marfrit/libva-v4l2-request-fourier
1
0
Fork 1
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

DEBUG: instrument surface CreateSurfaces2 + ExportSurfaceHandle for diagnosis

Browse Source 
Logs format_width/height + bytesperline + sizes from v4l2_get_format
in CreateSurfaces2, and the full VADRMPRIMESurfaceDescriptor in
ExportSurfaceHandle (fd, fourcc, width/height, num_objects/layers,
obj.size + drm_format/modifier, plane offsets/pitches).

Diagnostic for the surface-export bug surfaced by Phase 7 (mpv
--hwdec=vaapi --vo=gpu shows solid blue, Firefox falls back to SW
after frame 0 — both consumers GL-import the DMA-BUF, both fail
to render correctly while vaapi-copy works).

Phase 5 review (sonnet) suggested format_height might be 1080
(stream) vs 1088 (MB-aligned), miscomputing UV offset by 15360
bytes. Earlier ftrace shows kernel returns height=1088 — the
hypothesis is likely false but verifying in-driver to confirm.

Will compare with mpv --msg-level=vd=v --msg-level=vo=v output to
identify the import-side discrepancy.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Markus Fritsche
2026-05-04 14:00:13 +00:00
parent 6be3f3b120
commit 2517a1206b
1 changed files with 44 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/surface.c
+44
View File
@@ -146,6 +146,23 @@ VAStatus RequestCreateSurfaces2(VADriverContextP context, unsigned int format,
destination_planes_count = video_format->planes_count;
/*
* DEBUG INSTRUMENTATION (surface-export diagnosis 2026-05-04):
* dump what v4l2_get_format returned. Sonnet's Phase 5 review
* hypothesis #4 was that format_height might be 1080 (stream-
* signaled) vs 1088 (MB-aligned), causing UV offset to land
* 15360 bytes early. Earlier ftrace shows hantro returns
* height=1088 — but verify in-driver to be sure.
*/
request_log("CreateSurfaces2: surf_width=%u surf_height=%u "
"fmt_width=%u fmt_height=%u bytesperline[0]=%u "
"sizes[0]=%u sizes[1]=%u planes_count=%u "
"v4l2_buffers_count=%u\n",
width, height, format_width, format_height,
destination_bytesperlines[0],
destination_sizes[0], destination_sizes[1],
destination_planes_count, video_format->v4l2_buffers_count);
rc = v4l2_create_buffers(driver_data->video_fd, capture_type,
surfaces_count, &index_base);
if (rc < 0)
@@ -619,6 +636,33 @@ VAStatus RequestExportSurfaceHandle(VADriverContextP context,
surface_object->destination_bytesperlines[i];
}
/*
* DEBUG INSTRUMENTATION (surface-export diagnosis 2026-05-04):
* dump the full descriptor so we can compare against what mpv
* reports importing via --msg-level=vd=v --msg-level=vo=v.
* Phase 5 review identified DMA-BUF surface export as the
* likely root cause of the solid-blue render in mpv vaapi mode.
*/
request_log("ExportSurfaceHandle: surf=%u fd[0]=%d fourcc=0x%x "
"w=%u h=%u num_objects=%u num_layers=%u "
"obj[0].size=%u drm_fmt=0x%x drm_mod=0x%llx num_planes=%u "
"p[0].off=%u pitch=%u p[1].off=%u pitch=%u\n",
surface_id,
export_fds_count > 0 ? export_fds[0] : -1,
surface_descriptor->fourcc,
surface_descriptor->width,
surface_descriptor->height,
surface_descriptor->num_objects,
surface_descriptor->num_layers,
surface_descriptor->objects[0].size,
surface_descriptor->layers[0].drm_format,
(unsigned long long)surface_descriptor->objects[0].drm_format_modifier,
surface_descriptor->layers[0].num_planes,
surface_descriptor->layers[0].offset[0],
surface_descriptor->layers[0].pitch[0],
planes_count > 1 ? surface_descriptor->layers[0].offset[1] : 0,
planes_count > 1 ? surface_descriptor->layers[0].pitch[1] : 0);
status = VA_STATUS_SUCCESS;
goto complete;