| Commit-Queue | +1 |
bool HeapSnapshotGenerator::GenerateSnapshotAfterGC() {I want to refactor away this method in a follow up. It does almost the same thing is the regular `GenerateSnapshot()`, without the GC.
(There should be not progress counter without it being passed to the generator in first place.)
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Code-Review | +1 |
bool HeapSnapshotGenerator::GenerateSnapshotAfterGC() {I want to refactor away this method in a follow up. It does almost the same thing is the regular `GenerateSnapshot()`, without the GC.
(There should be not progress counter without it being passed to the generator in first place.)
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
// Flags: --expose-gc --shared-heap --heap-snapshot-on-gc=0This flag makes us create snapshots on every GC, in the current working directory.
@mache...@chromium.org: I have flagged the test as slow already. Should we avoid it in first place?
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
// Flags: --expose-gc --shared-heap --heap-snapshot-on-gc=0This flag makes us create snapshots on every GC, in the current working directory.
@mache...@chromium.org: I have flagged the test as slow already. Should we avoid it in first place?
Sounds like something we'd rather avoid. The bots won't care much, but I assume this will negatively affect developers? Is the file .gitignored in V8? Also, would it ever clash if we run this test concurrently and both processes want to write to this file?
Would it be possible to add yet another flag, and if that is passed, write the file to /dev/null or skip writing and dump to stdout instead?
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Commit-Queue | +1 |
// Flags: --expose-gc --shared-heap --heap-snapshot-on-gc=0Michael AchenbachThis flag makes us create snapshots on every GC, in the current working directory.
@mache...@chromium.org: I have flagged the test as slow already. Should we avoid it in first place?
Sounds like something we'd rather avoid. The bots won't care much, but I assume this will negatively affect developers? Is the file .gitignored in V8? Also, would it ever clash if we run this test concurrently and both processes want to write to this file?
Would it be possible to add yet another flag, and if that is passed, write the file to /dev/null or skip writing and dump to stdout instead?
Thanks, we already had a flag for the filename, so I can just pass that to `/dev/null` and only enable on Linux.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Code-Review | +1 |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Code-Review | +1 |
if (v8_flags.heap_snapshot_on_oom && !isolate->has_active_deserializer()) {I think it would be nice in general if we would be able to take a snapshot every time we are able to GC. Are some verifications in the heap snapshot generator too strict here that we could relax to allow this again?
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Commit-Queue | +2 |
if (v8_flags.heap_snapshot_on_oom && !isolate->has_active_deserializer()) {I think it would be nice in general if we would be able to take a snapshot every time we are able to GC. Are some verifications in the heap snapshot generator too strict here that we could relax to allow this again?
That's already a TODO at the site where the deserializer is recorded. Not sure how much work it actually is. The state is certainly inconsistent though.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
[heap] Enter global safepoint scope for snapshot creation after GC
The regular snapshot creation already enters the global safepoint which
is necessary to to shared space handling.
This basically is the same as http://crrev.com/c/6994961
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |