[Intel-gfx] [PATCH 3/3] intel: Add a function for the new register read ioctl.

[Eric Anholt]

---
I'm not sure if this is the API we want or not.  Getting your value
back through an out parameter sucks.  On the other hand, the 3d driver
wants to be able to use the error value to detect a supported kernel.

My other API thought would be for reg_read() to return the value and
whine if an error occurs, and to have a separate can_reg_read() call.

 intel/intel_bufmgr.h     |    3 +++
 intel/intel_bufmgr_gem.c |   18 ++++++++++++++++++
 2 files changed, 21 insertions(+)

diff --git a/intel/intel_bufmgr.h b/intel/intel_bufmgr.h
index 2167e43..8d7f239 100644
--- a/intel/intel_bufmgr.h
+++ b/intel/intel_bufmgr.h
@@ -241,6 +241,9 @@ void drm_intel_decode_set_head_tail(struct drm_intel_decode *ctx,
 void drm_intel_decode_set_output_file(struct drm_intel_decode *ctx, FILE *out);
 void drm_intel_decode(struct drm_intel_decode *ctx);
 
+int drm_intel_reg_read(drm_intel_bufmgr *bufmgr,
+		       uint32_t offset,
+		       uint64_t *result);
 
 /** @{ Compatibility defines to keep old code building despite the symbol rename
  * from dri_* to drm_intel_*
diff --git a/intel/intel_bufmgr_gem.c b/intel/intel_bufmgr_gem.c
index a484b12..a170813 100644
--- a/intel/intel_bufmgr_gem.c
+++ b/intel/intel_bufmgr_gem.c
@@ -2947,6 +2947,24 @@ drm_intel_gem_context_destroy(drm_intel_context *ctx)
 	free(ctx);
 }
 
+int
+drm_intel_reg_read(drm_intel_bufmgr *bufmgr,
+		   uint32_t offset,
+		   uint64_t *result)
+{
+	drm_intel_bufmgr_gem *bufmgr_gem = (drm_intel_bufmgr_gem *)bufmgr;
+	struct drm_i915_reg_read reg_read;
+	int ret;
+
+	VG_CLEAR(reg_read);
+	reg_read.offset = offset;
+
+	ret = drmIoctl(bufmgr_gem->fd, DRM_IOCTL_I915_REG_READ, &reg_read);
+
+	*result = reg_read.val;
+	return ret;
+}
+
 
 /**
  * Annotate the given bo for use in aub dumping.
-- 
1.7.10.4