Re: [PATCH 1/2] drm/i915: drop one bit on the hw_id when using guc

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 31/05/18 21:33, Michel Thierry wrote:
On 5/31/2018 12:56 PM, Lionel Landwerlin wrote:
We currently using GuC as a proxy to the hardware. When Guc is used in
such mode, it consumes the bit 20 of the hw_id to indicate that the
workload was submitted by proxy.

So far we probably haven't seen the issue because we need to allocate
1048576+ contexts to hit this issue. Still, we should avoid allocating
the hw_id on that bit and restriction to bits [0:19] (i.e 20bits
instead of 21).

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@xxxxxxxxx>
BSpec: 1237
---
  drivers/gpu/drm/i915/i915_drv.h         | 18 ++++++++++++++++++
  drivers/gpu/drm/i915/i915_gem_context.c |  7 +------
  drivers/gpu/drm/i915/intel_lrc.c        |  2 +-
  3 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 64659d4efeda..58ab9259fb73 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1841,6 +1841,7 @@ struct drm_i915_private {
           */
          struct ida hw_ida;
  #define MAX_CONTEXT_HW_ID (1<<21) /* exclusive */
+#define MAX_GUC_CONTEXT_HW_ID (1<<20) /* exclusive */
  #define GEN11_MAX_CONTEXT_HW_ID (1<<11) /* exclusive */
            bool dynamic_sseu;
@@ -3262,6 +3263,23 @@ i915_gem_context_lookup(struct drm_i915_file_private *file_priv, u32 id)
      return ctx;
  }
  +static inline u32
+i915_gem_context_max_hw_id(struct drm_i915_private *i915)
+{
+    /* TODO: Confirm the max number on Gen11 with GuC */
+    if (INTEL_GEN(i915) >= 11)
+        return GEN11_MAX_CONTEXT_HW_ID;
+
+    /*
+     * When using GuC in proxy submission, GuC consumes the highest bit in
+     * the context id to indicate proxy submission.
+     */
+    if (USES_GUC_SUBMISSION(i915))
+        return MAX_GUC_CONTEXT_HW_ID;
+
+    return MAX_CONTEXT_HW_ID;
+}
+

What was the reason of moving this out of i915_gem_context.c? I don't see any other user.

Yeah, I was initially thinking I would reuse that in i915_perf.c, but that's not the case.
I can send a v2 to put it back in.


Everything else looks good to me so

Reviewed-by: Michel Thierry <michel.thierry@xxxxxxxxx>

  int i915_gem_contexts_set_dynamic_sseu(struct drm_i915_private *i915,
                         bool allowed);
  bool i915_gem_contexts_get_dynamic_sseu(struct drm_i915_private *i915); diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index ff08515d0c67..b686c04d98a9 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -205,13 +205,8 @@ static void context_close(struct i915_gem_context *ctx)     static int assign_hw_id(struct drm_i915_private *dev_priv, unsigned *out)
  {
+    unsigned int max = i915_gem_context_max_hw_id(dev_priv);
      int ret;
-    unsigned int max;
-
-    if (INTEL_GEN(dev_priv) >= 11)
-        max = GEN11_MAX_CONTEXT_HW_ID;
-    else
-        max = MAX_CONTEXT_HW_ID;
        ret = ida_simple_get(&dev_priv->contexts.hw_ida,
                   0, max, GFP_KERNEL);
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 26e43eb0ef31..94f9c4795190 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -200,7 +200,7 @@ static inline bool need_preempt(const struct intel_engine_cs *engine,
   *
   *      bits  0-11:    flags, GEN8_CTX_* (cached in ctx->desc_template)
   *      bits 12-31:    LRCA, GTT address of (the HWSP of) this context
- *      bits 32-52:    ctx ID, a globally unique tag
+ *      bits 32-52:    ctx ID, a globally unique tag (highest bit used by GuC)
   *      bits 53-54:    mbz, reserved for use by hardware
   *      bits 55-63:    group ID, currently unused and set to 0
   *



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




[Index of Archives]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux