[maglev non-eager inlining] Update phi types after inlining [v8/v8 : main]
0 views
Skip to first unread message
Marja Hölttä (Gerrit)
Jan 9, 2026, 8:20:17 AM (2 days ago) Jan 9
to Victor Gomes, chrom...@appspot.gserviceaccount.com, V8 LUCI CQ, leszek...@chromium.org, v8-re...@googlegroups.com, verwaes...@chromium.org, victorgo...@chromium.org
Attention needed from Victor Gomes
Marja Hölttä added 1 comment![]( "Open in Gerrit")
Patchset-level comments
Related details
Attention is currently required from:
- Victor Gomes
Submit Requirements:
Code-Owners
Code-Review
Review-Enforcement
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
chromeperf@appspot.gserviceaccount.com (Gerrit)
Jan 9, 2026, 9:17:03 AM (2 days ago) Jan 9
to Marja Hölttä, Victor Gomes, V8 LUCI CQ, leszek...@chromium.org, v8-re...@googlegroups.com, verwaes...@chromium.org, victorgo...@chromium.org
Attention needed from Marja Hölttä and Victor Gomes
Message from chrom...@appspot.gserviceaccount.com![]( "Open in Gerrit")
📍 Job mac-m1_mini_2020-perf/jetstream-main.crossbench complete.
See results at: https://pinpoint-dot-chromeperf.appspot.com/job/15b56e7db10000
Related details
Attention is currently required from:
- Marja Hölttä
- Victor Gomes
Submit Requirements:
Code-Owners
Code-Review
Review-Enforcement
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Victor Gomes (Gerrit)
Jan 9, 2026, 9:30:44 AM (2 days ago) Jan 9
to Marja Hölttä, chrom...@appspot.gserviceaccount.com, V8 LUCI CQ, leszek...@chromium.org, v8-re...@googlegroups.com, verwaes...@chromium.org, victorgo...@chromium.org
Attention needed from Marja Hölttä
Victor Gomes added 3 comments![]( "Open in Gerrit")
File src/maglev/maglev-ir.h
Line 8263, Patchset 2 (Latest):
kHeapObject, // Can be a HeapNumberVictor Gomes . unresolved
Nit: should we follow the same name as NodeType and call it AnyHeapObject?
File src/maglev/maglev-kna-processor.h
Line 196, Patchset 2 (Latest):
for (int i = 0; i < phi->input_count(); ++i) {
ValueNode* input = phi->input_node(i)->UnwrapIdentities();
NodeType input_type =
block->state()->known_node_aspects()->GetType(broker(), input);
if (i == 0) {
new_type = input_type;
} else {
new_type = UnionType(new_type, input_type);
}
}Victor Gomes . unresolved
I am a bit confused. My understand is that we are merging a **single** predecessor here. What I would expect is: getting the predecessor_id from block using current_block_ then you can get a the input and the type in the previous/current block:
```
ValueNode* input = phi->input_node(predecessor_id);
NodeType input_type = known_node_aspects()->GetType(broker(), input);
```
Otherwise this code will run (loop) for all predecessors.
Line 194, Patchset 2 (Latest):
for (Phi* phi : *block->state()->phis()) {Victor Gomes . unresolved
Add a little comment before explaining what this is doing.
Related details
Attention is currently required from:
- Marja Hölttä
Submit Requirements:
Code-Owners
Code-Review
No-Unresolved-Comments
Review-Enforcement
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Reply all
Reply to author
Forward
0 new messages