.lock, flags);
-
 	if (!found) {
 		CT_ERROR(ct, "Unsolicited response (fence %u)\n", fence);
-		return -ENOKEY;
+		CT_ERROR(ct, "Could not find fence=%u, last_fence=%u\n", fence,
+			 ct->requests.last_fence);
+		list_for_each_entry(req, &ct->requests.pending, link)
+			CT_ERROR(ct, "request %u awaits response\n",
+				 req->fence);
+		err = -ENOKEY;
 	}
+	spin_unlock_irqrestore(&ct->requests.lock, flags);
 
 	if (unlikely(err))
 		return err;
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 3/7] drm/i915/guc: Increase size of CTB buffers

[Matthew Brost]

With the introduction of non-blocking CTBs more than one CTB can be in
flight at a time. Increasing the size of the CTBs should reduce how
often software hits the case where no space is available in the CTB
buffer.

Cc: John Harrison <john.c.harrison@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 80db59b45c45..43e03aa2dde8 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -58,11 +58,16 @@ static inline struct drm_device *ct_to_drm(struct intel_guc_ct *ct)
  *      +--------+-----------------------------------------------+------+
  *
  * Size of each `CT Buffer`_ must be multiple of 4K.
- * As we don't expect too many messages, for now use minimum sizes.
+ * We don't expect too many messages in flight at any time, unless we are
+ * using the GuC submission. In that case each request requires a minimum
+ * 2 dwords which gives us a maximum 256 queue'd requests. Hopefully this
+ * enough space to avoid backpressure on the driver. We increase the size
+ * of the receive buffer (relative to the send) to ensure a G2H response
+ * CTB has a landing spot.
  */
 #define CTB_DESC_SIZE		ALIGN(sizeof(struct guc_ct_buffer_desc), SZ_2K)
 #define CTB_H2G_BUFFER_SIZE	(SZ_4K)
-#define CTB_G2H_BUFFER_SIZE	(SZ_4K)
+#define CTB_G2H_BUFFER_SIZE	(4 * CTB_H2G_BUFFER_SIZE)
 
 struct ct_request {
 	struct list_head link;
@@ -643,7 +648,7 @@ static int ct_read(struct intel_guc_ct *ct, struct ct_incoming_msg **msg)
 	/* beware of buffer wrap case */
 	if (unlikely(available < 0))
 		available += size;
-	CT_DEBUG(ct, "available %d (%u:%u)\n", available, head, tail);
+	CT_DEBUG(ct, "available %d (%u:%u:%u)\n", available, head, tail, size);
 	GEM_BUG_ON(available < 0);
 
 	header = cmds[head];
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 4/7] drm/i915/guc: Add non blocking CTB send function

[Matthew Brost]

Add non blocking CTB send function, intel_guc_send_nb. GuC submission
will send CTBs in the critical path and does not need to wait for these
CTBs to complete before moving on, hence the need for this new function.

The non-blocking CTB now must have a flow control mechanism to ensure
the buffer isn't overrun. A lazy spin wait is used as we believe the
flow control condition should be rare with a properly sized buffer.

The function, intel_guc_send_nb, is exported in this patch but unused.
Several patches later in the series make use of this function.

v2:
 (Michal)
  - Use define for H2G room calculations
  - Move INTEL_GUC_SEND_NB define
 (Daniel Vetter)
  - Use msleep_interruptible rather than cond_resched

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 .../gt/uc/abi/guc_communication_ctb_abi.h     |  3 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.h        | 11 ++-
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 90 +++++++++++++++++--
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  4 +-
 4 files changed, 94 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
index e933ca02d0eb..99e1fad5ca20 100644
--- a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
+++ b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
@@ -79,7 +79,8 @@ static_assert(sizeof(struct guc_ct_buffer_desc) == 64);
  *  +---+-------+--------------------------------------------------------------+
  */
 
-#define GUC_CTB_MSG_MIN_LEN			1u
+#define GUC_CTB_HDR_LEN				1u
+#define GUC_CTB_MSG_MIN_LEN			GUC_CTB_HDR_LEN
 #define GUC_CTB_MSG_MAX_LEN			256u
 #define GUC_CTB_MSG_0_FENCE			(0xffff << 16)
 #define GUC_CTB_MSG_0_FORMAT			(0xf << 12)
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
index 4abc59f6f3cd..efc690fc8fb1 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
@@ -74,7 +74,14 @@ static inline struct intel_guc *log_to_guc(struct intel_guc_log *log)
 static
 inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
 {
-	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0);
+	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0, 0);
+}
+
+static
+inline int intel_guc_send_nb(struct intel_guc *guc, const u32 *action, u32 len)
+{
+	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0,
+				 INTEL_GUC_SEND_NB);
 }
 
 static inline int
@@ -82,7 +89,7 @@ intel_guc_send_and_receive(struct intel_guc *guc, const u32 *action, u32 len,
 			   u32 *response_buf, u32 response_buf_size)
 {
 	return intel_guc_ct_send(&guc->ct, action, len,
-				 response_buf, response_buf_size);
+				 response_buf, response_buf_size, 0);
 }
 
 static inline void intel_guc_to_host_event_handler(struct intel_guc *guc)
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 43e03aa2dde8..90ee95a240e8 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -3,6 +3,11 @@
  * Copyright © 2016-2019 Intel Corporation
  */
 
+#include <linux/circ_buf.h>
+#include <linux/ktime.h>
+#include <linux/time64.h>
+#include <linux/timekeeping.h>
+
 #include "i915_drv.h"
 #include "intel_guc_ct.h"
 #include "gt/intel_gt.h"
@@ -373,7 +378,7 @@ static void write_barrier(struct intel_guc_ct *ct)
 static int ct_write(struct intel_guc_ct *ct,
 		    const u32 *action,
 		    u32 len /* in dwords */,
-		    u32 fence)
+		    u32 fence, u32 flags)
 {
 	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
 	struct guc_ct_buffer_desc *desc = ctb->desc;
@@ -409,7 +414,7 @@ static int ct_write(struct intel_guc_ct *ct,
 		used = tail - head;
 
 	/* make sure there is a space including extra dw for the fence */
-	if (unlikely(used + len + 1 >= size))
+	if (unlikely(used + len + GUC_CTB_HDR_LEN >= size))
 		return -ENOSPC;
 
 	/*
@@ -421,9 +426,13 @@ static int ct_write(struct intel_guc_ct *ct,
 		 FIELD_PREP(GUC_CTB_MSG_0_NUM_DWORDS, len) |
 		 FIELD_PREP(GUC_CTB_MSG_0_FENCE, fence);
 
-	hxg = FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
-	      FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
-			 GUC_HXG_REQUEST_MSG_0_DATA0, action[0]);
+	hxg = (flags & INTEL_GUC_SEND_NB) ?
+		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_EVENT) |
+		 FIELD_PREP(GUC_HXG_EVENT_MSG_0_ACTION |
+			    GUC_HXG_EVENT_MSG_0_DATA0, action[0])) :
+		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
+		 FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
+			    GUC_HXG_REQUEST_MSG_0_DATA0, action[0]));
 
 	CT_DEBUG(ct, "writing (tail %u) %*ph %*ph %*ph\n",
 		 tail, 4, &header, 4, &hxg, 4 * (len - 1), &action[1]);
@@ -500,6 +509,48 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	return err;
 }
 
+static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
+{
+	struct guc_ct_buffer_desc *desc = ctb->desc;
+	u32 head = READ_ONCE(desc->head);
+	u32 space;
+
+	space = CIRC_SPACE(desc->tail, head, ctb->size);
+
+	return space >= len_dw;
+}
+
+static int ct_send_nb(struct intel_guc_ct *ct,
+		      const u32 *action,
+		      u32 len,
+		      u32 flags)
+{
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+	unsigned long spin_flags;
+	u32 fence;
+	int ret;
+
+	spin_lock_irqsave(&ctb->lock, spin_flags);
+
+	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
+	if (unlikely(!ret)) {
+		ret = -EBUSY;
+		goto out;
+	}
+
+	fence = ct_get_next_fence(ct);
+	ret = ct_write(ct, action, len, fence, flags);
+	if (unlikely(ret))
+		goto out;
+
+	intel_guc_notify(ct_to_guc(ct));
+
+out:
+	spin_unlock_irqrestore(&ctb->lock, spin_flags);
+
+	return ret;
+}
+
 static int ct_send(struct intel_guc_ct *ct,
 		   const u32 *action,
 		   u32 len,
@@ -507,8 +558,10 @@ static int ct_send(struct intel_guc_ct *ct,
 		   u32 response_buf_size,
 		   u32 *status)
 {
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
 	struct ct_request request;
 	unsigned long flags;
+	unsigned int sleep_period_ms = 1;
 	u32 fence;
 	int err;
 
@@ -516,8 +569,24 @@ static int ct_send(struct intel_guc_ct *ct,
 	GEM_BUG_ON(!len);
 	GEM_BUG_ON(len & ~GUC_CT_MSG_LEN_MASK);
 	GEM_BUG_ON(!response_buf && response_buf_size);
+	might_sleep();
+
+	/*
+	 * We use a lazy spin wait loop here as we believe that if the CT
+	 * buffers are sized correctly the flow control condition should be
+	 * rare.
+	 */
+retry:
+	spin_lock_irqsave(&ctb->lock, flags);
+	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+		spin_unlock_irqrestore(&ctb->lock, flags);
 
-	spin_lock_irqsave(&ct->ctbs.send.lock, flags);
+		if (msleep_interruptible(sleep_period_ms))
+			return -EINTR;
+		sleep_period_ms = sleep_period_ms << 1;
+
+		goto retry;
+	}
 
 	fence = ct_get_next_fence(ct);
 	request.fence = fence;
@@ -529,9 +598,9 @@ static int ct_send(struct intel_guc_ct *ct,
 	list_add_tail(&request.link, &ct->requests.pending);
 	spin_unlock(&ct->requests.lock);
 
-	err = ct_write(ct, action, len, fence);
+	err = ct_write(ct, action, len, fence, 0);
 
-	spin_unlock_irqrestore(&ct->ctbs.send.lock, flags);
+	spin_unlock_irqrestore(&ctb->lock, flags);
 
 	if (unlikely(err))
 		goto unlink;
@@ -571,7 +640,7 @@ static int ct_send(struct intel_guc_ct *ct,
  * Command Transport (CT) buffer based GuC send function.
  */
 int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
-		      u32 *response_buf, u32 response_buf_size)
+		      u32 *response_buf, u32 response_buf_size, u32 flags)
 {
 	u32 status = ~0; /* undefined */
 	int ret;
@@ -581,6 +650,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
 		return -ENODEV;
 	}
 
+	if (flags & INTEL_GUC_SEND_NB)
+		return ct_send_nb(ct, action, len, flags);
+
 	ret = ct_send(ct, action, len, response_buf, response_buf_size, &status);
 	if (unlikely(ret < 0)) {
 		CT_ERROR(ct, "Sending action %#x failed (err=%d status=%#X)\n",
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index 1ae2dde6db93..f6a4d5b33467 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -42,7 +42,6 @@ struct intel_guc_ct_buffer {
 	bool broken;
 };
 
-
 /** Top-level structure for Command Transport related data
  *
  * Includes a pair of CT buffers for bi-directional communication and tracking
@@ -87,8 +86,9 @@ static inline bool intel_guc_ct_enabled(struct intel_guc_ct *ct)
 	return ct->enabled;
 }
 
+#define INTEL_GUC_SEND_NB		BIT(31)
 int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
-		      u32 *response_buf, u32 response_buf_size);
+		      u32 *response_buf, u32 response_buf_size, u32 flags);
 void intel_guc_ct_event_handler(struct intel_guc_ct *ct);
 
 #endif /* _INTEL_GUC_CT_H_ */
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH 4/7] drm/i915/guc: Add non blocking CTB send function

[Michal Wajdeczko]

On 28.06.2021 01:14, Matthew Brost wrote:
> Add non blocking CTB send function, intel_guc_send_nb. GuC submission
> will send CTBs in the critical path and does not need to wait for these
> CTBs to complete before moving on, hence the need for this new function.
> 
> The non-blocking CTB now must have a flow control mechanism to ensure
> the buffer isn't overrun. A lazy spin wait is used as we believe the
> flow control condition should be rare with a properly sized buffer.
> 
> The function, intel_guc_send_nb, is exported in this patch but unused.
> Several patches later in the series make use of this function.
> 
> v2:
>  (Michal)
>   - Use define for H2G room calculations
>   - Move INTEL_GUC_SEND_NB define
>  (Daniel Vetter)
>   - Use msleep_interruptible rather than cond_resched
> 
> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> ---
>  .../gt/uc/abi/guc_communication_ctb_abi.h     |  3 +-
>  drivers/gpu/drm/i915/gt/uc/intel_guc.h        | 11 ++-
>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 90 +++++++++++++++++--
>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  4 +-
>  4 files changed, 94 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> index e933ca02d0eb..99e1fad5ca20 100644
> --- a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> +++ b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> @@ -79,7 +79,8 @@ static_assert(sizeof(struct guc_ct_buffer_desc) == 64);
>   *  +---+-------+--------------------------------------------------------------+
>   */
>  
> -#define GUC_CTB_MSG_MIN_LEN			1u
> +#define GUC_CTB_HDR_LEN				1u
> +#define GUC_CTB_MSG_MIN_LEN			GUC_CTB_HDR_LEN

if you insist to use dedicated macro for the CTB header len then to be
consistent you need rename header bitfield macros as well, thus
sections/tables will look like:

/**
 * DOC: CTB Message
 *
 *  +---+-------+-------------------------------------+
 *  |   | Bits  | Description                         |
 *  +===+=======+=====================================+
 *  | 0 |  31:0 | `CTB Header`_                       |
 *  +---+-------+-------------------------------------+
 *  | 1 |  31:0 |  +-------------------------------+  |
 *  +---+-------+  |                               |  |
 *  |...|       |  |  CTB Payload                  |  |
 *  +---+-------+  |                               |  |
 *  | n |  31:0 |  +-------------------------------+  |
 *  +---+-------+-------------------------------------+
 */

/**
 * DOC: CTB Header
 *
 *  +---+-------+-------------------------------------+
 *  |   | Bits  | Description                         |
 *  +===+=======+=====================================+
 *  | 0 | 31:16 | **FENCE** - ...                     |
 *  |   +-------+-------------------------------------+
 *  |   | 15:12 | **FORMAT** - ...                    |
 *  |   +-------+-------------------------------------+
 *  |   |  11:8 | **RESERVED**                        |
 *  |   +-------+-------------------------------------+
 *  |   |   7:0 | **NUM_DWORDS** - ...                |
 *  +---+-------+-------------------------------------+
 */

#define GUC_CTB_HDR_0_FENCE			(0xffff << 16)
#define GUC_CTB_HDR_0_FORMAT			(0xf << 12)
#define   GUC_CTB_FORMAT_HXG			0u
#define GUC_CTB_HDR_0_RESERVED			(0xf << 8)
#define GUC_CTB_HDR_0_NUM_DWORDS		(0xff << 0)
#define   GUC_CTB_MAX_DWORDS			255u

