DF-5575 — ppb_MS_loop() emits MS_DBRA(-1) whose branch lands on the DBRA itself — the loop body executes exactly once regardless of iter

Severity: Info · Confidence: certain · File: sys/bus/ppbus/ppb_msq.c (pass 2)

Reproduce

1. sh build.sh — compiles the verbatim production ppb_msq.c into msqhx.ko
2. sh run.sh — case 0 (non-fatal)
3. Observe dmesg (dmesg | grep MSQHX) / the serial console.

Expected

BASELINE (unfixed): 'DF-5575 iter=5 -> prolog=1 body=1 epilog=1 dbra_spins=4' — body ran ONCE. PATCHED: 'DF-5575 iter=5 -> prolog=1 body=5 epilog=1' — body runs iter times.

Harness (shared by DF-5573..DF-5577)

msqhx.c + Makefile compile the verbatim production /usr/src/sys/bus/ppbus/ppb_msq.c (sha256 223e7f9184c81ab83f49228ae7d1f99a 6e3936b9a452b01c6bb6176b150dda67, identical to the repo sys/ copy) into a KLD, together with a fake ppbus controller on nexus whose ppbus_exec_microseq method implements the exact dispatch semantics of ppc_exec_microseq() (sys/dev/misc/ppc/ppc.c:1332-1542) for the exercised opcodes. The interpreter target device's softc is a real struct ppb_data and its child's ivars are a real struct ppb_device, so mode2xfer(), ppb_MS_init(), ppb_MS_exec(), ppb_MS_microseq(), ppb_MS_loop() and ppb_MS_init_msq() all run their production code paths. Evidence class: "KLD-harness-with-production-code" (no parallel port in the QEMU rig) — family standard, cf. DF-5418.

Build

# inside the guest, from this directory
sh build.sh          # -> /tmp/msqhx-build/msqhx.ko

Run

sh run.sh            # runs this pack's case via sysctl hw.msqhx_case
sh run.sh <case>     # 0=suite(5575+5573) 2=5574 4=5576a 5=5576b 6=5577

Cases 2/4/5 PANIC the guest on unfixed production code (that is the finding); the panics land on the serial console (dfbsd-qemu/vm.sh log).

Fix validation

fix.diff (sha256 8e3dfc90e6049ce1704c3cef833cba0904b6404f11360bdd299e37980e9327bb) patches sys/bus/ppbus/ppb_msq.c + sys/bus/ppbus/ppbconf.h (all five findings; never applied to the repo's read-only sys/ tree). Validation procedure: copy the fixed files into the guest's /usr/src/sys/bus/ppbus/, sh build.sh again (recompiles the production source), re-run the same case. Baseline vs patched outputs are in run.log / fix_run.log (plus panic*.txt / fix_run.log for the fatal cases).

DF-5575 — verdict narrative

Finding

ppb_MS_loop() emits MS_DBRA(-1) whose branch lands on the DBRA itself — the loop body executes exactly once regardless of iter

Bottom line (status: reproduced, impact: none)

Reproduced on the QEMU guest (DragonFly 6.5-DEVELOPMENT x86_64). The adapter-level branch contract (ppc.c:1447-1451 'mi += offset; mi++') resolves a branch target to mi+offset+1, corroborated by every working in-tree sequence (nibble_inbyte DBRA(-7) at index 6 targets index 0; ps2_inbyte DBRA(-4) at index 4 targets index 1; vpoio.c:195/210/226/239). ppb_MS_loop()'s generated sequence places the body CALL at index 2 and MS_DBRA(-1) at index 3 (ppb_msq.c:160-170): -1 from index 3 lands on index 3 — the DBRA itself — so the branch spins in place decrementing the accumulator (harness counter dbra_spins=4 for iter=5) while the body CALL is never re-executed: prolog=1 body=1 epilog=1 regardless of iter. The documented API contract ('execute a microseq loop', body iter times) is broken; the correct offset is -2. ppb_MS_loop() has zero in-tree callers (grep across sys/ finds only the definition/declaration), so this is latent dead-API logic damage: severity Info. Fix validated: MS_DBRA(-2) makes the same run print body=5 with prolog/epilog unchanged.

Exploit chain

n/a

Evidence

run.log ('DF-5575 iter=5 -> prolog=1 body=1 epilog=1 dbra_spins=4 ... BODY EXECUTED ONCE (off-by-one CONFIRMED)') on pristine production code; fix_run.log ('observed body=5 -> loop works') on the fix.diff-patched source; the adapter-contract derivation is documented in VERDICT.md

Kernel references

• sys/bus/ppbus/ppb_msq.c:160
• sys/bus/ppbus/ppb_msq.c:166
• sys/dev/misc/ppc/ppc.c:1447
• sys/dev/disk/vpo/vpoio.c:195
• sys/dev/disk/vpo/vpoio.c:210

Fix validation (fixed)

MS_DBRA(-1) -> MS_DBRA(-2): iter=5 now yields body=5 (prolog=1 epilog=1), matching the documented ppb_MS_loop() semantics; no other behavior changed.

fix.diff (sha256 8e3dfc90e6049ce1704c3cef833cba0904b6404f11360bdd299e37980e9327bb) carries all five ppb_msq.c fixes for this pass; the hunks relevant to this finding are marked with its DF id in the comments. Fix validation followed the family standard (cf. DF-5418): the fixed sources were installed into the guest's /usr/src/sys/bus/ppbus/, the harness was rebuilt (recompiling the production file), and the exact same case was re-run — baseline vs patched outputs are in run.log / fix_run.log (+ panic*.txt). The repo's sys/ tree was never modified; the guest was vm.sh reset with-src afterwards (sources restored+verified by snapshot).

PoC changes vs seed

No seed PoC existed; written fresh. Non-fatal counters via MS_C_CALL hooks + SUBRET, executed by the fake adapter interpreter replicating ppc semantics.

Guest: DragonFly dfbsd 6.5-DEVELOPMENT DragonFly 6.5-DEVELOPMENT #0: Thu Jul 2 06:02:54 UTC 2026 root@dfbsd:/usr/obj/usr/src/sys/X86_64_GENERIC x86_64 Harness sha256: 4b728315b40157934f62c99e847f386b30202eef77083dc1a5c065b7b0448ded

Fix status (batch validation)

Applied with the 36 sibling fix.diffs of this batch in findings/poc/_batch41/all_fixes.patch; single in-guest make -j6 nativekernel KERNCONF=X86_64_GENERIC completed RC=0 with zero errors under -Werror (full log: findings/poc/_batch41/fix_build8.log). Behavior A/B intentionally skipped for this Info-tier item.