Re: [Intel-gfx] [PATCH v4 21/38] drm/i915: Added a module parameter for allowing scheduler overrides

2016-01-12 Thread Chris Wilson 
On Tue, Jan 12, 2016 at 11:34:45AM +, John Harrison wrote:
> On 11/01/2016 22:24, Chris Wilson wrote:
> >On Mon, Jan 11, 2016 at 06:42:50PM +, john.c.harri...@intel.com wrote:
> >>From: John Harrison 
> >>
> >>It can be useful to be able to disable certain features (e.g. the
> >>entire scheduler) via a module parameter for debugging purposes. A
> >>parameter has the advantage of not being a compile time switch but
> >>without implying that it can be changed dynamically at runtime.
> >>+module_param_named(scheduler_override, i915.scheduler_override, int, 0600);
> >>+MODULE_PARM_DESC(scheduler_override, "Scheduler override mask (0 = none, 1 
> >>= direct submission [default])");
> >Is this consistent with the other *enable* booleans?
> Initially there were a whole bunch of override flags for
> disabling/tweaking specific bits of the scheduler's operation. Since
> these extras now only exist in an internal debugging patch that is
> not going to be upstreamed, I guess it probably should be simplified
> to a bool rather than a flags word.

Yes, later on I saw the unsigned flags. Those should be restricted to a
debugfs interface, not a user parameter. module options are an
invitation for the world to fiddle, and they will.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH v4 21/38] drm/i915: Added a module parameter for allowing scheduler overrides

2016-01-12 Thread John Harrison 

On 11/01/2016 22:24, Chris Wilson wrote:

On Mon, Jan 11, 2016 at 06:42:50PM +, john.c.harri...@intel.com wrote:

From: John Harrison 

It can be useful to be able to disable certain features (e.g. the
entire scheduler) via a module parameter for debugging purposes. A
parameter has the advantage of not being a compile time switch but
without implying that it can be changed dynamically at runtime.
+module_param_named(scheduler_override, i915.scheduler_override, int, 0600);
+MODULE_PARM_DESC(scheduler_override, "Scheduler override mask (0 = none, 1 = direct 
submission [default])");

Is this consistent with the other *enable* booleans?
Initially there were a whole bunch of override flags for 
disabling/tweaking specific bits of the scheduler's operation. Since 
these extras now only exist in an internal debugging patch that is not 
going to be upstreamed, I guess it probably should be simplified to a 
bool rather than a flags word.




-Chris



___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH v4 21/38] drm/i915: Added a module parameter for allowing scheduler overrides

2016-01-11 Thread Chris Wilson 
On Mon, Jan 11, 2016 at 06:42:50PM +, john.c.harri...@intel.com wrote:
> From: John Harrison 
> 
> It can be useful to be able to disable certain features (e.g. the
> entire scheduler) via a module parameter for debugging purposes. A
> parameter has the advantage of not being a compile time switch but
> without implying that it can be changed dynamically at runtime.

> +module_param_named(scheduler_override, i915.scheduler_override, int, 0600);
> +MODULE_PARM_DESC(scheduler_override, "Scheduler override mask (0 = none, 1 = 
> direct submission [default])");

Is this consistent with the other *enable* booleans?
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

[Intel-gfx] [PATCH v4 21/38] drm/i915: Added a module parameter for allowing scheduler overrides

2016-01-11 Thread John . C . Harrison 
From: John Harrison 

It can be useful to be able to disable certain features (e.g. the
entire scheduler) via a module parameter for debugging purposes. A
parameter has the advantage of not being a compile time switch but
without implying that it can be changed dynamically at runtime.

For: VIZ-1587
Signed-off-by: John Harrison 
---
 drivers/gpu/drm/i915/i915_drv.h   | 1 +
 drivers/gpu/drm/i915/i915_params.c| 4 
 drivers/gpu/drm/i915/i915_scheduler.c | 5 -
 drivers/gpu/drm/i915/i915_scheduler.h | 5 +
 4 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 2906bb0..8d6d07a 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -2709,6 +2709,7 @@ struct i915_params {
bool verbose_state_checks;
bool nuclear_pageflip;
int edp_vswing;
+   int scheduler_override;
 };
 extern struct i915_params i915 __read_mostly;
 
diff --git a/drivers/gpu/drm/i915/i915_params.c 
b/drivers/gpu/drm/i915/i915_params.c
index ba7274a..be2e16c 100644
--- a/drivers/gpu/drm/i915/i915_params.c
+++ b/drivers/gpu/drm/i915/i915_params.c
@@ -54,6 +54,7 @@ struct i915_params i915 __read_mostly = {
.edp_vswing = 0,
.enable_guc_submission = true,
.guc_log_level = -1,
+   .scheduler_override = 1,
 };
 
 module_param_named(modeset, i915.modeset, int, 0400);
@@ -191,3 +192,6 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC 
submission (default:false)")
 module_param_named(guc_log_level, i915.guc_log_level, int, 0400);
 MODULE_PARM_DESC(guc_log_level,
"GuC firmware logging level (-1:disabled (default), 0-3:enabled)");
+
+module_param_named(scheduler_override, i915.scheduler_override, int, 0600);
+MODULE_PARM_DESC(scheduler_override, "Scheduler override mask (0 = none, 1 = 
direct submission [default])");
diff --git a/drivers/gpu/drm/i915/i915_scheduler.c 
b/drivers/gpu/drm/i915/i915_scheduler.c
index bd52752..bfde3a2 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.c
+++ b/drivers/gpu/drm/i915/i915_scheduler.c
@@ -44,6 +44,9 @@ bool i915_scheduler_is_enabled(struct drm_device *dev)
 {
struct drm_i915_private *dev_priv = dev->dev_private;
 
+   if (i915.scheduler_override & i915_so_direct_submit)
+   return false;
+
return dev_priv->scheduler != NULL;
 }
 
@@ -144,7 +147,7 @@ int i915_scheduler_queue_execbuffer(struct 
i915_scheduler_queue_entry *qe)
 
WARN_ON(!scheduler);
 
-   if (1/*i915.scheduler_override & i915_so_direct_submit*/) {
+   if (i915.scheduler_override & i915_so_direct_submit) {
int ret;
 
intel_ring_reserved_space_cancel(qe->params.request->ringbuf);
diff --git a/drivers/gpu/drm/i915/i915_scheduler.h 
b/drivers/gpu/drm/i915/i915_scheduler.h
index 7654013..cc9205b 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.h
+++ b/drivers/gpu/drm/i915/i915_scheduler.h
@@ -84,6 +84,11 @@ enum {
i915_sf_submitting  = (1 << 1),
 };
 
+/* Options for 'scheduler_override' module parameter: */
+enum {
+   i915_so_direct_submit   = (1 << 0),
+};
+
 booli915_scheduler_is_enabled(struct drm_device *dev);
 int i915_scheduler_init(struct drm_device *dev);
 int i915_scheduler_closefile(struct drm_device *dev,
-- 
1.9.1

___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx