On 11/01/18 11:31, Tvrtko Ursulin wrote:
On 18/12/2017 15:35, Lionel Landwerlin wrote:
While the end goal is to make this information available to userspace
through a new ioctl, there is no reason we can't display it in a human
readable fashion through debugfs.
slice0 (subslice_mask=0x7):
I'd add a subslice count while at it, since the eu lines have counts.
Bike-shedding on whether counts or masks are typically more important?
Slice0: 3 slices (0x7):
Subslice 0: 8 EUs (0xff)
Subslice 1: 8 EUs (0xff)
...
?
Yeah, sure.
subslice0:
eu_mask: 0xff (8)
subslice1:
eu_mask: 0xff (8)
subslice2:
eu_mask: 0xff (8)
subslice3:
eu_mask: 0x0 (0)
slice1 (subslice_mask=0x7):
subslice0:
eu_mask: 0xff (8)
subslice1:
eu_mask: 0xff (8)
subslice2:
eu_mask: 0xff (8)
subslice3:
eu_mask: 0x0 (0)
slice2 (subslice_mask=0x7):
subslice0:
eu_mask: 0xff (8)
subslice1:
eu_mask: 0xff (8)
subslice2:
eu_mask: 0xff (8)
subslice3:
eu_mask: 0x0 (0)
Suggested-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@xxxxxxxxx>
---
drivers/gpu/drm/i915/i915_debugfs.c | 37
+++++++++++++++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c
b/drivers/gpu/drm/i915/i915_debugfs.c
index 6ec7543e698f..79ca6e9f9ec9 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -3173,6 +3173,42 @@ static int i915_engine_info(struct seq_file
*m, void *unused)
return 0;
}
+static int i915_rcs_topology(struct seq_file *m, void *unused)
+{
+ struct drm_i915_private *dev_priv = node_to_i915(m->private);
+ const struct sseu_dev_info *sseu = &INTEL_INFO(dev_priv)->sseu;
+ int s, ss;
+ int subslice_stride = ALIGN(sseu->max_eus_per_subslice, 8) / 8;
+ int slice_stride = sseu->max_subslices * subslice_stride;
Another case for the before mentioned helper for indexing into eu_mask
array?
Done.
+
+ if (sseu->max_slices == 0) {
+ seq_printf(m, "Unavailable\n");
+ return 0;
+ }
Is this possible?
Yeah, there are no registers to read on anything < gen8.
I've decided not to generate data there. Most userspaces already has the
numbers from a table by pci-id.
+
+ for (s = 0; s < sseu->max_slices; s++) {
+ seq_printf(m, "slice%i (subslice_mask=0x%x):\n",
%i always confuses me. Googling shows it is equivalent to %d for
printing? Or is it something different in kernel space? If it is
equivalent I would go with a more standard one. And I would even
change to unsigned variables for iterators but I realize some people
have a different opinion so up to you.
Yeah, I'm always using %i, but I must be the only one.
It's been easier to remember int -> %i.
+ s, sseu->subslices_mask[s]);
+
+ for (ss = 0; ss < slice_stride / subslice_stride; ss++) {
With the indexing helpers hopefully it would be possible to simply
iterate to hweight8(sseu->sublice_mask[s]) ?
Actually I should use sseu->max_subslices.
+ int eu, n_subslice_eus = 0;
+
+ seq_printf(m, "\tsubslice%i:\n", ss);
+
+ seq_printf(m, "\t\teu_mask:");
+ for (eu = 0; eu < subslice_stride; eu++) {
+ u8 val = sseu->eu_mask[s * slice_stride +
+ ss * subslice_stride + eu];
+ seq_printf(m, " 0x%x", val);
+ n_subslice_eus += hweight8(val);
+ }
+ seq_printf(m, " (%i)\n", n_subslice_eus);
+ }
+ }
+
+ return 0;
+}
+
static int i915_shrinker_info(struct seq_file *m, void *unused)
{
struct drm_i915_private *i915 = node_to_i915(m->private);
@@ -4658,6 +4694,7 @@ static const struct drm_info_list
i915_debugfs_list[] = {
{"i915_dmc_info", i915_dmc_info, 0},
{"i915_display_info", i915_display_info, 0},
{"i915_engine_info", i915_engine_info, 0},
+ {"i915_rcs_topology", i915_rcs_topology, 0},
{"i915_shrinker_info", i915_shrinker_info, 0},
{"i915_shared_dplls_info", i915_shared_dplls_info, 0},
{"i915_dp_mst_info", i915_dp_mst_info, 0},
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
