From d69300abc2dec930fd280a013984150ae7c6a386 Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Wed, 28 Aug 2024 23:08:08 +0200
Subject: [PATCH 01/16] drm/xe/pf: Drop GuC notifications for non-existing VF
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

It is unlikely that GuC will ever send a G2H notification with an
invalid VFID and it is currently harmless if that actually happen.
But in upcoming patches we will start using that VFID as an index
and we must be sure it is a valid to avoid a crash due to a buggy
firmware or a currupted G2H message.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240828210809.1528-4-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
index d11839d9e7d4..ad447d867e51 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
@@ -8,6 +8,7 @@
 #include "xe_device.h"
 #include "xe_gt.h"
 #include "xe_gt_sriov_pf_control.h"
+#include "xe_gt_sriov_pf_helpers.h"
 #include "xe_gt_sriov_printk.h"
 #include "xe_guc_ct.h"
 #include "xe_sriov.h"
@@ -212,6 +213,11 @@ static void pf_handle_vf_flr_done(struct xe_gt *gt, u32 vfid)
 
 static int pf_handle_vf_event(struct xe_gt *gt, u32 vfid, u32 eventid)
 {
+	xe_gt_sriov_dbg_verbose(gt, "received VF%u event %#x\n", vfid, eventid);
+
+	if (vfid > xe_gt_sriov_pf_get_totalvfs(gt))
+		return -EPROTO;
+
 	switch (eventid) {
 	case GUC_PF_NOTIFY_VF_FLR:
 		pf_handle_vf_flr(gt, vfid);
-- 
2.51.0


From 2bd87f0fc24ae29ead9808d341497f465b339c0a Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Wed, 28 Aug 2024 23:08:09 +0200
Subject: [PATCH 02/16] drm/xe/pf: Improve VF control
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

Our initial VF control implementation was focused on providing
a very minimal support for the VF_STATE_NOTIFY events just to
meet GuC requirements, without tracking a VF state or doing any
expected actions (like cleanup in case of the FLR notification).

Try to improve this by defining set of VF state machines, each
responsible for processing one activity (PAUSE, RESUME, STOP or
FLR). All required steps defined by the VF state machine are then
executed by the PF worker from the dedicated workqueue.

Any external requests or notifications simply try to transition
between the states to trigger a work and then wait for that work
to finish. Some predefined default timeouts are used to avoid
changing existing API calls, but it should be easy to extend the
control API to also accept specific timeout values.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Piotr Piórkowski <piotr.piorkowski@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240828210809.1528-5-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_gt_sriov_pf.c           |    6 +
 drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c   | 1198 ++++++++++++++++-
 drivers/gpu/drm/xe/xe_gt_sriov_pf_control.h   |    3 +
 .../gpu/drm/xe/xe_gt_sriov_pf_control_types.h |  107 ++
 drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h     |    6 +
 5 files changed, 1305 insertions(+), 15 deletions(-)
 create mode 100644 drivers/gpu/drm/xe/xe_gt_sriov_pf_control_types.h

diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
index ef239440963c..905f409db74b 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
@@ -9,6 +9,7 @@
 
 #include "xe_gt_sriov_pf.h"
 #include "xe_gt_sriov_pf_config.h"
+#include "xe_gt_sriov_pf_control.h"
 #include "xe_gt_sriov_pf_helpers.h"
 #include "xe_gt_sriov_pf_service.h"
 #include "xe_mmio.h"
@@ -57,6 +58,10 @@ int xe_gt_sriov_pf_init_early(struct xe_gt *gt)
 	if (err)
 		return err;
 
+	err = xe_gt_sriov_pf_control_init(gt);
+	if (err)
+		return err;
+
 	return 0;
 }
 
@@ -93,4 +98,5 @@ void xe_gt_sriov_pf_init_hw(struct xe_gt *gt)
 void xe_gt_sriov_pf_restart(struct xe_gt *gt)
 {
 	xe_gt_sriov_pf_config_restart(gt);
+	xe_gt_sriov_pf_control_restart(gt);
 }
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
index ad447d867e51..02f7328bd6ce 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
@@ -3,12 +3,17 @@
  * Copyright © 2023-2024 Intel Corporation
  */
 
+#include <drm/drm_managed.h>
+
 #include "abi/guc_actions_sriov_abi.h"
 
 #include "xe_device.h"
 #include "xe_gt.h"
+#include "xe_gt_sriov_pf_config.h"
 #include "xe_gt_sriov_pf_control.h"
 #include "xe_gt_sriov_pf_helpers.h"
+#include "xe_gt_sriov_pf_monitor.h"
+#include "xe_gt_sriov_pf_service.h"
 #include "xe_gt_sriov_printk.h"
 #include "xe_guc_ct.h"
 #include "xe_sriov.h"
@@ -42,10 +47,6 @@ static int guc_action_vf_control_cmd(struct xe_guc *guc, u32 vfid, u32 cmd)
 	};
 	int ret;
 
-	/* XXX those two commands are now sent from the G2H handler */
-	if (cmd == GUC_PF_TRIGGER_VF_FLR_START || cmd == GUC_PF_TRIGGER_VF_FLR_FINISH)
-		return xe_guc_ct_send_g2h_handler(&guc->ct, request, ARRAY_SIZE(request));
-
 	ret = xe_guc_ct_send_block(&guc->ct, request, ARRAY_SIZE(request));
 	return ret > 0 ? -EPROTO : ret;
 }
@@ -55,6 +56,8 @@ static int pf_send_vf_control_cmd(struct xe_gt *gt, unsigned int vfid, u32 cmd)
 	int err;
 
 	xe_gt_assert(gt, vfid != PFID);
+	xe_gt_sriov_dbg_verbose(gt, "sending VF%u control command %s\n",
+				vfid, control_cmd_to_string(cmd));
 
 	err = guc_action_vf_control_cmd(&gt->uc.guc, vfid, cmd);
 	if (unlikely(err))
@@ -88,6 +91,456 @@ static int pf_send_vf_flr_finish(struct xe_gt *gt, unsigned int vfid)
 	return pf_send_vf_control_cmd(gt, vfid, GUC_PF_TRIGGER_VF_FLR_FINISH);
 }
 
+/**
+ * DOC: The VF state machine
+ *
+ * The simplified VF state machine could be presented as::
+ *
+ *	               pause--------------------------o
+ *	              /                               |
+ *	             /                                v
+ *	      (READY)<------------------resume-----(PAUSED)
+ *	         ^   \                             /    /
+ *	         |    \                           /    /
+ *	         |     stop---->(STOPPED)<----stop    /
+ *	         |                  /                /
+ *	         |                 /                /
+ *	         o--------<-----flr                /
+ *	          \                               /
+ *	           o------<--------------------flr
+ *
+ * Where:
+ *
+ * * READY - represents a state in which VF is fully operable
+ * * PAUSED - represents a state in which VF activity is temporarily suspended
+ * * STOPPED - represents a state in which VF activity is definitely halted
+ * * pause - represents a request to temporarily suspend VF activity
+ * * resume - represents a request to resume VF activity
+ * * stop - represents a request to definitely halt VF activity
+ * * flr - represents a request to perform VF FLR to restore VF activity
+ *
+ * However, each state transition requires additional steps that involves
+ * communication with GuC that might fail or be interrupted by other requests::
+ *
+ *	                   .................................WIP....
+ *	                   :                                      :
+ *	          pause--------------------->PAUSE_WIP----------------------------o
+ *	         /         :                /         \           :               |
+ *	        /          :    o----<---stop          flr--o     :               |
+ *	       /           :    |           \         /     |     :               V
+ *	(READY,RESUMED)<--------+------------RESUME_WIP<----+--<-----resume--(PAUSED)
+ *	  ^ \  \           :    |                           |     :          /   /
+ *	  |  \  \          :    |                           |     :         /   /
+ *	  |   \  \         :    |                           |     :        /   /
+ *	  |    \  \        :    o----<----------------------+--<-------stop   /
+ *	  |     \  \       :    |                           |     :          /
+ *	  |      \  \      :    V                           |     :         /
+ *	  |       \  stop----->STOP_WIP---------flr--->-----o     :        /
+ *	  |        \       :    |                           |     :       /
+ *	  |         \      :    |                           V     :      /
+ *	  |          flr--------+----->----------------->FLR_WIP<-----flr
+ *	  |                :    |                        /  ^     :
+ *	  |                :    |                       /   |     :
+ *	  o--------<-------:----+-----<----------------o    |     :
+ *	                   :    |                           |     :
+ *	                   :....|...........................|.....:
+ *	                        |                           |
+ *	                        V                           |
+ *	                     (STOPPED)--------------------flr
+ *
+ * For details about each internal WIP state machine see:
+ *
+ * * `The VF PAUSE state machine`_
+ * * `The VF RESUME state machine`_
+ * * `The VF STOP state machine`_
+ * * `The VF FLR state machine`_
+ */
+
+#ifdef CONFIG_DRM_XE_DEBUG_SRIOV
+static const char *control_bit_to_string(enum xe_gt_sriov_control_bits bit)
+{
+	switch (bit) {
+#define CASE2STR(_X) \
+	case XE_GT_SRIOV_STATE_##_X: return #_X
+	CASE2STR(WIP);
+	CASE2STR(FLR_WIP);
+	CASE2STR(FLR_SEND_START);
+	CASE2STR(FLR_WAIT_GUC);
+	CASE2STR(FLR_GUC_DONE);
+	CASE2STR(FLR_RESET_CONFIG);
+	CASE2STR(FLR_RESET_DATA);
+	CASE2STR(FLR_RESET_MMIO);
+	CASE2STR(FLR_SEND_FINISH);
+	CASE2STR(FLR_FAILED);
+	CASE2STR(PAUSE_WIP);
+	CASE2STR(PAUSE_SEND_PAUSE);
+	CASE2STR(PAUSE_WAIT_GUC);
+	CASE2STR(PAUSE_GUC_DONE);
+	CASE2STR(PAUSE_FAILED);
+	CASE2STR(PAUSED);
+	CASE2STR(RESUME_WIP);
+	CASE2STR(RESUME_SEND_RESUME);
+	CASE2STR(RESUME_FAILED);
+	CASE2STR(RESUMED);
+	CASE2STR(STOP_WIP);
+	CASE2STR(STOP_SEND_STOP);
+	CASE2STR(STOP_FAILED);
+	CASE2STR(STOPPED);
+	CASE2STR(MISMATCH);
+#undef  CASE2STR
+	default: return "?";
+	}
+}
+#endif
+
+static unsigned long pf_get_default_timeout(enum xe_gt_sriov_control_bits bit)
+{
+	switch (bit) {
+	case XE_GT_SRIOV_STATE_FLR_WAIT_GUC:
+	case XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC:
+		return HZ / 2;
+	case XE_GT_SRIOV_STATE_FLR_WIP:
+	case XE_GT_SRIOV_STATE_FLR_RESET_CONFIG:
+		return 5 * HZ;
+	default:
+		return HZ;
+	}
+}
+
+static struct xe_gt_sriov_control_state *pf_pick_vf_control(struct xe_gt *gt, unsigned int vfid)
+{
+	xe_gt_assert(gt, IS_SRIOV_PF(gt_to_xe(gt)));
+	xe_gt_assert(gt, vfid <= xe_gt_sriov_pf_get_totalvfs(gt));
+
+	return &gt->sriov.pf.vfs[vfid].control;
+}
+
+static unsigned long *pf_peek_vf_state(struct xe_gt *gt, unsigned int vfid)
+{
+	struct xe_gt_sriov_control_state *cs = pf_pick_vf_control(gt, vfid);
+
+	return &cs->state;
+}
+
+static bool pf_check_vf_state(struct xe_gt *gt, unsigned int vfid,
+			      enum xe_gt_sriov_control_bits bit)
+{
+	return test_bit(bit, pf_peek_vf_state(gt, vfid));
+}
+
+static void pf_dump_vf_state(struct xe_gt *gt, unsigned int vfid)
+{
+	unsigned long state = *pf_peek_vf_state(gt, vfid);
+	enum xe_gt_sriov_control_bits bit;
+
+	if (state) {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u state %#lx%s%*pbl\n",
+					vfid, state, state ? " bits " : "",
+					(int)BITS_PER_LONG, &state);
+		for_each_set_bit(bit, &state, BITS_PER_LONG)
+			xe_gt_sriov_dbg_verbose(gt, "VF%u state %s(%d)\n",
+						vfid, control_bit_to_string(bit), bit);
+	} else {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u state READY\n", vfid);
+	}
+}
+
+static bool pf_expect_vf_state(struct xe_gt *gt, unsigned int vfid,
+			       enum xe_gt_sriov_control_bits bit)
+{
+	bool result = pf_check_vf_state(gt, vfid, bit);
+
+	if (unlikely(!result))
+		pf_dump_vf_state(gt, vfid);
+
+	return result;
+}
+
+static bool pf_expect_vf_not_state(struct xe_gt *gt, unsigned int vfid,
+				   enum xe_gt_sriov_control_bits bit)
+{
+	bool result = !pf_check_vf_state(gt, vfid, bit);
+
+	if (unlikely(!result))
+		pf_dump_vf_state(gt, vfid);
+
+	return result;
+}
+
+static bool pf_enter_vf_state(struct xe_gt *gt, unsigned int vfid,
+			      enum xe_gt_sriov_control_bits bit)
+{
+	if (!test_and_set_bit(bit, pf_peek_vf_state(gt, vfid))) {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u state %s(%d) enter\n",
+					vfid, control_bit_to_string(bit), bit);
+		return true;
+	}
+	return false;
+}
+
+static bool pf_exit_vf_state(struct xe_gt *gt, unsigned int vfid,
+			     enum xe_gt_sriov_control_bits bit)
+{
+	if (test_and_clear_bit(bit, pf_peek_vf_state(gt, vfid))) {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u state %s(%d) exit\n",
+					vfid, control_bit_to_string(bit), bit);
+		return true;
+	}
+	return false;
+}
+
+static void pf_escape_vf_state(struct xe_gt *gt, unsigned int vfid,
+			       enum xe_gt_sriov_control_bits bit)
+{
+	if (pf_exit_vf_state(gt, vfid, bit))
+		xe_gt_sriov_dbg_verbose(gt, "VF%u state %s(%d) escaped by %ps\n",
+					vfid, control_bit_to_string(bit), bit,
+					__builtin_return_address(0));
+}
+
+static void pf_enter_vf_mismatch(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_MISMATCH)) {
+		xe_gt_sriov_dbg(gt, "VF%u state mismatch detected by %ps\n",
+				vfid, __builtin_return_address(0));
+		pf_dump_vf_state(gt, vfid);
+	}
+}
+
+static void pf_exit_vf_mismatch(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_MISMATCH))
+		xe_gt_sriov_dbg(gt, "VF%u state mismatch cleared by %ps\n",
+				vfid, __builtin_return_address(0));
+
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_FAILED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_FAILED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_FAILED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_FAILED);
+}
+
+#define pf_enter_vf_state_machine_bug(gt, vfid) ({	\
+	pf_enter_vf_mismatch((gt), (vfid));		\
+})
+
+static void pf_queue_control_worker(struct xe_gt *gt)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+
+	xe_gt_assert(gt, IS_SRIOV_PF(xe));
+
+	queue_work(xe->sriov.wq, &gt->sriov.pf.control.worker);
+}
+
+static void pf_queue_vf(struct xe_gt *gt, unsigned int vfid)
+{
+	struct xe_gt_sriov_pf_control *pfc = &gt->sriov.pf.control;
+
+	xe_gt_assert(gt, IS_SRIOV_PF(gt_to_xe(gt)));
+
+	spin_lock(&pfc->lock);
+	list_move_tail(&gt->sriov.pf.vfs[vfid].control.link, &pfc->list);
+	spin_unlock(&pfc->lock);
+
+	pf_queue_control_worker(gt);
+}
+
+static void pf_exit_vf_flr_wip(struct xe_gt *gt, unsigned int vfid);
+static void pf_exit_vf_stop_wip(struct xe_gt *gt, unsigned int vfid);
+static void pf_exit_vf_pause_wip(struct xe_gt *gt, unsigned int vfid);
+static void pf_exit_vf_resume_wip(struct xe_gt *gt, unsigned int vfid);
+
+static bool pf_enter_vf_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_WIP)) {
+		struct xe_gt_sriov_control_state *cs = pf_pick_vf_control(gt, vfid);
+
+		reinit_completion(&cs->done);
+		return true;
+	}
+	return false;
+}
+
+static void pf_exit_vf_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_WIP)) {
+		struct xe_gt_sriov_control_state *cs = pf_pick_vf_control(gt, vfid);
+
+		pf_exit_vf_flr_wip(gt, vfid);
+		pf_exit_vf_stop_wip(gt, vfid);
+		pf_exit_vf_pause_wip(gt, vfid);
+		pf_exit_vf_resume_wip(gt, vfid);
+
+		complete_all(&cs->done);
+	}
+}
+
+static int pf_wait_vf_wip_done(struct xe_gt *gt, unsigned int vfid, unsigned long timeout)
+{
+	struct xe_gt_sriov_control_state *cs = pf_pick_vf_control(gt, vfid);
+
+	return wait_for_completion_timeout(&cs->done, timeout) ? 0 : -ETIMEDOUT;
+}
+
+static void pf_enter_vf_ready(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOPPED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUMED);
+	pf_exit_vf_mismatch(gt, vfid);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+/**
+ * DOC: The VF PAUSE state machine
+ *
+ * The VF PAUSE state machine looks like::
+ *
+ *	 (READY,RESUMED)<-------------<---------------------o---------o
+ *	    |                                                \         \
+ *	   pause                                              \         \
+ *	    |                                                  \         \
+ *	....V...........................PAUSE_WIP........       \         \
+ *	:    \                                          :        o         \
+ *	:     \   o------<-----busy                     :        |          \
+ *	:      \ /              /                       :        |           |
+ *	:       PAUSE_SEND_PAUSE ---failed--->----------o--->(PAUSE_FAILED)  |
+ *	:        |              \                       :        |           |
+ *	:      acked             rejected---->----------o--->(MISMATCH)     /
+ *	:        |                                      :                  /
+ *	:        v                                      :                 /
+ *	:       PAUSE_WAIT_GUC                          :                /
+ *	:        |                                      :               /
+ *	:       done                                    :              /
+ *	:        |                                      :             /
+ *	:        v                                      :            /
+ *	:       PAUSE_GUC_DONE                          o-----restart
+ *	:      /                                        :
+ *	:     /                                         :
+ *	:....o..............o...............o...........:
+ *	     |              |               |
+ *	  completed        flr             stop
+ *	     |              |               |
+ *	     V         .....V.....    ......V.....
+ *	 (PAUSED)      : FLR_WIP :    : STOP_WIP :
+ *	               :.........:    :..........:
+ *
+ * For the full state machine view, see `The VF state machine`_.
+ */
+
+static void pf_exit_vf_pause_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WIP)) {
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_SEND_PAUSE);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_GUC_DONE);
+	}
+}
+
+static void pf_enter_vf_paused(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUMED);
+	pf_exit_vf_mismatch(gt, vfid);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_pause_completed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_paused(gt, vfid);
+}
+
+static void pf_enter_vf_pause_failed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_FAILED);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_pause_rejected(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_mismatch(gt, vfid);
+	pf_enter_vf_pause_failed(gt, vfid);
+}
+
+static bool pf_exit_vf_pause_guc_done(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_GUC_DONE))
+		return false;
+
+	pf_enter_vf_pause_completed(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_pause_guc_done(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_GUC_DONE))
+		pf_queue_vf(gt, vfid);
+}
+
+static void pf_enter_pause_wait_guc(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+}
+
+static bool pf_exit_pause_wait_guc(struct xe_gt *gt, unsigned int vfid)
+{
+	return pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC);
+}
+
+static void pf_enter_vf_pause_send_pause(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_SEND_PAUSE))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_pause_send_pause(struct xe_gt *gt, unsigned int vfid)
+{
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_SEND_PAUSE))
+		return false;
+
+	/* GuC may actually send a PAUSE_DONE before we get a RESPONSE */
+	pf_enter_pause_wait_guc(gt, vfid);
+
+	err = pf_send_vf_pause(gt, vfid);
+	if (err) {
+		/* send failed, so we shouldn't expect PAUSE_DONE from GuC */
+		pf_exit_pause_wait_guc(gt, vfid);
+
+		if (err == -EBUSY)
+			pf_enter_vf_pause_send_pause(gt, vfid);
+		else if (err == -EIO)
+			pf_enter_vf_pause_rejected(gt, vfid);
+		else
+			pf_enter_vf_pause_failed(gt, vfid);
+	} else {
+		/*
+		 * we have already moved to WAIT_GUC, maybe even to GUC_DONE
+		 * but since GuC didn't complain, we may clear MISMATCH
+		 */
+		pf_exit_vf_mismatch(gt, vfid);
+	}
+
+	return true;
+}
+
+static bool pf_enter_vf_pause_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WIP)) {
+		pf_enter_vf_wip(gt, vfid);
+		pf_enter_vf_pause_send_pause(gt, vfid);
+		return true;
+	}
+
+	return false;
+}
+
 /**
  * xe_gt_sriov_pf_control_pause_vf - Pause a VF.
  * @gt: the &xe_gt
@@ -99,7 +552,140 @@ static int pf_send_vf_flr_finish(struct xe_gt *gt, unsigned int vfid)
  */
 int xe_gt_sriov_pf_control_pause_vf(struct xe_gt *gt, unsigned int vfid)
 {
-	return pf_send_vf_pause(gt, vfid);
+	unsigned long timeout = pf_get_default_timeout(XE_GT_SRIOV_STATE_PAUSE_WIP);
+	int err;
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOPPED)) {
+		xe_gt_sriov_dbg(gt, "VF%u is stopped!\n", vfid);
+		return -EPERM;
+	}
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED)) {
+		xe_gt_sriov_dbg(gt, "VF%u was already paused!\n", vfid);
+		return -ESTALE;
+	}
+
+	if (!pf_enter_vf_pause_wip(gt, vfid)) {
+		xe_gt_sriov_dbg(gt, "VF%u pause already in progress!\n", vfid);
+		return -EALREADY;
+	}
+
+	err = pf_wait_vf_wip_done(gt, vfid, timeout);
+	if (err) {
+		xe_gt_sriov_dbg(gt, "VF%u pause didn't finish in %u ms (%pe)\n",
+				vfid, jiffies_to_msecs(timeout), ERR_PTR(err));
+		return err;
+	}
+
+	if (pf_expect_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED)) {
+		xe_gt_sriov_info(gt, "VF%u paused!\n", vfid);
+		return 0;
+	}
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_FAILED)) {
+		xe_gt_sriov_dbg(gt, "VF%u pause failed!\n", vfid);
+		return -EIO;
+	}
+
+	xe_gt_sriov_dbg(gt, "VF%u pause was canceled!\n", vfid);
+	return -ECANCELED;
+}
+
+/**
+ * DOC: The VF RESUME state machine
+ *
+ * The VF RESUME state machine looks like::
+ *
+ *	 (PAUSED)<-----------------<------------------------o
+ *	    |                                                \
+ *	   resume                                             \
+ *	    |                                                  \
+ *	....V............................RESUME_WIP......       \
+ *	:    \                                          :        o
+ *	:     \   o-------<-----busy                    :        |
+ *	:      \ /                /                     :        |
+ *	:       RESUME_SEND_RESUME ---failed--->--------o--->(RESUME_FAILED)
+ *	:       /                \                      :        |
+ *	:    acked                rejected---->---------o--->(MISMATCH)
+ *	:     /                                         :
+ *	:....o..............o...............o.....o.....:
+ *	     |              |               |      \
+ *	  completed        flr            stop      restart-->(READY)
+ *	     |              |               |
+ *	     V         .....V.....    ......V.....
+ *	 (RESUMED)     : FLR_WIP :    : STOP_WIP :
+ *	               :.........:    :..........:
+ *
+ * For the full state machine view, see `The VF state machine`_.
+ */
+
+static void pf_exit_vf_resume_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_WIP))
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_SEND_RESUME);
+}
+
+static void pf_enter_vf_resumed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUMED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED);
+	pf_exit_vf_mismatch(gt, vfid);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_resume_completed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_resumed(gt, vfid);
+}
+
+static void pf_enter_vf_resume_failed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_FAILED);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_resume_rejected(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_mismatch(gt, vfid);
+	pf_enter_vf_resume_failed(gt, vfid);
+}
+
+static void pf_enter_vf_resume_send_resume(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_SEND_RESUME))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_resume_send_resume(struct xe_gt *gt, unsigned int vfid)
+{
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_SEND_RESUME))
+		return false;
+
+	err = pf_send_vf_resume(gt, vfid);
+	if (err == -EBUSY)
+		pf_enter_vf_resume_send_resume(gt, vfid);
+	else if (err == -EIO)
+		pf_enter_vf_resume_rejected(gt, vfid);
+	else if (err)
+		pf_enter_vf_resume_failed(gt, vfid);
+	else
+		pf_enter_vf_resume_completed(gt, vfid);
+	return true;
+}
+
+static bool pf_enter_vf_resume_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_WIP)) {
+		pf_enter_vf_wip(gt, vfid);
+		pf_enter_vf_resume_send_resume(gt, vfid);
+		return true;
+	}
+
+	return false;
 }
 
 /**
@@ -113,7 +699,134 @@ int xe_gt_sriov_pf_control_pause_vf(struct xe_gt *gt, unsigned int vfid)
  */
 int xe_gt_sriov_pf_control_resume_vf(struct xe_gt *gt, unsigned int vfid)
 {
-	return pf_send_vf_resume(gt, vfid);
+	unsigned long timeout = pf_get_default_timeout(XE_GT_SRIOV_STATE_RESUME_WIP);
+	int err;
+
+	if (!pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED)) {
+		xe_gt_sriov_dbg(gt, "VF%u is not paused!\n", vfid);
+		return -EPERM;
+	}
+
+	if (!pf_enter_vf_resume_wip(gt, vfid)) {
+		xe_gt_sriov_dbg(gt, "VF%u resume already in progress!\n", vfid);
+		return -EALREADY;
+	}
+
+	err = pf_wait_vf_wip_done(gt, vfid, timeout);
+	if (err)
+		return err;
+
+	if (pf_expect_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUMED)) {
+		xe_gt_sriov_info(gt, "VF%u resumed!\n", vfid);
+		return 0;
+	}
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUME_FAILED)) {
+		xe_gt_sriov_dbg(gt, "VF%u resume failed!\n", vfid);
+		return -EIO;
+	}
+
+	xe_gt_sriov_dbg(gt, "VF%u resume was canceled!\n", vfid);
+	return -ECANCELED;
+}
+
+/**
+ * DOC: The VF STOP state machine
+ *
+ * The VF STOP state machine looks like::
+ *
+ *	 (READY,PAUSED,RESUMED)<-------<--------------------o
+ *	    |                                                \
+ *	   stop                                               \
+ *	    |                                                  \
+ *	....V..............................STOP_WIP......       \
+ *	:    \                                          :        o
+ *	:     \   o----<----busy                        :        |
+ *	:      \ /            /                         :        |
+ *	:       STOP_SEND_STOP--------failed--->--------o--->(STOP_FAILED)
+ *	:       /             \                         :        |
+ *	:    acked             rejected-------->--------o--->(MISMATCH)
+ *	:     /                                         :
+ *	:....o..............o...............o...........:
+ *	     |              |               |
+ *	  completed        flr            restart
+ *	     |              |               |
+ *	     V         .....V.....          V
+ *	 (STOPPED)     : FLR_WIP :       (READY)
+ *	               :.........:
+ *
+ * For the full state machine view, see `The VF state machine`_.
+ */
+
+static void pf_exit_vf_stop_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_WIP))
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_SEND_STOP);
+}
+
+static void pf_enter_vf_stopped(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOPPED))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_RESUMED);
+	pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSED);
+	pf_exit_vf_mismatch(gt, vfid);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_stop_completed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_stopped(gt, vfid);
+}
+
+static void pf_enter_vf_stop_failed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_FAILED);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_stop_rejected(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_mismatch(gt, vfid);
+	pf_enter_vf_stop_failed(gt, vfid);
+}
+
+static void pf_enter_vf_stop_send_stop(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_SEND_STOP))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_stop_send_stop(struct xe_gt *gt, unsigned int vfid)
+{
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_SEND_STOP))
+		return false;
+
+	err = pf_send_vf_stop(gt, vfid);
+	if (err == -EBUSY)
+		pf_enter_vf_stop_send_stop(gt, vfid);
+	else if (err == -EIO)
+		pf_enter_vf_stop_rejected(gt, vfid);
+	else if (err)
+		pf_enter_vf_stop_failed(gt, vfid);
+	else
+		pf_enter_vf_stop_completed(gt, vfid);
+	return true;
+}
+
+static bool pf_enter_vf_stop_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_WIP)) {
+		pf_enter_vf_wip(gt, vfid);
+		pf_enter_vf_stop_send_stop(gt, vfid);
+		return true;
+	}
+	return false;
 }
 
 /**
@@ -127,7 +840,280 @@ int xe_gt_sriov_pf_control_resume_vf(struct xe_gt *gt, unsigned int vfid)
  */
 int xe_gt_sriov_pf_control_stop_vf(struct xe_gt *gt, unsigned int vfid)
 {
-	return pf_send_vf_stop(gt, vfid);
+	unsigned long timeout = pf_get_default_timeout(XE_GT_SRIOV_STATE_STOP_WIP);
+	int err;
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOPPED)) {
+		xe_gt_sriov_dbg(gt, "VF%u was already stopped!\n", vfid);
+		return -ESTALE;
+	}
+
+	if (!pf_enter_vf_stop_wip(gt, vfid)) {
+		xe_gt_sriov_dbg(gt, "VF%u stop already in progress!\n", vfid);
+		return -EALREADY;
+	}
+
+	err = pf_wait_vf_wip_done(gt, vfid, timeout);
+	if (err)
+		return err;
+
+	if (pf_expect_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOPPED)) {
+		xe_gt_sriov_info(gt, "VF%u stopped!\n", vfid);
+		return 0;
+	}
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_STOP_FAILED)) {
+		xe_gt_sriov_dbg(gt, "VF%u stop failed!\n", vfid);
+		return -EIO;
+	}
+
+	xe_gt_sriov_dbg(gt, "VF%u stop was canceled!\n", vfid);
+	return -ECANCELED;
+}
+
+/**
+ * DOC: The VF FLR state machine
+ *
+ * The VF FLR state machine looks like::
+ *
+ *	 (READY,PAUSED,STOPPED)<------------<--------------o
+ *	    |                                               \
+ *	   flr                                               \
+ *	    |                                                 \
+ *	....V..........................FLR_WIP...........      \
+ *	:    \                                          :       \
+ *	:     \   o----<----busy                        :        |
+ *	:      \ /            /                         :        |
+ *	:       FLR_SEND_START---failed----->-----------o--->(FLR_FAILED)<---o
+ *	:        |            \                         :        |           |
+ *	:      acked           rejected----->-----------o--->(MISMATCH)      |
+ *	:        |                                      :        ^           |
+ *	:        v                                      :        |           |
+ *	:       FLR_WAIT_GUC                            :        |           |
+ *	:        |                                      :        |           |
+ *	:       done                                    :        |           |
+ *	:        |                                      :        |           |
+ *	:        v                                      :        |           |
+ *	:       FLR_GUC_DONE                            :        |           |
+ *	:        |                                      :        |           |
+ *	:       FLR_RESET_CONFIG---failed--->-----------o--------+-----------o
+ *	:        |                                      :        |           |
+ *	:       FLR_RESET_DATA                          :        |           |
+ *	:        |                                      :        |           |
+ *	:       FLR_RESET_MMIO                          :        |           |
+ *	:        |                                      :        |           |
+ *	:        | o----<----busy                       :        |           |
+ *	:        |/            /                        :        |           |
+ *	:       FLR_SEND_FINISH----failed--->-----------o--------+-----------o
+ *	:       /             \                         :        |
+ *	:     acked            rejected----->-----------o--------o
+ *	:     /                                         :
+ *	:....o..............................o...........:
+ *	     |                              |
+ *	  completed                       restart
+ *	     |                             /
+ *	     V                            /
+ *	  (READY)<----------<------------o
+ *
+ * For the full state machine view, see `The VF state machine`_.
+ */
+
+static void pf_enter_vf_flr_send_start(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_START))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static void pf_enter_vf_flr_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WIP)) {
+		xe_gt_sriov_dbg(gt, "VF%u FLR is already in progress\n", vfid);
+		return;
+	}
+
+	pf_enter_vf_wip(gt, vfid);
+	pf_enter_vf_flr_send_start(gt, vfid);
+}
+
+static void pf_exit_vf_flr_wip(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WIP)) {
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_FINISH);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_MMIO);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_DATA);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_CONFIG);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_GUC_DONE);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WAIT_GUC);
+		pf_escape_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_START);
+	}
+}
+
+static void pf_enter_vf_flr_completed(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_ready(gt, vfid);
+}
+
+static void pf_enter_vf_flr_failed(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_FAILED))
+		xe_gt_sriov_notice(gt, "VF%u FLR failed!\n", vfid);
+	pf_exit_vf_wip(gt, vfid);
+}
+
+static void pf_enter_vf_flr_rejected(struct xe_gt *gt, unsigned int vfid)
+{
+	pf_enter_vf_mismatch(gt, vfid);
+	pf_enter_vf_flr_failed(gt, vfid);
+}
+
+static void pf_enter_vf_flr_send_finish(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_FINISH))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_flr_send_finish(struct xe_gt *gt, unsigned int vfid)
+{
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_FINISH))
+		return false;
+
+	err = pf_send_vf_flr_finish(gt, vfid);
+	if (err == -EBUSY)
+		pf_enter_vf_flr_send_finish(gt, vfid);
+	else if (err == -EIO)
+		pf_enter_vf_flr_rejected(gt, vfid);
+	else if (err)
+		pf_enter_vf_flr_failed(gt, vfid);
+	else
+		pf_enter_vf_flr_completed(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_flr_reset_mmio(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_MMIO))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_flr_reset_mmio(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_MMIO))
+		return false;
+
+	/* XXX: placeholder */
+
+	pf_enter_vf_flr_send_finish(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_flr_reset_data(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_DATA))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_flr_reset_data(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_DATA))
+		return false;
+
+	xe_gt_sriov_pf_service_reset(gt, vfid);
+	xe_gt_sriov_pf_monitor_flr(gt, vfid);
+
+	pf_enter_vf_flr_reset_mmio(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_flr_reset_config(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_CONFIG))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+
+	pf_queue_vf(gt, vfid);
+}
+
+static bool pf_exit_vf_flr_reset_config(struct xe_gt *gt, unsigned int vfid)
+{
+	unsigned long timeout = pf_get_default_timeout(XE_GT_SRIOV_STATE_FLR_RESET_CONFIG);
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_CONFIG))
+		return false;
+
+	err = xe_gt_sriov_pf_config_sanitize(gt, vfid, timeout);
+	if (err)
+		pf_enter_vf_flr_failed(gt, vfid);
+	else
+		pf_enter_vf_flr_reset_data(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_flr_wait_guc(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WAIT_GUC))
+		pf_enter_vf_state_machine_bug(gt, vfid);
+}
+
+static bool pf_exit_vf_flr_wait_guc(struct xe_gt *gt, unsigned int vfid)
+{
+	return pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WAIT_GUC);
+}
+
+static bool pf_exit_vf_flr_send_start(struct xe_gt *gt, unsigned int vfid)
+{
+	int err;
+
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_SEND_START))
+		return false;
+
+	/* GuC may actually send a FLR_DONE before we get a RESPONSE */
+	pf_enter_vf_flr_wait_guc(gt, vfid);
+
+	err = pf_send_vf_flr_start(gt, vfid);
+	if (err) {
+		/* send failed, so we shouldn't expect FLR_DONE from GuC */
+		pf_exit_vf_flr_wait_guc(gt, vfid);
+
+		if (err == -EBUSY)
+			pf_enter_vf_flr_send_start(gt, vfid);
+		else if (err == -EIO)
+			pf_enter_vf_flr_rejected(gt, vfid);
+		else
+			pf_enter_vf_flr_failed(gt, vfid);
+	} else {
+		/*
+		 * we have already moved to WAIT_GUC, maybe even to GUC_DONE
+		 * but since GuC didn't complain, we may clear MISMATCH
+		 */
+		pf_exit_vf_mismatch(gt, vfid);
+	}
+
+	return true;
+}
+
+static bool pf_exit_vf_flr_guc_done(struct xe_gt *gt, unsigned int vfid)
+{
+	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_GUC_DONE))
+		return false;
+
+	pf_enter_vf_flr_reset_config(gt, vfid);
+	return true;
+}
+
+static void pf_enter_vf_flr_guc_done(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_enter_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_GUC_DONE))
+		pf_queue_vf(gt, vfid);
 }
 
 /**
@@ -141,14 +1127,22 @@ int xe_gt_sriov_pf_control_stop_vf(struct xe_gt *gt, unsigned int vfid)
  */
 int xe_gt_sriov_pf_control_trigger_flr(struct xe_gt *gt, unsigned int vfid)
 {
+	unsigned long timeout = pf_get_default_timeout(XE_GT_SRIOV_STATE_FLR_WIP);
 	int err;
 
-	/* XXX pf_send_vf_flr_start() expects ct->lock */
-	mutex_lock(&gt->uc.guc.ct.lock);
-	err = pf_send_vf_flr_start(gt, vfid);
-	mutex_unlock(&gt->uc.guc.ct.lock);
+	pf_enter_vf_flr_wip(gt, vfid);
 
-	return err;
+	err = pf_wait_vf_wip_done(gt, vfid, timeout);
+	if (err) {
+		xe_gt_sriov_notice(gt, "VF%u FLR didn't finish in %u ms (%pe)\n",
+				   vfid, jiffies_to_msecs(timeout), ERR_PTR(err));
+		return err;
+	}
+
+	if (!pf_expect_vf_not_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_FAILED))
+		return -EIO;
+
+	return 0;
 }
 
 /**
@@ -200,15 +1194,32 @@ static void pf_handle_vf_flr(struct xe_gt *gt, u32 vfid)
 
 	if (needs_dispatch_flr(xe)) {
 		for_each_gt(gtit, xe, gtid)
-			pf_send_vf_flr_start(gtit, vfid);
+			pf_enter_vf_flr_wip(gtit, vfid);
 	} else {
-		pf_send_vf_flr_start(gt, vfid);
+		pf_enter_vf_flr_wip(gt, vfid);
 	}
 }
 
 static void pf_handle_vf_flr_done(struct xe_gt *gt, u32 vfid)
 {
-	pf_send_vf_flr_finish(gt, vfid);
+	if (!pf_exit_vf_flr_wait_guc(gt, vfid)) {
+		xe_gt_sriov_dbg(gt, "Received out of order 'VF%u FLR done'\n", vfid);
+		pf_enter_vf_mismatch(gt, vfid);
+		return;
+	}
+
+	pf_enter_vf_flr_guc_done(gt, vfid);
+}
+
+static void pf_handle_vf_pause_done(struct xe_gt *gt, u32 vfid)
+{
+	if (!pf_exit_pause_wait_guc(gt, vfid)) {
+		xe_gt_sriov_dbg(gt, "Received out of order 'VF%u PAUSE done'\n", vfid);
+		pf_enter_vf_mismatch(gt, vfid);
+		return;
+	}
+
+	pf_enter_vf_pause_guc_done(gt, vfid);
 }
 
 static int pf_handle_vf_event(struct xe_gt *gt, u32 vfid, u32 eventid)
@@ -226,6 +1237,7 @@ static int pf_handle_vf_event(struct xe_gt *gt, u32 vfid, u32 eventid)
 		pf_handle_vf_flr_done(gt, vfid);
 		break;
 	case GUC_PF_NOTIFY_VF_PAUSE_DONE:
+		pf_handle_vf_pause_done(gt, vfid);
 		break;
 	case GUC_PF_NOTIFY_VF_FIXUP_DONE:
 		break;
@@ -284,3 +1296,159 @@ int xe_gt_sriov_pf_control_process_guc2pf(struct xe_gt *gt, const u32 *msg, u32
 
 	return vfid ? pf_handle_vf_event(gt, vfid, eventid) : pf_handle_pf_event(gt, eventid);
 }
+
+static bool pf_process_vf_state_machine(struct xe_gt *gt, unsigned int vfid)
+{
+	if (pf_exit_vf_flr_send_start(gt, vfid))
+		return true;
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_WAIT_GUC)) {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u in %s\n", vfid,
+					control_bit_to_string(XE_GT_SRIOV_STATE_FLR_WAIT_GUC));
+		return false;
+	}
+
+	if (pf_exit_vf_flr_guc_done(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_flr_reset_config(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_flr_reset_data(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_flr_reset_mmio(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_flr_send_finish(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_stop_send_stop(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_pause_send_pause(gt, vfid))
+		return true;
+
+	if (pf_check_vf_state(gt, vfid, XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC)) {
+		xe_gt_sriov_dbg_verbose(gt, "VF%u in %s\n", vfid,
+					control_bit_to_string(XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC));
+		return true;
+	}
+
+	if (pf_exit_vf_pause_guc_done(gt, vfid))
+		return true;
+
+	if (pf_exit_vf_resume_send_resume(gt, vfid))
+		return true;
+
+	return false;
+}
+
+static unsigned int pf_control_state_index(struct xe_gt *gt,
+					   struct xe_gt_sriov_control_state *cs)
+{
+	return container_of(cs, struct xe_gt_sriov_metadata, control) - gt->sriov.pf.vfs;
+}
+
+static void pf_worker_find_work(struct xe_gt *gt)
+{
+	struct xe_gt_sriov_pf_control *pfc = &gt->sriov.pf.control;
+	struct xe_gt_sriov_control_state *cs;
+	unsigned int vfid;
+	bool empty;
+	bool more;
+
+	spin_lock(&pfc->lock);
+	cs = list_first_entry_or_null(&pfc->list, struct xe_gt_sriov_control_state, link);
+	if (cs)
+		list_del_init(&cs->link);
+	empty = list_empty(&pfc->list);
+	spin_unlock(&pfc->lock);
+
+	if (!cs)
+		return;
+
+	/* VF metadata structures are indexed by the VFID */
+	vfid = pf_control_state_index(gt, cs);
+	xe_gt_assert(gt, vfid <= xe_gt_sriov_pf_get_totalvfs(gt));
+
+	more = pf_process_vf_state_machine(gt, vfid);
+	if (more)
+		pf_queue_vf(gt, vfid);
+	else if (!empty)
+		pf_queue_control_worker(gt);
+}
+
+static void control_worker_func(struct work_struct *w)
+{
+	struct xe_gt *gt = container_of(w, struct xe_gt, sriov.pf.control.worker);
+
+	xe_gt_assert(gt, IS_SRIOV_PF(gt_to_xe(gt)));
+	pf_worker_find_work(gt);
+}
+
+static void pf_stop_worker(struct xe_gt *gt)
+{
+	xe_gt_assert(gt, IS_SRIOV_PF(gt_to_xe(gt)));
+	cancel_work_sync(&gt->sriov.pf.control.worker);
+}
+
+static void control_fini_action(struct drm_device *dev, void *data)
+{
+	struct xe_gt *gt = data;
+
+	pf_stop_worker(gt);
+}
+
+/**
+ * xe_gt_sriov_pf_control_init() - Initialize PF's control data.
+ * @gt: the &xe_gt
+ *
+ * This function is for PF only.
+ *
+ * Return: 0 on success or a negative error code on failure.
+ */
+int xe_gt_sriov_pf_control_init(struct xe_gt *gt)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+	unsigned int n, totalvfs;
+
+	xe_gt_assert(gt, IS_SRIOV_PF(xe));
+
+	totalvfs = xe_sriov_pf_get_totalvfs(xe);
+	for (n = 0; n <= totalvfs; n++) {
+		struct xe_gt_sriov_control_state *cs = pf_pick_vf_control(gt, n);
+
+		init_completion(&cs->done);
+		INIT_LIST_HEAD(&cs->link);
+	}
+
+	spin_lock_init(&gt->sriov.pf.control.lock);
+	INIT_LIST_HEAD(&gt->sriov.pf.control.list);
+	INIT_WORK(&gt->sriov.pf.control.worker, control_worker_func);
+
+	return drmm_add_action_or_reset(&xe->drm, control_fini_action, gt);
+}
+
+/**
+ * xe_gt_sriov_pf_control_restart() - Restart SR-IOV control data after a GT reset.
+ * @gt: the &xe_gt
+ *
+ * Any per-VF status maintained by the PF or any ongoing VF control activity
+ * performed by the PF must be reset or cancelled when the GT is reset.
+ *
+ * This function is for PF only.
+ */
+void xe_gt_sriov_pf_control_restart(struct xe_gt *gt)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+	unsigned int n, totalvfs;
+
+	xe_gt_assert(gt, IS_SRIOV_PF(xe));
+
+	pf_stop_worker(gt);
+
+	totalvfs = xe_sriov_pf_get_totalvfs(xe);
+	for (n = 1; n <= totalvfs; n++)
+		pf_enter_vf_ready(gt, n);
+}
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.h
index 405d1586f991..c85e64f099cc 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.h
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.h
@@ -11,6 +11,9 @@
 
 struct xe_gt;
 
+int xe_gt_sriov_pf_control_init(struct xe_gt *gt);
+void xe_gt_sriov_pf_control_restart(struct xe_gt *gt);
+
 int xe_gt_sriov_pf_control_pause_vf(struct xe_gt *gt, unsigned int vfid);
 int xe_gt_sriov_pf_control_resume_vf(struct xe_gt *gt, unsigned int vfid);
 int xe_gt_sriov_pf_control_stop_vf(struct xe_gt *gt, unsigned int vfid);
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control_types.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control_types.h
new file mode 100644
index 000000000000..11830aafea45
--- /dev/null
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control_types.h
@@ -0,0 +1,107 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2024 Intel Corporation
+ */
+
+#ifndef _XE_GT_SRIOV_PF_CONTROL_TYPES_H_
+#define _XE_GT_SRIOV_PF_CONTROL_TYPES_H_
+
+#include <linux/completion.h>
+#include <linux/spinlock.h>
+#include <linux/workqueue_types.h>
+
+/**
+ * enum xe_gt_sriov_control_bits - Various bits used by the PF to represent a VF state
+ *
+ * @XE_GT_SRIOV_STATE_WIP: indicates that some operations are in progress.
+ * @XE_GT_SRIOV_STATE_FLR_WIP: indicates that a VF FLR is in progress.
+ * @XE_GT_SRIOV_STATE_FLR_SEND_START: indicates that the PF wants to send a FLR START command.
+ * @XE_GT_SRIOV_STATE_FLR_WAIT_GUC: indicates that the PF awaits for a response from the GuC.
+ * @XE_GT_SRIOV_STATE_FLR_GUC_DONE: indicates that the PF has received a response from the GuC.
+ * @XE_GT_SRIOV_STATE_FLR_RESET_CONFIG: indicates that the PF needs to clear VF's resources.
+ * @XE_GT_SRIOV_STATE_FLR_RESET_DATA: indicates that the PF needs to clear VF's data.
+ * @XE_GT_SRIOV_STATE_FLR_RESET_MMIO: indicates that the PF needs to reset VF's registers.
+ * @XE_GT_SRIOV_STATE_FLR_SEND_FINISH: indicates that the PF wants to send a FLR FINISH message.
+ * @XE_GT_SRIOV_STATE_FLR_FAILED: indicates that VF FLR sequence failed.
+ * @XE_GT_SRIOV_STATE_PAUSE_WIP: indicates that a VF pause operation is in progress.
+ * @XE_GT_SRIOV_STATE_PAUSE_SEND_PAUSE: indicates that the PF is about to send a PAUSE command.
+ * @XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC: indicates that the PF awaits for a response from the GuC.
+ * @XE_GT_SRIOV_STATE_PAUSE_GUC_DONE: indicates that the PF has received a response from the GuC.
+ * @XE_GT_SRIOV_STATE_PAUSE_FAILED: indicates that a VF pause operation has failed.
+ * @XE_GT_SRIOV_STATE_PAUSED: indicates that the VF is paused.
+ * @XE_GT_SRIOV_STATE_RESUME_WIP: indicates the a VF resume operation is in progress.
+ * @XE_GT_SRIOV_STATE_RESUME_SEND_RESUME: indicates that the PF is about to send RESUME command.
+ * @XE_GT_SRIOV_STATE_RESUME_FAILED: indicates that a VF resume operation has failed.
+ * @XE_GT_SRIOV_STATE_RESUMED: indicates that the VF was resumed.
+ * @XE_GT_SRIOV_STATE_STOP_WIP: indicates that a VF stop operation is in progress.
+ * @XE_GT_SRIOV_STATE_STOP_SEND_STOP: indicates that the PF wants to send a STOP command.
+ * @XE_GT_SRIOV_STATE_STOP_FAILED: indicates that the VF stop operation has failed
+ * @XE_GT_SRIOV_STATE_STOPPED: indicates that the VF was stopped.
+ * @XE_GT_SRIOV_STATE_MISMATCH: indicates that the PF has detected a VF state mismatch.
+ */
+enum xe_gt_sriov_control_bits {
+	XE_GT_SRIOV_STATE_WIP = 1,
+
+	XE_GT_SRIOV_STATE_FLR_WIP,
+	XE_GT_SRIOV_STATE_FLR_SEND_START,
+	XE_GT_SRIOV_STATE_FLR_WAIT_GUC,
+	XE_GT_SRIOV_STATE_FLR_GUC_DONE,
+	XE_GT_SRIOV_STATE_FLR_RESET_CONFIG,
+	XE_GT_SRIOV_STATE_FLR_RESET_DATA,
+	XE_GT_SRIOV_STATE_FLR_RESET_MMIO,
+	XE_GT_SRIOV_STATE_FLR_SEND_FINISH,
+	XE_GT_SRIOV_STATE_FLR_FAILED,
+
+	XE_GT_SRIOV_STATE_PAUSE_WIP,
+	XE_GT_SRIOV_STATE_PAUSE_SEND_PAUSE,
+	XE_GT_SRIOV_STATE_PAUSE_WAIT_GUC,
+	XE_GT_SRIOV_STATE_PAUSE_GUC_DONE,
+	XE_GT_SRIOV_STATE_PAUSE_FAILED,
+	XE_GT_SRIOV_STATE_PAUSED,
+
+	XE_GT_SRIOV_STATE_RESUME_WIP,
+	XE_GT_SRIOV_STATE_RESUME_SEND_RESUME,
+	XE_GT_SRIOV_STATE_RESUME_FAILED,
+	XE_GT_SRIOV_STATE_RESUMED,
+
+	XE_GT_SRIOV_STATE_STOP_WIP,
+	XE_GT_SRIOV_STATE_STOP_SEND_STOP,
+	XE_GT_SRIOV_STATE_STOP_FAILED,
+	XE_GT_SRIOV_STATE_STOPPED,
+
+	XE_GT_SRIOV_STATE_MISMATCH = BITS_PER_LONG - 1,
+};
+
+/**
+ * struct xe_gt_sriov_control_state - GT-level per-VF control state.
+ *
+ * Used by the PF driver to maintain per-VF control data.
+ */
+struct xe_gt_sriov_control_state {
+	/** @state: VF state bits */
+	unsigned long state;
+
+	/** @done: completion of async operations */
+	struct completion done;
+
+	/** @link: link into worker list */
+	struct list_head link;
+};
+
+/**
+ * struct xe_gt_sriov_pf_control - GT-level control data.
+ *
+ * Used by the PF driver to maintain its data.
+ */
+struct xe_gt_sriov_pf_control {
+	/** @worker: worker that executes a VF operations */
+	struct work_struct worker;
+
+	/** @list: list of VF entries that have a pending work */
+	struct list_head list;
+
+	/** @lock: protects VF pending list */
+	spinlock_t lock;
+};
+
+#endif
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h
index 40cbaea3ef44..28e1b130bf87 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h
@@ -9,6 +9,7 @@
 #include <linux/types.h>
 
 #include "xe_gt_sriov_pf_config_types.h"
+#include "xe_gt_sriov_pf_control_types.h"
 #include "xe_gt_sriov_pf_monitor_types.h"
 #include "xe_gt_sriov_pf_policy_types.h"
 #include "xe_gt_sriov_pf_service_types.h"
@@ -23,6 +24,9 @@ struct xe_gt_sriov_metadata {
 	/** @monitor: per-VF monitoring data. */
 	struct xe_gt_sriov_monitor monitor;
 
+	/** @control: per-VF control data. */
+	struct xe_gt_sriov_control_state control;
+
 	/** @version: negotiated VF/PF ABI version */
 	struct xe_gt_sriov_pf_service_version version;
 };
@@ -30,12 +34,14 @@ struct xe_gt_sriov_metadata {
 /**
  * struct xe_gt_sriov_pf - GT level PF virtualization data.
  * @service: service data.
+ * @control: control data.
  * @policy: policy data.
  * @spare: PF-only provisioning configuration.
  * @vfs: metadata for all VFs.
  */
 struct xe_gt_sriov_pf {
 	struct xe_gt_sriov_pf_service service;
+	struct xe_gt_sriov_pf_control control;
 	struct xe_gt_sriov_pf_policy policy;
 	struct xe_gt_sriov_spare_config spare;
 	struct xe_gt_sriov_metadata *vfs;
-- 
2.51.0


From c5f728de696caa35481fd84202dfbc9fecc18e0b Mon Sep 17 00:00:00 2001
From: Nirmoy Das <nirmoy.das@intel.com>
Date: Mon, 26 Aug 2024 18:20:35 +0200
Subject: [PATCH 03/16] drm/xe: Fix memory leak on xe_alloc_pf_queue failure

Simplify memory unwinding on error also fixing current memory
leak that can happen on error.

v2: use devm_kcalloc(Matt A)

Fixes: 3338e4f90c14 ("drm/xe: Use topology to determine page fault queue size")
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Stuart Summers <stuart.summers@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240826162035.20462-1-nirmoy.das@intel.com
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
---
 drivers/gpu/drm/xe/xe_gt_pagefault.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c
index 0be4687bfc20..730eec07795e 100644
--- a/drivers/gpu/drm/xe/xe_gt_pagefault.c
+++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c
@@ -388,20 +388,17 @@ static void pagefault_fini(void *arg)
 {
 	struct xe_gt *gt = arg;
 	struct xe_device *xe = gt_to_xe(gt);
-	int i;
 
 	if (!xe->info.has_usm)
 		return;
 
 	destroy_workqueue(gt->usm.acc_wq);
 	destroy_workqueue(gt->usm.pf_wq);
-
-	for (i = 0; i < NUM_PF_QUEUE; ++i)
-		kfree(gt->usm.pf_queue[i].data);
 }
 
 static int xe_alloc_pf_queue(struct xe_gt *gt, struct pf_queue *pf_queue)
 {
+	struct xe_device *xe = gt_to_xe(gt);
 	xe_dss_mask_t all_dss;
 	int num_dss, num_eus;
 
