jan/rrt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Narrow peer-site selector source lanes

Browse source
This commit is contained in:
Jan Petykiewicz 2026-04-18 18:49:16 -07:00
commit 359350f6c5
2 changed files with 13 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

12
crates/rrt-runtime/src/smp.rs
View file

@ -4111,6 +4111,8 @@ fn build_periodic_company_service_trace_report(
.to_string(),
"0x0045c36e / 0x0045da65 / 0x0045e0fc dense 0x00456100 caller family over stack-backed buffers and default scalar lanes"
.to_string(),
"0x0045c36e feeds 0x00456100 selector arg 1 from [owner+0x23e], 0x0045da65 feeds zero, and 0x0045e0fc feeds [ebp+0x08]"
.to_string(),
"0x00485819 typed placed-structure caller of 0x0052edf0 via 0x530640-style argument bundle"
.to_string(),
"0x00490a79 chooser-side caller of 0x00455b70 with literal selector 0x005cfd74 and fallback seed 0x005c87a8"
@ -4119,7 +4121,7 @@ fn build_periodic_company_service_trace_report(
"0x00409950 linked-transit roster sibling owner".to_string(),
];
let next_owner_questions = vec![
"Whether the dense 0x0045c36e / 0x0045da65 / 0x0045e0fc caller family over 0x00456100 is the actual live class-0 site-row restore strip that 0x00420030 / 0x00420280 / 0x0047efe0 / 0x0047fd50 depend on, and if so which persisted lane from those stack-backed buffers feeds the selector arg 3 into 0x00455b70 (with arg 1 as fallback)?".to_string(),
"Whether the dense 0x0045c36e / 0x0045da65 / 0x0045e0fc caller family over 0x00456100 is the actual live class-0 site-row restore strip that 0x00420030 / 0x00420280 / 0x0047efe0 / 0x0047fd50 depend on; the wrapper mapping is now grounded, so the remaining question is which of the sampled selector-source lanes ([owner+0x23e], literal zero, or [ebp+0x08]) corresponds to the persisted site-row selector.".to_string(),
"How much of the linked-peer refresh path is strictly post-load versus recurring runtime maintenance now that 0x004133b0 reaches 0x0040ee10 -> 0x0040edf6 -> 0x00480710 during bring-up and 0x004160aa also re-enters 0x0040ee10 later?".to_string(),
"Which persisted linkage fields behind 0x0047efe0, [region+0x2a4], and [region+0x276] are sufficient to rehost the 0x004014b0 proximity-and-news branch without shell state?".to_string(),
"Which infrastructure consumer above the grounded 0x38a5 seam actually drives the linked-transit branch that 0x00409950 follows?".to_string(),
@ -4284,7 +4286,7 @@ fn build_periodic_company_service_trace_report(
"Direct disassembly now closes the negative persistence side too: the direct 0x36b1 per-record callbacks serialize the shared base scalar triplets rooted at [this+0x206/+0x20a/+0x20e] plus the subordinate payload callback strip, while the 0x4a9d/0x4a3a/0x4a3b side-buffer owner only persists route-entry lists, three byte arrays, five proximity buckets, and the sampled-cell list. That means neither checked-in save owner seam currently persists the core peer-site identity fields [site+0x04], [site+0x2a8], or [peer+0x08] directly.".to_string(),
);
notes.push(
"The replay strip is tighter now too. 0x00444690 is the current late world bring-up caller of 0x004133b0, that outer owner drains queued site ids through 0x0040e450 and then sweeps every live placed structure through 0x0040ee10, and 0x0040ee10 itself reaches 0x0040edf6 -> 0x00480710 plus the later 0x0040e360 follow-on. A separate runtime path at 0x004160aa also re-enters 0x0040ee10 later. So [peer+0x08] replay is no longer the open question, and [site+0x04] itself is no longer an owner mystery either: 0x00455b70 feeds 0x0052edf0 with arg 3 as the primary selector and arg 1 as fallback, while grounded caller shapes now include the local wrapper 0x00456100, the fixed 0x55f2 callback 0x00456072, and the chooser-side path 0x00490a79. The current leading restore-family hypothesis is the dense 0x0045c36e / 0x0045da65 / 0x0045e0fc caller strip above 0x00456100, because those callers repeatedly allocate 0x23a rows, pass stack-backed buffers, and use the same default scalar lanes. The remaining non-hook target is to confirm whether that family seeds the live site rows that 0x00420030 / 0x00420280 / 0x0047efe0 / 0x0047fd50 consume, and to name the persisted selector lane inside that wrapper bundle.".to_string(),
"The replay strip is tighter now too. 0x00444690 is the current late world bring-up caller of 0x004133b0, that outer owner drains queued site ids through 0x0040e450 and then sweeps every live placed structure through 0x0040ee10, and 0x0040ee10 itself reaches 0x0040edf6 -> 0x00480710 plus the later 0x0040e360 follow-on. A separate runtime path at 0x004160aa also re-enters 0x0040ee10 later. So [peer+0x08] replay is no longer the open question, and [site+0x04] itself is no longer an owner mystery either: 0x00455b70 feeds 0x0052edf0 with arg 3 as the primary selector and arg 1 as fallback, while grounded caller shapes now include the local wrapper 0x00456100, the fixed 0x55f2 callback 0x00456072, and the chooser-side path 0x00490a79. The current leading restore-family hypothesis is the dense 0x0045c36e / 0x0045da65 / 0x0045e0fc caller strip above 0x00456100, because those callers repeatedly allocate 0x23a rows, pass stack-backed buffers, and use the same default scalar lanes; the wrapper mapping is now grounded far enough to say that the sampled selector-source lanes are [owner+0x23e], literal zero, and [ebp+0x08] respectively. The remaining non-hook target is to confirm which of those sampled lanes corresponds to the persisted selector for the live site rows that 0x00420030 / 0x00420280 / 0x0047efe0 / 0x0047fd50 consume.".to_string(),
);
SmpPeriodicCompanyServiceTraceReport {
@ -25945,7 +25947,7 @@ mod tests {
let trace = build_periodic_company_service_trace_report(&analysis);
assert_eq!(trace.selected_company_id, Some(7));
assert_eq!(trace.atlas_candidate_consumers.len(), 7);
assert_eq!(trace.known_bridge_helpers.len(), 35);
assert_eq!(trace.known_bridge_helpers.len(), 36);
assert_eq!(trace.next_owner_questions.len(), 4);
assert_eq!(trace.companies.len(), 1);
assert!(trace.atlas_candidate_consumers.iter().any(|line| {
@ -25954,7 +25956,9 @@ mod tests {
&& line.contains("peer-site boolean/selector pair")
}));
assert!(trace.next_owner_questions.iter().any(|line| {
line.contains("0x0045c36e") && line.contains("0x00456100") && line.contains("arg 3")
line.contains("0x0045c36e")
&& line.contains("0x00456100")
&& line.contains("[owner+0x23e]")
}));
assert!(trace.next_owner_questions.iter().any(|line| {
line.contains("0x004160aa")

6
docs/rehost-queue.md
View file

@ -75,6 +75,9 @@ Working rule:
- `0x0045c36e / 0x0045da65 / 0x0045e0fc` are concrete callers of `0x00456100`, repeatedly
allocating `0x23a` rows, forwarding stack-backed buffers, and using the same default scalar
lanes
- the `0x00456100 -> 0x00455b70` wrapper mapping is now grounded far enough to say the sampled
selector-source lanes are `[owner+0x23e]` at `0x0045c36e`, literal zero at `0x0045da65`, and
`[ebp+0x08]` at `0x0045e0fc`
- `0x00485819` is one typed placed-structure caller of `0x0052edf0` through the generic
three-arg wrapper `0x00530640`
- `0x00490a79` is one chooser-side caller of `0x00455b70`, feeding literal selector
@ -97,7 +100,8 @@ Working rule:
`0x00420030 / 0x00420280 / 0x0047efe0 / 0x0047fd50` consume the resulting selector, with the
current first target being the dense `0x0045c36e / 0x0045da65 / 0x0045e0fc -> 0x00456100`
family
- name the persisted lane that reaches selector arg `3` inside that wrapper bundle
- name which of the currently sampled selector-source lanes (`[owner+0x23e]`, literal zero, or
`[ebp+0x08]`) reaches selector arg `3` for the live site rows
- Use the higher-layer probes as the standard entry point for the current blocked frontier instead
of generic save scans:
`runtime inspect-periodic-company-service-trace <save.gms>`,