Conversation
roshs189
force-pushed
the
add/iq10-rrd
branch
2 times, most recently
from
August 22, 2026 17:40
874230a to
f10899e
Compare
roshs189
force-pushed
the
add/iq10-rrd
branch
3 times, most recently
from
August 28, 2026 16:03
95db71b to
3c8f5b0
Compare
roshs189
added a commit
to roshs189/meta-qcom
that referenced
this pull request
Aug 29, 2026
Re-points qcom-ptool.inc SRCREV at the current head of roshs189/qcom-ptool#14 (add/iq10-rrd on qcom-ptool) so the build picks up its partition files. Signed-off-by: Roshan Sivakumar <roshs@qti.qualcomm.com>
Signed-off-by: Roshan Sivakumar <roshs@qti.qualcomm.com>
roshs189
added a commit
to roshs189/meta-qcom
that referenced
this pull request
Aug 29, 2026
…sed head PR roshs189/qcom-ptool#14 was rebased to a single commit; its head moved to 734ac25a. Reconcile the SRCREV pin to the current PR head so the ptool source built for iq10-rrd tracks the open PR.
Auto-correct the generated ufs partitions.conf so it only references binaries that actually exist at flash time (bootbins union ∪ flat-build produced), leaving partition.xml / board_partition_meta.yaml read-only. - substitute stale devcfg name -> shipped binary (devcfg_auto_sgvm.mbn -> devcfg_iot.mbn) in both slots, per board_partition_meta.yaml rename_map - line-drop (comment whole --partition) the 8 firmware-only partitions whose payload is neither in the bootbins zips nor produced by the flat build - token-drop (comment only --filename=, keep the GPT slot) the 34 absent-payload entries so LUN/GPT geometry is preserved - rootfs.img / efi.bin / dtb.bin kept (flat-build produced) Each removal is a "# RECONCILE-DROP(...)" comment (never a silent delete) so PR review sees exactly what was dropped and why. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ion.xml Regenerate the reconciled UFS partitions.conf directly from the current spec partition.xml filtered by the flashable set (bootbins union ∪ build_produced), dropping the rename_map concept entirely. partitions.conf is always derive(partition.xml) — a rename map only papered over a stale input. Net effect vs previous commit, all spec-authoritative corrections: - drop keymaster_a/b + kmvirt_a/b: commented-out (not live) in partition.xml - correct uefisecapp_a/b to the live spec entry (type-guid + uefi_sec.mbn), replacing the stale bare slots - devcfg_iot.mbn now a direct KEEP from the union (never a substitution) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…er-safe
Two coupled fixes for a flash-time "Failed to Open File adsp2_dtbs.elf"
regression on iq10-rrd.
Root cause: loaders/conf.read_conf only skipped lines whose first
non-space char was '#'; it never stripped *inline* comments. A reconcile
token-drop line of the form
--partition ... --type-guid=... # RECONCILE-DROP(no-bootbin): --filename=adsp2_dtbs.elf
therefore survived, and parse_partition_lines' naive line.split(" ")
handed the commented "--filename=adsp2_dtbs.elf" token to getopt, which
re-added the deliberately-dropped filename. gen_partition then emitted a
rawprogram entry for a binary the flat build never produces, so the
flasher failed opening the non-existent file.
Fix 1 (loader, true root cause): split each conf line on the first '#'
before tokenising, so inline comments never leak option tokens. Whole-line
comments and blanks still skip. --disk/--partition values never contain
'#', so this is safe. 107 unit tests pass.
Fix 2 (defense in depth, emit format): regenerate iq10-rrd
partitions.conf so a token-drop emits NO --filename token on the live
--partition line (the canonical stock idiom for an image-less GPT slot)
and moves the reconcile reason to a SEPARATE preceding comment line. This
keeps the committed conf safe even against an unpatched upstream loader.
Verified end-to-end (partitions.conf -> gen_partition -> partition.xml ->
ptool -> rawprogram*.xml): all 20 reconcile-dropped binaries, including
adsp2_dtbs.elf, are absent from every rawprogram[0-9].xml, while kept
binaries (cpucp_dtbs.elf, uefi_dtbs.elf, rootfs.img, efi.bin, dtb.bin)
remain. GPT slots for the dropped audio_dsp/compute_dsp regions are
preserved (token-drop), matching the board layout.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
roshs189
added a commit
to roshs189/meta-qcom
that referenced
this pull request
Sep 1, 2026
… + firmware two-zip rework Points qcom-ptool.inc at the head of roshs189/qcom-ptool#14 (add/iq10-rrd on qcom-ptool) so the build picks up its partition files. Reworks firmware-qcom-boot-nord to fetch and deploy the plain and immutable NORD_bootbinaries zips separately (per spec firmware_boot.notes), replacing the stale artifactory-las flat bundle with the two versioned zips reconstructed from FW_ARTIFACTORY/PV, and adds the immutable set's sha256sum alongside the updated plain-zip sha256sum. Signed-off-by: Roshan Sivakumar <roshs@qti.qualcomm.com>
shrm_b was placed on LUN5 (per iq10-rrd partition.xml block comments), which fails the XBL/Sahara pre-load storage read with 0xAE001002. On the booting NORD reference board (kaanapali-mtp) both shrm_a and shrm_b live on LUN4. Relocate shrm_b to LUN4 alongside shrm_a via the board spec's lun_reference opt-in. Only shrm_b moves; every other partition keeps its block-index LUN, and all RECONCILE-DROP entries are preserved verbatim. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
roshs189
added a commit
that referenced
this pull request
Sep 6, 2026
…strip Two coupled durable fixes pinned together for iq10-rrd flashing: 1. partitions.conf: drop --filename=tools.fv from the toolsfv partition. No recipe produces a tools.fv binary and none exists in the qcomflash deploy tree; the reference-board template left the filename behind, so gen_partition emitted a rawprogram entry for a non-existent file and the flasher failed with 'Failed to Open File ... tools.fv: Invalid argument'. toolsfv keeps its GPT slot (filename=""), matching its siblings quantumcontentfv/quantumfv/quantumsdk. 2. loaders/conf.read_conf: strip inline comments before tokenising (ported from PR #14 959845e). The 5ee8fff line this fix was based on predated that loader fix; re-pinning without it would silently reopen the inline-comment trap (a '# ... --filename=foo' comment leaking a dropped filename token back into getopt). 107 unit tests pass.
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
1 participant
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.
Summary
partitions.confandcontents.xml.infor theiq10-rrdboard (UFS storage, SA8797P/NORD chipset family)partition.xmlspec (8 physical_partition/LUN blocks);vendor_aexcluded as it has no_bcounterpart (unpaired A/B label)Validation
make all: generates partitions.xml, GPT images, rawprogram/patch XML, and contents.xml cleanly for this boardtests/integration/check-missing-files: 0 Missing (no co-location issues); 25 unique Unknown binaries — all are genuinely new/board-specific boot images not yet in the allowlist, not typos (list available on request; a follow-up PR is expected to extend the allowlist separately)Test plan
make allsucceedscheck-missing-filesshows 0 Missingpartition.xmlpartitions.conf+contents.xml.incommitted (generated artifacts cleaned up)