@@ -417,7 +414,8 @@ static int xe_alloc_pf_queue(struct xe_gt *gt, struct pf_queue *pf_queue)
 		(num_eus + XE_NUM_HW_ENGINES) * PF_MSG_LEN_DW;
 
 	pf_queue->gt = gt;
-	pf_queue->data = kcalloc(pf_queue->num_dw, sizeof(u32), GFP_KERNEL);
+	pf_queue->data = devm_kcalloc(xe->drm.dev, pf_queue->num_dw,
+				      sizeof(u32), GFP_KERNEL);
 	if (!pf_queue->data)
 		return -ENOMEM;
 
-- 
2.51.0


From cad08fa7760ace5e01b24c6f9e1f918fe72324bf Mon Sep 17 00:00:00 2001
From: Matt Roper <matthew.d.roper@intel.com>
Date: Thu, 29 Aug 2024 16:03:08 -0700
Subject: [PATCH 04/16] drm/xe/display: Drop unnecessary xe_gt.h includes

None of the Xe display files work directly with the GT or need anything
from xe_gt.h.  Drop the unnecessary include.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829230307.886233-2-matthew.d.roper@intel.com
---
 drivers/gpu/drm/xe/display/intel_fbdev_fb.c | 1 -
 drivers/gpu/drm/xe/display/xe_dsb_buffer.c  | 1 -
 drivers/gpu/drm/xe/display/xe_fb_pin.c      | 1 -
 drivers/gpu/drm/xe/display/xe_hdcp_gsc.c    | 1 -
 4 files changed, 4 deletions(-)

diff --git a/drivers/gpu/drm/xe/display/intel_fbdev_fb.c b/drivers/gpu/drm/xe/display/intel_fbdev_fb.c
index cd8948c08661..99499d6c0256 100644
--- a/drivers/gpu/drm/xe/display/intel_fbdev_fb.c
+++ b/drivers/gpu/drm/xe/display/intel_fbdev_fb.c
@@ -8,7 +8,6 @@
 #include "intel_display_types.h"
 #include "intel_fbdev_fb.h"
 #include "xe_bo.h"
-#include "xe_gt.h"
 #include "xe_ttm_stolen_mgr.h"
 #include "xe_wa.h"
 
diff --git a/drivers/gpu/drm/xe/display/xe_dsb_buffer.c b/drivers/gpu/drm/xe/display/xe_dsb_buffer.c
index ccd0d87d438a..f99d901a3214 100644
--- a/drivers/gpu/drm/xe/display/xe_dsb_buffer.c
+++ b/drivers/gpu/drm/xe/display/xe_dsb_buffer.c
@@ -9,7 +9,6 @@
 #include "xe_bo.h"
 #include "xe_device.h"
 #include "xe_device_types.h"
-#include "xe_gt.h"
 
 u32 intel_dsb_buffer_ggtt_offset(struct intel_dsb_buffer *dsb_buf)
 {
diff --git a/drivers/gpu/drm/xe/display/xe_fb_pin.c b/drivers/gpu/drm/xe/display/xe_fb_pin.c
index d650c5ac41a4..b58fc4ba2aac 100644
--- a/drivers/gpu/drm/xe/display/xe_fb_pin.c
+++ b/drivers/gpu/drm/xe/display/xe_fb_pin.c
@@ -12,7 +12,6 @@
 #include "xe_bo.h"
 #include "xe_device.h"
 #include "xe_ggtt.h"
-#include "xe_gt.h"
 #include "xe_pm.h"
 
 static void
diff --git a/drivers/gpu/drm/xe/display/xe_hdcp_gsc.c b/drivers/gpu/drm/xe/display/xe_hdcp_gsc.c
index 0af667ebebf9..6619a40aed15 100644
--- a/drivers/gpu/drm/xe/display/xe_hdcp_gsc.c
+++ b/drivers/gpu/drm/xe/display/xe_hdcp_gsc.c
@@ -16,7 +16,6 @@
 #include "xe_force_wake.h"
 #include "xe_gsc_proxy.h"
 #include "xe_gsc_submit.h"
-#include "xe_gt.h"
 #include "xe_map.h"
 #include "xe_pm.h"
 #include "xe_uc_fw.h"
-- 
2.51.0


From 3034cc8107b8d0c7d1b56584394e215dab57f8a3 Mon Sep 17 00:00:00 2001
From: Matt Roper <matthew.d.roper@intel.com>
Date: Thu, 29 Aug 2024 15:06:21 -0700
Subject: [PATCH 05/16] drm/xe/pcode: Treat pcode as per-tile rather than
 per-GT

There's only one instance of the pcode per tile, and for GT-related
accesses both the primary and media GT share the same register
interface.  Since Xe was using per-GT locking, the pcode mutex wasn't
actually protecting everything that it should since concurrent accesses
related to a tile's primary GT and media GT were possible.

Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829220619.789159-5-matthew.d.roper@intel.com
---
 .../drm/xe/compat-i915-headers/intel_pcode.h  |   8 +-
 .../drm/xe/compat-i915-headers/intel_uncore.h |   7 ++
 drivers/gpu/drm/xe/xe_device_types.h          |   6 +
 drivers/gpu/drm/xe/xe_gt.c                    |   2 -
 drivers/gpu/drm/xe/xe_gt_types.h              |   6 -
 drivers/gpu/drm/xe/xe_guc_pc.c                |   2 +-
 drivers/gpu/drm/xe/xe_hwmon.c                 |   4 +-
 drivers/gpu/drm/xe/xe_pcode.c                 | 104 +++++++++---------
 drivers/gpu/drm/xe/xe_pcode.h                 |  16 +--
 drivers/gpu/drm/xe/xe_tile.c                  |   3 +
 drivers/gpu/drm/xe/xe_vram_freq.c             |   6 +-
 11 files changed, 85 insertions(+), 79 deletions(-)

diff --git a/drivers/gpu/drm/xe/compat-i915-headers/intel_pcode.h b/drivers/gpu/drm/xe/compat-i915-headers/intel_pcode.h
index 0c47661bdc6a..a473aa6697d0 100644
--- a/drivers/gpu/drm/xe/compat-i915-headers/intel_pcode.h
+++ b/drivers/gpu/drm/xe/compat-i915-headers/intel_pcode.h
@@ -13,7 +13,7 @@ static inline int
 snb_pcode_write_timeout(struct intel_uncore *uncore, u32 mbox, u32 val,
 			int fast_timeout_us, int slow_timeout_ms)
 {
-	return xe_pcode_write_timeout(__compat_uncore_to_gt(uncore), mbox, val,
+	return xe_pcode_write_timeout(__compat_uncore_to_tile(uncore), mbox, val,
 				      slow_timeout_ms ?: 1);
 }
 
@@ -21,13 +21,13 @@ static inline int
 snb_pcode_write(struct intel_uncore *uncore, u32 mbox, u32 val)
 {
 
-	return xe_pcode_write(__compat_uncore_to_gt(uncore), mbox, val);
+	return xe_pcode_write(__compat_uncore_to_tile(uncore), mbox, val);
 }
 
 static inline int
 snb_pcode_read(struct intel_uncore *uncore, u32 mbox, u32 *val, u32 *val1)
 {
-	return xe_pcode_read(__compat_uncore_to_gt(uncore), mbox, val, val1);
+	return xe_pcode_read(__compat_uncore_to_tile(uncore), mbox, val, val1);
 }
 
 static inline int
@@ -35,7 +35,7 @@ skl_pcode_request(struct intel_uncore *uncore, u32 mbox,
 		  u32 request, u32 reply_mask, u32 reply,
 		  int timeout_base_ms)
 {
-	return xe_pcode_request(__compat_uncore_to_gt(uncore), mbox, request, reply_mask, reply,
+	return xe_pcode_request(__compat_uncore_to_tile(uncore), mbox, request, reply_mask, reply,
 				timeout_base_ms);
 }
 
diff --git a/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h b/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h
index 083c4da2ea41..eb5b5f0e4bd9 100644
--- a/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h
+++ b/drivers/gpu/drm/xe/compat-i915-headers/intel_uncore.h
@@ -17,6 +17,13 @@ static inline struct xe_gt *__compat_uncore_to_gt(struct intel_uncore *uncore)
 	return xe_root_mmio_gt(xe);
 }
 
+static inline struct xe_tile *__compat_uncore_to_tile(struct intel_uncore *uncore)
+{
+	struct xe_device *xe = container_of(uncore, struct xe_device, uncore);
+
+	return xe_device_get_root_tile(xe);
+}
+
 static inline u32 intel_uncore_read(struct intel_uncore *uncore,
 				    i915_reg_t i915_reg)
 {
diff --git a/drivers/gpu/drm/xe/xe_device_types.h b/drivers/gpu/drm/xe/xe_device_types.h
index 856db4020048..9c2ae57ecf99 100644
--- a/drivers/gpu/drm/xe/xe_device_types.h
+++ b/drivers/gpu/drm/xe/xe_device_types.h
@@ -208,6 +208,12 @@ struct xe_tile {
 		} vf;
 	} sriov;
 
+	/** @pcode: tile's PCODE */
+	struct {
+		/** @pcode.lock: protecting tile's PCODE mailbox data */
+		struct mutex lock;
+	} pcode;
+
 	/** @migrate: Migration helper for vram blits and clearing */
 	struct xe_migrate *migrate;
 
diff --git a/drivers/gpu/drm/xe/xe_gt.c b/drivers/gpu/drm/xe/xe_gt.c
index bb612f089e49..ffacddb587b2 100644
--- a/drivers/gpu/drm/xe/xe_gt.c
+++ b/drivers/gpu/drm/xe/xe_gt.c
@@ -46,7 +46,6 @@
 #include "xe_migrate.h"
 #include "xe_mmio.h"
 #include "xe_pat.h"
-#include "xe_pcode.h"
 #include "xe_pm.h"
 #include "xe_mocs.h"
 #include "xe_reg_sr.h"
@@ -386,7 +385,6 @@ int xe_gt_init_early(struct xe_gt *gt)
 	xe_tuning_process_gt(gt);
 
 	xe_force_wake_init_gt(gt, gt_to_fw(gt));
-	xe_pcode_init(gt);
 	spin_lock_init(&gt->global_invl_lock);
 
 	return 0;
diff --git a/drivers/gpu/drm/xe/xe_gt_types.h b/drivers/gpu/drm/xe/xe_gt_types.h
index 31946d7fe701..3d1c51de0268 100644
--- a/drivers/gpu/drm/xe/xe_gt_types.h
+++ b/drivers/gpu/drm/xe/xe_gt_types.h
@@ -329,12 +329,6 @@ struct xe_gt {
 	/** @eclass: per hardware engine class interface on the GT */
 	struct xe_hw_engine_class_intf  eclass[XE_ENGINE_CLASS_MAX];
 
-	/** @pcode: GT's PCODE */
-	struct {
-		/** @pcode.lock: protecting GT's PCODE mailbox data */
-		struct mutex lock;
-	} pcode;
-
 	/** @sysfs: sysfs' kobj used by xe_gt_sysfs */
 	struct kobject *sysfs;
 
diff --git a/drivers/gpu/drm/xe/xe_guc_pc.c b/drivers/gpu/drm/xe/xe_guc_pc.c
index def503abeed5..034b29984d5e 100644
--- a/drivers/gpu/drm/xe/xe_guc_pc.c
+++ b/drivers/gpu/drm/xe/xe_guc_pc.c
@@ -915,7 +915,7 @@ static void pc_init_pcode_freq(struct xe_guc_pc *pc)
 	u32 min = DIV_ROUND_CLOSEST(pc->rpn_freq, GT_FREQUENCY_MULTIPLIER);
 	u32 max = DIV_ROUND_CLOSEST(pc->rp0_freq, GT_FREQUENCY_MULTIPLIER);
 
-	XE_WARN_ON(xe_pcode_init_min_freq_table(pc_to_gt(pc), min, max));
+	XE_WARN_ON(xe_pcode_init_min_freq_table(gt_to_tile(pc_to_gt(pc)), min, max));
 }
 
 static int pc_init_freqs(struct xe_guc_pc *pc)
diff --git a/drivers/gpu/drm/xe/xe_hwmon.c b/drivers/gpu/drm/xe/xe_hwmon.c
index 1faeca70900e..98e3ec08279e 100644
--- a/drivers/gpu/drm/xe/xe_hwmon.c
+++ b/drivers/gpu/drm/xe/xe_hwmon.c
@@ -441,14 +441,14 @@ static int xe_hwmon_pcode_read_i1(struct xe_gt *gt, u32 *uval)
 	if (gt_to_xe(gt)->info.platform == XE_DG2)
 		return -ENXIO;
 
-	return xe_pcode_read(gt, PCODE_MBOX(PCODE_POWER_SETUP,
+	return xe_pcode_read(gt_to_tile(gt), PCODE_MBOX(PCODE_POWER_SETUP,
 			     POWER_SETUP_SUBCOMMAND_READ_I1, 0),
 			     uval, NULL);
 }
 
 static int xe_hwmon_pcode_write_i1(struct xe_gt *gt, u32 uval)
 {
-	return xe_pcode_write(gt, PCODE_MBOX(PCODE_POWER_SETUP,
+	return xe_pcode_write(gt_to_tile(gt), PCODE_MBOX(PCODE_POWER_SETUP,
 			      POWER_SETUP_SUBCOMMAND_WRITE_I1, 0),
 			      (uval & POWER_SETUP_I1_DATA_MASK));
 }
diff --git a/drivers/gpu/drm/xe/xe_pcode.c b/drivers/gpu/drm/xe/xe_pcode.c
index 9c4eefdf6642..7397d556996a 100644
--- a/drivers/gpu/drm/xe/xe_pcode.c
+++ b/drivers/gpu/drm/xe/xe_pcode.c
@@ -12,7 +12,6 @@
 
 #include "xe_assert.h"
 #include "xe_device.h"
-#include "xe_gt.h"
 #include "xe_mmio.h"
 #include "xe_pcode_api.h"
 
@@ -30,7 +29,7 @@
  * - PCODE for display operations
  */
 
-static int pcode_mailbox_status(struct xe_gt *gt)
+static int pcode_mailbox_status(struct xe_tile *tile)
 {
 	u32 err;
 	static const struct pcode_err_decode err_decode[] = {
@@ -45,9 +44,9 @@ static int pcode_mailbox_status(struct xe_gt *gt)
 		[PCODE_ERROR_MASK] = {-EPROTO, "Unknown"},
 	};
 
-	err = xe_mmio_read32(gt, PCODE_MAILBOX) & PCODE_ERROR_MASK;
+	err = xe_mmio_read32(tile->primary_gt, PCODE_MAILBOX) & PCODE_ERROR_MASK;
 	if (err) {
-		drm_err(&gt_to_xe(gt)->drm, "PCODE Mailbox failed: %d %s", err,
+		drm_err(&tile_to_xe(tile)->drm, "PCODE Mailbox failed: %d %s", err,
 			err_decode[err].str ?: "Unknown");
 		return err_decode[err].errno ?: -EPROTO;
 	}
@@ -55,84 +54,85 @@ static int pcode_mailbox_status(struct xe_gt *gt)
 	return 0;
 }
 
-static int __pcode_mailbox_rw(struct xe_gt *gt, u32 mbox, u32 *data0, u32 *data1,
+static int __pcode_mailbox_rw(struct xe_tile *tile, u32 mbox, u32 *data0, u32 *data1,
 			      unsigned int timeout_ms, bool return_data,
 			      bool atomic)
 {
+	struct xe_gt *mmio = tile->primary_gt;
 	int err;
 
-	if (gt_to_xe(gt)->info.skip_pcode)
+	if (tile_to_xe(tile)->info.skip_pcode)
 		return 0;
 
-	if ((xe_mmio_read32(gt, PCODE_MAILBOX) & PCODE_READY) != 0)
+	if ((xe_mmio_read32(mmio, PCODE_MAILBOX) & PCODE_READY) != 0)
 		return -EAGAIN;
 
-	xe_mmio_write32(gt, PCODE_DATA0, *data0);
-	xe_mmio_write32(gt, PCODE_DATA1, data1 ? *data1 : 0);
-	xe_mmio_write32(gt, PCODE_MAILBOX, PCODE_READY | mbox);
+	xe_mmio_write32(mmio, PCODE_DATA0, *data0);
+	xe_mmio_write32(mmio, PCODE_DATA1, data1 ? *data1 : 0);
+	xe_mmio_write32(mmio, PCODE_MAILBOX, PCODE_READY | mbox);
 
-	err = xe_mmio_wait32(gt, PCODE_MAILBOX, PCODE_READY, 0,
+	err = xe_mmio_wait32(mmio, PCODE_MAILBOX, PCODE_READY, 0,
 			     timeout_ms * USEC_PER_MSEC, NULL, atomic);
 	if (err)
 		return err;
 
 	if (return_data) {
-		*data0 = xe_mmio_read32(gt, PCODE_DATA0);
+		*data0 = xe_mmio_read32(mmio, PCODE_DATA0);
 		if (data1)
-			*data1 = xe_mmio_read32(gt, PCODE_DATA1);
+			*data1 = xe_mmio_read32(mmio, PCODE_DATA1);
 	}
 
-	return pcode_mailbox_status(gt);
+	return pcode_mailbox_status(tile);
 }
 
-static int pcode_mailbox_rw(struct xe_gt *gt, u32 mbox, u32 *data0, u32 *data1,
+static int pcode_mailbox_rw(struct xe_tile *tile, u32 mbox, u32 *data0, u32 *data1,
 			    unsigned int timeout_ms, bool return_data,
 			    bool atomic)
 {
-	if (gt_to_xe(gt)->info.skip_pcode)
+	if (tile_to_xe(tile)->info.skip_pcode)
 		return 0;
 
-	lockdep_assert_held(&gt->pcode.lock);
+	lockdep_assert_held(&tile->pcode.lock);
 
-	return __pcode_mailbox_rw(gt, mbox, data0, data1, timeout_ms, return_data, atomic);
+	return __pcode_mailbox_rw(tile, mbox, data0, data1, timeout_ms, return_data, atomic);
 }
 
-int xe_pcode_write_timeout(struct xe_gt *gt, u32 mbox, u32 data, int timeout)
+int xe_pcode_write_timeout(struct xe_tile *tile, u32 mbox, u32 data, int timeout)
 {
 	int err;
 
-	mutex_lock(&gt->pcode.lock);
-	err = pcode_mailbox_rw(gt, mbox, &data, NULL, timeout, false, false);
-	mutex_unlock(&gt->pcode.lock);
+	mutex_lock(&tile->pcode.lock);
+	err = pcode_mailbox_rw(tile, mbox, &data, NULL, timeout, false, false);
+	mutex_unlock(&tile->pcode.lock);
 
 	return err;
 }
 
-int xe_pcode_read(struct xe_gt *gt, u32 mbox, u32 *val, u32 *val1)
+int xe_pcode_read(struct xe_tile *tile, u32 mbox, u32 *val, u32 *val1)
 {
 	int err;
 
-	mutex_lock(&gt->pcode.lock);
-	err = pcode_mailbox_rw(gt, mbox, val, val1, 1, true, false);
-	mutex_unlock(&gt->pcode.lock);
+	mutex_lock(&tile->pcode.lock);
+	err = pcode_mailbox_rw(tile, mbox, val, val1, 1, true, false);
+	mutex_unlock(&tile->pcode.lock);
 
 	return err;
 }
 
-static int pcode_try_request(struct xe_gt *gt, u32 mbox,
+static int pcode_try_request(struct xe_tile *tile, u32 mbox,
 			     u32 request, u32 reply_mask, u32 reply,
 			     u32 *status, bool atomic, int timeout_us, bool locked)
 {
 	int slept, wait = 10;
 
-	xe_gt_assert(gt, timeout_us > 0);
+	xe_tile_assert(tile, timeout_us > 0);
 
 	for (slept = 0; slept < timeout_us; slept += wait) {
 		if (locked)
-			*status = pcode_mailbox_rw(gt, mbox, &request, NULL, 1, true,
+			*status = pcode_mailbox_rw(tile, mbox, &request, NULL, 1, true,
 						   atomic);
 		else
-			*status = __pcode_mailbox_rw(gt, mbox, &request, NULL, 1, true,
+			*status = __pcode_mailbox_rw(tile, mbox, &request, NULL, 1, true,
 						     atomic);
 		if ((*status == 0) && ((request & reply_mask) == reply))
 			return 0;
@@ -149,7 +149,7 @@ static int pcode_try_request(struct xe_gt *gt, u32 mbox,
 
 /**
  * xe_pcode_request - send PCODE request until acknowledgment
- * @gt: gt
+ * @tile: tile
  * @mbox: PCODE mailbox ID the request is targeted for
  * @request: request ID
  * @reply_mask: mask used to check for request acknowledgment
@@ -166,17 +166,17 @@ static int pcode_try_request(struct xe_gt *gt, u32 mbox,
  * Returns 0 on success, %-ETIMEDOUT in case of a timeout, <0 in case of some
  * other error as reported by PCODE.
  */
-int xe_pcode_request(struct xe_gt *gt, u32 mbox, u32 request,
-		      u32 reply_mask, u32 reply, int timeout_base_ms)
+int xe_pcode_request(struct xe_tile *tile, u32 mbox, u32 request,
+		     u32 reply_mask, u32 reply, int timeout_base_ms)
 {
 	u32 status;
 	int ret;
 
-	xe_gt_assert(gt, timeout_base_ms <= 3);
+	xe_tile_assert(tile, timeout_base_ms <= 3);
 
-	mutex_lock(&gt->pcode.lock);
+	mutex_lock(&tile->pcode.lock);
 
-	ret = pcode_try_request(gt, mbox, request, reply_mask, reply, &status,
+	ret = pcode_try_request(tile, mbox, request, reply_mask, reply, &status,
 				false, timeout_base_ms * 1000, true);
 	if (!ret)
 		goto out;
@@ -191,20 +191,20 @@ int xe_pcode_request(struct xe_gt *gt, u32 mbox, u32 request,
 	 * requests, and for any quirks of the PCODE firmware that delays
 	 * the request completion.
 	 */
-	drm_err(&gt_to_xe(gt)->drm,
+	drm_err(&tile_to_xe(tile)->drm,
 		"PCODE timeout, retrying with preemption disabled\n");
 	preempt_disable();
-	ret = pcode_try_request(gt, mbox, request, reply_mask, reply, &status,
+	ret = pcode_try_request(tile, mbox, request, reply_mask, reply, &status,
 				true, 50 * 1000, true);
 	preempt_enable();
 
 out:
-	mutex_unlock(&gt->pcode.lock);
+	mutex_unlock(&tile->pcode.lock);
 	return status ? status : ret;
 }
 /**
  * xe_pcode_init_min_freq_table - Initialize PCODE's QOS frequency table
- * @gt: gt instance
+ * @tile: tile instance
  * @min_gt_freq: Minimal (RPn) GT frequency in units of 50MHz.
  * @max_gt_freq: Maximal (RP0) GT frequency in units of 50MHz.
  *
@@ -227,30 +227,30 @@ out:
  * - -EACCES, "PCODE Rejected"
  * - -EPROTO, "Unknown"
  */
-int xe_pcode_init_min_freq_table(struct xe_gt *gt, u32 min_gt_freq,
+int xe_pcode_init_min_freq_table(struct xe_tile *tile, u32 min_gt_freq,
 				 u32 max_gt_freq)
 {
 	int ret;
 	u32 freq;
 
-	if (!gt_to_xe(gt)->info.has_llc)
+	if (!tile_to_xe(tile)->info.has_llc)
 		return 0;
 
 	if (max_gt_freq <= min_gt_freq)
 		return -EINVAL;
 
-	mutex_lock(&gt->pcode.lock);
+	mutex_lock(&tile->pcode.lock);
 	for (freq = min_gt_freq; freq <= max_gt_freq; freq++) {
 		u32 data = freq << PCODE_FREQ_RING_RATIO_SHIFT | freq;
 
-		ret = pcode_mailbox_rw(gt, PCODE_WRITE_MIN_FREQ_TABLE,
+		ret = pcode_mailbox_rw(tile, PCODE_WRITE_MIN_FREQ_TABLE,
 				       &data, NULL, 1, false, false);
 		if (ret)
 			goto unlock;
 	}
 
 unlock:
-	mutex_unlock(&gt->pcode.lock);
+	mutex_unlock(&tile->pcode.lock);
 	return ret;
 }
 
@@ -270,7 +270,7 @@ unlock:
 int xe_pcode_ready(struct xe_device *xe, bool locked)
 {
 	u32 status, request = DGFX_GET_INIT_STATUS;
-	struct xe_gt *gt = xe_root_mmio_gt(xe);
+	struct xe_tile *tile = xe_device_get_root_tile(xe);
 	int timeout_us = 180000000; /* 3 min */
 	int ret;
 
@@ -281,15 +281,15 @@ int xe_pcode_ready(struct xe_device *xe, bool locked)
 		return 0;
 
 	if (locked)
-		mutex_lock(&gt->pcode.lock);
+		mutex_lock(&tile->pcode.lock);
 
-	ret = pcode_try_request(gt, DGFX_PCODE_STATUS, request,
+	ret = pcode_try_request(tile, DGFX_PCODE_STATUS, request,
 				DGFX_INIT_STATUS_COMPLETE,
 				DGFX_INIT_STATUS_COMPLETE,
 				&status, false, timeout_us, locked);
 
 	if (locked)
-		mutex_unlock(&gt->pcode.lock);
+		mutex_unlock(&tile->pcode.lock);
 
 	if (ret)
 		drm_err(&xe->drm,
@@ -300,14 +300,14 @@ int xe_pcode_ready(struct xe_device *xe, bool locked)
 
 /**
  * xe_pcode_init: initialize components of PCODE
- * @gt: gt instance
+ * @tile: tile instance
  *
  * This function initializes the xe_pcode component.
  * To be called once only during probe.
  */
-void xe_pcode_init(struct xe_gt *gt)
+void xe_pcode_init(struct xe_tile *tile)
 {
-	drmm_mutex_init(&gt_to_xe(gt)->drm, &gt->pcode.lock);
+	drmm_mutex_init(&tile_to_xe(tile)->drm, &tile->pcode.lock);
 }
 
 /**
diff --git a/drivers/gpu/drm/xe/xe_pcode.h b/drivers/gpu/drm/xe/xe_pcode.h
index 3f54c6d2a57d..ba33991d72a7 100644
--- a/drivers/gpu/drm/xe/xe_pcode.h
+++ b/drivers/gpu/drm/xe/xe_pcode.h
@@ -7,21 +7,21 @@
 #define _XE_PCODE_H_
 
 #include <linux/types.h>
-struct xe_gt;
+struct xe_tile;
 struct xe_device;
 
-void xe_pcode_init(struct xe_gt *gt);
+void xe_pcode_init(struct xe_tile *tile);
 int xe_pcode_probe_early(struct xe_device *xe);
 int xe_pcode_ready(struct xe_device *xe, bool locked);
-int xe_pcode_init_min_freq_table(struct xe_gt *gt, u32 min_gt_freq,
+int xe_pcode_init_min_freq_table(struct xe_tile *tile, u32 min_gt_freq,
 				 u32 max_gt_freq);
-int xe_pcode_read(struct xe_gt *gt, u32 mbox, u32 *val, u32 *val1);
-int xe_pcode_write_timeout(struct xe_gt *gt, u32 mbox, u32 val,
+int xe_pcode_read(struct xe_tile *tile, u32 mbox, u32 *val, u32 *val1);
+int xe_pcode_write_timeout(struct xe_tile *tile, u32 mbox, u32 val,
 			   int timeout_ms);
-#define xe_pcode_write(gt, mbox, val) \
-	xe_pcode_write_timeout(gt, mbox, val, 1)
+#define xe_pcode_write(tile, mbox, val) \
+	xe_pcode_write_timeout(tile, mbox, val, 1)
 
-int xe_pcode_request(struct xe_gt *gt, u32 mbox, u32 request,
+int xe_pcode_request(struct xe_tile *tile, u32 mbox, u32 request,
 		     u32 reply_mask, u32 reply, int timeout_ms);
 
 #define PCODE_MBOX(mbcmd, param1, param2)\
diff --git a/drivers/gpu/drm/xe/xe_tile.c b/drivers/gpu/drm/xe/xe_tile.c
index 15ea0a942f67..dda5268507d8 100644
--- a/drivers/gpu/drm/xe/xe_tile.c
+++ b/drivers/gpu/drm/xe/xe_tile.c
@@ -9,6 +9,7 @@
 #include "xe_ggtt.h"
 #include "xe_gt.h"
 #include "xe_migrate.h"
+#include "xe_pcode.h"
 #include "xe_sa.h"
 #include "xe_tile.h"
 #include "xe_tile_sysfs.h"
@@ -124,6 +125,8 @@ int xe_tile_init_early(struct xe_tile *tile, struct xe_device *xe, u8 id)
 	if (IS_ERR(tile->primary_gt))
 		return PTR_ERR(tile->primary_gt);
 
+	xe_pcode_init(tile);
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/xe/xe_vram_freq.c b/drivers/gpu/drm/xe/xe_vram_freq.c
index 99ff95e408e0..b26e26d73dae 100644
--- a/drivers/gpu/drm/xe/xe_vram_freq.c
+++ b/drivers/gpu/drm/xe/xe_vram_freq.c
@@ -34,7 +34,6 @@ static ssize_t max_freq_show(struct device *dev, struct device_attribute *attr,
 			     char *buf)
 {
 	struct xe_tile *tile = dev_to_tile(dev);
-	struct xe_gt *gt = tile->primary_gt;
 	u32 val, mbox;
 	int err;
 
@@ -42,7 +41,7 @@ static ssize_t max_freq_show(struct device *dev, struct device_attribute *attr,
 		| REG_FIELD_PREP(PCODE_MB_PARAM1, PCODE_MBOX_FC_SC_READ_FUSED_P0)
 		| REG_FIELD_PREP(PCODE_MB_PARAM2, PCODE_MBOX_DOMAIN_HBM);
 
-	err = xe_pcode_read(gt, mbox, &val, NULL);
+	err = xe_pcode_read(tile, mbox, &val, NULL);
 	if (err)
 		return err;
 
@@ -57,7 +56,6 @@ static ssize_t min_freq_show(struct device *dev, struct device_attribute *attr,
 			     char *buf)
 {
 	struct xe_tile *tile = dev_to_tile(dev);
-	struct xe_gt *gt = tile->primary_gt;
 	u32 val, mbox;
 	int err;
 
@@ -65,7 +63,7 @@ static ssize_t min_freq_show(struct device *dev, struct device_attribute *attr,
 		| REG_FIELD_PREP(PCODE_MB_PARAM1, PCODE_MBOX_FC_SC_READ_FUSED_PN)
 		| REG_FIELD_PREP(PCODE_MB_PARAM2, PCODE_MBOX_DOMAIN_HBM);
 
-	err = xe_pcode_read(gt, mbox, &val, NULL);
+	err = xe_pcode_read(tile, mbox, &val, NULL);
 	if (err)
 		return err;
 
-- 
2.51.0


From 20f61c1ead989e5d16f35bd0a546c6233b703b69 Mon Sep 17 00:00:00 2001
From: Matt Roper <matthew.d.roper@intel.com>
Date: Thu, 29 Aug 2024 15:06:22 -0700
Subject: [PATCH 06/16] drm/xe/hwmon: Treat hwmon as a per-device concept

There's only one instance of hwmon per device, and MMIO access to it is
always done through the root tile.  The code has been passing around a
pointer to the root tile's primary GT, which is confusing since this
isn't really a GT-level concept.  Replace that pointer with an xe_device
pointer and use xe_root_mmio_gt(xe) to get a pointer when we need to do
register MMIO.  This makes things easier to follow, and also cleans up
the code in preparation for a much larger MMIO register access overhaul
that's coming soon.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829220619.789159-6-matthew.d.roper@intel.com
---
 drivers/gpu/drm/xe/xe_hwmon.c | 95 +++++++++++++++++++----------------
 1 file changed, 53 insertions(+), 42 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_hwmon.c b/drivers/gpu/drm/xe/xe_hwmon.c
index 98e3ec08279e..aa11728e7e79 100644
--- a/drivers/gpu/drm/xe/xe_hwmon.c
+++ b/drivers/gpu/drm/xe/xe_hwmon.c
@@ -12,7 +12,6 @@
 #include "regs/xe_mchbar_regs.h"
 #include "regs/xe_pcode_regs.h"
 #include "xe_device.h"
-#include "xe_gt.h"
 #include "xe_hwmon.h"
 #include "xe_mmio.h"
 #include "xe_pcode.h"
@@ -65,8 +64,8 @@ struct xe_hwmon_energy_info {
 struct xe_hwmon {
 	/** @hwmon_dev: hwmon device for xe */
 	struct device *hwmon_dev;
-	/** @gt: primary gt */
-	struct xe_gt *gt;
+	/** @xe: Xe device */
+	struct xe_device *xe;
 	/** @hwmon_lock: lock for rw attributes*/
 	struct mutex hwmon_lock;
 	/** @scl_shift_power: pkg power unit */
@@ -82,7 +81,7 @@ struct xe_hwmon {
 static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg hwmon_reg,
 				      int channel)
 {
-	struct xe_device *xe = gt_to_xe(hwmon->gt);
+	struct xe_device *xe = hwmon->xe;
 
 	switch (hwmon_reg) {
 	case REG_PKG_RAPL_LIMIT:
@@ -148,8 +147,9 @@ static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg
 static void xe_hwmon_power_max_read(struct xe_hwmon *hwmon, int channel, long *value)
 {
 	u64 reg_val, min, max;
-	struct xe_device *xe = gt_to_xe(hwmon->gt);
+	struct xe_device *xe = hwmon->xe;
 	struct xe_reg rapl_limit, pkg_power_sku;
+	struct xe_gt *mmio = xe_root_mmio_gt(xe);
 
 	rapl_limit = xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel);
 	pkg_power_sku = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel);
@@ -166,7 +166,7 @@ static void xe_hwmon_power_max_read(struct xe_hwmon *hwmon, int channel, long *v
 
 	mutex_lock(&hwmon->hwmon_lock);
 
-	reg_val = xe_mmio_read32(hwmon->gt, rapl_limit);
+	reg_val = xe_mmio_read32(mmio, rapl_limit);
 	/* Check if PL1 limit is disabled */
 	if (!(reg_val & PKG_PWR_LIM_1_EN)) {
 		*value = PL1_DISABLE;
@@ -176,7 +176,7 @@ static void xe_hwmon_power_max_read(struct xe_hwmon *hwmon, int channel, long *v
 	reg_val = REG_FIELD_GET(PKG_PWR_LIM_1, reg_val);
 	*value = mul_u64_u32_shr(reg_val, SF_POWER, hwmon->scl_shift_power);
 
-	reg_val = xe_mmio_read64_2x32(hwmon->gt, pkg_power_sku);
+	reg_val = xe_mmio_read64_2x32(mmio, pkg_power_sku);
 	min = REG_FIELD_GET(PKG_MIN_PWR, reg_val);
 	min = mul_u64_u32_shr(min, SF_POWER, hwmon->scl_shift_power);
 	max = REG_FIELD_GET(PKG_MAX_PWR, reg_val);
@@ -190,6 +190,7 @@ unlock:
 
 static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, int channel, long value)
 {
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	int ret = 0;
 	u64 reg_val;
 	struct xe_reg rapl_limit;
@@ -200,10 +201,10 @@ static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, int channel, long va
 
 	/* Disable PL1 limit and verify, as limit cannot be disabled on all platforms */
 	if (value == PL1_DISABLE) {
-		reg_val = xe_mmio_rmw32(hwmon->gt, rapl_limit, PKG_PWR_LIM_1_EN, 0);
-		reg_val = xe_mmio_read32(hwmon->gt, rapl_limit);
+		reg_val = xe_mmio_rmw32(mmio, rapl_limit, PKG_PWR_LIM_1_EN, 0);
+		reg_val = xe_mmio_read32(mmio, rapl_limit);
 		if (reg_val & PKG_PWR_LIM_1_EN) {
-			drm_warn(&gt_to_xe(hwmon->gt)->drm, "PL1 disable is not supported!\n");
+			drm_warn(&hwmon->xe->drm, "PL1 disable is not supported!\n");
 			ret = -EOPNOTSUPP;
 		}
 		goto unlock;
@@ -212,7 +213,7 @@ static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, int channel, long va
 	/* Computation in 64-bits to avoid overflow. Round to nearest. */
 	reg_val = DIV_ROUND_CLOSEST_ULL((u64)value << hwmon->scl_shift_power, SF_POWER);
 	reg_val = PKG_PWR_LIM_1_EN | REG_FIELD_PREP(PKG_PWR_LIM_1, reg_val);
-	reg_val = xe_mmio_rmw32(hwmon->gt, rapl_limit, PKG_PWR_LIM_1_EN | PKG_PWR_LIM_1, reg_val);
+	reg_val = xe_mmio_rmw32(mmio, rapl_limit, PKG_PWR_LIM_1_EN | PKG_PWR_LIM_1, reg_val);
 
 unlock:
 	mutex_unlock(&hwmon->hwmon_lock);
@@ -221,6 +222,7 @@ unlock:
 
 static void xe_hwmon_power_rated_max_read(struct xe_hwmon *hwmon, int channel, long *value)
 {
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	struct xe_reg reg = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel);
 	u64 reg_val;
 
@@ -229,7 +231,7 @@ static void xe_hwmon_power_rated_max_read(struct xe_hwmon *hwmon, int channel, l
 	 * for this register can be skipped.
 	 * See xe_hwmon_power_is_visible.
 	 */
-	reg_val = xe_mmio_read32(hwmon->gt, reg);
+	reg_val = xe_mmio_read32(mmio, reg);
 	reg_val = REG_FIELD_GET(PKG_TDP, reg_val);
 	*value = mul_u64_u32_shr(reg_val, SF_POWER, hwmon->scl_shift_power);
 }
@@ -257,11 +259,12 @@ static void xe_hwmon_power_rated_max_read(struct xe_hwmon *hwmon, int channel, l
 static void
 xe_hwmon_energy_get(struct xe_hwmon *hwmon, int channel, long *energy)
 {
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	struct xe_hwmon_energy_info *ei = &hwmon->ei[channel];
 	u64 reg_val;
 
-	reg_val = xe_mmio_read32(hwmon->gt, xe_hwmon_get_reg(hwmon, REG_PKG_ENERGY_STATUS,
-							     channel));
+	reg_val = xe_mmio_read32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_ENERGY_STATUS,
+							channel));
 
 	if (reg_val >= ei->reg_val_prev)
 		ei->accum_energy += reg_val - ei->reg_val_prev;
@@ -279,19 +282,20 @@ xe_hwmon_power_max_interval_show(struct device *dev, struct device_attribute *at
 				 char *buf)
 {
 	struct xe_hwmon *hwmon = dev_get_drvdata(dev);
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	u32 x, y, x_w = 2; /* 2 bits */
 	u64 r, tau4, out;
 	int sensor_index = to_sensor_dev_attr(attr)->index;
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	mutex_lock(&hwmon->hwmon_lock);
 
-	r = xe_mmio_read32(hwmon->gt, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index));
+	r = xe_mmio_read32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index));
 
 	mutex_unlock(&hwmon->hwmon_lock);
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	x = REG_FIELD_GET(PKG_PWR_LIM_1_TIME_X, r);
 	y = REG_FIELD_GET(PKG_PWR_LIM_1_TIME_Y, r);
@@ -319,6 +323,7 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a
 				  const char *buf, size_t count)
 {
 	struct xe_hwmon *hwmon = dev_get_drvdata(dev);
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	u32 x, y, rxy, x_w = 2; /* 2 bits */
 	u64 tau4, r, max_win;
 	unsigned long val;
@@ -371,16 +376,16 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a
 
 	rxy = REG_FIELD_PREP(PKG_PWR_LIM_1_TIME_X, x) | REG_FIELD_PREP(PKG_PWR_LIM_1_TIME_Y, y);
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	mutex_lock(&hwmon->hwmon_lock);
 
-	r = xe_mmio_rmw32(hwmon->gt, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index),
+	r = xe_mmio_rmw32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index),
 			  PKG_PWR_LIM_1_TIME, rxy);
 
 	mutex_unlock(&hwmon->hwmon_lock);
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	return count;
 }
@@ -406,11 +411,11 @@ static umode_t xe_hwmon_attributes_visible(struct kobject *kobj,
 	struct xe_hwmon *hwmon = dev_get_drvdata(dev);
 	int ret = 0;
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	ret = xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, index)) ? attr->mode : 0;
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	return ret;
 }
@@ -435,20 +440,24 @@ static const struct hwmon_channel_info * const hwmon_info[] = {
 };
 
 /* I1 is exposed as power_crit or as curr_crit depending on bit 31 */
-static int xe_hwmon_pcode_read_i1(struct xe_gt *gt, u32 *uval)
+static int xe_hwmon_pcode_read_i1(const struct xe_hwmon *hwmon, u32 *uval)
 {
+	struct xe_tile *root_tile = xe_device_get_root_tile(hwmon->xe);
+
 	/* Avoid Illegal Subcommand error */
-	if (gt_to_xe(gt)->info.platform == XE_DG2)
+	if (hwmon->xe->info.platform == XE_DG2)
 		return -ENXIO;
 
-	return xe_pcode_read(gt_to_tile(gt), PCODE_MBOX(PCODE_POWER_SETUP,
+	return xe_pcode_read(root_tile, PCODE_MBOX(PCODE_POWER_SETUP,
 			     POWER_SETUP_SUBCOMMAND_READ_I1, 0),
 			     uval, NULL);
 }
 
-static int xe_hwmon_pcode_write_i1(struct xe_gt *gt, u32 uval)
+static int xe_hwmon_pcode_write_i1(const struct xe_hwmon *hwmon, u32 uval)
 {
-	return xe_pcode_write(gt_to_tile(gt), PCODE_MBOX(PCODE_POWER_SETUP,
+	struct xe_tile *root_tile = xe_device_get_root_tile(hwmon->xe);
+
+	return xe_pcode_write(root_tile, PCODE_MBOX(PCODE_POWER_SETUP,
 			      POWER_SETUP_SUBCOMMAND_WRITE_I1, 0),
 			      (uval & POWER_SETUP_I1_DATA_MASK));
 }
@@ -461,7 +470,7 @@ static int xe_hwmon_power_curr_crit_read(struct xe_hwmon *hwmon, int channel,
 
 	mutex_lock(&hwmon->hwmon_lock);
 
-	ret = xe_hwmon_pcode_read_i1(hwmon->gt, &uval);
+	ret = xe_hwmon_pcode_read_i1(hwmon, &uval);
 	if (ret)
 		goto unlock;
 
@@ -481,7 +490,7 @@ static int xe_hwmon_power_curr_crit_write(struct xe_hwmon *hwmon, int channel,
 	mutex_lock(&hwmon->hwmon_lock);
 
 	uval = DIV_ROUND_CLOSEST_ULL(value << POWER_SETUP_I1_SHIFT, scale_factor);
-	ret = xe_hwmon_pcode_write_i1(hwmon->gt, uval);
+	ret = xe_hwmon_pcode_write_i1(hwmon, uval);
 
 	mutex_unlock(&hwmon->hwmon_lock);
 	return ret;
@@ -489,9 +498,10 @@ static int xe_hwmon_power_curr_crit_write(struct xe_hwmon *hwmon, int channel,
 
 static void xe_hwmon_get_voltage(struct xe_hwmon *hwmon, int channel, long *value)
 {
+	struct xe_gt *mmio = xe_root_mmio_gt(hwmon->xe);
 	u64 reg_val;
 
-	reg_val = xe_mmio_read32(hwmon->gt, xe_hwmon_get_reg(hwmon, REG_GT_PERF_STATUS, channel));
+	reg_val = xe_mmio_read32(mmio, xe_hwmon_get_reg(hwmon, REG_GT_PERF_STATUS, channel));
 	/* HW register value in units of 2.5 millivolt */
 	*value = DIV_ROUND_CLOSEST(REG_FIELD_GET(VOLTAGE_MASK, reg_val) * 2500, SF_VOLTAGE);
 }
@@ -510,7 +520,7 @@ xe_hwmon_power_is_visible(struct xe_hwmon *hwmon, u32 attr, int channel)
 				       channel)) ? 0444 : 0;
 	case hwmon_power_crit:
 		if (channel == CHANNEL_PKG)
-			return (xe_hwmon_pcode_read_i1(hwmon->gt, &uval) ||
+			return (xe_hwmon_pcode_read_i1(hwmon, &uval) ||
 				!(uval & POWER_SETUP_I1_WATTS)) ? 0 : 0644;
 		break;
 	case hwmon_power_label:
@@ -563,10 +573,10 @@ xe_hwmon_curr_is_visible(const struct xe_hwmon *hwmon, u32 attr, int channel)
 
 	switch (attr) {
 	case hwmon_curr_crit:
-			return (xe_hwmon_pcode_read_i1(hwmon->gt, &uval) ||
+			return (xe_hwmon_pcode_read_i1(hwmon, &uval) ||
 				(uval & POWER_SETUP_I1_WATTS)) ? 0 : 0644;
 	case hwmon_curr_label:
-			return (xe_hwmon_pcode_read_i1(hwmon->gt, &uval) ||
+			return (xe_hwmon_pcode_read_i1(hwmon, &uval) ||
 				(uval & POWER_SETUP_I1_WATTS)) ? 0 : 0444;
 		break;
 	default:
@@ -654,7 +664,7 @@ xe_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_types type,
 	struct xe_hwmon *hwmon = (struct xe_hwmon *)drvdata;
 	int ret;
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	switch (type) {
 	case hwmon_power:
@@ -674,7 +684,7 @@ xe_hwmon_is_visible(const void *drvdata, enum hwmon_sensor_types type,
 		break;
 	}
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	return ret;
 }
@@ -686,7 +696,7 @@ xe_hwmon_read(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 	struct xe_hwmon *hwmon = dev_get_drvdata(dev);
 	int ret;
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	switch (type) {
 	case hwmon_power:
@@ -706,7 +716,7 @@ xe_hwmon_read(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 		break;
 	}
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	return ret;
 }
@@ -718,7 +728,7 @@ xe_hwmon_write(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 	struct xe_hwmon *hwmon = dev_get_drvdata(dev);
 	int ret;
 
-	xe_pm_runtime_get(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_get(hwmon->xe);
 
 	switch (type) {
 	case hwmon_power:
@@ -732,7 +742,7 @@ xe_hwmon_write(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 		break;
 	}
 
-	xe_pm_runtime_put(gt_to_xe(hwmon->gt));
+	xe_pm_runtime_put(hwmon->xe);
 
 	return ret;
 }
@@ -771,6 +781,7 @@ static const struct hwmon_chip_info hwmon_chip_info = {
 static void
 xe_hwmon_get_preregistration_info(struct xe_device *xe)
 {
+	struct xe_gt *mmio = xe_root_mmio_gt(xe);
 	struct xe_hwmon *hwmon = xe->hwmon;
 	long energy;
 	u64 val_sku_unit = 0;
@@ -783,7 +794,7 @@ xe_hwmon_get_preregistration_info(struct xe_device *xe)
 	 */
 	pkg_power_sku_unit = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU_UNIT, 0);
 	if (xe_reg_is_valid(pkg_power_sku_unit)) {
-		val_sku_unit = xe_mmio_read32(hwmon->gt, pkg_power_sku_unit);
+		val_sku_unit = xe_mmio_read32(mmio, pkg_power_sku_unit);
 		hwmon->scl_shift_power = REG_FIELD_GET(PKG_PWR_UNIT, val_sku_unit);
 		hwmon->scl_shift_energy = REG_FIELD_GET(PKG_ENERGY_UNIT, val_sku_unit);
 		hwmon->scl_shift_time = REG_FIELD_GET(PKG_TIME_UNIT, val_sku_unit);
@@ -828,8 +839,8 @@ void xe_hwmon_register(struct xe_device *xe)
 	if (devm_add_action_or_reset(dev, xe_hwmon_mutex_destroy, hwmon))
 		return;
 
-	/* primary GT to access device level properties */
-	hwmon->gt = xe->tiles[0].primary_gt;
+	/* There's only one instance of hwmon per device */
+	hwmon->xe = xe;
 
 	xe_hwmon_get_preregistration_info(xe);
 
-- 
2.51.0


From ddc94d0b17e8ea8179ecbbefacac3fba0fb77265 Mon Sep 17 00:00:00 2001
From: Matthew Brost <matthew.brost@intel.com>
Date: Mon, 26 Aug 2024 10:01:43 -0700
Subject: [PATCH 07/16] dma-buf: Split out dma fence array create into alloc
 and arm functions
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

Useful to preallocate dma fence array and then arm in path of reclaim or
a dma fence.

v2:
 - s/arm/init (Christian)
 - Drop !array warn (Christian)
v3:
 - Fix kernel doc typos (dim)

Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Christian König <christian.koenig@amd.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240826170144.2492062-2-matthew.brost@intel.com
---
 drivers/dma-buf/dma-fence-array.c | 78 ++++++++++++++++++++++---------
 include/linux/dma-fence-array.h   |  6 +++
 2 files changed, 63 insertions(+), 21 deletions(-)

diff --git a/drivers/dma-buf/dma-fence-array.c b/drivers/dma-buf/dma-fence-array.c
index c74ac197d5fe..8a08ffde31e7 100644
--- a/drivers/dma-buf/dma-fence-array.c
+++ b/drivers/dma-buf/dma-fence-array.c
@@ -144,37 +144,38 @@ const struct dma_fence_ops dma_fence_array_ops = {
 EXPORT_SYMBOL(dma_fence_array_ops);
 
 /**
- * dma_fence_array_create - Create a custom fence array
+ * dma_fence_array_alloc - Allocate a custom fence array
+ * @num_fences:		[in]	number of fences to add in the array
+ *
+ * Return dma fence array on success, NULL on failure
+ */
+struct dma_fence_array *dma_fence_array_alloc(int num_fences)
+{
+	struct dma_fence_array *array;
+
+	return kzalloc(struct_size(array, callbacks, num_fences), GFP_KERNEL);
+}
+EXPORT_SYMBOL(dma_fence_array_alloc);
+
+/**
+ * dma_fence_array_init - Init a custom fence array
+ * @array:		[in]	dma fence array to arm
  * @num_fences:		[in]	number of fences to add in the array
  * @fences:		[in]	array containing the fences
  * @context:		[in]	fence context to use
  * @seqno:		[in]	sequence number to use
  * @signal_on_any:	[in]	signal on any fence in the array
  *
- * Allocate a dma_fence_array object and initialize the base fence with
- * dma_fence_init().
- * In case of error it returns NULL.
- *
- * The caller should allocate the fences array with num_fences size
- * and fill it with the fences it wants to add to the object. Ownership of this
- * array is taken and dma_fence_put() is used on each fence on release.
- *
- * If @signal_on_any is true the fence array signals if any fence in the array
- * signals, otherwise it signals when all fences in the array signal.
+ * Implementation of @dma_fence_array_create without allocation. Useful to init
+ * a preallocated dma fence array in the path of reclaim or dma fence signaling.
  */
-struct dma_fence_array *dma_fence_array_create(int num_fences,
-					       struct dma_fence **fences,
-					       u64 context, unsigned seqno,
-					       bool signal_on_any)
+void dma_fence_array_init(struct dma_fence_array *array,
+			  int num_fences, struct dma_fence **fences,
+			  u64 context, unsigned seqno,
+			  bool signal_on_any)
 {
-	struct dma_fence_array *array;
-
 	WARN_ON(!num_fences || !fences);
 
-	array = kzalloc(struct_size(array, callbacks, num_fences), GFP_KERNEL);
-	if (!array)
-		return NULL;
-
 	array->num_fences = num_fences;
 
 	spin_lock_init(&array->lock);
@@ -200,6 +201,41 @@ struct dma_fence_array *dma_fence_array_create(int num_fences,
 	 */
 	while (num_fences--)
 		WARN_ON(dma_fence_is_container(fences[num_fences]));
+}
+EXPORT_SYMBOL(dma_fence_array_init);
+
+/**
+ * dma_fence_array_create - Create a custom fence array
+ * @num_fences:		[in]	number of fences to add in the array
+ * @fences:		[in]	array containing the fences
+ * @context:		[in]	fence context to use
+ * @seqno:		[in]	sequence number to use
+ * @signal_on_any:	[in]	signal on any fence in the array
+ *
+ * Allocate a dma_fence_array object and initialize the base fence with
+ * dma_fence_init().
+ * In case of error it returns NULL.
+ *
+ * The caller should allocate the fences array with num_fences size
+ * and fill it with the fences it wants to add to the object. Ownership of this
+ * array is taken and dma_fence_put() is used on each fence on release.
+ *
+ * If @signal_on_any is true the fence array signals if any fence in the array
+ * signals, otherwise it signals when all fences in the array signal.
+ */
+struct dma_fence_array *dma_fence_array_create(int num_fences,
+					       struct dma_fence **fences,
+					       u64 context, unsigned seqno,
+					       bool signal_on_any)
+{
+	struct dma_fence_array *array;
+
+	array = dma_fence_array_alloc(num_fences);
+	if (!array)
+		return NULL;
+
+	dma_fence_array_init(array, num_fences, fences,
+			     context, seqno, signal_on_any);
 
 	return array;
 }
diff --git a/include/linux/dma-fence-array.h b/include/linux/dma-fence-array.h
index 29c5650c1038..079b3dec0a16 100644
--- a/include/linux/dma-fence-array.h
+++ b/include/linux/dma-fence-array.h
@@ -79,6 +79,12 @@ to_dma_fence_array(struct dma_fence *fence)
 	for (index = 0, fence = dma_fence_array_first(head); fence;	\
 	     ++(index), fence = dma_fence_array_next(head, index))
 
+struct dma_fence_array *dma_fence_array_alloc(int num_fences);
+void dma_fence_array_init(struct dma_fence_array *array,
+			  int num_fences, struct dma_fence **fences,
+			  u64 context, unsigned seqno,
+			  bool signal_on_any);
+
 struct dma_fence_array *dma_fence_array_create(int num_fences,
 					       struct dma_fence **fences,
 					       u64 context, unsigned seqno,
-- 
2.51.0


From bf758226c7e6927f1c2c34fd6eb2cb580b77f543 Mon Sep 17 00:00:00 2001
From: Matthew Brost <matthew.brost@intel.com>
Date: Mon, 26 Aug 2024 10:01:44 -0700
Subject: [PATCH 08/16] drm/xe: Invalidate media_gt TLBs in PT code
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

Testing on LNL has shown media GT's TLBs need to be invalidated via the
GuC, update PT code appropriately.

v2:
 - Do dma_fence_get before first call of invalidation_fence_init (Himal)
 - No need to check for valid chain fence (Himal)
v3:
 - Use dma-fence-array

Fixes: 3330361543fc ("drm/xe/lnl: Add LNL platform definition")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240826170144.2492062-3-matthew.brost@intel.com
---
 drivers/gpu/drm/xe/xe_pt.c | 117 ++++++++++++++++++++++++++++++-------
 1 file changed, 96 insertions(+), 21 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c
index 579ed31b46db..d6353e8969f0 100644
--- a/drivers/gpu/drm/xe/xe_pt.c
+++ b/drivers/gpu/drm/xe/xe_pt.c
@@ -3,6 +3,8 @@
  * Copyright © 2022 Intel Corporation
  */
 
+#include <linux/dma-fence-array.h>
+
 #include "xe_pt.h"
 
 #include "regs/xe_gtt_defs.h"
@@ -1627,9 +1629,11 @@ xe_pt_update_ops_rfence_interval(struct xe_vm_pgtable_update_ops *pt_update_ops,
 
 static int vma_reserve_fences(struct xe_device *xe, struct xe_vma *vma)
 {
+	int shift = xe_device_get_root_tile(xe)->media_gt ? 1 : 0;
+
 	if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
 		return dma_resv_reserve_fences(xe_vma_bo(vma)->ttm.base.resv,
-					       xe->info.tile_count);
+					       xe->info.tile_count << shift);
 
 	return 0;
 }
@@ -1816,6 +1820,7 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
 	struct xe_vm_pgtable_update_ops *pt_update_ops =
 		&vops->pt_update_ops[tile->id];
 	struct xe_vma_op *op;
+	int shift = tile->media_gt ? 1 : 0;
 	int err;
 
 	lockdep_assert_held(&vops->vm->lock);
@@ -1824,7 +1829,7 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
 	xe_pt_update_ops_init(pt_update_ops);
 
 	err = dma_resv_reserve_fences(xe_vm_resv(vops->vm),
-				      tile_to_xe(tile)->info.tile_count);
+				      tile_to_xe(tile)->info.tile_count << shift);
 	if (err)
 		return err;
 
@@ -1849,13 +1854,20 @@ int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops)
 
 static void bind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
 			   struct xe_vm_pgtable_update_ops *pt_update_ops,
-			   struct xe_vma *vma, struct dma_fence *fence)
+			   struct xe_vma *vma, struct dma_fence *fence,
+			   struct dma_fence *fence2)
 {
-	if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+	if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm) {
 		dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence,
 				   pt_update_ops->wait_vm_bookkeep ?
 				   DMA_RESV_USAGE_KERNEL :
 				   DMA_RESV_USAGE_BOOKKEEP);
+		if (fence2)
+			dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence2,
+					   pt_update_ops->wait_vm_bookkeep ?
+					   DMA_RESV_USAGE_KERNEL :
+					   DMA_RESV_USAGE_BOOKKEEP);
+	}
 	vma->tile_present |= BIT(tile->id);
 	vma->tile_staged &= ~BIT(tile->id);
 	if (xe_vma_is_userptr(vma)) {
@@ -1875,13 +1887,20 @@ static void bind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
 
 static void unbind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
 			     struct xe_vm_pgtable_update_ops *pt_update_ops,
-			     struct xe_vma *vma, struct dma_fence *fence)
+			     struct xe_vma *vma, struct dma_fence *fence,
+			     struct dma_fence *fence2)
 {
-	if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm)
+	if (!xe_vma_has_no_bo(vma) && !xe_vma_bo(vma)->vm) {
 		dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence,
 				   pt_update_ops->wait_vm_bookkeep ?
 				   DMA_RESV_USAGE_KERNEL :
 				   DMA_RESV_USAGE_BOOKKEEP);
+		if (fence2)
+			dma_resv_add_fence(xe_vma_bo(vma)->ttm.base.resv, fence2,
+					   pt_update_ops->wait_vm_bookkeep ?
+					   DMA_RESV_USAGE_KERNEL :
+					   DMA_RESV_USAGE_BOOKKEEP);
+	}
 	vma->tile_present &= ~BIT(tile->id);
 	if (!vma->tile_present) {
 		list_del_init(&vma->combined_links.rebind);
@@ -1898,7 +1917,8 @@ static void unbind_op_commit(struct xe_vm *vm, struct xe_tile *tile,
 static void op_commit(struct xe_vm *vm,
 		      struct xe_tile *tile,
 		      struct xe_vm_pgtable_update_ops *pt_update_ops,
-		      struct xe_vma_op *op, struct dma_fence *fence)
+		      struct xe_vma_op *op, struct dma_fence *fence,
+		      struct dma_fence *fence2)
 {
 	xe_vm_assert_held(vm);
 
@@ -1907,26 +1927,28 @@ static void op_commit(struct xe_vm *vm,
 		if (!op->map.immediate && xe_vm_in_fault_mode(vm))
 			break;
 
-		bind_op_commit(vm, tile, pt_update_ops, op->map.vma, fence);
+		bind_op_commit(vm, tile, pt_update_ops, op->map.vma, fence,
+			       fence2);
 		break;
 	case DRM_GPUVA_OP_REMAP:
 		unbind_op_commit(vm, tile, pt_update_ops,
-				 gpuva_to_vma(op->base.remap.unmap->va), fence);
+				 gpuva_to_vma(op->base.remap.unmap->va), fence,
+				 fence2);
 
 		if (op->remap.prev)
 			bind_op_commit(vm, tile, pt_update_ops, op->remap.prev,
-				       fence);
+				       fence, fence2);
 		if (op->remap.next)
 			bind_op_commit(vm, tile, pt_update_ops, op->remap.next,
-				       fence);
+				       fence, fence2);
 		break;
 	case DRM_GPUVA_OP_UNMAP:
 		unbind_op_commit(vm, tile, pt_update_ops,
-				 gpuva_to_vma(op->base.unmap.va), fence);
+				 gpuva_to_vma(op->base.unmap.va), fence, fence2);
 		break;
 	case DRM_GPUVA_OP_PREFETCH:
 		bind_op_commit(vm, tile, pt_update_ops,
-			       gpuva_to_vma(op->base.prefetch.va), fence);
+			       gpuva_to_vma(op->base.prefetch.va), fence, fence2);
 		break;
 	default:
 		drm_warn(&vm->xe->drm, "NOT POSSIBLE");
@@ -1963,7 +1985,9 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
 	struct xe_vm_pgtable_update_ops *pt_update_ops =
 		&vops->pt_update_ops[tile->id];
 	struct dma_fence *fence;
-	struct invalidation_fence *ifence = NULL;
+	struct invalidation_fence *ifence = NULL, *mfence = NULL;
+	struct dma_fence **fences = NULL;
+	struct dma_fence_array *cf = NULL;
 	struct xe_range_fence *rfence;
 	struct xe_vma_op *op;
 	int err = 0, i;
@@ -1996,6 +2020,23 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
 			err = -ENOMEM;
 			goto kill_vm_tile1;
 		}
+		if (tile->media_gt) {
+			mfence = kzalloc(sizeof(*ifence), GFP_KERNEL);
+			if (!mfence) {
+				err = -ENOMEM;
+				goto free_ifence;
+			}
+			fences = kmalloc_array(2, sizeof(*fences), GFP_KERNEL);
+			if (!fences) {
+				err = -ENOMEM;
+				goto free_ifence;
+			}
+			cf = dma_fence_array_alloc(2);
+			if (!cf) {
+				err = -ENOMEM;
+				goto free_ifence;
+			}
+		}
 	}
 
 	rfence = kzalloc(sizeof(*rfence), GFP_KERNEL);
@@ -2027,19 +2068,50 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
 
 	/* tlb invalidation must be done before signaling rebind */
 	if (ifence) {
+		if (mfence)
+			dma_fence_get(fence);
 		invalidation_fence_init(tile->primary_gt, ifence, fence,
 					pt_update_ops->start,
 					pt_update_ops->last, vm->usm.asid);
-		fence = &ifence->base.base;
+		if (mfence) {
+			invalidation_fence_init(tile->media_gt, mfence, fence,
+						pt_update_ops->start,
+						pt_update_ops->last, vm->usm.asid);
+			fences[0] = &ifence->base.base;
+			fences[1] = &mfence->base.base;
+			dma_fence_array_init(cf, 2, fences,
+					     vm->composite_fence_ctx,
+					     vm->composite_fence_seqno++,
+					     false);
+			fence = &cf->base;
+		} else {
+			fence = &ifence->base.base;
+		}
 	}
 
-	dma_resv_add_fence(xe_vm_resv(vm), fence,
-			   pt_update_ops->wait_vm_bookkeep ?
-			   DMA_RESV_USAGE_KERNEL :
-			   DMA_RESV_USAGE_BOOKKEEP);
+	if (!mfence) {
+		dma_resv_add_fence(xe_vm_resv(vm), fence,
+				   pt_update_ops->wait_vm_bookkeep ?
+				   DMA_RESV_USAGE_KERNEL :
+				   DMA_RESV_USAGE_BOOKKEEP);
 
-	list_for_each_entry(op, &vops->list, link)
-		op_commit(vops->vm, tile, pt_update_ops, op, fence);
+		list_for_each_entry(op, &vops->list, link)
+			op_commit(vops->vm, tile, pt_update_ops, op, fence, NULL);
+	} else {
+		dma_resv_add_fence(xe_vm_resv(vm), &ifence->base.base,
+				   pt_update_ops->wait_vm_bookkeep ?
+				   DMA_RESV_USAGE_KERNEL :
+				   DMA_RESV_USAGE_BOOKKEEP);
+
+		dma_resv_add_fence(xe_vm_resv(vm), &mfence->base.base,
+				   pt_update_ops->wait_vm_bookkeep ?
+				   DMA_RESV_USAGE_KERNEL :
+				   DMA_RESV_USAGE_BOOKKEEP);
+
+		list_for_each_entry(op, &vops->list, link)
+			op_commit(vops->vm, tile, pt_update_ops, op,
+				  &ifence->base.base, &mfence->base.base);
+	}
 
 	if (pt_update_ops->needs_userptr_lock)
 		up_read(&vm->userptr.notifier_lock);
@@ -2049,6 +2121,9 @@ xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops)
 free_rfence:
 	kfree(rfence);
 free_ifence:
+	kfree(cf);
+	kfree(fences);
+	kfree(mfence);
 	kfree(ifence);
 kill_vm_tile1:
 	if (err != -EAGAIN && tile->id)
-- 
2.51.0


From a1498ab229ca16dfc35331dfb309f8e28b84bb5d Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Fri, 30 Aug 2024 15:20:59 +0200
Subject: [PATCH 09/16] drm/xe/pf: Add thresholds to the VF KLV config
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

We are pushing threshold KLV to the GuC immediately during the
threshold provisioning, but those configs will be lost during a
GT reset.  Include threshold KLVs while encoding full VF config
buffer to make sure the GuC receives all of the config KLVs.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240830132100.1704-2-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
index 497af3949b19..151eb69ab4ae 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
@@ -277,6 +277,14 @@ static u32 encode_config(u32 *cfg, const struct xe_gt_sriov_config *config)
 	cfg[n++] = PREP_GUC_KLV_TAG(VF_CFG_PREEMPT_TIMEOUT);
 	cfg[n++] = config->preempt_timeout;
 
+#define encode_threshold_config(TAG, ...) ({					\
+	cfg[n++] = PREP_GUC_KLV_TAG(VF_CFG_THRESHOLD_##TAG);			\
+	cfg[n++] = config->thresholds[MAKE_XE_GUC_KLV_THRESHOLD_INDEX(TAG)];	\
+});
+
+	MAKE_XE_GUC_KLV_THRESHOLDS_SET(encode_threshold_config);
+#undef encode_threshold_config
+
 	return n;
 }
 
-- 
2.51.0


From da6ec743397702e0b551bcacfeaa48829d303a29 Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Fri, 30 Aug 2024 15:21:00 +0200
Subject: [PATCH 10/16] drm/xe/pf: Reset thresholds when releasing a VF config
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

As part of the VF config release, we should reset all parameters,
including thresholds, to always start with the clean VF config.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240830132100.1704-3-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
index 151eb69ab4ae..a95e546b7744 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c
@@ -1842,6 +1842,18 @@ u32 xe_gt_sriov_pf_config_get_threshold(struct xe_gt *gt, unsigned int vfid,
 	return value;
 }
 
+static void pf_reset_config_thresholds(struct xe_gt *gt, struct xe_gt_sriov_config *config)
+{
+	lockdep_assert_held(xe_gt_sriov_pf_master_mutex(gt));
+
+#define reset_threshold_config(TAG, ...) ({				\
+	config->thresholds[MAKE_XE_GUC_KLV_THRESHOLD_INDEX(TAG)] = 0;	\
+});
+
+	MAKE_XE_GUC_KLV_THRESHOLDS_SET(reset_threshold_config);
+#undef reset_threshold_config
+}
+
 static void pf_release_vf_config(struct xe_gt *gt, unsigned int vfid)
 {
 	struct xe_gt_sriov_config *config = pf_pick_vf_config(gt, vfid);
@@ -1857,6 +1869,7 @@ static void pf_release_vf_config(struct xe_gt *gt, unsigned int vfid)
 	pf_release_config_ctxs(gt, config);
 	pf_release_config_dbs(gt, config);
 	pf_reset_config_sched(gt, config);
+	pf_reset_config_thresholds(gt, config);
 }
 
 /**
-- 
2.51.0


From 9f6b47907e2d01bfa90c3003e42b6dfeefd8d03a Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Mon, 2 Sep 2024 21:07:26 +0200
Subject: [PATCH 11/16] drm/xe: Remove redundant [drm] tag from xe_assert()
 message

Since commit 178c0a33c421 ("drm/print: Add generic drm dev printk
function") the output from drm_WARN() includes previously missing
the [drm] tag, so now xe_assert() is printing it twice:

  [ ] xe 0000:00:02.0: [drm] [drm] Assertion `false` failed!

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240902190726.1748-1-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_assert.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_assert.h b/drivers/gpu/drm/xe/xe_assert.h
index 8b0cc1bc9327..e22bbf57fca7 100644
--- a/drivers/gpu/drm/xe/xe_assert.h
+++ b/drivers/gpu/drm/xe/xe_assert.h
@@ -81,7 +81,7 @@
 
 #if IS_ENABLED(CONFIG_DRM_XE_DEBUG)
 #define __xe_assert_msg(xe, condition, msg, arg...) ({						\
-	(void)drm_WARN(&(xe)->drm, !(condition), "[" DRM_NAME "] Assertion `%s` failed!\n" msg,	\
+	(void)drm_WARN(&(xe)->drm, !(condition), "Assertion `%s` failed!\n" msg,		\
 		       __stringify(condition), ## arg);						\
 })
 #else
-- 
2.51.0


From bc947d9a8c3ebd207e52c0e35cfc88f3e1abe54f Mon Sep 17 00:00:00 2001
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Date: Fri, 30 Aug 2024 14:35:06 -0400
Subject: [PATCH 12/16] drm/xe: Add missing runtime reference to wedged upon
 gt_reset

Fixes this missed case:

xe 0000:00:02.0: [drm] Missing outer runtime PM protection
WARNING: CPU: 99 PID: 1455 at drivers/gpu/drm/xe/xe_pm.c:564 xe_pm_runtime_get_noresume+0x48/0x60 [xe]
Call Trace:
<TASK>
? show_regs+0x67/0x70
? __warn+0x94/0x1b0
? xe_pm_runtime_get_noresume+0x48/0x60 [xe]
? report_bug+0x1b7/0x1d0
? handle_bug+0x46/0x80
? exc_invalid_op+0x19/0x70
? asm_exc_invalid_op+0x1b/0x20
? xe_pm_runtime_get_noresume+0x48/0x60 [xe]
xe_device_declare_wedged+0x91/0x280 [xe]
gt_reset_worker+0xa2/0x250 [xe]

v2: Also move get and get the right Fixes tag (Himal, Brost)

Fixes: fb74b205cdd2 ("drm/xe: Introduce a simple wedged state")
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240830183507.298351-1-rodrigo.vivi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
---
 drivers/gpu/drm/xe/xe_gt.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_gt.c b/drivers/gpu/drm/xe/xe_gt.c
index ffacddb587b2..f0dc2bf24c7b 100644
--- a/drivers/gpu/drm/xe/xe_gt.c
+++ b/drivers/gpu/drm/xe/xe_gt.c
@@ -752,12 +752,13 @@ static int gt_reset(struct xe_gt *gt)
 
 	xe_gt_info(gt, "reset started\n");
 
+	xe_pm_runtime_get(gt_to_xe(gt));
+
 	if (xe_fault_inject_gt_reset()) {
 		err = -ECANCELED;
 		goto err_fail;
 	}
 
-	xe_pm_runtime_get(gt_to_xe(gt));
 	xe_gt_sanitize(gt);
 
 	err = xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL);
@@ -792,11 +793,11 @@ err_out:
 	XE_WARN_ON(xe_force_wake_put(gt_to_fw(gt), XE_FORCEWAKE_ALL));
 err_msg:
 	XE_WARN_ON(xe_uc_start(&gt->uc));
-	xe_pm_runtime_put(gt_to_xe(gt));
 err_fail:
 	xe_gt_err(gt, "reset failed (%pe)\n", ERR_PTR(err));
 
 	xe_device_declare_wedged(gt_to_xe(gt));
+	xe_pm_runtime_put(gt_to_xe(gt));
 
 	return err;
 }
-- 
2.51.0


From 8da19441d0a02b53e362df81843bb20db3a8006a Mon Sep 17 00:00:00 2001
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Date: Fri, 30 Aug 2024 14:35:07 -0400
Subject: [PATCH 13/16] drm/xe/display: Avoid encoder_suspend at runtime
 suspend

Fix circular locking dependency on runtime suspend.

