Skip to content

Refresh x86_64 checksums and patch idempotency#1

Open
maximdeclercq wants to merge 2 commits into
ichergui:wip-l4t-r39.2.0from
maximdeclercq:wip-l4t-r39.2.0
Open

Refresh x86_64 checksums and patch idempotency#1
maximdeclercq wants to merge 2 commits into
ichergui:wip-l4t-r39.2.0from
maximdeclercq:wip-l4t-r39.2.0

Conversation

@maximdeclercq

Copy link
Copy Markdown

Two fixes found building the R39.2.0 / JetPack 7.2 x86_64 host toolchain.

  1. NVIDIA rebuilt + re-signed the whole x86_64 cuda 13.2 .deb set under the same version strings (last-modified 2026-06-03 20:03 UTC, confirmed against the published binary-amd64/Packages index), so all 35 pinned *SUM:x86-64 values fail do_fetch "Checksum mismatch" on an x86_64 build host. Refreshed to the current published sha256s.
  2. patch_nv_modprobe applies its patch with a raw patch in a do_patch postfunc, not idempotent: on a re-run against an already-patched WORKDIR (devtool modify, -c patch -f, sstate-miss re-unpack) it fails "Reversed (or previously applied) patch detected". Guard on the .download_stamp it already writes (the marker mk/nvidia-modprobe.mk uses).

@ichergui ichergui force-pushed the wip-l4t-r39.2.0 branch 3 times, most recently from 0fb60ac to c534132 Compare June 6, 2026 01:19
@maximdeclercq maximdeclercq force-pushed the wip-l4t-r39.2.0 branch 3 times, most recently from 1511e05 to 8211e22 Compare June 8, 2026 11:12
The T264 flash helper reads these from the generated flashvars file, but they were missing from TEGRA_FLASHVARS, so tegra-flashvars left them empty.

Signed-off-by: Maxim De Clercq <maximdeclercq00@gmail.com>
@maximdeclercq maximdeclercq force-pushed the wip-l4t-r39.2.0 branch 2 times, most recently from c41562e to db28ce3 Compare June 8, 2026 11:23
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants