summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/xe/xe_rtp.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/xe/xe_rtp.c')
-rw-r--r--drivers/gpu/drm/xe/xe_rtp.c158
1 files changed, 110 insertions, 48 deletions
diff --git a/drivers/gpu/drm/xe/xe_rtp.c b/drivers/gpu/drm/xe/xe_rtp.c
index fb44cc7521d8..ed509b1c8cfc 100644
--- a/drivers/gpu/drm/xe/xe_rtp.c
+++ b/drivers/gpu/drm/xe/xe_rtp.c
@@ -7,12 +7,14 @@
#include <kunit/visibility.h>
-#include <drm/xe_drm.h>
+#include <uapi/drm/xe_drm.h>
+#include "xe_configfs.h"
#include "xe_gt.h"
#include "xe_gt_topology.h"
#include "xe_macros.h"
#include "xe_reg_sr.h"
+#include "xe_sriov.h"
/**
* DOC: Register Table Processing
@@ -35,11 +37,18 @@ static bool rule_matches(const struct xe_device *xe,
unsigned int n_rules)
{
const struct xe_rtp_rule *r;
- unsigned int i;
+ unsigned int i, rcount = 0;
bool match;
for (r = rules, i = 0; i < n_rules; r = &rules[++i]) {
switch (r->match_type) {
+ case XE_RTP_MATCH_OR:
+ /*
+ * This is only reached if a complete set of
+ * rules passed or none were evaluated. For both cases,
+ * shortcut the other rules and return the proper value.
+ */
+ goto done;
case XE_RTP_MATCH_PLATFORM:
match = xe->info.platform == r->platform;
break;
@@ -48,33 +57,63 @@ static bool rule_matches(const struct xe_device *xe,
xe->info.subplatform == r->subplatform;
break;
case XE_RTP_MATCH_GRAPHICS_VERSION:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.graphics_verx100 == r->ver_start &&
(!has_samedia(xe) || !xe_gt_is_media_type(gt));
break;
case XE_RTP_MATCH_GRAPHICS_VERSION_RANGE:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.graphics_verx100 >= r->ver_start &&
xe->info.graphics_verx100 <= r->ver_end &&
(!has_samedia(xe) || !xe_gt_is_media_type(gt));
break;
+ case XE_RTP_MATCH_GRAPHICS_VERSION_ANY_GT:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
+ match = xe->info.graphics_verx100 == r->ver_start;
+ break;
case XE_RTP_MATCH_GRAPHICS_STEP:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.step.graphics >= r->step_start &&
xe->info.step.graphics < r->step_end &&
(!has_samedia(xe) || !xe_gt_is_media_type(gt));
break;
case XE_RTP_MATCH_MEDIA_VERSION:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.media_verx100 == r->ver_start &&
(!has_samedia(xe) || xe_gt_is_media_type(gt));
break;
case XE_RTP_MATCH_MEDIA_VERSION_RANGE:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.media_verx100 >= r->ver_start &&
xe->info.media_verx100 <= r->ver_end &&
(!has_samedia(xe) || xe_gt_is_media_type(gt));
break;
case XE_RTP_MATCH_MEDIA_STEP:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
match = xe->info.step.media >= r->step_start &&
xe->info.step.media < r->step_end &&
(!has_samedia(xe) || xe_gt_is_media_type(gt));
break;
+ case XE_RTP_MATCH_MEDIA_VERSION_ANY_GT:
+ if (drm_WARN_ON(&xe->drm, !gt))
+ return false;
+
+ match = xe->info.media_verx100 == r->ver_start;
+ break;
case XE_RTP_MATCH_INTEGRATED:
match = !xe->info.is_dgfx;
break;
@@ -94,7 +133,7 @@ static bool rule_matches(const struct xe_device *xe,
match = hwe->class != r->engine_class;
break;
case XE_RTP_MATCH_FUNC:
- match = r->match_func(gt, hwe);
+ match = r->match_func(xe, gt, hwe);
break;
default:
drm_warn(&xe->drm, "Invalid RTP match %u\n",
@@ -102,10 +141,27 @@ static bool rule_matches(const struct xe_device *xe,
match = false;
}
- if (!match)
- return false;
+ if (!match) {
+ /*
+ * Advance rules until we find XE_RTP_MATCH_OR to check
+ * if there's another set of conditions to check
+ */
+ while (++i < n_rules && rules[i].match_type != XE_RTP_MATCH_OR)
+ ;
+
+ if (i >= n_rules)
+ return false;
+
+ rcount = 0;
+ } else {
+ rcount++;
+ }
}
+done:
+ if (drm_WARN_ON(&xe->drm, !rcount))
+ return false;
+
return true;
}
@@ -155,6 +211,11 @@ static void rtp_get_context(struct xe_rtp_process_ctx *ctx,
struct xe_device **xe)
{
switch (ctx->type) {
+ case XE_RTP_PROCESS_TYPE_DEVICE:
+ *hwe = NULL;
+ *gt = NULL;
+ *xe = ctx->xe;
+ break;
case XE_RTP_PROCESS_TYPE_GT:
*hwe = NULL;
*gt = ctx->gt;
@@ -165,7 +226,7 @@ static void rtp_get_context(struct xe_rtp_process_ctx *ctx,
*gt = (*hwe)->gt;
*xe = gt_to_xe(*gt);
break;
- };
+ }
}
/**
@@ -186,21 +247,19 @@ void xe_rtp_process_ctx_enable_active_tracking(struct xe_rtp_process_ctx *ctx,
ctx->active_entries = active_entries;
ctx->n_entries = n_entries;
}
+EXPORT_SYMBOL_IF_KUNIT(xe_rtp_process_ctx_enable_active_tracking);
static void rtp_mark_active(struct xe_device *xe,
struct xe_rtp_process_ctx *ctx,
- unsigned int first, unsigned int last)
+ unsigned int idx)
{
if (!ctx->active_entries)
return;
- if (drm_WARN_ON(&xe->drm, last > ctx->n_entries))
+ if (drm_WARN_ON(&xe->drm, idx >= ctx->n_entries))
return;
- if (first == last)
- bitmap_set(ctx->active_entries, first, 1);
- else
- bitmap_set(ctx->active_entries, first, last - first + 2);
+ bitmap_set(ctx->active_entries, idx, 1);
}
/**
@@ -208,6 +267,7 @@ static void rtp_mark_active(struct xe_device *xe,
* the save-restore argument.
* @ctx: The context for processing the table, with one of device, gt or hwe
* @entries: Table with RTP definitions
+ * @n_entries: Number of entries to process, usually ARRAY_SIZE(entries)
* @sr: Save-restore struct where matching rules execute the action. This can be
* viewed as the "coalesced view" of multiple the tables. The bits for each
* register set are expected not to collide with previously added entries
@@ -218,6 +278,7 @@ static void rtp_mark_active(struct xe_device *xe,
*/
void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx,
const struct xe_rtp_entry_sr *entries,
+ size_t n_entries,
struct xe_reg_sr *sr)
{
const struct xe_rtp_entry_sr *entry;
@@ -227,7 +288,9 @@ void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx,
rtp_get_context(ctx, &hwe, &gt, &xe);
- for (entry = entries; entry && entry->name; entry++) {
+ xe_assert(xe, entries);
+
+ for (entry = entries; entry - entries < n_entries; entry++) {
bool match = false;
if (entry->flags & XE_RTP_ENTRY_FLAG_FOREACH_ENGINE) {
@@ -242,8 +305,7 @@ void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx,
}
if (match)
- rtp_mark_active(xe, ctx, entry - entries,
- entry - entries);
+ rtp_mark_active(xe, ctx, entry - entries);
}
}
EXPORT_SYMBOL_IF_KUNIT(xe_rtp_process_to_sr);
@@ -254,52 +316,39 @@ EXPORT_SYMBOL_IF_KUNIT(xe_rtp_process_to_sr);
* @entries: Table with RTP definitions
*
* Walk the table pointed by @entries (with an empty sentinel), executing the
- * rules. A few differences from xe_rtp_process_to_sr():
- *
- * 1. There is no action associated with each entry since this uses
- * struct xe_rtp_entry. Its main use is for marking active workarounds via
- * xe_rtp_process_ctx_enable_active_tracking().
- * 2. There is support for OR operations by having entries with no name.
+ * rules. One difference from xe_rtp_process_to_sr(): there is no action
+ * associated with each entry since this uses struct xe_rtp_entry. Its main use
+ * is for marking active workarounds via
+ * xe_rtp_process_ctx_enable_active_tracking().
*/
void xe_rtp_process(struct xe_rtp_process_ctx *ctx,
const struct xe_rtp_entry *entries)
{
- const struct xe_rtp_entry *entry, *first_entry;
+ const struct xe_rtp_entry *entry;
struct xe_hw_engine *hwe;
struct xe_gt *gt;
struct xe_device *xe;
rtp_get_context(ctx, &hwe, &gt, &xe);
- first_entry = entries;
- if (drm_WARN_ON(&xe->drm, !first_entry->name))
- return;
-
for (entry = entries; entry && entry->rules; entry++) {
- if (entry->name)
- first_entry = entry;
-
if (!rule_matches(xe, gt, hwe, entry->rules, entry->n_rules))
continue;
- /* Fast-forward entry, eliminating the OR'ed entries */
- for (entry++; entry && entry->rules; entry++)
- if (entry->name)
- break;
- entry--;
-
- rtp_mark_active(xe, ctx, first_entry - entries,
- entry - entries);
+ rtp_mark_active(xe, ctx, entry - entries);
}
}
+EXPORT_SYMBOL_IF_KUNIT(xe_rtp_process);
-bool xe_rtp_match_even_instance(const struct xe_gt *gt,
+bool xe_rtp_match_even_instance(const struct xe_device *xe,
+ const struct xe_gt *gt,
const struct xe_hw_engine *hwe)
{
return hwe->instance % 2 == 0;
}
-bool xe_rtp_match_first_render_or_compute(const struct xe_gt *gt,
+bool xe_rtp_match_first_render_or_compute(const struct xe_device *xe,
+ const struct xe_gt *gt,
const struct xe_hw_engine *hwe)
{
u64 render_compute_mask = gt->info.engine_mask &
@@ -309,17 +358,30 @@ bool xe_rtp_match_first_render_or_compute(const struct xe_gt *gt,
hwe->engine_id == __ffs(render_compute_mask);
}
-bool xe_rtp_match_first_gslice_fused_off(const struct xe_gt *gt,
- const struct xe_hw_engine *hwe)
+bool xe_rtp_match_not_sriov_vf(const struct xe_device *xe,
+ const struct xe_gt *gt,
+ const struct xe_hw_engine *hwe)
{
- unsigned int dss_per_gslice = 4;
- unsigned int dss;
+ return !IS_SRIOV_VF(xe);
+}
- if (drm_WARN(&gt_to_xe(gt)->drm, xe_dss_mask_empty(gt->fuse_topo.g_dss_mask),
- "Checking gslice for platform without geometry pipeline\n"))
- return false;
+bool xe_rtp_match_psmi_enabled(const struct xe_device *xe,
+ const struct xe_gt *gt,
+ const struct xe_hw_engine *hwe)
+{
+ return xe_configfs_get_psmi_enabled(to_pci_dev(xe->drm.dev));
+}
- dss = xe_dss_mask_group_ffs(gt->fuse_topo.g_dss_mask, 0, 0);
+bool xe_rtp_match_gt_has_discontiguous_dss_groups(const struct xe_device *xe,
+ const struct xe_gt *gt,
+ const struct xe_hw_engine *hwe)
+{
+ return xe_gt_has_discontiguous_dss_groups(gt);
+}
- return dss >= dss_per_gslice;
+bool xe_rtp_match_has_flat_ccs(const struct xe_device *xe,
+ const struct xe_gt *gt,
+ const struct xe_hw_engine *hwe)
+{
+ return xe->info.has_flat_ccs;
}
generated by cgit (git 2.34.1) at 2025-12-26 01:18:24 +0000