<4> [74.952215] ======================================================
<4> [74.952217] WARNING: possible circular locking dependency detected
<4> [74.952219] 6.10.0-rc7-xe #1 Not tainted
<4> [74.952221] ------------------------------------------------------
<4> [74.952223] kworker/7:1/82 is trying to acquire lock:
<4> [74.952226] ffff888120548488 (&dev->mode_config.mutex){+.+.}-{3:3}, at: drm_modeset_lock_all+0x40/0x1e0 [drm]
<4> [74.952260]
but task is already holding lock:
<4> [74.952262] ffffffffa0ae59c0 (xe_pm_runtime_lockdep_map){+.+.}-{0:0}, at: xe_pm_runtime_suspend+0x2f/0x340 [xe]
<4> [74.952322]
which lock already depends on the new lock.

The commit 'b1d90a86 ("drm/xe: Use the encoder suspend helper also used
by the i915 driver")' didn't do anything wrong. It actually fixed a
critical bug, because the encoder_suspend was never getting actually
called because it was returning if (has_display(xe)) instead of
if (!has_display(xe)). However, this ended up introducing the encoder
suspend calls in the runtime routines as well, causing the circular
locking dependency.

Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2304
Fixes: b1d90a862c89 ("drm/xe: Use the encoder suspend helper also used by the i915 driver")
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240830183507.298351-2-rodrigo.vivi@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
---
 drivers/gpu/drm/xe/display/xe_display.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c
index ae92012253f8..75736faf2a80 100644
--- a/drivers/gpu/drm/xe/display/xe_display.c
+++ b/drivers/gpu/drm/xe/display/xe_display.c
@@ -345,10 +345,10 @@ void xe_display_pm_suspend(struct xe_device *xe, bool runtime)
 
 	intel_hpd_cancel_work(xe);
 
-	if (!runtime && has_display(xe))
+	if (!runtime && has_display(xe)) {
 		intel_display_driver_suspend_access(xe);
-
-	intel_encoder_suspend_all(&xe->display);
+		intel_encoder_suspend_all(&xe->display);
+	}
 
 	intel_opregion_suspend(display, s2idle ? PCI_D1 : PCI_D3cold);
 
-- 
2.51.0


From 34bb7b813ab398106f700b0a6b218509bb0b904c Mon Sep 17 00:00:00 2001
From: =?utf8?q?Thomas=20Hellstr=C3=B6m?= <thomas.hellstrom@linux.intel.com>
Date: Tue, 3 Sep 2024 11:42:32 +0200
Subject: [PATCH 14/16] drm/xe: Use xe_pm_runtime_get in xe_bo_move() if
 reclaim-safe.
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

xe_bo_move() might be called in the TTM swapout path from validation
by another TTM device. If so, we are not likely to have a RPM
reference. So iff xe_pm_runtime_get() is safe to call from reclaim,
use it instead of xe_pm_runtime_get_noresume().

Strictly this is currently needed only if handle_system_ccs is true,
but use xe_pm_runtime_get() if possible anyway to increase test
coverage.

At the same time warn if handle_system_ccs is true and we can't
call xe_pm_runtime_get() from reclaim context. This will likely trip
if someone tries to enable SRIOV on LNL, without fixing Xe SRIOV
runtime resume / suspend.

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240903094232.166342-1-thomas.hellstrom@linux.intel.com
---
 drivers/gpu/drm/xe/xe_bo.c | 11 ++++++++++-
 drivers/gpu/drm/xe/xe_pm.c |  9 ++++++++-
 drivers/gpu/drm/xe/xe_pm.h |  1 +
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
index 1c18ba9bd099..0c64d3b3155e 100644
--- a/drivers/gpu/drm/xe/xe_bo.c
+++ b/drivers/gpu/drm/xe/xe_bo.c
@@ -758,7 +758,16 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 
 	xe_assert(xe, migrate);
 	trace_xe_bo_move(bo, new_mem->mem_type, old_mem_type, move_lacks_source);
-	xe_pm_runtime_get_noresume(xe);
+	if (xe_rpm_reclaim_safe(xe)) {
+		/*
+		 * We might be called through swapout in the validation path of
+		 * another TTM device, so unconditionally acquire rpm here.
+		 */
+		xe_pm_runtime_get(xe);
+	} else {
+		drm_WARN_ON(&xe->drm, handle_system_ccs);
+		xe_pm_runtime_get_noresume(xe);
+	}
 
 	if (xe_bo_is_pinned(bo) && !xe_bo_is_user(bo)) {
 		/*
diff --git a/drivers/gpu/drm/xe/xe_pm.c b/drivers/gpu/drm/xe/xe_pm.c
index 2600c936527e..e518557e0eec 100644
--- a/drivers/gpu/drm/xe/xe_pm.c
+++ b/drivers/gpu/drm/xe/xe_pm.c
@@ -79,7 +79,14 @@ static struct lockdep_map xe_pm_runtime_nod3cold_map = {
 };
 #endif
 
-static bool __maybe_unused xe_rpm_reclaim_safe(const struct xe_device *xe)
+/**
+ * xe_rpm_reclaim_safe() - Whether runtime resume can be done from reclaim context
+ * @xe: The xe device.
+ *
+ * Return: true if it is safe to runtime resume from reclaim context.
+ * false otherwise.
+ */
+bool xe_rpm_reclaim_safe(const struct xe_device *xe)
 {
 	return !xe->d3cold.capable && !xe->info.has_sriov;
 }
diff --git a/drivers/gpu/drm/xe/xe_pm.h b/drivers/gpu/drm/xe/xe_pm.h
index 9aef673b1c8a..998d1ed64556 100644
--- a/drivers/gpu/drm/xe/xe_pm.h
+++ b/drivers/gpu/drm/xe/xe_pm.h
@@ -31,6 +31,7 @@ bool xe_pm_runtime_resume_and_get(struct xe_device *xe);
 void xe_pm_assert_unbounded_bridge(struct xe_device *xe);
 int xe_pm_set_vram_threshold(struct xe_device *xe, u32 threshold);
 void xe_pm_d3cold_allowed_toggle(struct xe_device *xe);
+bool xe_rpm_reclaim_safe(const struct xe_device *xe);
 struct task_struct *xe_pm_read_callback_task(struct xe_device *xe);
 int xe_pm_module_init(void);
 
-- 
2.51.0


From 13a48a0fa52352f9fe58e2e1927670dcfea64c3a Mon Sep 17 00:00:00 2001
From: Michal Wajdeczko <michal.wajdeczko@intel.com>
Date: Mon, 2 Sep 2024 21:29:53 +0200
Subject: [PATCH 15/16] drm/xe/pf: Sanitize VF scratch registers on FLR
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

Some VF accessible registers (like GuC scratch registers) must be
explicitly reset during the FLR. While this is today done by the GuC
firmware, according to the design, this should be responsibility of
the PF driver, as future platforms may require more registers to be
reset. Likewise GuC, the PF can access VFs registers by adding some
platform specific offset to the original register address.

Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240902192953.1792-1-michal.wajdeczko@intel.com
---
 drivers/gpu/drm/xe/xe_gt_sriov_pf.c         | 52 +++++++++++++++++++++
 drivers/gpu/drm/xe/xe_gt_sriov_pf.h         |  1 +
 drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c |  3 +-
 3 files changed, 55 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
index 905f409db74b..919d960165d5 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c
@@ -5,8 +5,10 @@
 
 #include <drm/drm_managed.h>
 
+#include "regs/xe_guc_regs.h"
 #include "regs/xe_regs.h"
 
+#include "xe_gt.h"
 #include "xe_gt_sriov_pf.h"
 #include "xe_gt_sriov_pf_config.h"
 #include "xe_gt_sriov_pf_control.h"
@@ -89,6 +91,56 @@ void xe_gt_sriov_pf_init_hw(struct xe_gt *gt)
 	xe_gt_sriov_pf_service_update(gt);
 }
 
+static u32 pf_get_vf_regs_stride(struct xe_device *xe)
+{
+	return GRAPHICS_VERx100(xe) > 1200 ? 0x400 : 0x1000;
+}
+
+static struct xe_reg xe_reg_vf_to_pf(struct xe_reg vf_reg, unsigned int vfid, u32 stride)
+{
+	struct xe_reg pf_reg = vf_reg;
+
+	pf_reg.vf = 0;
+	pf_reg.addr += stride * vfid;
+
+	return pf_reg;
+}
+
+static void pf_clear_vf_scratch_regs(struct xe_gt *gt, unsigned int vfid)
+{
+	u32 stride = pf_get_vf_regs_stride(gt_to_xe(gt));
+	struct xe_reg scratch;
+	int n, count;
+
+	if (xe_gt_is_media_type(gt)) {
+		count = MED_VF_SW_FLAG_COUNT;
+		for (n = 0; n < count; n++) {
+			scratch = xe_reg_vf_to_pf(MED_VF_SW_FLAG(n), vfid, stride);
+			xe_mmio_write32(gt, scratch, 0);
+		}
+	} else {
+		count = VF_SW_FLAG_COUNT;
+		for (n = 0; n < count; n++) {
+			scratch = xe_reg_vf_to_pf(VF_SW_FLAG(n), vfid, stride);
+			xe_mmio_write32(gt, scratch, 0);
+		}
+	}
+}
+
+/**
+ * xe_gt_sriov_pf_sanitize_hw() - Reset hardware state related to a VF.
+ * @gt: the &xe_gt
+ * @vfid: the VF identifier
+ *
+ * This function can only be called on PF.
+ */
+void xe_gt_sriov_pf_sanitize_hw(struct xe_gt *gt, unsigned int vfid)
+{
+	xe_gt_assert(gt, IS_SRIOV_PF(gt_to_xe(gt)));
+
+	pf_clear_vf_scratch_regs(gt, vfid);
+}
+
 /**
  * xe_gt_sriov_pf_restart - Restart SR-IOV support after a GT reset.
  * @gt: the &xe_gt
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf.h
index f0cb726a6919..96fab779a906 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf.h
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf.h
@@ -11,6 +11,7 @@ struct xe_gt;
 #ifdef CONFIG_PCI_IOV
 int xe_gt_sriov_pf_init_early(struct xe_gt *gt);
 void xe_gt_sriov_pf_init_hw(struct xe_gt *gt);
+void xe_gt_sriov_pf_sanitize_hw(struct xe_gt *gt, unsigned int vfid);
 void xe_gt_sriov_pf_restart(struct xe_gt *gt);
 #else
 static inline int xe_gt_sriov_pf_init_early(struct xe_gt *gt)
diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
index 02f7328bd6ce..b4fd5a81aff1 100644
--- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
+++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_control.c
@@ -9,6 +9,7 @@
 
 #include "xe_device.h"
 #include "xe_gt.h"
+#include "xe_gt_sriov_pf.h"
 #include "xe_gt_sriov_pf_config.h"
 #include "xe_gt_sriov_pf_control.h"
 #include "xe_gt_sriov_pf_helpers.h"
@@ -1008,7 +1009,7 @@ static bool pf_exit_vf_flr_reset_mmio(struct xe_gt *gt, unsigned int vfid)
 	if (!pf_exit_vf_state(gt, vfid, XE_GT_SRIOV_STATE_FLR_RESET_MMIO))
 		return false;
 
-	/* XXX: placeholder */
+	xe_gt_sriov_pf_sanitize_hw(gt, vfid);
 
 	pf_enter_vf_flr_send_finish(gt, vfid);
 	return true;
-- 
2.51.0


From cd89de14bbacce1fc060fdfab75bacf95b1c5d40 Mon Sep 17 00:00:00 2001
From: Nitin Gote <nitin.r.gote@intel.com>
Date: Fri, 23 Aug 2024 13:36:43 +0530
Subject: [PATCH 16/16] drm/xe: Replace double space with single space after
 comma

Avoid using double space, ",  " in function or macro parameters
where it's not required by any alignment purpose. Replace it with
a single space, ", ".

Signed-off-by: Nitin Gote <nitin.r.gote@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240823080643.2461992-1-nitin.r.gote@intel.com
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
---
 drivers/gpu/drm/xe/regs/xe_reg_defs.h | 2 +-
 drivers/gpu/drm/xe/xe_guc.c           | 2 +-
 drivers/gpu/drm/xe/xe_guc_ct.c        | 4 ++--
 drivers/gpu/drm/xe/xe_irq.c           | 4 ++--
 drivers/gpu/drm/xe/xe_trace_bo.h      | 2 +-
 5 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/xe/regs/xe_reg_defs.h b/drivers/gpu/drm/xe/regs/xe_reg_defs.h
index 23f7dc5bbe99..51fd40ffafcb 100644
--- a/drivers/gpu/drm/xe/regs/xe_reg_defs.h
+++ b/drivers/gpu/drm/xe/regs/xe_reg_defs.h
@@ -128,7 +128,7 @@ struct xe_reg_mcr {
  *       options.
  */
 #define XE_REG_MCR(r_, ...)	((const struct xe_reg_mcr){					\
-				 .__reg = XE_REG_INITIALIZER(r_,  ##__VA_ARGS__, .mcr = 1)	\
+				 .__reg = XE_REG_INITIALIZER(r_, ##__VA_ARGS__, .mcr = 1)	\
 				 })
 
 static inline bool xe_reg_is_valid(struct xe_reg r)
diff --git a/drivers/gpu/drm/xe/xe_guc.c b/drivers/gpu/drm/xe/xe_guc.c
index 52df28032a6f..c67d4807f37d 100644
--- a/drivers/gpu/drm/xe/xe_guc.c
+++ b/drivers/gpu/drm/xe/xe_guc.c
@@ -985,7 +985,7 @@ timeout:
 		BUILD_BUG_ON(FIELD_MAX(GUC_HXG_MSG_0_TYPE) != GUC_HXG_TYPE_RESPONSE_SUCCESS);
 		BUILD_BUG_ON((GUC_HXG_TYPE_RESPONSE_SUCCESS ^ GUC_HXG_TYPE_RESPONSE_FAILURE) != 1);
 
-		ret = xe_mmio_wait32(gt, reply_reg,  resp_mask, resp_mask,
+		ret = xe_mmio_wait32(gt, reply_reg, resp_mask, resp_mask,
 				     1000000, &header, false);
 
 		if (unlikely(FIELD_GET(GUC_HXG_MSG_0_ORIGIN, header) !=
diff --git a/drivers/gpu/drm/xe/xe_guc_ct.c b/drivers/gpu/drm/xe/xe_guc_ct.c
index f24dd5223926..4b95f75b1546 100644
--- a/drivers/gpu/drm/xe/xe_guc_ct.c
+++ b/drivers/gpu/drm/xe/xe_guc_ct.c
@@ -182,7 +182,7 @@ int xe_guc_ct_init(struct xe_guc_ct *ct)
 	spin_lock_init(&ct->fast_lock);
 	xa_init(&ct->fence_lookup);
 	INIT_WORK(&ct->g2h_worker, g2h_worker_func);
-	INIT_DELAYED_WORK(&ct->safe_mode_worker,  safe_mode_worker_func);
+	INIT_DELAYED_WORK(&ct->safe_mode_worker, safe_mode_worker_func);
 	init_waitqueue_head(&ct->wq);
 	init_waitqueue_head(&ct->g2h_fence_wq);
 
@@ -852,7 +852,7 @@ static bool retry_failure(struct xe_guc_ct *ct, int ret)
 #define ct_alive(ct)	\
 	(xe_guc_ct_enabled(ct) && !ct->ctbs.h2g.info.broken && \
 	 !ct->ctbs.g2h.info.broken)
-	if (!wait_event_interruptible_timeout(ct->wq, ct_alive(ct),  HZ * 5))
+	if (!wait_event_interruptible_timeout(ct->wq, ct_alive(ct), HZ * 5))
 		return false;
 #undef ct_alive
 
diff --git a/drivers/gpu/drm/xe/xe_irq.c b/drivers/gpu/drm/xe/xe_irq.c
index 5f2c368c35ad..14c3a476597a 100644
--- a/drivers/gpu/drm/xe/xe_irq.c
+++ b/drivers/gpu/drm/xe/xe_irq.c
@@ -173,7 +173,7 @@ void xe_irq_enable_hwe(struct xe_gt *gt)
 		if (ccs_mask & (BIT(0)|BIT(1)))
 			xe_mmio_write32(gt, CCS0_CCS1_INTR_MASK, ~dmask);
 		if (ccs_mask & (BIT(2)|BIT(3)))
-			xe_mmio_write32(gt,  CCS2_CCS3_INTR_MASK, ~dmask);
+			xe_mmio_write32(gt, CCS2_CCS3_INTR_MASK, ~dmask);
 	}
 
 	if (xe_gt_is_media_type(gt) || MEDIA_VER(xe) < 13) {
@@ -504,7 +504,7 @@ static void gt_irq_reset(struct xe_tile *tile)
 	if (ccs_mask & (BIT(0)|BIT(1)))
 		xe_mmio_write32(mmio, CCS0_CCS1_INTR_MASK, ~0);
 	if (ccs_mask & (BIT(2)|BIT(3)))
-		xe_mmio_write32(mmio,  CCS2_CCS3_INTR_MASK, ~0);
+		xe_mmio_write32(mmio, CCS2_CCS3_INTR_MASK, ~0);
 
 	if ((tile->media_gt &&
 	     xe_hw_engine_mask_per_class(tile->media_gt, XE_ENGINE_CLASS_OTHER)) ||
diff --git a/drivers/gpu/drm/xe/xe_trace_bo.h b/drivers/gpu/drm/xe/xe_trace_bo.h
index 9b1a1d4304ae..30a3cfbaaa09 100644
--- a/drivers/gpu/drm/xe/xe_trace_bo.h
+++ b/drivers/gpu/drm/xe/xe_trace_bo.h
@@ -189,7 +189,7 @@ DECLARE_EVENT_CLASS(xe_vm,
 			   ),
 
 		    TP_printk("dev=%s, vm=%p, asid=0x%05x", __get_str(dev),
-			      __entry->vm,  __entry->asid)
+			      __entry->vm, __entry->asid)
 );
 
 DEFINE_EVENT(xe_vm, xe_vm_kill,
-- 
2.51.0