@@ -7774,8 +7774,9 @@ enum {
#define FORCEWAKE_ACK_MEDIA_GEN9 _MMIO(0x0D88)
#define FORCEWAKE_ACK_RENDER_GEN9 _MMIO(0x0D84)
#define FORCEWAKE_ACK_BLITTER_GEN9 _MMIO(0x130044)
-#define FORCEWAKE_KERNEL 0x1
-#define FORCEWAKE_USER 0x2
+#define FORCEWAKE_KERNEL BIT(0)
+#define FORCEWAKE_USER BIT(1)
+#define FORCEWAKE_KERNEL_FALLBACK BIT(15)
#define FORCEWAKE_MT_ACK _MMIO(0x130040)
#define ECOBUS _MMIO(0xa180)
#define FORCEWAKE_MT_ENABLE (1<<5)
@@ -69,17 +69,104 @@ fw_domain_arm_timer(struct intel_uncore_forcewake_domain *d)
HRTIMER_MODE_REL);
}
+static inline int
+__wait_for_ack(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d,
+ const u32 ack,
+ const u32 value)
+{
+ return wait_for_atomic((__raw_i915_read32(i915, d->reg_ack) & ack) == value,
+ FORCEWAKE_ACK_TIMEOUT_MS);
+}
+
+static inline int
+wait_ack_clear(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d,
+ const u32 ack)
+{
+ return __wait_for_ack(i915, d, ack, 0);
+}
+
+static inline int
+wait_ack_set(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d,
+ const u32 ack)
+{
+ return __wait_for_ack(i915, d, ack, ack);
+}
+
static inline void
fw_domain_wait_ack_clear(const struct drm_i915_private *i915,
const struct intel_uncore_forcewake_domain *d)
{
- if (wait_for_atomic((__raw_i915_read32(i915, d->reg_ack) &
- FORCEWAKE_KERNEL) == 0,
- FORCEWAKE_ACK_TIMEOUT_MS))
+ if (wait_ack_clear(i915, d, FORCEWAKE_KERNEL))
DRM_ERROR("%s: timed out waiting for forcewake ack to clear.\n",
intel_uncore_forcewake_domain_to_str(d->id));
}
+enum ack_type {
+ ACK_CLEAR = 0,
+ ACK_SET
+};
+
+static int
+fw_domain_wait_ack_with_fallback(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d,
+ const enum ack_type type)
+{
+ const u32 ack_bit = FORCEWAKE_KERNEL;
+ const u32 value = type == ACK_SET ? ack_bit : 0;
+ unsigned int pass;
+ bool ack_detected;
+
+ /*
+ * There is a possibility of driver's wake request colliding
+ * with hardware's own wake requests and that can cause
+ * hardware to not deliver the driver's ack message.
+ *
+ * Use a fallback bit toggle to kick the gpu state machine
+ * in the hope that the original ack will be delivered along with
+ * the fallback ack.
+ *
+ * This workaround is described in HSDES #1604254524
+ */
+
+ pass = 1;
+ do {
+ wait_ack_clear(i915, d, FORCEWAKE_KERNEL_FALLBACK);
+
+ __raw_i915_write32(i915, d->reg_set,
+ _MASKED_BIT_ENABLE(FORCEWAKE_KERNEL_FALLBACK));
+ /* Give gt some time to relax before the polling frenzy */
+ udelay(10 * pass);
+ wait_ack_set(i915, d, FORCEWAKE_KERNEL_FALLBACK);
+
+ ack_detected = (__raw_i915_read32(i915, d->reg_ack) & ack_bit) == value;
+
+ __raw_i915_write32(i915, d->reg_set,
+ _MASKED_BIT_DISABLE(FORCEWAKE_KERNEL_FALLBACK));
+ } while (!ack_detected && pass++ < 10);
+
+ DRM_DEBUG_DRIVER("%s had to use fallback to %s ack, 0x%x (passes %u)\n",
+ intel_uncore_forcewake_domain_to_str(d->id),
+ type == ACK_SET ? "set" : "clear",
+ __raw_i915_read32(i915, d->reg_ack),
+ pass);
+
+ return ack_detected ? 0 : -ETIMEDOUT;
+}
+
+static inline void
+fw_domain_wait_ack_clear_fallback(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d)
+{
+ if (likely(!wait_ack_clear(i915, d, FORCEWAKE_KERNEL)))
+ return;
+
+ if (fw_domain_wait_ack_with_fallback(i915, d, ACK_CLEAR))
+ fw_domain_wait_ack_clear(i915, d);
+}
+
static inline void
fw_domain_get(struct drm_i915_private *i915,
const struct intel_uncore_forcewake_domain *d)
@@ -88,17 +175,26 @@ fw_domain_get(struct drm_i915_private *i915,
}
static inline void
-fw_domain_wait_ack(const struct drm_i915_private *i915,
- const struct intel_uncore_forcewake_domain *d)
+fw_domain_wait_ack_set(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d)
{
- if (wait_for_atomic((__raw_i915_read32(i915, d->reg_ack) &
- FORCEWAKE_KERNEL),
- FORCEWAKE_ACK_TIMEOUT_MS))
+ if (wait_ack_set(i915, d, FORCEWAKE_KERNEL))
DRM_ERROR("%s: timed out waiting for forcewake ack request.\n",
intel_uncore_forcewake_domain_to_str(d->id));
}
static inline void
+fw_domain_wait_ack_set_fallback(const struct drm_i915_private *i915,
+ const struct intel_uncore_forcewake_domain *d)
+{
+ if (likely(!wait_ack_set(i915, d, FORCEWAKE_KERNEL)))
+ return;
+
+ if (fw_domain_wait_ack_with_fallback(i915, d, ACK_SET))
+ fw_domain_wait_ack_set(i915, d);
+}
+
+static inline void
fw_domain_put(const struct drm_i915_private *i915,
const struct intel_uncore_forcewake_domain *d)
{
@@ -119,7 +215,27 @@ fw_domains_get(struct drm_i915_private *i915, enum forcewake_domains fw_domains)
}
for_each_fw_domain_masked(d, fw_domains, i915, tmp)
- fw_domain_wait_ack(i915, d);
+ fw_domain_wait_ack_set(i915, d);
+
+ i915->uncore.fw_domains_active |= fw_domains;
+}
+
+static void
+fw_domains_get_with_fallback(struct drm_i915_private *i915,
+ enum forcewake_domains fw_domains)
+{
+ struct intel_uncore_forcewake_domain *d;
+ unsigned int tmp;
+
+ GEM_BUG_ON(fw_domains & ~i915->uncore.fw_domains);
+
+ for_each_fw_domain_masked(d, fw_domains, i915, tmp) {
+ fw_domain_wait_ack_clear_fallback(i915, d);
+ fw_domain_get(i915, d);
+ }
+
+ for_each_fw_domain_masked(d, fw_domains, i915, tmp)
+ fw_domain_wait_ack_set_fallback(i915, d);
i915->uncore.fw_domains_active |= fw_domains;
}
@@ -1142,7 +1258,8 @@ static void intel_uncore_fw_domains_init(struct drm_i915_private *dev_priv)
}
if (INTEL_GEN(dev_priv) >= 9) {
- dev_priv->uncore.funcs.force_wake_get = fw_domains_get;
+ dev_priv->uncore.funcs.force_wake_get =
+ fw_domains_get_with_fallback;
dev_priv->uncore.funcs.force_wake_put = fw_domains_put;
fw_domain_init(dev_priv, FW_DOMAIN_ID_RENDER,
FORCEWAKE_RENDER_GEN9,
There is a possibility on gen9 hardware to miss the forcewake ack message. The recommended workaround is to use another free bit and toggle it until original bit is successfully acknowledged. Some future gen9 revs might or might not fix the underlying issue but the fallback to reserve bit dance can be considered as harmless: without the ack timeout we never reach the reserve bit forcewake. Thus as of now we adopt a blanket approach for all gen9 and leave the bypassing the reserve bit approach for future patches if corresponding hw revisions do appear. Commit 83e3337204b2 ("drm/i915: Increase maximum polling time to 50ms for forcewake request/clear ack") did increase the forcewake timeout. If the issue was a delayed ack, future work could include finding a suitable timeout value both for primary ack and reserve toggle to reduce the worst case latency. v2: use bit 15, naming, comment (Chris), only wait fallback ack v3: fix return on fallback, backoff after fallback write (Chris) v4: udelay on first pass, grammar (Chris) References: HSDES #1604254524 References: https://bugs.freedesktop.org/show_bug.cgi?id=102051 Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com> Signed-off-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> --- drivers/gpu/drm/i915/i915_reg.h | 5 +- drivers/gpu/drm/i915/intel_uncore.c | 137 +++++++++++++++++++++++++++++++++--- 2 files changed, 130 insertions(+), 12 deletions(-)