#define GUC_CTB_MSG_MIN_LEN	GUC_CTB_HDR_LEN
#define GUC_CTB_MSG_MAX_LEN	(GUC_CTB_HDR_LEN + GUC_CTB_MAX_DWORDS)


alternatively leave ABI defs as-as and just move your HDR definition out
of ABI headers to inteL_guc_fwif.h as:

#define GUC_CTB_HDR_LEN GUC_CTB_MSG_MIN_LEN


>  #define GUC_CTB_MSG_MAX_LEN			256u
>  #define GUC_CTB_MSG_0_FENCE			(0xffff << 16)
>  #define GUC_CTB_MSG_0_FORMAT			(0xf << 12)
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> index 4abc59f6f3cd..efc690fc8fb1 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> @@ -74,7 +74,14 @@ static inline struct intel_guc *log_to_guc(struct intel_guc_log *log)
>  static
>  inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
>  {
> -	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0);
> +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0, 0);
> +}
> +
> +static
> +inline int intel_guc_send_nb(struct intel_guc *guc, const u32 *action, u32 len)
> +{
> +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0,
> +				 INTEL_GUC_SEND_NB);
>  }
>  
>  static inline int
> @@ -82,7 +89,7 @@ intel_guc_send_and_receive(struct intel_guc *guc, const u32 *action, u32 len,
>  			   u32 *response_buf, u32 response_buf_size)
>  {
>  	return intel_guc_ct_send(&guc->ct, action, len,
> -				 response_buf, response_buf_size);
> +				 response_buf, response_buf_size, 0);
>  }
>  
>  static inline void intel_guc_to_host_event_handler(struct intel_guc *guc)
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> index 43e03aa2dde8..90ee95a240e8 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> @@ -3,6 +3,11 @@
>   * Copyright © 2016-2019 Intel Corporation
>   */
>  
> +#include <linux/circ_buf.h>

> +#include <linux/ktime.h>
> +#include <linux/time64.h>
> +#include <linux/timekeeping.h>

these headers should likely be part of patch 5/7

> +
>  #include "i915_drv.h"
>  #include "intel_guc_ct.h"
>  #include "gt/intel_gt.h"
> @@ -373,7 +378,7 @@ static void write_barrier(struct intel_guc_ct *ct)
>  static int ct_write(struct intel_guc_ct *ct,
>  		    const u32 *action,
>  		    u32 len /* in dwords */,
> -		    u32 fence)
> +		    u32 fence, u32 flags)
>  {
>  	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
>  	struct guc_ct_buffer_desc *desc = ctb->desc;
> @@ -409,7 +414,7 @@ static int ct_write(struct intel_guc_ct *ct,
>  		used = tail - head;
                ^^^^
code above for blocking path is likely calculating free space in old
fashion, while below you are using CIRC_BUF ...

so maybe switch to use circ_buf could be done as separate earlier step
(not part of intro of non-blocking send) so then below use of CIRC_BUF
in non-blocking path will look natural

>  
>  	/* make sure there is a space including extra dw for the fence */

this extra DW is now "header" not just fence

> -	if (unlikely(used + len + 1 >= size))
> +	if (unlikely(used + len + GUC_CTB_HDR_LEN >= size))
>  		return -ENOSPC;
>  
>  	/*
> @@ -421,9 +426,13 @@ static int ct_write(struct intel_guc_ct *ct,
>  		 FIELD_PREP(GUC_CTB_MSG_0_NUM_DWORDS, len) |
>  		 FIELD_PREP(GUC_CTB_MSG_0_FENCE, fence);
>  
> -	hxg = FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> -	      FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> -			 GUC_HXG_REQUEST_MSG_0_DATA0, action[0]);
> +	hxg = (flags & INTEL_GUC_SEND_NB) ?
> +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_EVENT) |
> +		 FIELD_PREP(GUC_HXG_EVENT_MSG_0_ACTION |
> +			    GUC_HXG_EVENT_MSG_0_DATA0, action[0])) :
> +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> +		 FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> +			    GUC_HXG_REQUEST_MSG_0_DATA0, action[0]));
>  
>  	CT_DEBUG(ct, "writing (tail %u) %*ph %*ph %*ph\n",
>  		 tail, 4, &header, 4, &hxg, 4 * (len - 1), &action[1]);
> @@ -500,6 +509,48 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
>  	return err;
>  }
>  
> +static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)

why h2g prefix ? function is taking pure *ctb, so maybe:


static bool ctb_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
{ ... }

static bool ct_can_send(struct intel_guc_ct *ct, u32 len_dw)
{
	return ctb_has_room(&ct->ctbs.send, len_dw + CTB_HDR_LEN);
}


as we use send/recv, not h2g/g2h and to avoid mistakes add extra header
len here (not by caller - same as in call to ct_write)

> +{
> +	struct guc_ct_buffer_desc *desc = ctb->desc;
> +	u32 head = READ_ONCE(desc->head);
> +	u32 space;
> +
> +	space = CIRC_SPACE(desc->tail, head, ctb->size);
> +
> +	return space >= len_dw;
> +}
> +
> +static int ct_send_nb(struct intel_guc_ct *ct,
> +		      const u32 *action,
> +		      u32 len,
> +		      u32 flags)
> +{
> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> +	unsigned long spin_flags;
> +	u32 fence;
> +	int ret;
> +
> +	spin_lock_irqsave(&ctb->lock, spin_flags);
> +
> +	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
> +	if (unlikely(!ret)) {
> +		ret = -EBUSY;
> +		goto out;
> +	}
> +
> +	fence = ct_get_next_fence(ct);
> +	ret = ct_write(ct, action, len, fence, flags);
> +	if (unlikely(ret))
> +		goto out;
> +
> +	intel_guc_notify(ct_to_guc(ct));
> +
> +out:
> +	spin_unlock_irqrestore(&ctb->lock, spin_flags);
> +
> +	return ret;
> +}
> +
>  static int ct_send(struct intel_guc_ct *ct,
>  		   const u32 *action,
>  		   u32 len,
> @@ -507,8 +558,10 @@ static int ct_send(struct intel_guc_ct *ct,
>  		   u32 response_buf_size,
>  		   u32 *status)
>  {
> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
>  	struct ct_request request;
>  	unsigned long flags;
> +	unsigned int sleep_period_ms = 1;
>  	u32 fence;
>  	int err;
>  
> @@ -516,8 +569,24 @@ static int ct_send(struct intel_guc_ct *ct,
>  	GEM_BUG_ON(!len);
>  	GEM_BUG_ON(len & ~GUC_CT_MSG_LEN_MASK);
>  	GEM_BUG_ON(!response_buf && response_buf_size);
> +	might_sleep();
> +
> +	/*
> +	 * We use a lazy spin wait loop here as we believe that if the CT
> +	 * buffers are sized correctly the flow control condition should be
> +	 * rare.
> +	 */
> +retry:
> +	spin_lock_irqsave(&ctb->lock, flags);
> +	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
> +		spin_unlock_irqrestore(&ctb->lock, flags);
>  
> -	spin_lock_irqsave(&ct->ctbs.send.lock, flags);
> +		if (msleep_interruptible(sleep_period_ms))
> +			return -EINTR;
> +		sleep_period_ms = sleep_period_ms << 1;
> +
> +		goto retry;
> +	}

I'm still not convinced that this chunk should be exactly in the same
patch that adds non-blocking path, it's not that with very first use of
one NB call we will hit such problem

so IMHO it can be done earlier, in by anticipating such problem, or
later as fix for potential problem due to stress

note that in the commit message you admitted that new NB variant is not
used yet, so we can't be hit

in other words, try to make patches focused on one item at the time.

>  
>  	fence = ct_get_next_fence(ct);
>  	request.fence = fence;
> @@ -529,9 +598,9 @@ static int ct_send(struct intel_guc_ct *ct,
>  	list_add_tail(&request.link, &ct->requests.pending);
>  	spin_unlock(&ct->requests.lock);
>  
> -	err = ct_write(ct, action, len, fence);
> +	err = ct_write(ct, action, len, fence, 0);
>  
> -	spin_unlock_irqrestore(&ct->ctbs.send.lock, flags);
> +	spin_unlock_irqrestore(&ctb->lock, flags);
>  
>  	if (unlikely(err))
>  		goto unlink;
> @@ -571,7 +640,7 @@ static int ct_send(struct intel_guc_ct *ct,
>   * Command Transport (CT) buffer based GuC send function.
>   */
>  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> -		      u32 *response_buf, u32 response_buf_size)
> +		      u32 *response_buf, u32 response_buf_size, u32 flags)
>  {
>  	u32 status = ~0; /* undefined */
>  	int ret;
> @@ -581,6 +650,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
>  		return -ENODEV;
>  	}
>  
> +	if (flags & INTEL_GUC_SEND_NB)
> +		return ct_send_nb(ct, action, len, flags);
> +
>  	ret = ct_send(ct, action, len, response_buf, response_buf_size, &status);
>  	if (unlikely(ret < 0)) {
>  		CT_ERROR(ct, "Sending action %#x failed (err=%d status=%#X)\n",
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> index 1ae2dde6db93..f6a4d5b33467 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> @@ -42,7 +42,6 @@ struct intel_guc_ct_buffer {
>  	bool broken;
>  };
>  
> -
>  /** Top-level structure for Command Transport related data
>   *
>   * Includes a pair of CT buffers for bi-directional communication and tracking
> @@ -87,8 +86,9 @@ static inline bool intel_guc_ct_enabled(struct intel_guc_ct *ct)
>  	return ct->enabled;
>  }
>  
> +#define INTEL_GUC_SEND_NB		BIT(31)

INTEL_GUC_CT_SEND_NB ?
         ^^^^

>  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> -		      u32 *response_buf, u32 response_buf_size);
> +		      u32 *response_buf, u32 response_buf_size, u32 flags);
>  void intel_guc_ct_event_handler(struct intel_guc_ct *ct);
>  
>  #endif /* _INTEL_GUC_CT_H_ */
> 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH 4/7] drm/i915/guc: Add non blocking CTB send function

[Matthew Brost]

On Thu, Jul 01, 2021 at 12:52:58AM +0200, Michal Wajdeczko wrote:
> 
> 
> On 28.06.2021 01:14, Matthew Brost wrote:
> > Add non blocking CTB send function, intel_guc_send_nb. GuC submission
> > will send CTBs in the critical path and does not need to wait for these
> > CTBs to complete before moving on, hence the need for this new function.
> > 
> > The non-blocking CTB now must have a flow control mechanism to ensure
> > the buffer isn't overrun. A lazy spin wait is used as we believe the
> > flow control condition should be rare with a properly sized buffer.
> > 
> > The function, intel_guc_send_nb, is exported in this patch but unused.
> > Several patches later in the series make use of this function.
> > 
> > v2:
> >  (Michal)
> >   - Use define for H2G room calculations
> >   - Move INTEL_GUC_SEND_NB define
> >  (Daniel Vetter)
> >   - Use msleep_interruptible rather than cond_resched
> > 
> > Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> > Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> > ---
> >  .../gt/uc/abi/guc_communication_ctb_abi.h     |  3 +-
> >  drivers/gpu/drm/i915/gt/uc/intel_guc.h        | 11 ++-
> >  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 90 +++++++++++++++++--
> >  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  4 +-
> >  4 files changed, 94 insertions(+), 14 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> > index e933ca02d0eb..99e1fad5ca20 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> > +++ b/drivers/gpu/drm/i915/gt/uc/abi/guc_communication_ctb_abi.h
> > @@ -79,7 +79,8 @@ static_assert(sizeof(struct guc_ct_buffer_desc) == 64);
> >   *  +---+-------+--------------------------------------------------------------+
> >   */
> >  
> > -#define GUC_CTB_MSG_MIN_LEN			1u
> > +#define GUC_CTB_HDR_LEN				1u
> > +#define GUC_CTB_MSG_MIN_LEN			GUC_CTB_HDR_LEN
> 
> if you insist to use dedicated macro for the CTB header len then to be
> consistent you need rename header bitfield macros as well, thus
> sections/tables will look like:
>

Kernel doc can't link to defines, right? So what does it matter? 

This example is also a reason why I think the kernel doc included is too
verbose to hand generate. Either we auto-gen this or just don't include
it.
 
> /**
>  * DOC: CTB Message
>  *
>  *  +---+-------+-------------------------------------+
>  *  |   | Bits  | Description                         |
>  *  +===+=======+=====================================+
>  *  | 0 |  31:0 | `CTB Header`_                       |
>  *  +---+-------+-------------------------------------+
>  *  | 1 |  31:0 |  +-------------------------------+  |
>  *  +---+-------+  |                               |  |
>  *  |...|       |  |  CTB Payload                  |  |
>  *  +---+-------+  |                               |  |
>  *  | n |  31:0 |  +-------------------------------+  |
>  *  +---+-------+-------------------------------------+
>  */
> 
> /**
>  * DOC: CTB Header
>  *
>  *  +---+-------+-------------------------------------+
>  *  |   | Bits  | Description                         |
>  *  +===+=======+=====================================+
>  *  | 0 | 31:16 | **FENCE** - ...                     |
>  *  |   +-------+-------------------------------------+
>  *  |   | 15:12 | **FORMAT** - ...                    |
>  *  |   +-------+-------------------------------------+
>  *  |   |  11:8 | **RESERVED**                        |
>  *  |   +-------+-------------------------------------+
>  *  |   |   7:0 | **NUM_DWORDS** - ...                |
>  *  +---+-------+-------------------------------------+
>  */
> 
> #define GUC_CTB_HDR_0_FENCE			(0xffff << 16)
> #define GUC_CTB_HDR_0_FORMAT			(0xf << 12)
> #define   GUC_CTB_FORMAT_HXG			0u
> #define GUC_CTB_HDR_0_RESERVED			(0xf << 8)
> #define GUC_CTB_HDR_0_NUM_DWORDS		(0xff << 0)
> #define   GUC_CTB_MAX_DWORDS			255u
> 
> #define GUC_CTB_MSG_MIN_LEN	GUC_CTB_HDR_LEN
> #define GUC_CTB_MSG_MAX_LEN	(GUC_CTB_HDR_LEN + GUC_CTB_MAX_DWORDS)
> 
> 
> alternatively leave ABI defs as-as and just move your HDR definition out
> of ABI headers to inteL_guc_fwif.h as:
> 
> #define GUC_CTB_HDR_LEN GUC_CTB_MSG_MIN_LEN

This is backwards. The minimum length of a message is the header length.

> 
> 
> >  #define GUC_CTB_MSG_MAX_LEN			256u
> >  #define GUC_CTB_MSG_0_FENCE			(0xffff << 16)
> >  #define GUC_CTB_MSG_0_FORMAT			(0xf << 12)
> > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > index 4abc59f6f3cd..efc690fc8fb1 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> > @@ -74,7 +74,14 @@ static inline struct intel_guc *log_to_guc(struct intel_guc_log *log)
> >  static
> >  inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
> >  {
> > -	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0);
> > +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0, 0);
> > +}
> > +
> > +static
> > +inline int intel_guc_send_nb(struct intel_guc *guc, const u32 *action, u32 len)
> > +{
> > +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0,
> > +				 INTEL_GUC_SEND_NB);
> >  }
> >  
> >  static inline int
> > @@ -82,7 +89,7 @@ intel_guc_send_and_receive(struct intel_guc *guc, const u32 *action, u32 len,
> >  			   u32 *response_buf, u32 response_buf_size)
> >  {
> >  	return intel_guc_ct_send(&guc->ct, action, len,
> > -				 response_buf, response_buf_size);
> > +				 response_buf, response_buf_size, 0);
> >  }
> >  
> >  static inline void intel_guc_to_host_event_handler(struct intel_guc *guc)
> > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > index 43e03aa2dde8..90ee95a240e8 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > @@ -3,6 +3,11 @@
> >   * Copyright © 2016-2019 Intel Corporation
> >   */
> >  
> > +#include <linux/circ_buf.h>
> 
> > +#include <linux/ktime.h>
> > +#include <linux/time64.h>
> > +#include <linux/timekeeping.h>
> 
> these headers should likely be part of patch 5/7
> 

Yes, will fix.

> > +
> >  #include "i915_drv.h"
> >  #include "intel_guc_ct.h"
> >  #include "gt/intel_gt.h"
> > @@ -373,7 +378,7 @@ static void write_barrier(struct intel_guc_ct *ct)
> >  static int ct_write(struct intel_guc_ct *ct,
> >  		    const u32 *action,
> >  		    u32 len /* in dwords */,
> > -		    u32 fence)
> > +		    u32 fence, u32 flags)
> >  {
> >  	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> >  	struct guc_ct_buffer_desc *desc = ctb->desc;
> > @@ -409,7 +414,7 @@ static int ct_write(struct intel_guc_ct *ct,
> >  		used = tail - head;
>                 ^^^^
> code above for blocking path is likely calculating free space in old
> fashion, while below you are using CIRC_BUF ...
> 
> so maybe switch to use circ_buf could be done as separate earlier step
> (not part of intro of non-blocking send) so then below use of CIRC_BUF
> in non-blocking path will look natural
>

This is existing code that deleted 2 patches from now. I don't think we
need to rework it to just delete it.
 
> >  
> >  	/* make sure there is a space including extra dw for the fence */
> 
> this extra DW is now "header" not just fence
> 

Correct but existing code, right? We should've fixed this with the CTB
interface change boondoggle. I suppose I'll fix this if I respin.

> > -	if (unlikely(used + len + 1 >= size))
> > +	if (unlikely(used + len + GUC_CTB_HDR_LEN >= size))
> >  		return -ENOSPC;
> >  
> >  	/*
> > @@ -421,9 +426,13 @@ static int ct_write(struct intel_guc_ct *ct,
> >  		 FIELD_PREP(GUC_CTB_MSG_0_NUM_DWORDS, len) |
> >  		 FIELD_PREP(GUC_CTB_MSG_0_FENCE, fence);
> >  
> > -	hxg = FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> > -	      FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> > -			 GUC_HXG_REQUEST_MSG_0_DATA0, action[0]);
> > +	hxg = (flags & INTEL_GUC_SEND_NB) ?
> > +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_EVENT) |
> > +		 FIELD_PREP(GUC_HXG_EVENT_MSG_0_ACTION |
> > +			    GUC_HXG_EVENT_MSG_0_DATA0, action[0])) :
> > +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> > +		 FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> > +			    GUC_HXG_REQUEST_MSG_0_DATA0, action[0]));
> >  
> >  	CT_DEBUG(ct, "writing (tail %u) %*ph %*ph %*ph\n",
> >  		 tail, 4, &header, 4, &hxg, 4 * (len - 1), &action[1]);
> > @@ -500,6 +509,48 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
> >  	return err;
> >  }
> >  
> > +static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
> 
> why h2g prefix ? function is taking pure *ctb, so maybe:
>

This was called ctb_has_room in the prior to this rev. You literally
suggested this change in previous revs comments. A follow up patch
basically does what you suggest below once we start checking the g2h
credits too.
 
> 
> static bool ctb_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
> { ... }
> 
> static bool ct_can_send(struct intel_guc_ct *ct, u32 len_dw)
> {
> 	return ctb_has_room(&ct->ctbs.send, len_dw + CTB_HDR_LEN);
> }
> 
> 
> as we use send/recv, not h2g/g2h and to avoid mistakes add extra header
> len here (not by caller - same as in call to ct_write)
>

Don't agree. I prefer h2g / g2h here. Send and recv and both relative to
the observer while h2g / g2h are not.
 
> > +{
> > +	struct guc_ct_buffer_desc *desc = ctb->desc;
> > +	u32 head = READ_ONCE(desc->head);
> > +	u32 space;
> > +
> > +	space = CIRC_SPACE(desc->tail, head, ctb->size);
> > +
> > +	return space >= len_dw;
> > +}
> > +
> > +static int ct_send_nb(struct intel_guc_ct *ct,
> > +		      const u32 *action,
> > +		      u32 len,
> > +		      u32 flags)
> > +{
> > +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> > +	unsigned long spin_flags;
> > +	u32 fence;
> > +	int ret;
> > +
> > +	spin_lock_irqsave(&ctb->lock, spin_flags);
> > +
> > +	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
> > +	if (unlikely(!ret)) {
> > +		ret = -EBUSY;
> > +		goto out;
> > +	}
> > +
> > +	fence = ct_get_next_fence(ct);
> > +	ret = ct_write(ct, action, len, fence, flags);
> > +	if (unlikely(ret))
> > +		goto out;
> > +
> > +	intel_guc_notify(ct_to_guc(ct));
> > +
> > +out:
> > +	spin_unlock_irqrestore(&ctb->lock, spin_flags);
> > +
> > +	return ret;
> > +}
> > +
> >  static int ct_send(struct intel_guc_ct *ct,
> >  		   const u32 *action,
> >  		   u32 len,
> > @@ -507,8 +558,10 @@ static int ct_send(struct intel_guc_ct *ct,
> >  		   u32 response_buf_size,
> >  		   u32 *status)
> >  {
> > +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> >  	struct ct_request request;
> >  	unsigned long flags;
> > +	unsigned int sleep_period_ms = 1;
> >  	u32 fence;
> >  	int err;
> >  
> > @@ -516,8 +569,24 @@ static int ct_send(struct intel_guc_ct *ct,
> >  	GEM_BUG_ON(!len);
> >  	GEM_BUG_ON(len & ~GUC_CT_MSG_LEN_MASK);
> >  	GEM_BUG_ON(!response_buf && response_buf_size);
> > +	might_sleep();
> > +
> > +	/*
> > +	 * We use a lazy spin wait loop here as we believe that if the CT
> > +	 * buffers are sized correctly the flow control condition should be
> > +	 * rare.
> > +	 */
> > +retry:
> > +	spin_lock_irqsave(&ctb->lock, flags);
> > +	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
> > +		spin_unlock_irqrestore(&ctb->lock, flags);
> >  
> > -	spin_lock_irqsave(&ct->ctbs.send.lock, flags);
> > +		if (msleep_interruptible(sleep_period_ms))
> > +			return -EINTR;
> > +		sleep_period_ms = sleep_period_ms << 1;
> > +
> > +		goto retry;
> > +	}
> 
> I'm still not convinced that this chunk should be exactly in the same
> patch that adds non-blocking path, it's not that with very first use of
> one NB call we will hit such problem
>

Once you have non-blocking calls the credits can be exhausted.
 
> so IMHO it can be done earlier, in by anticipating such problem, or
> later as fix for potential problem due to stress
>

What?
 
> note that in the commit message you admitted that new NB variant is not
> used yet, so we can't be hit
>

If we introduce a function that can cause a problem for another function
we likely should make both work in a single patch.
 
> in other words, try to make patches focused on one item at the time.
>

This patch is quite simple and all the changes follow together. For
example if you did a 'git blame' on the above lines and the changes were
only by themselves it wouldn't make any sense at all. With the patch as
is it makes sense.
 
> >  
> >  	fence = ct_get_next_fence(ct);
> >  	request.fence = fence;
> > @@ -529,9 +598,9 @@ static int ct_send(struct intel_guc_ct *ct,
> >  	list_add_tail(&request.link, &ct->requests.pending);
> >  	spin_unlock(&ct->requests.lock);
> >  
> > -	err = ct_write(ct, action, len, fence);
> > +	err = ct_write(ct, action, len, fence, 0);
> >  
> > -	spin_unlock_irqrestore(&ct->ctbs.send.lock, flags);
> > +	spin_unlock_irqrestore(&ctb->lock, flags);
> >  
> >  	if (unlikely(err))
> >  		goto unlink;
> > @@ -571,7 +640,7 @@ static int ct_send(struct intel_guc_ct *ct,
> >   * Command Transport (CT) buffer based GuC send function.
> >   */
> >  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> > -		      u32 *response_buf, u32 response_buf_size)
> > +		      u32 *response_buf, u32 response_buf_size, u32 flags)
> >  {
> >  	u32 status = ~0; /* undefined */
> >  	int ret;
> > @@ -581,6 +650,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> >  		return -ENODEV;
> >  	}
> >  
> > +	if (flags & INTEL_GUC_SEND_NB)
> > +		return ct_send_nb(ct, action, len, flags);
> > +
> >  	ret = ct_send(ct, action, len, response_buf, response_buf_size, &status);
> >  	if (unlikely(ret < 0)) {
> >  		CT_ERROR(ct, "Sending action %#x failed (err=%d status=%#X)\n",
> > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > index 1ae2dde6db93..f6a4d5b33467 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > @@ -42,7 +42,6 @@ struct intel_guc_ct_buffer {
> >  	bool broken;
> >  };
> >  
> > -
> >  /** Top-level structure for Command Transport related data
> >   *
> >   * Includes a pair of CT buffers for bi-directional communication and tracking
> > @@ -87,8 +86,9 @@ static inline bool intel_guc_ct_enabled(struct intel_guc_ct *ct)
> >  	return ct->enabled;
> >  }
> >  
> > +#define INTEL_GUC_SEND_NB		BIT(31)
> 
> INTEL_GUC_CT_SEND_NB ?
>          ^^^^
> 

I guess, but does it really matter? I could see it either way.

Again please sign off with your name when done with your comments so I
know if it is your last comment.

Matt
 

> >  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> > -		      u32 *response_buf, u32 response_buf_size);
> > +		      u32 *response_buf, u32 response_buf_size, u32 flags);
> >  void intel_guc_ct_event_handler(struct intel_guc_ct *ct);
> >  
> >  #endif /* _INTEL_GUC_CT_H_ */
> > 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Matthew Brost]

Implement a stall timer which fails H2G CTBs once a period of time
with no forward progress is reached to prevent deadlock.

v2:
 (Michal)
  - Improve error message in ct_deadlock()
  - Set broken when ct_deadlock() returns true
  - Return -EPIPE on ct_deadlock()

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
 2 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 90ee95a240e8..8f553f7f9619 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -319,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
 		goto err_deregister;
 
 	ct->enabled = true;
+	ct->stall_time = KTIME_MAX;
 
 	return 0;
 
@@ -391,9 +392,6 @@ static int ct_write(struct intel_guc_ct *ct,
 	u32 *cmds = ctb->cmds;
 	unsigned int i;
 
-	if (unlikely(ctb->broken))
-		return -EPIPE;
-
 	if (unlikely(desc->status))
 		goto corrupted;
 
@@ -509,6 +507,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	return err;
 }
 
+#define GUC_CTB_TIMEOUT_MS	1500
+static inline bool ct_deadlocked(struct intel_guc_ct *ct)
+{
+	long timeout = GUC_CTB_TIMEOUT_MS;
+	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
+
+	if (unlikely(ret)) {
+		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
+		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
+
+		CT_ERROR(ct, "Communication stalled for %lld, desc status=%#x,%#x\n",
+			 ktime_ms_delta(ktime_get(), ct->stall_time),
+			 send->status, recv->status);
+		ct->ctbs.send.broken = true;
+	}
+
+	return ret;
+}
+
 static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 {
 	struct guc_ct_buffer_desc *desc = ctb->desc;
@@ -520,6 +537,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 	return space >= len_dw;
 }
 
+static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
+{
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+
+	lockdep_assert_held(&ct->ctbs.send.lock);
+
+	if (unlikely(!h2g_has_room(ctb, len_dw))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
+
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+		else
+			return -EBUSY;
+	}
+
+	ct->stall_time = KTIME_MAX;
+	return 0;
+}
+
 static int ct_send_nb(struct intel_guc_ct *ct,
 		      const u32 *action,
 		      u32 len,
@@ -530,13 +567,14 @@ static int ct_send_nb(struct intel_guc_ct *ct,
 	u32 fence;
 	int ret;
 
+	if (unlikely(ctb->broken))
+		return -EPIPE;
+
 	spin_lock_irqsave(&ctb->lock, spin_flags);
 
-	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
-	if (unlikely(!ret)) {
-		ret = -EBUSY;
+	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
+	if (unlikely(ret))
 		goto out;
-	}
 
 	fence = ct_get_next_fence(ct);
 	ret = ct_write(ct, action, len, fence, flags);
@@ -571,6 +609,9 @@ static int ct_send(struct intel_guc_ct *ct,
 	GEM_BUG_ON(!response_buf && response_buf_size);
 	might_sleep();
 
+	if (unlikely(ctb->broken))
+		return -EPIPE;
+
 	/*
 	 * We use a lazy spin wait loop here as we believe that if the CT
 	 * buffers are sized correctly the flow control condition should be
@@ -579,8 +620,13 @@ static int ct_send(struct intel_guc_ct *ct,
 retry:
 	spin_lock_irqsave(&ctb->lock, flags);
 	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
 		spin_unlock_irqrestore(&ctb->lock, flags);
 
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+
 		if (msleep_interruptible(sleep_period_ms))
 			return -EINTR;
 		sleep_period_ms = sleep_period_ms << 1;
@@ -588,6 +634,8 @@ static int ct_send(struct intel_guc_ct *ct,
 		goto retry;
 	}
 
+	ct->stall_time = KTIME_MAX;
+
 	fence = ct_get_next_fence(ct);
 	request.fence = fence;
 	request.status = 0;
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index f6a4d5b33467..c9d6ae7848a7 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -9,6 +9,7 @@
 #include <linux/interrupt.h>
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
+#include <linux/ktime.h>
 
 #include "intel_guc_fwif.h"
 
@@ -68,6 +69,9 @@ struct intel_guc_ct {
 		struct list_head incoming; /* incoming requests */
 		struct work_struct worker; /* handler for incoming requests */
 	} requests;
+
+	/** @stall_time: time of first time a CTB submission is stalled */
+	ktime_t stall_time;
 };
 
 void intel_guc_ct_init_early(struct intel_guc_ct *ct);
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Michal Wajdeczko]

On 28.06.2021 01:14, Matthew Brost wrote:
> Implement a stall timer which fails H2G CTBs once a period of time
> with no forward progress is reached to prevent deadlock.
> 
> v2:
>  (Michal)
>   - Improve error message in ct_deadlock()
>   - Set broken when ct_deadlock() returns true
>   - Return -EPIPE on ct_deadlock()
> 
> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> ---
>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
>  2 files changed, 59 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> index 90ee95a240e8..8f553f7f9619 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> @@ -319,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
>  		goto err_deregister;
>  
>  	ct->enabled = true;
> +	ct->stall_time = KTIME_MAX;
>  
>  	return 0;
>  
> @@ -391,9 +392,6 @@ static int ct_write(struct intel_guc_ct *ct,
>  	u32 *cmds = ctb->cmds;
>  	unsigned int i;
>  
> -	if (unlikely(ctb->broken))
> -		return -EPIPE;
> -
>  	if (unlikely(desc->status))
>  		goto corrupted;
>  
> @@ -509,6 +507,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
>  	return err;
>  }
>  
> +#define GUC_CTB_TIMEOUT_MS	1500
> +static inline bool ct_deadlocked(struct intel_guc_ct *ct)
> +{
> +	long timeout = GUC_CTB_TIMEOUT_MS;
> +	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
> +
> +	if (unlikely(ret)) {
> +		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
> +		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
> +
> +		CT_ERROR(ct, "Communication stalled for %lld, desc status=%#x,%#x\n",

nit: missing unit in "stalled for ... ms"
                                     ^^^^

> +			 ktime_ms_delta(ktime_get(), ct->stall_time),
> +			 send->status, recv->status);
> +		ct->ctbs.send.broken = true;
> +	}
> +
> +	return ret;
> +}
> +
>  static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
>  {
>  	struct guc_ct_buffer_desc *desc = ctb->desc;
> @@ -520,6 +537,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
>  	return space >= len_dw;
>  }
>  
> +static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
> +{
> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> +
> +	lockdep_assert_held(&ct->ctbs.send.lock);
> +
> +	if (unlikely(!h2g_has_room(ctb, len_dw))) {
> +		if (ct->stall_time == KTIME_MAX)
> +			ct->stall_time = ktime_get();
> +
> +		if (unlikely(ct_deadlocked(ct)))
> +			return -EPIPE;
> +		else
> +			return -EBUSY;
> +	}
> +
> +	ct->stall_time = KTIME_MAX;
> +	return 0;
> +}
> +
>  static int ct_send_nb(struct intel_guc_ct *ct,
>  		      const u32 *action,
>  		      u32 len,
> @@ -530,13 +567,14 @@ static int ct_send_nb(struct intel_guc_ct *ct,
>  	u32 fence;
>  	int ret;
>  
> +	if (unlikely(ctb->broken))
> +		return -EPIPE;
> +
>  	spin_lock_irqsave(&ctb->lock, spin_flags);
>  
> -	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
> -	if (unlikely(!ret)) {
> -		ret = -EBUSY;
> +	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
> +	if (unlikely(ret))
>  		goto out;
> -	}
>  
>  	fence = ct_get_next_fence(ct);
>  	ret = ct_write(ct, action, len, fence, flags);
> @@ -571,6 +609,9 @@ static int ct_send(struct intel_guc_ct *ct,
>  	GEM_BUG_ON(!response_buf && response_buf_size);
>  	might_sleep();
>  
> +	if (unlikely(ctb->broken))
> +		return -EPIPE;

ok, but likely could be part of ct_can_send/has_room

> +
>  	/*
>  	 * We use a lazy spin wait loop here as we believe that if the CT
>  	 * buffers are sized correctly the flow control condition should be
> @@ -579,8 +620,13 @@ static int ct_send(struct intel_guc_ct *ct,
>  retry:
>  	spin_lock_irqsave(&ctb->lock, flags);
>  	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
> +		if (ct->stall_time == KTIME_MAX)
> +			ct->stall_time = ktime_get();
>  		spin_unlock_irqrestore(&ctb->lock, flags);
>  
> +		if (unlikely(ct_deadlocked(ct)))
> +			return -EPIPE;
> +

can't we really put all this into one place?

static int ct_can_send(struct intel_guc_ct *ct, u32 len_dw, bool wait)
{
	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;

	lockdep_assert_held(&ct->ctbs.send.lock);

retry:
	if (ct->broken)
		return -EPIPE;

	if (unlikely(!ctb_has_room(ctb, len_dw + GUC_CTB_HDR_LEN))) {
		if (ct->stall_time == KTIME_MAX)
			ct->stall_time = ktime_get();

		if (unlikely(ct_deadlocked(ct)))
			return -EPIPE;
		if (!wait)
			return -EBUSY;

		spin_unlock_irqrestore(&ctb->lock, flags);
		...
		spin_lock_irqrestore(&ctb->lock, flags);

		goto retry;
	}

	ct->stall_time = KTIME_MAX;
	return 0;
}

Michal

>  		if (msleep_interruptible(sleep_period_ms))
>  			return -EINTR;
>  		sleep_period_ms = sleep_period_ms << 1;
> @@ -588,6 +634,8 @@ static int ct_send(struct intel_guc_ct *ct,
>  		goto retry;
>  	}
>  
> +	ct->stall_time = KTIME_MAX;
> +
>  	fence = ct_get_next_fence(ct);
>  	request.fence = fence;
>  	request.status = 0;
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> index f6a4d5b33467..c9d6ae7848a7 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> @@ -9,6 +9,7 @@
>  #include <linux/interrupt.h>
>  #include <linux/spinlock.h>
>  #include <linux/workqueue.h>
> +#include <linux/ktime.h>
>  
>  #include "intel_guc_fwif.h"
>  
> @@ -68,6 +69,9 @@ struct intel_guc_ct {
>  		struct list_head incoming; /* incoming requests */
>  		struct work_struct worker; /* handler for incoming requests */
>  	} requests;
> +
> +	/** @stall_time: time of first time a CTB submission is stalled */
> +	ktime_t stall_time;
>  };
>  
>  void intel_guc_ct_init_early(struct intel_guc_ct *ct);
> 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Matthew Brost]

On Thu, Jul 01, 2021 at 01:23:36AM +0200, Michal Wajdeczko wrote:
> 
> 
> On 28.06.2021 01:14, Matthew Brost wrote:
> > Implement a stall timer which fails H2G CTBs once a period of time
> > with no forward progress is reached to prevent deadlock.
> > 
> > v2:
> >  (Michal)
> >   - Improve error message in ct_deadlock()
> >   - Set broken when ct_deadlock() returns true
> >   - Return -EPIPE on ct_deadlock()
> > 
> > Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> > Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
> > Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> > ---
> >  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
> >  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
> >  2 files changed, 59 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > index 90ee95a240e8..8f553f7f9619 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> > @@ -319,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
> >  		goto err_deregister;
> >  
> >  	ct->enabled = true;
> > +	ct->stall_time = KTIME_MAX;
> >  
> >  	return 0;
> >  
> > @@ -391,9 +392,6 @@ static int ct_write(struct intel_guc_ct *ct,
> >  	u32 *cmds = ctb->cmds;
> >  	unsigned int i;
> >  
> > -	if (unlikely(ctb->broken))
> > -		return -EPIPE;
> > -
> >  	if (unlikely(desc->status))
> >  		goto corrupted;
> >  
> > @@ -509,6 +507,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
> >  	return err;
> >  }
> >  
> > +#define GUC_CTB_TIMEOUT_MS	1500
> > +static inline bool ct_deadlocked(struct intel_guc_ct *ct)
> > +{
> > +	long timeout = GUC_CTB_TIMEOUT_MS;
> > +	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
> > +
> > +	if (unlikely(ret)) {
> > +		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
> > +		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
> > +
> > +		CT_ERROR(ct, "Communication stalled for %lld, desc status=%#x,%#x\n",
> 
> nit: missing unit in "stalled for ... ms"
>                                      ^^^^

Yep, will fix.

> 
> > +			 ktime_ms_delta(ktime_get(), ct->stall_time),
> > +			 send->status, recv->status);
> > +		ct->ctbs.send.broken = true;
> > +	}
> > +
> > +	return ret;
> > +}
> > +
> >  static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
> >  {
> >  	struct guc_ct_buffer_desc *desc = ctb->desc;
> > @@ -520,6 +537,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
> >  	return space >= len_dw;
> >  }
> >  
> > +static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
> > +{
> > +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> > +
> > +	lockdep_assert_held(&ct->ctbs.send.lock);
> > +
> > +	if (unlikely(!h2g_has_room(ctb, len_dw))) {
> > +		if (ct->stall_time == KTIME_MAX)
> > +			ct->stall_time = ktime_get();
> > +
> > +		if (unlikely(ct_deadlocked(ct)))
> > +			return -EPIPE;
> > +		else
> > +			return -EBUSY;
> > +	}
> > +
> > +	ct->stall_time = KTIME_MAX;
> > +	return 0;
> > +}
> > +
> >  static int ct_send_nb(struct intel_guc_ct *ct,
> >  		      const u32 *action,
> >  		      u32 len,
> > @@ -530,13 +567,14 @@ static int ct_send_nb(struct intel_guc_ct *ct,
> >  	u32 fence;
> >  	int ret;
> >  
> > +	if (unlikely(ctb->broken))
> > +		return -EPIPE;
> > +
> >  	spin_lock_irqsave(&ctb->lock, spin_flags);
> >  
> > -	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
> > -	if (unlikely(!ret)) {
> > -		ret = -EBUSY;
> > +	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
> > +	if (unlikely(ret))
> >  		goto out;
> > -	}
> >  
> >  	fence = ct_get_next_fence(ct);
> >  	ret = ct_write(ct, action, len, fence, flags);
> > @@ -571,6 +609,9 @@ static int ct_send(struct intel_guc_ct *ct,
> >  	GEM_BUG_ON(!response_buf && response_buf_size);
> >  	might_sleep();
> >  
> > +	if (unlikely(ctb->broken))
> > +		return -EPIPE;
> 
> ok, but likely could be part of ct_can_send/has_room
> 

No, this actually should be apart of 'intel_guc_ct_send'.

> > +
> >  	/*
> >  	 * We use a lazy spin wait loop here as we believe that if the CT
> >  	 * buffers are sized correctly the flow control condition should be
> > @@ -579,8 +620,13 @@ static int ct_send(struct intel_guc_ct *ct,
> >  retry:
> >  	spin_lock_irqsave(&ctb->lock, flags);
> >  	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
> > +		if (ct->stall_time == KTIME_MAX)
> > +			ct->stall_time = ktime_get();
> >  		spin_unlock_irqrestore(&ctb->lock, flags);
> >  
> > +		if (unlikely(ct_deadlocked(ct)))
> > +			return -EPIPE;
> > +
> 
> can't we really put all this into one place?
>

Maybe? IMO a helper with arguments might be worse that inline code
depending on how it looks in the end. Now that you mention this I
realize the patch that handles G2H credits is wrong as we really need to
reserve credits here too. When I rework that patch, I'll revisit this.

Matt
 
> static int ct_can_send(struct intel_guc_ct *ct, u32 len_dw, bool wait)
> {
> 	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> 
> 	lockdep_assert_held(&ct->ctbs.send.lock);
> 
> retry:
> 	if (ct->broken)
> 		return -EPIPE;
> 
> 	if (unlikely(!ctb_has_room(ctb, len_dw + GUC_CTB_HDR_LEN))) {
> 		if (ct->stall_time == KTIME_MAX)
> 			ct->stall_time = ktime_get();
> 
> 		if (unlikely(ct_deadlocked(ct)))
> 			return -EPIPE;
> 		if (!wait)
> 			return -EBUSY;
> 
> 		spin_unlock_irqrestore(&ctb->lock, flags);
> 		...
> 		spin_lock_irqrestore(&ctb->lock, flags);
> 
> 		goto retry;
> 	}
> 
> 	ct->stall_time = KTIME_MAX;
> 	return 0;
> }
> 
> Michal
> 
> >  		if (msleep_interruptible(sleep_period_ms))
> >  			return -EINTR;
> >  		sleep_period_ms = sleep_period_ms << 1;
> > @@ -588,6 +634,8 @@ static int ct_send(struct intel_guc_ct *ct,
> >  		goto retry;
> >  	}
> >  
> > +	ct->stall_time = KTIME_MAX;
> > +
> >  	fence = ct_get_next_fence(ct);
> >  	request.fence = fence;
> >  	request.status = 0;
> > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > index f6a4d5b33467..c9d6ae7848a7 100644
> > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> > @@ -9,6 +9,7 @@
> >  #include <linux/interrupt.h>
> >  #include <linux/spinlock.h>
> >  #include <linux/workqueue.h>
> > +#include <linux/ktime.h>
> >  
> >  #include "intel_guc_fwif.h"
> >  
> > @@ -68,6 +69,9 @@ struct intel_guc_ct {
> >  		struct list_head incoming; /* incoming requests */
> >  		struct work_struct worker; /* handler for incoming requests */
> >  	} requests;
> > +
> > +	/** @stall_time: time of first time a CTB submission is stalled */
> > +	ktime_t stall_time;
> >  };
> >  
> >  void intel_guc_ct_init_early(struct intel_guc_ct *ct);
> > 
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 6/7] drm/i915/guc: Optimize CTB writes and reads

[Matthew Brost]

CTB writes are now in the path of command submission and should be
optimized for performance. Rather than reading CTB descriptor values
(e.g. head, tail) which could result in accesses across the PCIe bus,
store shadow local copies and only read/write the descriptor values when
absolutely necessary. Also store the current space in the each channel
locally.

v2:
 (Michel)
  - Add additional sanity checks for head / tail pointers
  - Use GUC_CTB_HDR_LEN rather than magic 1

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 88 +++++++++++++++--------
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  6 ++
 2 files changed, 65 insertions(+), 29 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 8f553f7f9619..63bce4dc21fa 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -130,6 +130,10 @@ static void guc_ct_buffer_desc_init(struct guc_ct_buffer_desc *desc)
 static void guc_ct_buffer_reset(struct intel_guc_ct_buffer *ctb)
 {
 	ctb->broken = false;
+	ctb->tail = 0;
+	ctb->head = 0;
+	ctb->space = CIRC_SPACE(ctb->tail, ctb->head, ctb->size);
+
 	guc_ct_buffer_desc_init(ctb->desc);
 }
 
@@ -383,10 +387,8 @@ static int ct_write(struct intel_guc_ct *ct,
 {
 	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
 	struct guc_ct_buffer_desc *desc = ctb->desc;
-	u32 head = desc->head;
-	u32 tail = desc->tail;
+	u32 tail = ctb->tail;
 	u32 size = ctb->size;
-	u32 used;
 	u32 header;
 	u32 hxg;
 	u32 *cmds = ctb->cmds;
@@ -395,25 +397,22 @@ static int ct_write(struct intel_guc_ct *ct,
 	if (unlikely(desc->status))
 		goto corrupted;
 
-	if (unlikely((tail | head) >= size)) {
+	GEM_BUG_ON(tail > size);
+
+#ifdef CONFIG_DRM_I915_DEBUG_GUC
+	if (unlikely(tail != READ_ONCE(desc->tail))) {
+		CT_ERROR(ct, "Tail was modified %u != %u\n",
+			 desc->tail, ctb->tail);
+		desc->status |= GUC_CTB_STATUS_MISMATCH;
+		goto corrupted;
+	}
+	if (unlikely((desc->tail | desc->head) >= size)) {
 		CT_ERROR(ct, "Invalid offsets head=%u tail=%u (size=%u)\n",
-			 head, tail, size);
+			 desc->head, desc->tail, size);
 		desc->status |= GUC_CTB_STATUS_OVERFLOW;
 		goto corrupted;
 	}
-
-	/*
-	 * tail == head condition indicates empty. GuC FW does not support
-	 * using up the entire buffer to get tail == head meaning full.
-	 */
-	if (tail < head)
-		used = (size - head) + tail;
-	else
-		used = tail - head;
-
-	/* make sure there is a space including extra dw for the fence */
-	if (unlikely(used + len + GUC_CTB_HDR_LEN >= size))
-		return -ENOSPC;
+#endif
 
 	/*
 	 * dw0: CT header (including fence)
@@ -454,7 +453,9 @@ static int ct_write(struct intel_guc_ct *ct,
 	write_barrier(ct);
 
 	/* now update descriptor */
+	ctb->tail = tail;
 	WRITE_ONCE(desc->tail, tail);
+	ctb->space -= len + GUC_CTB_HDR_LEN;
 
 	return 0;
 
@@ -470,7 +471,7 @@ static int ct_write(struct intel_guc_ct *ct,
  * @req:	pointer to pending request
  * @status:	placeholder for status
  *
- * For each sent request, Guc shall send bac CT response message.
+ * For each sent request, GuC shall send back CT response message.
  * Our message handler will update status of tracked request once
  * response message with given fence is received. Wait here and
  * check for valid response status value.
@@ -526,24 +527,35 @@ static inline bool ct_deadlocked(struct intel_guc_ct *ct)
 	return ret;
 }
 
-static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
+static inline bool h2g_has_room(struct intel_guc_ct *ct, u32 len_dw)
 {
-	struct guc_ct_buffer_desc *desc = ctb->desc;
-	u32 head = READ_ONCE(desc->head);
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+	u32 head;
 	u32 space;
 
-	space = CIRC_SPACE(desc->tail, head, ctb->size);
+	if (ctb->space >= len_dw)
+		return true;
+
+	head = READ_ONCE(ctb->desc->head);
+	if (unlikely(head > ctb->size)) {
+		CT_ERROR(ct, "Corrupted descriptor head=%u tail=%u size=%u\n",
+			  ctb->desc->head, ctb->desc->tail, ctb->size);
+		ctb->desc->status |= GUC_CTB_STATUS_OVERFLOW;
+		ctb->broken = true;
+		return false;
+	}
+
+	space = CIRC_SPACE(ctb->tail, head, ctb->size);
+	ctb->space = space;
 
 	return space >= len_dw;
 }
 
 static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
 {
-	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
-
 	lockdep_assert_held(&ct->ctbs.send.lock);
 
-	if (unlikely(!h2g_has_room(ctb, len_dw))) {
+	if (unlikely(!h2g_has_room(ct, len_dw))) {
 		if (ct->stall_time == KTIME_MAX)
 			ct->stall_time = ktime_get();
 
@@ -619,7 +631,7 @@ static int ct_send(struct intel_guc_ct *ct,
 	 */
 retry:
 	spin_lock_irqsave(&ctb->lock, flags);
-	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+	if (unlikely(!h2g_has_room(ct, len + GUC_CTB_HDR_LEN))) {
 		if (ct->stall_time == KTIME_MAX)
 			ct->stall_time = ktime_get();
 		spin_unlock_irqrestore(&ctb->lock, flags);
@@ -736,7 +748,7 @@ static int ct_read(struct intel_guc_ct *ct, struct ct_incoming_msg **msg)
 {
 	struct intel_guc_ct_buffer *ctb = &ct->ctbs.recv;
 	struct guc_ct_buffer_desc *desc = ctb->desc;
-	u32 head = desc->head;
+	u32 head = ctb->head;
 	u32 tail = desc->tail;
 	u32 size = ctb->size;
 	u32 *cmds = ctb->cmds;
@@ -751,12 +763,29 @@ static int ct_read(struct intel_guc_ct *ct, struct ct_incoming_msg **msg)
 	if (unlikely(desc->status))
 		goto corrupted;
 
-	if (unlikely((tail | head) >= size)) {
+	GEM_BUG_ON(head > size);
+
+#ifdef CONFIG_DRM_I915_DEBUG_GUC
+	if (unlikely(head != READ_ONCE(desc->head))) {
+		CT_ERROR(ct, "Head was modified %u != %u\n",
+			 desc->head, ctb->head);
+		desc->status |= GUC_CTB_STATUS_MISMATCH;
+		goto corrupted;
+	}
+	if (unlikely((desc->tail | desc->head) >= size)) {
+		CT_ERROR(ct, "Invalid offsets head=%u tail=%u (size=%u)\n",
+			 head, tail, size);
+		desc->status |= GUC_CTB_STATUS_OVERFLOW;
+		goto corrupted;
+	}
+#else
+	if (unlikely((tail | ctb->head) >= size)) {
 		CT_ERROR(ct, "Invalid offsets head=%u tail=%u (size=%u)\n",
 			 head, tail, size);
 		desc->status |= GUC_CTB_STATUS_OVERFLOW;
 		goto corrupted;
 	}
+#endif
 
 	/* tail == head condition indicates empty */
 	available = tail - head;
@@ -806,6 +835,7 @@ static int ct_read(struct intel_guc_ct *ct, struct ct_incoming_msg **msg)
 	}
 	CT_DEBUG(ct, "received %*ph\n", 4 * len, (*msg)->msg);
 
+	ctb->head = head;
 	/* now update descriptor */
 	WRITE_ONCE(desc->head, head);
 
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index c9d6ae7848a7..a5b42b58483b 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -33,6 +33,9 @@ struct intel_guc;
  * @desc: pointer to the buffer descriptor
  * @cmds: pointer to the commands buffer
  * @size: size of the commands buffer in dwords
+ * @head: local shadow copy of head in dwords
+ * @tail: local shadow copy of tail in dwords
+ * @space: local shadow copy of space in dwords
  * @broken: flag to indicate if descriptor data is broken
  */
 struct intel_guc_ct_buffer {
@@ -40,6 +43,9 @@ struct intel_guc_ct_buffer {
 	struct guc_ct_buffer_desc *desc;
 	u32 *cmds;
 	u32 size;
+	u32 tail;
+	u32 head;
+	u32 space;
 	bool broken;
 };
 
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 7/7] drm/i915/guc: Module load failure test for CT buffer creation

[Matthew Brost]

From: John Harrison <John.C.Harrison@Intel.com>

Add several module failure load inject points in the CT buffer creation
code path.

Signed-off-by: John Harrison <john.c.harrison@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 63bce4dc21fa..44bb6d630150 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -175,6 +175,10 @@ static int ct_register_buffer(struct intel_guc_ct *ct, u32 type,
 {
 	int err;
 
+	err = i915_inject_probe_error(guc_to_gt(ct_to_guc(ct))->i915, -ENXIO);
+	if (unlikely(err))
+		return err;
+
 	err = guc_action_register_ct_buffer(ct_to_guc(ct), type,
 					    desc_addr, buff_addr, size);
 	if (unlikely(err))
@@ -226,6 +230,10 @@ int intel_guc_ct_init(struct intel_guc_ct *ct)
 	u32 *cmds;
 	int err;
 
+	err = i915_inject_probe_error(guc_to_gt(guc)->i915, -ENXIO);
+	if (err)
+		return err;
+
 	GEM_BUG_ON(ct->vma);
 
 	blob_size = 2 * CTB_DESC_SIZE + CTB_H2G_BUFFER_SIZE + CTB_G2H_BUFFER_SIZE;
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✓ Fi.CI.BAT: success for CT changes required for GuC submission

[Patchwork]

[-- Attachment #1.1: Type: text/plain, Size: 2600 bytes --]

== Series Details ==

Series: CT changes required for GuC submission
URL   : https://patchwork.freedesktop.org/series/91943/
State : success

== Summary ==

CI Bug Log - changes from CI_DRM_10282 -> Patchwork_20474
====================================================

Summary
-------

  **SUCCESS**

  No regressions found.

  External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/index.html

Known issues
------------

  Here are the changes found in Patchwork_20474 that come from known issues:

### IGT changes ###

#### Issues hit ####

  * igt@amdgpu/amd_basic@cs-gfx:
    - fi-kbl-soraka:      NOTRUN -> [SKIP][1] ([fdo#109271]) +16 similar issues
   [1]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/fi-kbl-soraka/igt@amdgpu/amd_basic@cs-gfx.html

  
#### Possible fixes ####

  * igt@i915_selftest@live@gt_mocs:
    - {fi-tgl-1115g4}:    [DMESG-WARN][2] ([i915#2867]) -> [PASS][3] +14 similar issues
   [2]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/fi-tgl-1115g4/igt@i915_selftest@live@gt_mocs.html
   [3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/fi-tgl-1115g4/igt@i915_selftest@live@gt_mocs.html

  
  {name}: This element is suppressed. This means it is ignored when computing
          the status of the difference (SUCCESS, WARNING, or FAILURE).

  [fdo#109271]: https://bugs.freedesktop.org/show_bug.cgi?id=109271
  [i915#2867]: https://gitlab.freedesktop.org/drm/intel/issues/2867


Participating hosts (43 -> 38)
------------------------------

  Missing    (5): fi-ilk-m540 fi-hsw-4200u fi-bsw-cyan fi-kbl-x1275 fi-bdw-samus 


Build changes
-------------

  * Linux: CI_DRM_10282 -> Patchwork_20474

  CI-20190529: 20190529
  CI_DRM_10282: 67f5a18128770817e4218a9e496d2bf5047c51e8 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_6119: a306810ebbc8984bde38a57ef0c33eea394f4e18 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
  Patchwork_20474: 8dbeeecc18bfe993044da10dd04268fbac31c637 @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

8dbeeecc18bf drm/i915/guc: Module load failure test for CT buffer creation
609b9d9541a4 drm/i915/guc: Optimize CTB writes and reads
af75da893bfe drm/i915/guc: Add stall timer to non blocking CTB send function
3a1d54d48cfe drm/i915/guc: Add non blocking CTB send function
7ed4a3e0276a drm/i915/guc: Increase size of CTB buffers
9074da160af0 drm/i915/guc: Improve error message for unsolicited CT response
4de6bd659c01 drm/i915/guc: Relax CTB response timeout

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/index.html

[-- Attachment #1.2: Type: text/html, Size: 3262 bytes --]

[-- Attachment #2: Type: text/plain, Size: 160 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] ✓ Fi.CI.IGT: success for CT changes required for GuC submission

[Patchwork]

[-- Attachment #1.1: Type: text/plain, Size: 24956 bytes --]

== Series Details ==

Series: CT changes required for GuC submission
URL   : https://patchwork.freedesktop.org/series/91943/
State : success

== Summary ==

CI Bug Log - changes from CI_DRM_10282_full -> Patchwork_20474_full
====================================================

Summary
-------

  **WARNING**

  Minor unknown changes coming with Patchwork_20474_full need to be verified
  manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_20474_full, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  

Possible new issues
-------------------

  Here are the unknown changes that may have been introduced in Patchwork_20474_full:

### IGT changes ###

#### Warnings ####

  * igt@kms_psr2_sf@overlay-plane-update-sf-dmg-area-4:
    - shard-iclb:         [SKIP][1] ([i915#658]) -> [SKIP][2] +1 similar issue
   [1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb7/igt@kms_psr2_sf@overlay-plane-update-sf-dmg-area-4.html
   [2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb2/igt@kms_psr2_sf@overlay-plane-update-sf-dmg-area-4.html

  
Known issues
------------

  Here are the changes found in Patchwork_20474_full that come from known issues:

### IGT changes ###

#### Issues hit ####

  * igt@gem_create@create-massive:
    - shard-snb:          NOTRUN -> [DMESG-WARN][3] ([i915#3002]) +1 similar issue
   [3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-snb5/igt@gem_create@create-massive.html
    - shard-kbl:          NOTRUN -> [DMESG-WARN][4] ([i915#3002])
   [4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl7/igt@gem_create@create-massive.html

  * igt@gem_exec_fair@basic-none-rrul@rcs0:
    - shard-glk:          [PASS][5] -> [FAIL][6] ([i915#2842]) +2 similar issues
   [5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk5/igt@gem_exec_fair@basic-none-rrul@rcs0.html
   [6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk7/igt@gem_exec_fair@basic-none-rrul@rcs0.html

  * igt@gem_exec_fair@basic-none@vecs0:
    - shard-apl:          [PASS][7] -> [FAIL][8] ([i915#2842] / [i915#3468])
   [7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl2/igt@gem_exec_fair@basic-none@vecs0.html
   [8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl1/igt@gem_exec_fair@basic-none@vecs0.html

  * igt@gem_exec_params@no-vebox:
    - shard-skl:          NOTRUN -> [SKIP][9] ([fdo#109271]) +23 similar issues
   [9]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl9/igt@gem_exec_params@no-vebox.html

  * igt@gem_huc_copy@huc-copy:
    - shard-apl:          NOTRUN -> [SKIP][10] ([fdo#109271] / [i915#2190])
   [10]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@gem_huc_copy@huc-copy.html

  * igt@gem_mmap_gtt@cpuset-medium-copy:
    - shard-iclb:         [PASS][11] -> [FAIL][12] ([i915#307])
   [11]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb4/igt@gem_mmap_gtt@cpuset-medium-copy.html
   [12]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb1/igt@gem_mmap_gtt@cpuset-medium-copy.html

  * igt@gem_userptr_blits@dmabuf-sync:
    - shard-kbl:          NOTRUN -> [SKIP][13] ([fdo#109271] / [i915#3323])
   [13]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl1/igt@gem_userptr_blits@dmabuf-sync.html

  * igt@gem_userptr_blits@input-checking:
    - shard-apl:          NOTRUN -> [DMESG-WARN][14] ([i915#3002]) +1 similar issue
   [14]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@gem_userptr_blits@input-checking.html

  * igt@gem_workarounds@suspend-resume-fd:
    - shard-apl:          NOTRUN -> [DMESG-WARN][15] ([i915#180])
   [15]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl1/igt@gem_workarounds@suspend-resume-fd.html

  * igt@i915_pm_lpsp@kms-lpsp@kms-lpsp-dp:
    - shard-apl:          NOTRUN -> [SKIP][16] ([fdo#109271] / [i915#1937])
   [16]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@i915_pm_lpsp@kms-lpsp@kms-lpsp-dp.html

  * igt@kms_big_fb@linear-32bpp-rotate-180:
    - shard-iclb:         [PASS][17] -> [DMESG-WARN][18] ([i915#3621])
   [17]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb7/igt@kms_big_fb@linear-32bpp-rotate-180.html
   [18]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb1/igt@kms_big_fb@linear-32bpp-rotate-180.html

  * igt@kms_chamelium@dp-hpd-storm-disable:
    - shard-skl:          NOTRUN -> [SKIP][19] ([fdo#109271] / [fdo#111827]) +1 similar issue
   [19]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl2/igt@kms_chamelium@dp-hpd-storm-disable.html

  * igt@kms_chamelium@hdmi-edid-read:
    - shard-snb:          NOTRUN -> [SKIP][20] ([fdo#109271] / [fdo#111827]) +5 similar issues
   [20]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-snb5/igt@kms_chamelium@hdmi-edid-read.html

  * igt@kms_chamelium@vga-edid-read:
    - shard-apl:          NOTRUN -> [SKIP][21] ([fdo#109271] / [fdo#111827]) +13 similar issues
   [21]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@kms_chamelium@vga-edid-read.html

  * igt@kms_chamelium@vga-hpd-with-enabled-mode:
    - shard-iclb:         NOTRUN -> [SKIP][22] ([fdo#109284] / [fdo#111827])
   [22]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_chamelium@vga-hpd-with-enabled-mode.html

  * igt@kms_color_chamelium@pipe-b-ctm-max:
    - shard-glk:          NOTRUN -> [SKIP][23] ([fdo#109271] / [fdo#111827])
   [23]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk9/igt@kms_color_chamelium@pipe-b-ctm-max.html

  * igt@kms_content_protection@uevent:
    - shard-apl:          NOTRUN -> [FAIL][24] ([i915#2105])
   [24]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl3/igt@kms_content_protection@uevent.html
    - shard-iclb:         NOTRUN -> [SKIP][25] ([fdo#109300] / [fdo#111066])
   [25]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_content_protection@uevent.html

  * igt@kms_cursor_crc@pipe-d-cursor-256x85-rapid-movement:
    - shard-glk:          NOTRUN -> [SKIP][26] ([fdo#109271]) +3 similar issues
   [26]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk9/igt@kms_cursor_crc@pipe-d-cursor-256x85-rapid-movement.html

  * igt@kms_cursor_legacy@pipe-d-single-move:
    - shard-iclb:         NOTRUN -> [SKIP][27] ([fdo#109278]) +2 similar issues
   [27]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_cursor_legacy@pipe-d-single-move.html

  * igt@kms_draw_crc@draw-method-rgb565-mmap-cpu-ytiled:
    - shard-skl:          [PASS][28] -> [DMESG-WARN][29] ([i915#1982])
   [28]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl7/igt@kms_draw_crc@draw-method-rgb565-mmap-cpu-ytiled.html
   [29]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl7/igt@kms_draw_crc@draw-method-rgb565-mmap-cpu-ytiled.html

  * igt@kms_flip@2x-wf_vblank-ts-check-interruptible:
    - shard-iclb:         NOTRUN -> [SKIP][30] ([fdo#109274])
   [30]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_flip@2x-wf_vblank-ts-check-interruptible.html

  * igt@kms_flip@flip-vs-suspend-interruptible@a-dp1:
    - shard-apl:          [PASS][31] -> [DMESG-WARN][32] ([i915#180]) +1 similar issue
   [31]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl2/igt@kms_flip@flip-vs-suspend-interruptible@a-dp1.html
   [32]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@kms_flip@flip-vs-suspend-interruptible@a-dp1.html

  * igt@kms_flip@flip-vs-suspend@c-dp1:
    - shard-kbl:          [PASS][33] -> [DMESG-WARN][34] ([i915#180]) +4 similar issues
   [33]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-kbl7/igt@kms_flip@flip-vs-suspend@c-dp1.html
   [34]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl6/igt@kms_flip@flip-vs-suspend@c-dp1.html

  * igt@kms_flip@flip-vs-suspend@c-edp1:
    - shard-skl:          [PASS][35] -> [INCOMPLETE][36] ([i915#198])
   [35]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl6/igt@kms_flip@flip-vs-suspend@c-edp1.html
   [36]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl8/igt@kms_flip@flip-vs-suspend@c-edp1.html

  * igt@kms_flip_scaled_crc@flip-64bpp-ytile-to-32bpp-ytilercccs:
    - shard-apl:          NOTRUN -> [SKIP][37] ([fdo#109271] / [i915#2672])
   [37]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@kms_flip_scaled_crc@flip-64bpp-ytile-to-32bpp-ytilercccs.html

  * igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-shrfb-draw-mmap-cpu:
    - shard-snb:          NOTRUN -> [SKIP][38] ([fdo#109271]) +105 similar issues
   [38]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-snb5/igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-shrfb-draw-mmap-cpu.html

  * igt@kms_frontbuffer_tracking@fbcpsr-2p-primscrn-spr-indfb-draw-mmap-cpu:
    - shard-kbl:          NOTRUN -> [SKIP][39] ([fdo#109271]) +12 similar issues
   [39]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl1/igt@kms_frontbuffer_tracking@fbcpsr-2p-primscrn-spr-indfb-draw-mmap-cpu.html

  * igt@kms_frontbuffer_tracking@fbcpsr-2p-shrfb-fliptrack-mmap-gtt:
    - shard-iclb:         NOTRUN -> [SKIP][40] ([fdo#109280]) +3 similar issues
   [40]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_frontbuffer_tracking@fbcpsr-2p-shrfb-fliptrack-mmap-gtt.html

  * igt@kms_hdr@bpc-switch-dpms:
    - shard-skl:          [PASS][41] -> [FAIL][42] ([i915#1188]) +1 similar issue
   [41]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl7/igt@kms_hdr@bpc-switch-dpms.html
   [42]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl7/igt@kms_hdr@bpc-switch-dpms.html

  * igt@kms_pipe_crc_basic@read-crc-pipe-d:
    - shard-apl:          NOTRUN -> [SKIP][43] ([fdo#109271] / [i915#533]) +1 similar issue
   [43]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl3/igt@kms_pipe_crc_basic@read-crc-pipe-d.html

  * igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b:
    - shard-kbl:          NOTRUN -> [DMESG-WARN][44] ([i915#180])
   [44]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl1/igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b.html

  * igt@kms_plane_alpha_blend@pipe-a-alpha-basic:
    - shard-apl:          NOTRUN -> [FAIL][45] ([fdo#108145] / [i915#265]) +2 similar issues
   [45]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@kms_plane_alpha_blend@pipe-a-alpha-basic.html

  * igt@kms_plane_alpha_blend@pipe-c-constant-alpha-min:
    - shard-skl:          [PASS][46] -> [FAIL][47] ([fdo#108145] / [i915#265]) +1 similar issue
   [46]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl1/igt@kms_plane_alpha_blend@pipe-c-constant-alpha-min.html
   [47]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl5/igt@kms_plane_alpha_blend@pipe-c-constant-alpha-min.html

  * igt@kms_psr2_sf@overlay-plane-update-sf-dmg-area-2:
    - shard-apl:          NOTRUN -> [SKIP][48] ([fdo#109271] / [i915#658]) +2 similar issues
   [48]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@kms_psr2_sf@overlay-plane-update-sf-dmg-area-2.html

  * igt@kms_psr2_sf@overlay-primary-update-sf-dmg-area-3:
    - shard-kbl:          NOTRUN -> [SKIP][49] ([fdo#109271] / [i915#658])
   [49]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl1/igt@kms_psr2_sf@overlay-primary-update-sf-dmg-area-3.html

  * igt@kms_psr@psr2_dpms:
    - shard-iclb:         NOTRUN -> [SKIP][50] ([fdo#109441])
   [50]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@kms_psr@psr2_dpms.html

  * igt@kms_psr@psr2_sprite_render:
    - shard-iclb:         [PASS][51] -> [SKIP][52] ([fdo#109441])
   [51]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb2/igt@kms_psr@psr2_sprite_render.html
   [52]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb4/igt@kms_psr@psr2_sprite_render.html

  * igt@kms_setmode@clone-exclusive-crtc:
    - shard-skl:          NOTRUN -> [WARN][53] ([i915#2100])
   [53]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl9/igt@kms_setmode@clone-exclusive-crtc.html

  * igt@kms_vblank@pipe-d-wait-forked-hang:
    - shard-apl:          NOTRUN -> [SKIP][54] ([fdo#109271]) +152 similar issues
   [54]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@kms_vblank@pipe-d-wait-forked-hang.html

  * igt@kms_writeback@writeback-invalid-parameters:
    - shard-apl:          NOTRUN -> [SKIP][55] ([fdo#109271] / [i915#2437])
   [55]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@kms_writeback@writeback-invalid-parameters.html

  * igt@nouveau_crc@pipe-c-source-rg:
    - shard-iclb:         NOTRUN -> [SKIP][56] ([i915#2530])
   [56]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@nouveau_crc@pipe-c-source-rg.html

  * igt@prime_nv_api@i915_nv_double_export:
    - shard-iclb:         NOTRUN -> [SKIP][57] ([fdo#109291])
   [57]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@prime_nv_api@i915_nv_double_export.html

  * igt@runner@aborted:
    - shard-snb:          NOTRUN -> ([FAIL][58], [FAIL][59]) ([i915#3002])
   [58]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-snb6/igt@runner@aborted.html
   [59]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-snb5/igt@runner@aborted.html

  * igt@sysfs_clients@split-50:
    - shard-apl:          NOTRUN -> [SKIP][60] ([fdo#109271] / [i915#2994]) +1 similar issue
   [60]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@sysfs_clients@split-50.html

  
#### Possible fixes ####

  * igt@gem_exec_fair@basic-flow@rcs0:
    - shard-tglb:         [FAIL][61] ([i915#2842]) -> [PASS][62]
   [61]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-tglb5/igt@gem_exec_fair@basic-flow@rcs0.html
   [62]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-tglb8/igt@gem_exec_fair@basic-flow@rcs0.html

  * igt@gem_exec_fair@basic-none@vcs0:
    - shard-apl:          [FAIL][63] ([i915#2842]) -> [PASS][64]
   [63]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl2/igt@gem_exec_fair@basic-none@vcs0.html
   [64]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl1/igt@gem_exec_fair@basic-none@vcs0.html

  * igt@gem_exec_fair@basic-pace-share@rcs0:
    - shard-glk:          [FAIL][65] ([i915#2842]) -> [PASS][66]
   [65]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk3/igt@gem_exec_fair@basic-pace-share@rcs0.html
   [66]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk1/igt@gem_exec_fair@basic-pace-share@rcs0.html

  * igt@gem_exec_gttfill@basic:
    - shard-iclb:         [INCOMPLETE][67] ([i915#2405]) -> [PASS][68]
   [67]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb4/igt@gem_exec_gttfill@basic.html
   [68]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb7/igt@gem_exec_gttfill@basic.html

  * igt@gem_exec_suspend@basic-s3:
    - shard-kbl:          [DMESG-WARN][69] ([i915#180]) -> [PASS][70] +2 similar issues
   [69]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-kbl6/igt@gem_exec_suspend@basic-s3.html
   [70]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-kbl1/igt@gem_exec_suspend@basic-s3.html

  * igt@gem_mmap_gtt@big-copy-xy:
    - shard-glk:          [FAIL][71] ([i915#307]) -> [PASS][72]
   [71]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk3/igt@gem_mmap_gtt@big-copy-xy.html
   [72]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk6/igt@gem_mmap_gtt@big-copy-xy.html

  * igt@gem_spin_batch@spin-each:
    - shard-apl:          [FAIL][73] ([i915#2898]) -> [PASS][74]
   [73]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl3/igt@gem_spin_batch@spin-each.html
   [74]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@gem_spin_batch@spin-each.html

  * igt@gen9_exec_parse@allowed-all:
    - shard-glk:          [DMESG-WARN][75] ([i915#1436] / [i915#716]) -> [PASS][76]
   [75]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk1/igt@gen9_exec_parse@allowed-all.html
   [76]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk9/igt@gen9_exec_parse@allowed-all.html

  * igt@kms_big_fb@linear-32bpp-rotate-0:
    - shard-glk:          [DMESG-WARN][77] ([i915#118] / [i915#95]) -> [PASS][78]
   [77]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk9/igt@kms_big_fb@linear-32bpp-rotate-0.html
   [78]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk7/igt@kms_big_fb@linear-32bpp-rotate-0.html

  * igt@kms_color@pipe-a-ctm-0-5:
    - shard-skl:          [DMESG-WARN][79] ([i915#1982]) -> [PASS][80]
   [79]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl7/igt@kms_color@pipe-a-ctm-0-5.html
   [80]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl7/igt@kms_color@pipe-a-ctm-0-5.html

  * igt@kms_flip@flip-vs-suspend@b-dp1:
    - shard-apl:          [DMESG-WARN][81] ([i915#180]) -> [PASS][82] +1 similar issue
   [81]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl1/igt@kms_flip@flip-vs-suspend@b-dp1.html
   [82]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@kms_flip@flip-vs-suspend@b-dp1.html

  * igt@kms_plane_alpha_blend@pipe-b-coverage-7efc:
    - shard-skl:          [FAIL][83] ([fdo#108145] / [i915#265]) -> [PASS][84]
   [83]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-skl1/igt@kms_plane_alpha_blend@pipe-b-coverage-7efc.html
   [84]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-skl8/igt@kms_plane_alpha_blend@pipe-b-coverage-7efc.html

  * igt@kms_psr2_su@frontbuffer:
    - shard-iclb:         [SKIP][85] ([fdo#109642] / [fdo#111068] / [i915#658]) -> [PASS][86]
   [85]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb1/igt@kms_psr2_su@frontbuffer.html
   [86]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb2/igt@kms_psr2_su@frontbuffer.html

  * igt@kms_psr@psr2_suspend:
    - shard-iclb:         [SKIP][87] ([fdo#109441]) -> [PASS][88] +2 similar issues
   [87]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb8/igt@kms_psr@psr2_suspend.html
   [88]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb2/igt@kms_psr@psr2_suspend.html

  
#### Warnings ####

  * igt@i915_pm_rc6_residency@rc6-idle:
    - shard-iclb:         [WARN][89] ([i915#1804] / [i915#2684]) -> [WARN][90] ([i915#2684])
   [89]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb4/igt@i915_pm_rc6_residency@rc6-idle.html
   [90]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb1/igt@i915_pm_rc6_residency@rc6-idle.html

  * igt@kms_big_fb@x-tiled-32bpp-rotate-180:
    - shard-glk:          [DMESG-WARN][91] ([i915#118] / [i915#95]) -> [DMESG-WARN][92] ([i915#118] / [i915#1982] / [i915#95])
   [91]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-glk9/igt@kms_big_fb@x-tiled-32bpp-rotate-180.html
   [92]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-glk7/igt@kms_big_fb@x-tiled-32bpp-rotate-180.html

  * igt@kms_psr2_sf@overlay-primary-update-sf-dmg-area-1:
    - shard-iclb:         [SKIP][93] -> [SKIP][94] ([i915#658])
   [93]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-iclb2/igt@kms_psr2_sf@overlay-primary-update-sf-dmg-area-1.html
   [94]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-iclb4/igt@kms_psr2_sf@overlay-primary-update-sf-dmg-area-1.html

  * igt@runner@aborted:
    - shard-apl:          ([FAIL][95], [FAIL][96], [FAIL][97]) ([fdo#109271] / [i915#180] / [i915#1814] / [i915#3363]) -> ([FAIL][98], [FAIL][99], [FAIL][100], [FAIL][101], [FAIL][102]) ([fdo#109271] / [i915#180] / [i915#1814] / [i915#3002] / [i915#3363])
   [95]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl8/igt@runner@aborted.html
   [96]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl1/igt@runner@aborted.html
   [97]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_10282/shard-apl2/igt@runner@aborted.html
   [98]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl1/igt@runner@aborted.html
   [99]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@runner@aborted.html
   [100]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl1/igt@runner@aborted.html
   [101]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl2/igt@runner@aborted.html
   [102]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/shard-apl8/igt@runner@aborted.html

  
  {name}: This element is suppressed. This means it is ignored when computing
          the status of the difference (SUCCESS, WARNING, or FAILURE).

  [fdo#108145]: https://bugs.freedesktop.org/show_bug.cgi?id=108145
  [fdo#109271]: https://bugs.freedesktop.org/show_bug.cgi?id=109271
  [fdo#109274]: https://bugs.freedesktop.org/show_bug.cgi?id=109274
  [fdo#109278]: https://bugs.freedesktop.org/show_bug.cgi?id=109278
  [fdo#109280]: https://bugs.freedesktop.org/show_bug.cgi?id=109280
  [fdo#109284]: https://bugs.freedesktop.org/show_bug.cgi?id=109284
  [fdo#109291]: https://bugs.freedesktop.org/show_bug.cgi?id=109291
  [fdo#109300]: https://bugs.freedesktop.org/show_bug.cgi?id=109300
  [fdo#109441]: https://bugs.freedesktop.org/show_bug.cgi?id=109441
  [fdo#109642]: https://bugs.freedesktop.org/show_bug.cgi?id=109642
  [fdo#111066]: https://bugs.freedesktop.org/show_bug.cgi?id=111066
  [fdo#111068]: https://bugs.freedesktop.org/show_bug.cgi?id=111068
  [fdo#111827]: https://bugs.freedesktop.org/show_bug.cgi?id=111827
  [i915#118]: https://gitlab.freedesktop.org/drm/intel/issues/118
  [i915#1188]: https://gitlab.freedesktop.org/drm/intel/issues/1188
  [i915#1436]: https://gitlab.freedesktop.org/drm/intel/issues/1436
  [i915#180]: https://gitlab.freedesktop.org/drm/intel/issues/180
  [i915#1804]: https://gitlab.freedesktop.org/drm/intel/issues/1804
  [i915#1814]: https://gitlab.freedesktop.org/drm/intel/issues/1814
  [i915#1937]: https://gitlab.freedesktop.org/drm/intel/issues/1937
  [i915#198]: https://gitlab.freedesktop.org/drm/intel/issues/198
  [i915#1982]: https://gitlab.freedesktop.org/drm/intel/issues/1982
  [i915#2100]: https://gitlab.freedesktop.org/drm/intel/issues/2100
  [i915#2105]: https://gitlab.freedesktop.org/drm/intel/issues/2105
  [i915#2190]: https://gitlab.freedesktop.org/drm/intel/issues/2190
  [i915#2405]: https://gitlab.freedesktop.org/drm/intel/issues/2405
  [i915#2437]: https://gitlab.freedesktop.org/drm/intel/issues/2437
  [i915#2530]: https://gitlab.freedesktop.org/drm/intel/issues/2530
  [i915#265]: https://gitlab.freedesktop.org/drm/intel/issues/265
  [i915#2672]: https://gitlab.freedesktop.org/drm/intel/issues/2672
  [i915#2684]: https://gitlab.freedesktop.org/drm/intel/issues/2684
  [i915#2842]: https://gitlab.freedesktop.org/drm/intel/issues/2842
  [i915#2898]: https://gitlab.freedesktop.org/drm/intel/issues/2898
  [i915#2994]: https://gitlab.freedesktop.org/drm/intel/issues/2994
  [i915#3002]: https://gitlab.freedesktop.org/drm/intel/issues/3002
  [i915#307]: https://gitlab.freedesktop.org/drm/intel/issues/307
  [i915#3323]: https://gitlab.freedesktop.org/drm/intel/issues/3323
  [i915#3363]: https://gitlab.freedesktop.org/drm/intel/issues/3363
  [i915#3468]: https://gitlab.freedesktop.org/drm/intel/issues/3468
  [i915#3621]: https://gitlab.freedesktop.org/drm/intel/issues/3621
  [i915#533]: https://gitlab.freedesktop.org/drm/intel/issues/533
  [i915#658]: https://gitlab.freedesktop.org/drm/intel/issues/658
  [i915#716]: https://gitlab.freedesktop.org/drm/intel/issues/716
  [i915#95]: https://gitlab.freedesktop.org/drm/intel/issues/95


Participating hosts (9 -> 9)
------------------------------

  No changes in participating hosts


Build changes
-------------

  * Linux: CI_DRM_10282 -> Patchwork_20474

  CI-20190529: 20190529
  CI_DRM_10282: 67f5a18128770817e4218a9e496d2bf5047c51e8 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_6119: a306810ebbc8984bde38a57ef0c33eea394f4e18 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
  Patchwork_20474: 8dbeeecc18bfe993044da10dd04268fbac31c637 @ git://anongit.freedesktop.org/gfx-ci/linux
  piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_20474/index.html

[-- Attachment #1.2: Type: text/html, Size: 31434 bytes --]

[-- Attachment #2: Type: text/plain, Size: 160 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 0/7] CT changes required for GuC submission

[Matthew Brost]

As part of enabling GuC submission discussed in [1], [2], and [3] we
need optimize and update the CT code as this is now in the critical
path of submission. This series includes the patches to do that which is
the first 7 patches from [3]. The patches should have addressed all the
feedback in [3] and should be ready to merge once CI returns a we get a
few more RBs.

v2: Fix checkpatch warning, address a couple of Michal's comments

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

[1] https://patchwork.freedesktop.org/series/89844/
[2] https://patchwork.freedesktop.org/series/91417/
[3] https://patchwork.freedesktop.org/series/91840/

John Harrison (1):
  drm/i915/guc: Module load failure test for CT buffer creation

Matthew Brost (6):
  drm/i915/guc: Relax CTB response timeout
  drm/i915/guc: Improve error message for unsolicited CT response
  drm/i915/guc: Increase size of CTB buffers
  drm/i915/guc: Add non blocking CTB send function
  drm/i915/guc: Add stall timer to non blocking CTB send function
  drm/i915/guc: Optimize CTB writes and reads

 .../gt/uc/abi/guc_communication_ctb_abi.h     |   3 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.h        |  11 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 250 +++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  14 +-
 4 files changed, 232 insertions(+), 46 deletions(-)

-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Matthew Brost]

Implement a stall timer which fails H2G CTBs once a period of time
with no forward progress is reached to prevent deadlock.

v2:
 (Michal)
  - Improve error message in ct_deadlock()
  - Set broken when ct_deadlock() returns true
  - Return -EPIPE on ct_deadlock()
v3:
 (Michal)
  - Add ms to stall timer comment
 (Matthew)
  - Move broken check to intel_guc_ct_send()

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
 2 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index fb825cc1d090..a9cb7b608520 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -4,6 +4,9 @@
  */
 
 #include <linux/circ_buf.h>
+#include <linux/ktime.h>
+#include <linux/time64.h>
+#include <linux/timekeeping.h>
 
 #include "i915_drv.h"
 #include "intel_guc_ct.h"
@@ -316,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
 		goto err_deregister;
 
 	ct->enabled = true;
+	ct->stall_time = KTIME_MAX;
 
 	return 0;
 
@@ -388,9 +392,6 @@ static int ct_write(struct intel_guc_ct *ct,
 	u32 *cmds = ctb->cmds;
 	unsigned int i;
 
-	if (unlikely(ctb->broken))
-		return -EPIPE;
-
 	if (unlikely(desc->status))
 		goto corrupted;
 
@@ -506,6 +507,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	return err;
 }
 
+#define GUC_CTB_TIMEOUT_MS	1500
+static inline bool ct_deadlocked(struct intel_guc_ct *ct)
+{
+	long timeout = GUC_CTB_TIMEOUT_MS;
+	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
+
+	if (unlikely(ret)) {
+		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
+		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
+
+		CT_ERROR(ct, "Communication stalled for %lld ms, desc status=%#x,%#x\n",
+			 ktime_ms_delta(ktime_get(), ct->stall_time),
+			 send->status, recv->status);
+		ct->ctbs.send.broken = true;
+	}
+
+	return ret;
+}
+
 static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 {
 	struct guc_ct_buffer_desc *desc = ctb->desc;
@@ -517,6 +537,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 	return space >= len_dw;
 }
 
+static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
+{
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+
+	lockdep_assert_held(&ct->ctbs.send.lock);
+
+	if (unlikely(!h2g_has_room(ctb, len_dw))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
+
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+		else
+			return -EBUSY;
+	}
+
+	ct->stall_time = KTIME_MAX;
+	return 0;
+}
+
 static int ct_send_nb(struct intel_guc_ct *ct,
 		      const u32 *action,
 		      u32 len,
@@ -529,11 +569,9 @@ static int ct_send_nb(struct intel_guc_ct *ct,
 
 	spin_lock_irqsave(&ctb->lock, spin_flags);
 
-	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
-	if (unlikely(!ret)) {
-		ret = -EBUSY;
+	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
+	if (unlikely(ret))
 		goto out;
-	}
 
 	fence = ct_get_next_fence(ct);
 	ret = ct_write(ct, action, len, fence, flags);
@@ -576,8 +614,13 @@ static int ct_send(struct intel_guc_ct *ct,
 retry:
 	spin_lock_irqsave(&ctb->lock, flags);
 	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
 		spin_unlock_irqrestore(&ctb->lock, flags);
 
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+
 		if (msleep_interruptible(sleep_period_ms))
 			return -EINTR;
 		sleep_period_ms = sleep_period_ms << 1;
@@ -585,6 +628,8 @@ static int ct_send(struct intel_guc_ct *ct,
 		goto retry;
 	}
 
+	ct->stall_time = KTIME_MAX;
+
 	fence = ct_get_next_fence(ct);
 	request.fence = fence;
 	request.status = 0;
@@ -647,6 +692,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
 		return -ENODEV;
 	}
 
+	if (unlikely(ct->ctbs.send.broken))
+		return -EPIPE;
+
 	if (flags & INTEL_GUC_CT_SEND_NB)
 		return ct_send_nb(ct, action, len, flags);
 
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index 5bb8bef024c8..bee03794c1eb 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -9,6 +9,7 @@
 #include <linux/interrupt.h>
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
+#include <linux/ktime.h>
 
 #include "intel_guc_fwif.h"
 
@@ -68,6 +69,9 @@ struct intel_guc_ct {
 		struct list_head incoming; /* incoming requests */
 		struct work_struct worker; /* handler for incoming requests */
 	} requests;
+
+	/** @stall_time: time of first time a CTB submission is stalled */
+	ktime_t stall_time;
 };
 
 void intel_guc_ct_init_early(struct intel_guc_ct *ct);
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[John Harrison]

On 7/1/2021 10:15, Matthew Brost wrote:
> Implement a stall timer which fails H2G CTBs once a period of time
> with no forward progress is reached to prevent deadlock.
>
> v2:
>   (Michal)
>    - Improve error message in ct_deadlock()
>    - Set broken when ct_deadlock() returns true
>    - Return -EPIPE on ct_deadlock()
> v3:
>   (Michal)
>    - Add ms to stall timer comment
>   (Matthew)
>    - Move broken check to intel_guc_ct_send()
>
> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Looks plausible to me.

Reviewed-by: John Harrison <John.C.Harrison@Intel.com>

> ---
>   drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
>   drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
>   2 files changed, 59 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> index fb825cc1d090..a9cb7b608520 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> @@ -4,6 +4,9 @@
>    */
>   
>   #include <linux/circ_buf.h>
> +#include <linux/ktime.h>
> +#include <linux/time64.h>
> +#include <linux/timekeeping.h>
>   
>   #include "i915_drv.h"
>   #include "intel_guc_ct.h"
> @@ -316,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
>   		goto err_deregister;
>   
>   	ct->enabled = true;
> +	ct->stall_time = KTIME_MAX;
>   
>   	return 0;
>   
> @@ -388,9 +392,6 @@ static int ct_write(struct intel_guc_ct *ct,
>   	u32 *cmds = ctb->cmds;
>   	unsigned int i;
>   
> -	if (unlikely(ctb->broken))
> -		return -EPIPE;
> -
>   	if (unlikely(desc->status))
>   		goto corrupted;
>   
> @@ -506,6 +507,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
>   	return err;
>   }
>   
> +#define GUC_CTB_TIMEOUT_MS	1500
> +static inline bool ct_deadlocked(struct intel_guc_ct *ct)
> +{
> +	long timeout = GUC_CTB_TIMEOUT_MS;
> +	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
> +
> +	if (unlikely(ret)) {
> +		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
> +		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
> +
> +		CT_ERROR(ct, "Communication stalled for %lld ms, desc status=%#x,%#x\n",
> +			 ktime_ms_delta(ktime_get(), ct->stall_time),
> +			 send->status, recv->status);
> +		ct->ctbs.send.broken = true;
> +	}
> +
> +	return ret;
> +}
> +
>   static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
>   {
>   	struct guc_ct_buffer_desc *desc = ctb->desc;
> @@ -517,6 +537,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
>   	return space >= len_dw;
>   }
>   
> +static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
> +{
> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> +
> +	lockdep_assert_held(&ct->ctbs.send.lock);
> +
> +	if (unlikely(!h2g_has_room(ctb, len_dw))) {
> +		if (ct->stall_time == KTIME_MAX)
> +			ct->stall_time = ktime_get();
> +
> +		if (unlikely(ct_deadlocked(ct)))
> +			return -EPIPE;
> +		else
> +			return -EBUSY;
> +	}
> +
> +	ct->stall_time = KTIME_MAX;
> +	return 0;
> +}
> +
>   static int ct_send_nb(struct intel_guc_ct *ct,
>   		      const u32 *action,
>   		      u32 len,
> @@ -529,11 +569,9 @@ static int ct_send_nb(struct intel_guc_ct *ct,
>   
>   	spin_lock_irqsave(&ctb->lock, spin_flags);
>   
> -	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
> -	if (unlikely(!ret)) {
> -		ret = -EBUSY;
> +	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
> +	if (unlikely(ret))
>   		goto out;
> -	}
>   
>   	fence = ct_get_next_fence(ct);
>   	ret = ct_write(ct, action, len, fence, flags);
> @@ -576,8 +614,13 @@ static int ct_send(struct intel_guc_ct *ct,
>   retry:
>   	spin_lock_irqsave(&ctb->lock, flags);
>   	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
> +		if (ct->stall_time == KTIME_MAX)
> +			ct->stall_time = ktime_get();
>   		spin_unlock_irqrestore(&ctb->lock, flags);
>   
> +		if (unlikely(ct_deadlocked(ct)))
> +			return -EPIPE;
> +
>   		if (msleep_interruptible(sleep_period_ms))
>   			return -EINTR;
>   		sleep_period_ms = sleep_period_ms << 1;
> @@ -585,6 +628,8 @@ static int ct_send(struct intel_guc_ct *ct,
>   		goto retry;
>   	}
>   
> +	ct->stall_time = KTIME_MAX;
> +
>   	fence = ct_get_next_fence(ct);
>   	request.fence = fence;
>   	request.status = 0;
> @@ -647,6 +692,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
>   		return -ENODEV;
>   	}
>   
> +	if (unlikely(ct->ctbs.send.broken))
> +		return -EPIPE;
> +
>   	if (flags & INTEL_GUC_CT_SEND_NB)
>   		return ct_send_nb(ct, action, len, flags);
>   
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> index 5bb8bef024c8..bee03794c1eb 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> @@ -9,6 +9,7 @@
>   #include <linux/interrupt.h>
>   #include <linux/spinlock.h>
>   #include <linux/workqueue.h>
> +#include <linux/ktime.h>
>   
>   #include "intel_guc_fwif.h"
>   
> @@ -68,6 +69,9 @@ struct intel_guc_ct {
>   		struct list_head incoming; /* incoming requests */
>   		struct work_struct worker; /* handler for incoming requests */
>   	} requests;
> +
> +	/** @stall_time: time of first time a CTB submission is stalled */
> +	ktime_t stall_time;
>   };
>   
>   void intel_guc_ct_init_early(struct intel_guc_ct *ct);

