KonstantinSeurer
/
mesa
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

intel/perf: Fix OA report accumulation on Gfx12+. 

The intel_perf_query path used for performance queries on GL was
passing a bogus "end" pointer to intel_perf_query_result_accumulate(),
causing it to accumulate garbage values.  This was causing the values
of many performance counters to be corrupted.

The "end" pointer was incorrect because the current code was assuming
that different OA reports were located TOTAL_QUERY_DATA_SIZE bytes
apart, which is a hard-coded preprocessor define.  However recent
(Gfx12+) hardware generations use a variable query size determined by
the query layout.  Use the size derived from it instead, and remove
the stale define.

Fixes: 3c51325025 ("intel/perf: switch query code to use query layout")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15783>
This commit is contained in:
Francisco Jerez 2022-04-04 15:45:54 -07:00 committed by Marge Bot
parent 6f98dc535a
commit e858da39e5
1 changed files with 4 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
src/intel/perf/intel_perf_query.c
View File

@ -48,19 +48,6 @@
#define ALIGN(x, y) (((x) + (y)-1) & ~((y)-1))
/* Align to 64bytes, requirement for OA report write address. */
#define TOTAL_QUERY_DATA_SIZE \
ALIGN(256 /* OA report */ + \
4 /* freq register */ + \
8 + 8 /* perf counter 1 & 2 */, \
64)
static uint32_t field_offset(bool end, uint32_t offset)
{
return (end ? TOTAL_QUERY_DATA_SIZE : 0) + offset;
}
#define MAP_READ (1 << 0)
#define MAP_WRITE (1 << 1)
@ -1089,8 +1076,8 @@ read_oa_samples_for_query(struct intel_perf_context *perf_ctx,
if (query->oa.map == NULL)
query->oa.map = perf_cfg->vtbl.bo_map(perf_ctx->ctx, query->oa.bo, MAP_READ);
start = last = query->oa.map + field_offset(false, 0);
end = query->oa.map + field_offset(true, 0);
start = last = query->oa.map;
end = query->oa.map + perf_ctx->perf->query_layout.size;
if (start[0] != query->oa.begin_report_id) {
DBG("Spurious start report id=%"PRIu32"\n", start[0]);
@ -1280,8 +1267,8 @@ accumulate_oa_reports(struct intel_perf_context *perf_ctx,
assert(query->oa.map != NULL);
start = last = query->oa.map + field_offset(false, 0);
end = query->oa.map + field_offset(true, 0);
start = last = query->oa.map;
end = query->oa.map + perf_ctx->perf->query_layout.size;
if (start[0] != query->oa.begin_report_id) {
DBG("Spurious start report id=%"PRIu32"\n", start[0]);