admin/LithosAnanake
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

G.4 (2h): fix dangling §G.4 reference in G.1 follow-up note

Browse Source
This commit is contained in:
Robert Allan James
2026-08-29 09:50:06 -04:00
parent dc2f38a1e1
commit aee1ecaa8f
1 changed files with 12 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
FABRIC-3.md
+12 -5
View File
@@ -3333,11 +3333,18 @@ failed"` gate, now with an else-branch for the recoverable-STALL case.
is deferred to hardware (v2.5.0/Artemis bare-metal)**, where a real bad transfer can be is deferred to hardware (v2.5.0/Artemis bare-metal)**, where a real bad transfer can be
staged. This is the last QEMU-verifiable storage-integrity gap and the recovery logic is in staged. This is the last QEMU-verifiable storage-integrity gap and the recovery logic is in
place; the one thing QEMU cannot prove is the live stall injection itself. place; the one thing QEMU cannot prove is the live stall injection itself.
- **Follow-up (pre-existing, NOT G.1): ROOT-CAUSED and FIXED 2026-08-29 — see §G.4 below.** - **Follow-up (pre-existing, NOT G.1): ROOT-CAUSED and FIXED 2026-08-29.** During G.1
During G.1 verification an intermittent boot-time attach race was observed (the verification an intermittent boot-time attach race was observed (the `sk_repl_idle()`
`sk_repl_idle()` `bot_msc_attach_pending` handoff occasionally does not progress on a cold `bot_msc_attach_pending` handoff occasionally does not progress on a cold QEMU boot,
QEMU boot, independent of source, with baseline `HEAD` exhibiting it too). Unrelated to G.1; independent of source, with baseline `HEAD` exhibiting it too). Unrelated to G.1. Root
root cause and fix are documented in §G.4, verified across six consecutive fresh boots. cause: `xhci_poll_events()`'s event-ring drain loop had no hard ceiling — ERDP is written
back only when the loop exits, so the controller cannot reclaim event TRBs mid-drain, and
on pathological controller behavior the head can chase the software dequeue pointer
indefinitely, causing the drain (and hence the attach handoff) to livelock. Fixed by bounding
the drain to one full ring (`XHCI_EVT_RING_MAX_DRAIN`, `xhci.c` `xhci_poll_events()`); the
loop always terminates and always writes ERDP each call, and unprocessed events keep their
cycle bit and are re-read next poll (nothing is dropped). Verified across six consecutive
fresh boots.
#### G.2 [v2.0.0] Real-hardware RNG driver plumbing, QEMU-verifiable slice (rest of it lands at v2.5.0) #### G.2 [v2.0.0] Real-hardware RNG driver plumbing, QEMU-verifiable slice (rest of it lands at v2.5.0)