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 0/7] CT changes required for GuC submission

[Matthew Brost]

As part of enabling GuC submission discussed in [1], [2], and [3] we
need optimize and update the CT code as this is now in the critical
path of submission. This series includes the patches to do that which is
the first 7 patches from [3]. The patches should have addressed all the
feedback in [3] and should be ready to merge once CI returns a we get a
few more RBs.

v2: Fix checkpatch warning, address a couple of Michal's comments
v3: Address John Harrison's comments

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

[1] https://patchwork.freedesktop.org/series/89844/
[2] https://patchwork.freedesktop.org/series/91417/
[3] https://patchwork.freedesktop.org/series/91840/

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

John Harrison (1):
  drm/i915/guc: Module load failure test for CT buffer creation

Matthew Brost (6):
  drm/i915/guc: Relax CTB response timeout
  drm/i915/guc: Improve error message for unsolicited CT response
  drm/i915/guc: Increase size of CTB buffers
  drm/i915/guc: Add non blocking CTB send function
  drm/i915/guc: Add stall timer to non blocking CTB send function
  drm/i915/guc: Optimize CTB writes and reads

 .../gt/uc/abi/guc_communication_ctb_abi.h     |   3 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.h        |  11 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 249 +++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  14 +-
 4 files changed, 231 insertions(+), 46 deletions(-)

-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Matthew Brost]

Implement a stall timer which fails H2G CTBs once a period of time
with no forward progress is reached to prevent deadlock.

v2:
 (Michal)
  - Improve error message in ct_deadlock()
  - Set broken when ct_deadlock() returns true
  - Return -EPIPE on ct_deadlock()
v3:
 (Michal)
  - Add ms to stall timer comment
 (Matthew)
  - Move broken check to intel_guc_ct_send()

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
 2 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 3d6cba8d91ad..db3e85b89573 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -4,6 +4,9 @@
  */
 
 #include <linux/circ_buf.h>
+#include <linux/ktime.h>
+#include <linux/time64.h>
+#include <linux/timekeeping.h>
 
 #include "i915_drv.h"
 #include "intel_guc_ct.h"
@@ -316,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
 		goto err_deregister;
 
 	ct->enabled = true;
+	ct->stall_time = KTIME_MAX;
 
 	return 0;
 
@@ -389,9 +393,6 @@ static int ct_write(struct intel_guc_ct *ct,
 	u32 *cmds = ctb->cmds;
 	unsigned int i;
 
-	if (unlikely(ctb->broken))
-		return -EPIPE;
-
 	if (unlikely(desc->status))
 		goto corrupted;
 
@@ -505,6 +506,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	return err;
 }
 
+#define GUC_CTB_TIMEOUT_MS	1500
+static inline bool ct_deadlocked(struct intel_guc_ct *ct)
+{
+	long timeout = GUC_CTB_TIMEOUT_MS;
+	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
+
+	if (unlikely(ret)) {
+		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
+		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
+
+		CT_ERROR(ct, "Communication stalled for %lld ms, desc status=%#x,%#x\n",
+			 ktime_ms_delta(ktime_get(), ct->stall_time),
+			 send->status, recv->status);
+		ct->ctbs.send.broken = true;
+	}
+
+	return ret;
+}
+
 static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 {
 	struct guc_ct_buffer_desc *desc = ctb->desc;
@@ -516,6 +536,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 	return space >= len_dw;
 }
 
+static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
+{
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+
+	lockdep_assert_held(&ct->ctbs.send.lock);
+
+	if (unlikely(!h2g_has_room(ctb, len_dw))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
+
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+		else
+			return -EBUSY;
+	}
+
+	ct->stall_time = KTIME_MAX;
+	return 0;
+}
+
 static int ct_send_nb(struct intel_guc_ct *ct,
 		      const u32 *action,
 		      u32 len,
@@ -528,11 +568,9 @@ static int ct_send_nb(struct intel_guc_ct *ct,
 
 	spin_lock_irqsave(&ctb->lock, spin_flags);
 
-	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
-	if (unlikely(!ret)) {
-		ret = -EBUSY;
+	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
+	if (unlikely(ret))
 		goto out;
-	}
 
 	fence = ct_get_next_fence(ct);
 	ret = ct_write(ct, action, len, fence, flags);
@@ -575,8 +613,13 @@ static int ct_send(struct intel_guc_ct *ct,
 retry:
 	spin_lock_irqsave(&ctb->lock, flags);
 	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
 		spin_unlock_irqrestore(&ctb->lock, flags);
 
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+
 		if (msleep_interruptible(sleep_period_ms))
 			return -EINTR;
 		sleep_period_ms = sleep_period_ms << 1;
@@ -584,6 +627,8 @@ static int ct_send(struct intel_guc_ct *ct,
 		goto retry;
 	}
 
+	ct->stall_time = KTIME_MAX;
+
 	fence = ct_get_next_fence(ct);
 	request.fence = fence;
 	request.status = 0;
@@ -646,6 +691,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
 		return -ENODEV;
 	}
 
+	if (unlikely(ct->ctbs.send.broken))
+		return -EPIPE;
+
 	if (flags & INTEL_GUC_CT_SEND_NB)
 		return ct_send_nb(ct, action, len, flags);
 
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index 5bb8bef024c8..bee03794c1eb 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -9,6 +9,7 @@
 #include <linux/interrupt.h>
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
+#include <linux/ktime.h>
 
 #include "intel_guc_fwif.h"
 
@@ -68,6 +69,9 @@ struct intel_guc_ct {
 		struct list_head incoming; /* incoming requests */
 		struct work_struct worker; /* handler for incoming requests */
 	} requests;
+
+	/** @stall_time: time of first time a CTB submission is stalled */
+	ktime_t stall_time;
 };
 
 void intel_guc_ct_init_early(struct intel_guc_ct *ct);
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 0/7] CT changes required for GuC submission

[Matthew Brost]

As part of enabling GuC submission discussed in [1], [2], and [3] we
need optimize and update the CT code as this is now in the critical
path of submission. This series includes the patches to do that which is
the first 7 patches from [3]. The patches should have addressed all the
feedback in [3] and should be ready to merge once CI returns a we get a
few more RBs.

v2: Fix checkpatch warning, address a couple of Michal's comments
v3: Address John Harrison's comments
v4: Address remaining comments, resend for patchworks to merge

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

[1] https://patchwork.freedesktop.org/series/89844/
[2] https://patchwork.freedesktop.org/series/91417/
[3] https://patchwork.freedesktop.org/series/91840/

Signed-off-by: Matthew Brost <matthew.brost@intel.com>

John Harrison (1):
  drm/i915/guc: Module load failure test for CT buffer creation

Matthew Brost (6):
  drm/i915/guc: Relax CTB response timeout
  drm/i915/guc: Improve error message for unsolicited CT response
  drm/i915/guc: Increase size of CTB buffers
  drm/i915/guc: Add non blocking CTB send function
  drm/i915/guc: Add stall timer to non blocking CTB send function
  drm/i915/guc: Optimize CTB writes and reads

 .../gt/uc/abi/guc_communication_ctb_abi.h     |   3 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc.h        |  11 +-
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c     | 256 +++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h     |  14 +-
 4 files changed, 234 insertions(+), 50 deletions(-)

-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH 5/7] drm/i915/guc: Add stall timer to non blocking CTB send function

[Matthew Brost]

Implement a stall timer which fails H2G CTBs once a period of time
with no forward progress is reached to prevent deadlock.

v2:
 (Michal)
  - Improve error message in ct_deadlock()
  - Set broken when ct_deadlock() returns true
  - Return -EPIPE on ct_deadlock()
v3:
 (Michal)
  - Add ms to stall timer comment
 (Matthew)
  - Move broken check to intel_guc_ct_send()

Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 62 ++++++++++++++++++++---
 drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  4 ++
 2 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
index 3d6cba8d91ad..db3e85b89573 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
@@ -4,6 +4,9 @@
  */
 
 #include <linux/circ_buf.h>
+#include <linux/ktime.h>
+#include <linux/time64.h>
+#include <linux/timekeeping.h>
 
 #include "i915_drv.h"
 #include "intel_guc_ct.h"
@@ -316,6 +319,7 @@ int intel_guc_ct_enable(struct intel_guc_ct *ct)
 		goto err_deregister;
 
 	ct->enabled = true;
+	ct->stall_time = KTIME_MAX;
 
 	return 0;
 
@@ -389,9 +393,6 @@ static int ct_write(struct intel_guc_ct *ct,
 	u32 *cmds = ctb->cmds;
 	unsigned int i;
 
-	if (unlikely(ctb->broken))
-		return -EPIPE;
-
 	if (unlikely(desc->status))
 		goto corrupted;
 
@@ -505,6 +506,25 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
 	return err;
 }
 
+#define GUC_CTB_TIMEOUT_MS	1500
+static inline bool ct_deadlocked(struct intel_guc_ct *ct)
+{
+	long timeout = GUC_CTB_TIMEOUT_MS;
+	bool ret = ktime_ms_delta(ktime_get(), ct->stall_time) > timeout;
+
+	if (unlikely(ret)) {
+		struct guc_ct_buffer_desc *send = ct->ctbs.send.desc;
+		struct guc_ct_buffer_desc *recv = ct->ctbs.send.desc;
+
+		CT_ERROR(ct, "Communication stalled for %lld ms, desc status=%#x,%#x\n",
+			 ktime_ms_delta(ktime_get(), ct->stall_time),
+			 send->status, recv->status);
+		ct->ctbs.send.broken = true;
+	}
+
+	return ret;
+}
+
 static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 {
 	struct guc_ct_buffer_desc *desc = ctb->desc;
@@ -516,6 +536,26 @@ static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
 	return space >= len_dw;
 }
 
+static int has_room_nb(struct intel_guc_ct *ct, u32 len_dw)
+{
+	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
+
+	lockdep_assert_held(&ct->ctbs.send.lock);
+
+	if (unlikely(!h2g_has_room(ctb, len_dw))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
+
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+		else
+			return -EBUSY;
+	}
+
+	ct->stall_time = KTIME_MAX;
+	return 0;
+}
+
 static int ct_send_nb(struct intel_guc_ct *ct,
 		      const u32 *action,
 		      u32 len,
@@ -528,11 +568,9 @@ static int ct_send_nb(struct intel_guc_ct *ct,
 
 	spin_lock_irqsave(&ctb->lock, spin_flags);
 
-	ret = h2g_has_room(ctb, len + GUC_CTB_HDR_LEN);
-	if (unlikely(!ret)) {
-		ret = -EBUSY;
+	ret = has_room_nb(ct, len + GUC_CTB_HDR_LEN);
+	if (unlikely(ret))
 		goto out;
-	}
 
 	fence = ct_get_next_fence(ct);
 	ret = ct_write(ct, action, len, fence, flags);
@@ -575,8 +613,13 @@ static int ct_send(struct intel_guc_ct *ct,
 retry:
 	spin_lock_irqsave(&ctb->lock, flags);
 	if (unlikely(!h2g_has_room(ctb, len + GUC_CTB_HDR_LEN))) {
+		if (ct->stall_time == KTIME_MAX)
+			ct->stall_time = ktime_get();
 		spin_unlock_irqrestore(&ctb->lock, flags);
 
+		if (unlikely(ct_deadlocked(ct)))
+			return -EPIPE;
+
 		if (msleep_interruptible(sleep_period_ms))
 			return -EINTR;
 		sleep_period_ms = sleep_period_ms << 1;
@@ -584,6 +627,8 @@ static int ct_send(struct intel_guc_ct *ct,
 		goto retry;
 	}
 
+	ct->stall_time = KTIME_MAX;
+
 	fence = ct_get_next_fence(ct);
 	request.fence = fence;
 	request.status = 0;
@@ -646,6 +691,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
 		return -ENODEV;
 	}
 
+	if (unlikely(ct->ctbs.send.broken))
+		return -EPIPE;
+
 	if (flags & INTEL_GUC_CT_SEND_NB)
 		return ct_send_nb(ct, action, len, flags);
 
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
index 5bb8bef024c8..bee03794c1eb 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
@@ -9,6 +9,7 @@
 #include <linux/interrupt.h>
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
+#include <linux/ktime.h>
 
 #include "intel_guc_fwif.h"
 
@@ -68,6 +69,9 @@ struct intel_guc_ct {
 		struct list_head incoming; /* incoming requests */
 		struct work_struct worker; /* handler for incoming requests */
 	} requests;
+
+	/** @stall_time: time of first time a CTB submission is stalled */
+	ktime_t stall_time;
 };
 
 void intel_guc_ct_init_early(struct intel_guc_ct *ct);
-- 
2.28.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx