Skip to content

camx: revision update for Lemans, Talos,Kodiak, Hamoa,Shikra - #3133

Merged
Dmitry Baryshkov (lumag) merged 5 commits into
qualcomm-linux:masterfrom
gkhose-qipl:camx_downstream
Sep 27, 2026
Merged

Dmitry Baryshkov (lumag) merged 5 commits into
qualcomm-linux:masterfrom
gkhose-qipl:camx_downstream

Conversation

@gkhose-qipl

Copy link
Copy Markdown
Contributor

refpolicy-targeted: Rename nhx.sh to camera-nhx

  • The camera test launcher was renamed from nhx.sh to camera-nhx to follow the conventional command naming used for executables installed in /usr/bin. Update its SELinux file context so the renamed launcher continues to receive qcom_nhx_launcher_exec_t.

camxlib-talos: upgrade v1.0.38 -> v1.0.45

  • Add IMX858 and OV13B10 support for Talos Lyra EVK
  • Add IMX577 support for Talos Lyra EVK

camxlib-hamoa: upgrade v1.0.41 -> v1.0.45

  • Updated IMX688 tuning for new schema support, ADRC enablement, and recalibrated AEC/AWB/LSC parameters. Optimized BLC, noise reduction, gamma, and LDC tuning to improve overall image quality and KPI compliance.
  • Updated IMX577 tuning for new schema support and ADRC enablement.
  • Refined AEC/AWB, shading, BLC, ABF, LTM, and gamma tuning to improve low-light performance and overall image quality.

camxlib-lemans: upgrade v1.0.38 -> v1.0.45

  • Add ox08d10 GMSL RB8 support for des1/des2/des3
  • Update package contents to install and package the NHX executable as 'camera-nhx' instead of 'nhx.sh', aligning with the current naming convention.
  • Replace the hardcoded camxtest version in SRC_URI with ${PV} so the artifact version is derived from the recipe version.
  • The libbitml_nsp_73nb_skel.so symlink is now provided as part of the tarball, so the explicit symlink creation in the recipe is no longer required.

camxlib-kodiak: upgrade v1.0.35 -> v1.0.41

  • Add PIRIS(Precision Iris Control) support by integrating a lens driver manager in the camx sensor module and extending AEC framework to support PIRIS-based aperture control and exposure tuning.
  • Base 2A sync enablement in Kodiak.
  • Add logical device entry to support 2 concurrent sensors mcx usecase.

camxcommon-headers: upgrade v1.0.38 -> v1.0.45.

  • Update the headers tar to align with the sources.

@gkhose-qipl

Copy link
Copy Markdown
Contributor Author

Dmitry Baryshkov (@lumag) Could you please add the backport label and GA milestone? Thanks in advance.

@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown

Test run workflow

Test jobs for commit e2fe595

qcom-distro_linux-qcom-6.18
Pass: 235 | Fail: 0 | Total: 259
nodistro
Pass: 10 | Fail: 0 | Total: 10
qcom-distro
Pass: 323 | Fail: 5 | Total: 357

@test-reporting-app

test-reporting-app Bot commented Sep 11, 2026 •

Copy link
Copy Markdown

Test Results

  119 files  ±0    715 suites  ±0   7h 34m 6s ⏱️ - 24m 47s
  171 tests  - 4    165 ✅ + 9   1 💤 ±0  5 ❌  - 13 
4 605 runs  +2  4 547 ✅ +19  53 💤 +2  5 ❌  - 19 

For more details on these failures, see this check.

Results for commit e2fe595. ± Comparison against base commit 047a924.

This pull request removes 4 tests.
lava ‑ auto-login-action
lava ‑ lava-test-retry
lava ‑ lava-test-shell
lava ‑ minimal-boot

♻️ This comment has been updated with latest results.

@gkhose-qipl
Ganesh Khose (gkhose-qipl) marked this pull request as draft September 24, 2026 15:56
@gkhose-qipl

Copy link
Copy Markdown
Contributor Author

Dmitry Baryshkov (@lumag) / Ricardo Salveti (@ricardosalveti) ,
need to check why different SHA in CLI and CI logs for same path
looks like BitBake is not hashing the same content that Artifactory is reporting

job:- https://github.com/qualcomm-linux/meta-qcom/actions/runs/36023613337/job/107715743255?pr=3133

The Artifactory checksum matches the recipe, so the mismatch may be due to BitBake downloading different content than Artifactory is reporting, or a cached/stale download. can we try clearing the download cache and re-fetching the artifact.

ERROR: camxlib-hamoa-1.0.48-r0 do_fetch: Fetcher failure for URL: 'https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz;name=camx'. Checksum mismatch!
File: '/downloads/camx-hamoa_1.0.48_armv8-2a.tar.gz.tmp' has sha256 checksum 'd43e2ab877de00693ed5bb59ca0adc1de9a3d365e486d8e40b4559a33d5117b6' when 'cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d' was expected

~$ curl -I --silent "https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz" | grep -i X-Checksum-Sha2
56
X-Checksum-Sha256: cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d

@ricardosalveti

Copy link
Copy Markdown
Contributor

Dmitry Baryshkov (Dmitry Baryshkov (@lumag)) / Ricardo Salveti (Ricardo Salveti (@ricardosalveti)) , need to check why different SHA in CLI and CI logs for same path looks like BitBake is not hashing the same content that Artifactory is reporting

job:- https://github.com/qualcomm-linux/meta-qcom/actions/runs/36023613337/job/107715743255?pr=3133

The Artifactory checksum matches the recipe, so the mismatch may be due to BitBake downloading different content than Artifactory is reporting, or a cached/stale download. can we try clearing the download cache and re-fetching the artifact.

ERROR: camxlib-hamoa-1.0.48-r0 do_fetch: Fetcher failure for URL: 'https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz;name=camx'. Checksum mismatch! File: '/downloads/camx-hamoa_1.0.48_armv8-2a.tar.gz.tmp' has sha256 checksum 'd43e2ab877de00693ed5bb59ca0adc1de9a3d365e486d8e40b4559a33d5117b6' when 'cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d' was expected

~$ curl -I --silent "https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz" | grep -i X-Checksum-Sha2 56 X-Checksum-Sha256: cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d

I'm retrying to see if it will move further.

Introduce camxlib recipe which delivers three components:
- camx: Core CamX engine for managing camera pipelines and interfacing
  with hardware.
- camxlib: Image processing algorithms and hardware support libraries
  extending CamX functionalities. Uses openCV version 3.1.
- chicdk: Camera hardware interface developer kit providing a
  configurable mechanism for use case selection and camera pipeline
  creation.

Signed-off-by: Ganesh Khose <gkhose@qti.qualcomm.com>
@quaresmajose

Copy link
Copy Markdown
Member

Dmitry Baryshkov (Dmitry Baryshkov (Dmitry Baryshkov (@lumag))) / Ricardo Salveti (Ricardo Salveti (Ricardo Salveti (@ricardosalveti))) , need to check why different SHA in CLI and CI logs for same path looks like BitBake is not hashing the same content that Artifactory is reporting
job:- https://github.com/qualcomm-linux/meta-qcom/actions/runs/36023613337/job/107715743255?pr=3133
The Artifactory checksum matches the recipe, so the mismatch may be due to BitBake downloading different content than Artifactory is reporting, or a cached/stale download. can we try clearing the download cache and re-fetching the artifact.
ERROR: camxlib-hamoa-1.0.48-r0 do_fetch: Fetcher failure for URL: 'https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz;name=camx'. Checksum mismatch! File: '/downloads/camx-hamoa_1.0.48_armv8-2a.tar.gz.tmp' has sha256 checksum 'd43e2ab877de00693ed5bb59ca0adc1de9a3d365e486d8e40b4559a33d5117b6' when 'cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d' was expected
~$ curl -I --silent "https://qartifactory-edge.qualcomm.com/artifactory/qsc_releases/software/chip/component/camx.qclinux.0.0/260923.1/prebuilt_yocto_master/camx-hamoa_1.0.48_armv8-2a.tar.gz" | grep -i X-Checksum-Sha2 56 X-Checksum-Sha256: cba9487e3ada7f9649685b81bf6ea17c8c40d7ce169284b2dae3fcf3cdbeec0d

I'm retrying to see if it will move further.

I have fixed a similar issue and I still have the branch around #3072. I will take a look on it.

@quaresmajose

Copy link
Copy Markdown
Member

Lets see if it get fixed with #3241

@ricardosalveti

Copy link
Copy Markdown
Contributor

I believe we are having sync issues with efsx when we have multiple runners all downloading the same file, almost like if the download lock is not really propagating fast enough, causing download issues. #3241 should proabbly help yeah.

@ricardosalveti

Copy link
Copy Markdown
Contributor

#3241 built fine, Jose Quaresma (@quaresmajose) just propose the changes.

@gkhose-qipl

Copy link
Copy Markdown
Contributor Author

#3241 built fine, Jose Quaresma (Jose Quaresma (@quaresmajose)) just propose the changes.

Ricardo Salveti (@ricardosalveti) ,
do we need to close 3133 and merge 3241?

@gkhose-qipl

Copy link
Copy Markdown
Contributor Author

Dmitry Baryshkov (@lumag) / Ricardo Salveti (@ricardosalveti) ,
can you please review PR

@quaresmajose

Jose Quaresma (quaresmajose) commented Sep 26, 2026 •

Copy link
Copy Markdown
Member

#3241 built fine, Jose Quaresma (Jose Quaresma (Jose Quaresma (@quaresmajose))) just propose the changes.

Ricardo Salveti (Ricardo Salveti (@ricardosalveti)) , do we need to close 3133 and merge 3241?

No, the #3241 was just to fix the download cache storage that gets corrupted.

I've already relaunch a new CI build for this PR that no longer has the previous issue.

@quaresmajose

Copy link
Copy Markdown
Member

I believe we are having sync issues with efsx when we have multiple runners all downloading the same file, almost like if the download lock is not really propagating fast enough, causing download issues. #3241 should proabbly help yeah.

Yes, we’ve been having some issues with the download cache since we switched the file system. This is the second time I’ve had to use that workaround #3241 to fix file system problems. Another interesting point is that this has always happened with files coming from Artifactory.

However, I've never heard of such a thing being reported on the autobuilder, even though they run several build jobs at the same time. I think there is an anomaly in our file system's implementation of locks.

@qcomlnxci

Copy link
Copy Markdown

Test Coral run workflow

Test jobs for commit e2fe595

  • qcomdistro: multimedia image-prop
    Pass: 32 | Fail: 0 | Total: 32
  • qcomdistro: multimedia image
    Pass: 10 | Fail: 0 | Total: 10

@ricardosalveti

Ricardo Salveti (ricardosalveti) commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

However, I've never heard of such a thing being reported on the autobuilder, even though they run several build jobs at the same time. I think there is an anomaly in our file system's implementation of locks.

Yes, it might be fsx specific, need to debug further, separately from this pr.

DEPENDS += " \
${@bb.utils.contains('DISTRO_FEATURES', 'opencl', 'virtual/libopencl1', '', d)} \
${@bb.utils.contains('DISTRO_FEATURES', 'opengl', 'virtual/egl virtual/libgles2', '', d)} \
"

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nit: what was wrong with the previous lines?


S = "${UNPACKDIR}"

DEPENDS += "glib-2.0 fastrpc protobuf libxml2 qmi-framework sensinghub qcom-sensors-binaries"

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nit: why?

@lumag

Copy link
Copy Markdown
Contributor

Known failures.

@lumag
Dmitry Baryshkov (lumag) merged commit d006b6a into qualcomm-linux:master Sep 27, 2026
220 of 228 checks passed
@quic-yocto-ci

Copy link
Copy Markdown
Contributor

Backport failed for wrynose, because it was unable to cherry-pick the commit(s).

Please cherry-pick the changes locally and resolve any conflicts.

git fetch origin wrynose
git worktree add -d .worktree/backport/3133-to-wrynose origin/wrynose
cd .worktree/backport/3133-to-wrynose
git switch --create backport/3133-to-wrynose
git cherry-pick -x 9e23f6c9fceb566b196137a741798c37143306eb ffbcf453132702f69b52a9f508a83542e43ac4d5 3e048ae24c7203613ede6586e8f214d760037d50 fcdea88073a880663844f07e5e546283d4628ba6 e2fe5951b912d2c451b667eeea32e46957c3df36

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

9 participants