Repository navigation
media: port OV13858 and VD55G0 research with explicit stop retries (draft) - #73
Merged
ooaklee merged 1 commit intoNov 12, 2026
Merged
Conversation
This was referenced Oct 20, 2026
ooaklee
added a commit
that referenced
this pull request
Nov 10, 2026
Preserve the original tools and historical validation commits, with scoped documentation and Linux Node 22 parser CI. Camera kernel foundations remain in PR #73.
ooaklee
changed the title
ooaklee
changed the base branch from
x/sp11-sam-qualification-20261015
to
sp11/ubuntu-qcom-x1e-7.3.y
November 10, 2026 14:47
ooaklee
force-pushed
the
x/sp11-v25-native-camera-20261016
branch
from
November 10, 2026 14:47
3cfa1bf to
2e03812
Compare
ooaklee
force-pushed
the
x/sp11-v25-native-camera-20261016
branch
from
November 12, 2026 03:25
2e03812 to
d0ad6e1
Compare
ooaklee
changed the title
ooaklee
force-pushed
the
x/sp11-v25-native-camera-20261016
branch
from
November 12, 2026 19:18
2dc0f03 to
c3999d5
Compare
Port the SP11 camera research from the v25 native camera branch onto the current 7.3 lifetime. Add the OmniVision OV13858 device-tree binding and native OF power management. Extend the ST VD55G1 driver for the VD55G0 global-shutter sensor: indirect hardware register addressing, model/revision feature abstraction, unified frame timing, corrected register access widths, bounded optional LED configuration with illumination gated pending qualification, GPIO inputs established after firmware boot, native exposure and HDR limit retention, stream ownership preserved until stop succeeds, legacy retries after retained stop errors, and safe line lengths with sensor-style asynchronous registration. Narrow the st,vd55g1 binding to avoid a family-wide fixed-address claim and track both bindings in MAINTAINERS. Validated with strict checkpatch, hermetic AArch64 W=1 builds, and behavioral fixtures for the stop/retry ownership (see sp11-driver-parity-artifacts/20261015/private/camera-v3*). The v4l2-subdev retained-stop opt-in this series depends on merged earlier as PR #80. Assisted-by: Codex:gpt-5.6 Assisted-by: OpenCode:omni-route/zai/glm-5.3 Signed-off-by: Leon Silcott <lnsilcott@gmail.com>
ooaklee
force-pushed
the
x/sp11-v25-native-camera-20261016
branch
from
November 12, 2026 20:06
c3999d5 to
7d134a4
Compare
ooaklee
marked this pull request as ready for review
November 12, 2026 20:07
ooaklee
added a commit
that referenced
this pull request
Nov 12, 2026
…config (#76) Port the Qualcomm USB4 host-router research frontend from PR #76 (head b888847) onto the current sp11/ubuntu-qcom-x1e-7.3.y head, which already carries the PHY prerequisite from #24 (squash-merged as 30d6b18) with its tunnel-clock ownership fix. Carried from the PR head, byte-identical: - drivers/thunderbolt/Kconfig: default-off CONFIG_USB4_QCOM_EXPERIMENTAL research scaffold option (depends on ARCH_QCOM && OF). - drivers/thunderbolt/Makefile: conditional qcom_usb4_hr.o linkage into the thunderbolt composite. - drivers/thunderbolt/pci.c: registration in domain -> PCI -> Qualcomm order with reverse error unwind and unload; disabled builds use no-op private stubs. - drivers/thunderbolt/qcom_usb4_hr.c/.h: the research backend with the whole-container firmware validator derived from the REA evidence against QcUsb4Filter8380.sys (FUN_140005530 consuming the 40816-byte embedded container at 0x14004d1c0), full pre-map MMIO bounds validation of all named regions, and serialised stop/restart ownership. One deliberate change on top of the PR blobs: the UCS0 mailbox mapping now first tries a named optional "ucsb" platform resource and only then falls back to the fixed ACPI-declared window (QCOM0CA4, USBC 0x81F20040/0x40). The mapping error path unwinds through the prepare devres group instead of being ignored. This addresses the review gate on the previously unconditional hardcoded devm_ioremap. Not carried: the PR's PHY blobs (superseded by #24's merged tunnel-clock fix; its com_init/com_exit would revert that fix), camera blobs (already merged through #73) and its changelog entry (replaced below). debian.qcom-x1e/changelog: new 7.3.0-jg-1sp11v5 entry. This remains a source-review scaffold: no DT activation, firmware upload still gated behind fw_load_allow=0, and no hardware qualification claimed. MMIO/firmware/activation semantics are unchanged from the reviewed PR apart from the UCSB mapping fix above.
This was referenced Nov 13, 2026
ooaklee
added a commit
that referenced
this pull request
Nov 14, 2026
…earch Port the DP cluster rework from the 7.2 overlay (507356d) to the current 7.3 tree: explicit AUX transfer ownership with fault draining, state-lock-serialized ctrl transitions, audio restart/stop entrypoints, and the rewritten dp_display host lifecycle. Adaptations for the 7.3 base: - config_ctrl keeps the RMW clear of stream-specific fields from the #73 stream split (commit 7de83c2), layered under the overlay rework. - bridge ops switch from .atomic_reset to .atomic_create_state (base commit 65dc3b4 removed atomic_reset support). - container_of and long-line wraps to keep checkpatch line-length clean. Build-verified on the Surface Pro 11 (native arm64 gcc 15.2, shipped 7.3.0-jg-1sp11v4 source + Ubuntu config): all eight dp/*.o plus the full msm/ directory compile with zero warnings or errors. Signed-off-by: Leon Silcott <lnsilcott@gmail.com>
1 of 2 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Port the native OV13858 / VD55G0 camera research series onto the current
sp11/ubuntu-qcom-x1e-7.3.ydefault. Remain draft: failed-start/remove ownership and broader qualification gates remain.The 26 existing camera author commits are preserved on the new base, with one focused follow-up: opt the modified VD55 driver into the retained-stop API merged by #80. Its failed stop retains the stream PM reference and grabbed controls; the legacy wrapper now returns that error and retains enabled state for a later explicit retry. Other helper users keep their existing behavior. This resolves the prior wrapper/resource-retention mismatch.
Scope remains five camera-only files: OV13858 implementation/binding, VD55 family implementation/binding, and MAINTAINERS. No board graph, illumination, configuration or activation change is included.
Validation: the standalone core change passed independent review, GitHub CI, 192 assertions/five negative controls and AArch64 compilation before merging. The private source-derived VD55/core integration fixture covers 1,612 assertions in 101 scenarios and four expected-failing controls; both complete AArch64 sensor objects (
ov13858.o,vd55g1.o) build at final headd0ad6e1fd427withW=1, GCC 15.2 / GNU ld 2.46, without warnings. The final composed VD55/core fixture also passes; these checks do not establish physical capture or power-state behavior.Remaining blockers:
The source findings are not claims of an observed runtime leak or physical power failure. The original pre-update head
2e03812da559is preserved atarchive/pr73-pre-stop-retry-20261112.