Switch wasm sandboxing Linux x64 and Mac x64 builds from lucet to wasm2c
Categories
(Core :: Security: Process Sandboxing, enhancement, P3)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox92 | --- | fixed |
People
(Reporter: shravanrn, Assigned: shravanrn)
References
Details
Attachments
(1 file)
|
(deleted),
text/x-phabricator-request
|
Details |
Now that wasm2c has landed in the source tree, this patch switches between using Lucet as the wasm library sandboxing tool in Firerfox to wasm2c.
Lucet is currently deprecated and we currently rely on modified version of Lucet which was forked a while back.
This change means we now rely on a recent (but forked) version of wasm2c. We will be attempting to upstream all the required changes of wasm2c to remove reliance on a fork.
|
Assignee | |
Comment 1•3 years ago
|
||
|
||
Updated•3 years ago
|
|
||
Comment 3•3 years ago
|
||
Backed out changeset 93c0dc15cc3b (Bug 1720828) for causing xpcshell failures in test_hunspell.js
Backout link: https://hg.mozilla.org/integration/autoland/rev/1618b5beb0bddea1331ba022d7bc9e14ac3c517f
Push with failures, failure log.
|
||
Comment 4•3 years ago
|
||
This bug caused an improvement but it was backed out.
== Change summary for alert #30757 (as of Tue, 03 Aug 2021 09:23:00 GMT) ==
Regressions:
| Ratio | Suite | Test | Platform | Options | Absolute values (old vs new) |
|---|---|---|---|---|---|
| 0.40% | installer size | osx-cross | 80,951,120.25 -> 81,275,174.83 |
Improvements:
| Ratio | