OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

panfrost: Handle Job VA cycles when decoding a dump file

Browse Source 
When a job loop is submitted to the GPU, as in IGT
panfrost_submit@pan-reset, this will trigger a DRM scheduler timeout and
eventually a devcoredump. However, when pandecode traverses the list of
jobs in a submit BO, it will iterate forever.

Fix it by adding already-visited CPU VA's into a mesa pointer set and
checking that the current job's CPU VA hasn't already been handled.

Signed-off-by: Adrián Larumbe <adrian.larumbe@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14034>
This commit is contained in:
Adrián Larumbe
2022-03-26 02:13:16 +00:00
committed by Marge Bot
parent d3642a0e02
commit 3da8c9193c
1 changed files with 19 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
src/panfrost/lib/genxml/decode.c
View File

@@ -33,6 +33,7 @@
#include <ctype.h>
#include "decode.h"
#include "util/set.h"
#include "midgard/disassemble.h"
#include "bifrost/disassemble.h"
#include "bifrost/valhall/disassemble.h"
@@ -1138,9 +1139,21 @@ GENX(pandecode_jc)(mali_ptr jc_gpu_va, unsigned gpu_id)
{
pandecode_dump_file_open();
struct set *va_set = _mesa_pointer_set_create(NULL);
struct set_entry *entry = NULL;
mali_ptr next_job = 0;
do {
struct pandecode_mapped_memory *mem =
pandecode_find_mapped_gpu_mem_containing(jc_gpu_va);
entry = _mesa_set_search(va_set, mem->addr);
if (entry != NULL) {
fprintf(stdout, "Job list has a cycle\n");
break;
}
pan_unpack(PANDECODE_PTR(jc_gpu_va, struct mali_job_header_packed),
JOB_HEADER, h);
next_job = h.next;
@@ -1189,8 +1202,14 @@ GENX(pandecode_jc)(mali_ptr jc_gpu_va, unsigned gpu_id)
default:
break;
}
/* Add the latest visited job GPU VA to avoid cycles */
_mesa_set_add(va_set, mem->addr);
} while ((jc_gpu_va = next_job));
_mesa_set_destroy(va_set, NULL);
fflush(pandecode_dump_stream);
pandecode_map_read_write();
}