Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Gustavo Sousa <gustavo.sousa@intel.com>
To: intel-gfx@lists.freedesktop.org, intel-xe@lists.freedesktop.org
Cc: Luca Coelho <luciano.coelho@intel.com>,
	Jani Nikula <jani.nikula@intel.com>
Subject: [PATCH v2 03/17] drm/i915/dmc_wl: Use non-sleeping variant of MMIO wait
Date: Wed,  6 Nov 2024 18:50:29 -0300	[thread overview]
Message-ID: <20241106215231.103474-4-gustavo.sousa@intel.com> (raw)
In-Reply-To: <20241106215231.103474-1-gustavo.sousa@intel.com>

Some display MMIO transactions for offsets in the range that requires
the DMC wakelock happen in atomic context (this has been confirmed
during tests on PTL). That means that we need to use a non-sleeping
variant of MMIO waiting function.

Implement __intel_de_wait_for_register_atomic_nowl() and use it when
waiting for acknowledgment of acquire/release.

v2:
  - No __intel_de_wait_for_register_atomic_nowl() wrapper to convert
    i915 to display. (Jani)
  - Add a quick explanation why DMC_WAKELOCK_CTL_TIMEOUT_US is defined
    in microseconds. (Luca)

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
---
 drivers/gpu/drm/i915/display/intel_de.h     | 10 +++++++++
 drivers/gpu/drm/i915/display/intel_dmc_wl.c | 24 ++++++++++++++-------
 2 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/i915/display/intel_de.h b/drivers/gpu/drm/i915/display/intel_de.h
index bb51f974e9e2..4561de5d5e10 100644
--- a/drivers/gpu/drm/i915/display/intel_de.h
+++ b/drivers/gpu/drm/i915/display/intel_de.h
@@ -117,6 +117,16 @@ __intel_de_wait_for_register_nowl(struct intel_display *display,
 				       value, timeout);
 }
 
+static inline int
+__intel_de_wait_for_register_atomic_nowl(struct intel_display *display,
+					 i915_reg_t reg,
+					 u32 mask, u32 value,
+					 unsigned int fast_timeout_us)
+{
+	return __intel_wait_for_register(__to_uncore(display), reg, mask,
+					 value, fast_timeout_us, 0, NULL);
+}
+
 static inline int
 intel_de_wait(struct intel_display *display, i915_reg_t reg,
 	      u32 mask, u32 value, unsigned int timeout)
diff --git a/drivers/gpu/drm/i915/display/intel_dmc_wl.c b/drivers/gpu/drm/i915/display/intel_dmc_wl.c
index 05892a237d3a..9255505437d5 100644
--- a/drivers/gpu/drm/i915/display/intel_dmc_wl.c
+++ b/drivers/gpu/drm/i915/display/intel_dmc_wl.c
@@ -39,7 +39,11 @@
  * potential future use.
  */
 
-#define DMC_WAKELOCK_CTL_TIMEOUT 5
+/*
+ * Define DMC_WAKELOCK_CTL_TIMEOUT_US in microseconds because we use the
+ * atomic variant of waiting MMIO.
+ */
+#define DMC_WAKELOCK_CTL_TIMEOUT_US 5000
 #define DMC_WAKELOCK_HOLD_TIME 50
 
 struct intel_dmc_wl_range {
@@ -78,9 +82,9 @@ static void intel_dmc_wl_work(struct work_struct *work)
 
 	__intel_de_rmw_nowl(display, DMC_WAKELOCK1_CTL, DMC_WAKELOCK_CTL_REQ, 0);
 
-	if (__intel_de_wait_for_register_nowl(display, DMC_WAKELOCK1_CTL,
-					      DMC_WAKELOCK_CTL_ACK, 0,
-					      DMC_WAKELOCK_CTL_TIMEOUT)) {
+	if (__intel_de_wait_for_register_atomic_nowl(display, DMC_WAKELOCK1_CTL,
+						     DMC_WAKELOCK_CTL_ACK, 0,
+						     DMC_WAKELOCK_CTL_TIMEOUT_US)) {
 		WARN_RATELIMIT(1, "DMC wakelock release timed out");
 		goto out_unlock;
 	}
@@ -217,10 +221,14 @@ void intel_dmc_wl_get(struct intel_display *display, i915_reg_t reg)
 		__intel_de_rmw_nowl(display, DMC_WAKELOCK1_CTL, 0,
 				    DMC_WAKELOCK_CTL_REQ);
 
-		if (__intel_de_wait_for_register_nowl(display, DMC_WAKELOCK1_CTL,
-						      DMC_WAKELOCK_CTL_ACK,
-						      DMC_WAKELOCK_CTL_ACK,
-						      DMC_WAKELOCK_CTL_TIMEOUT)) {
+		/*
+		 * We need to use the atomic variant of the waiting routine
+		 * because the DMC wakelock is also taken in atomic context.
+		 */
+		if (__intel_de_wait_for_register_atomic_nowl(display, DMC_WAKELOCK1_CTL,
+							     DMC_WAKELOCK_CTL_ACK,
+							     DMC_WAKELOCK_CTL_ACK,
+							     DMC_WAKELOCK_CTL_TIMEOUT_US)) {
 			WARN_RATELIMIT(1, "DMC wakelock ack timed out");
 			goto out_unlock;
 		}
-- 
2.47.0


  parent reply	other threads:[~2024-11-06 21:52 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-06 21:50 [PATCH v2 00/17] drm/i915/dmc_wl: Fixes and enablement for Xe3_LPD Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 01/17] drm/i915/dmc_wl: Use i915_mmio_reg_offset() instead of reg.reg Gustavo Sousa
2024-11-07  9:23   ` Jani Nikula
2024-11-07 10:39   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 02/17] drm/xe: Mimic i915 behavior for non-sleeping MMIO wait Gustavo Sousa
2024-11-07 10:40   ` Luca Coelho
2024-11-06 21:50 ` Gustavo Sousa [this message]
2024-11-07 10:42   ` [PATCH v2 03/17] drm/i915/dmc_wl: Use non-sleeping variant of " Luca Coelho
2024-11-06 21:50 ` [PATCH v2 04/17] drm/i915/dmc_wl: Check for non-zero refcount in release work Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 05/17] drm/i915/dmc_wl: Get wakelock when disabling dynamic DC states Gustavo Sousa
2024-11-07 10:43   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 06/17] drm/i915/dmc_wl: Use sentinel item for range tables Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 07/17] drm/i915/dmc_wl: Extract intel_dmc_wl_reg_in_range() Gustavo Sousa
2024-11-07 10:44   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 08/17] drm/i915/dmc_wl: Rename lnl_wl_range to powered_off_ranges Gustavo Sousa
2024-11-07 10:46   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 09/17] drm/i915/dmc_wl: Track registers touched by the DMC Gustavo Sousa
2024-11-07 10:47   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 10/17] drm/i915/dmc_wl: Allow simpler syntax for single reg in range tables Gustavo Sousa
2024-11-07 10:48   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 11/17] drm/i915/dmc_wl: Deal with existing references when disabling Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 12/17] drm/i915/dmc_wl: Couple enable/disable with dynamic DC states Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 13/17] drm/i915/dmc_wl: Add and use HAS_DMC_WAKELOCK() Gustavo Sousa
2024-11-07 10:49   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 14/17] drm/i915/dmc_wl: Init only after we have runtime device info Gustavo Sousa
2024-11-07 10:50   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 15/17] drm/i915/dmc_wl: Use HAS_DMC() in HAS_DMC_WAKELOCK() Gustavo Sousa
2024-11-07 10:51   ` Luca Coelho
2024-11-06 21:50 ` [PATCH v2 16/17] drm/i915/dmc_wl: Sanitize enable_dmc_wl according to hardware support Gustavo Sousa
2024-11-06 21:50 ` [PATCH v2 17/17] drm/i915/xe3lpd: Use DMC wakelock by default Gustavo Sousa
2024-11-07 10:51   ` Luca Coelho
2024-11-07  1:23 ` ✓ CI.Patch_applied: success for drm/i915/dmc_wl: Fixes and enablement for Xe3_LPD (rev2) Patchwork
2024-11-07  1:24 ` ✗ CI.checkpatch: warning " Patchwork
2024-11-07  1:25 ` ✓ CI.KUnit: success " Patchwork
2024-11-07  1:37 ` ✓ CI.Build: " Patchwork
2024-11-07  1:39 ` ✓ CI.Hooks: " Patchwork
2024-11-07  1:40 ` ✗ CI.checksparse: warning " Patchwork
2024-11-07  2:00 ` ✓ CI.BAT: success " Patchwork
2024-11-08  7:06 ` ✗ CI.FULL: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20241106215231.103474-4-gustavo.sousa@intel.com \
    --to=gustavo.sousa@intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=jani.nikula@intel.com \
    --cc=luciano.coelho@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox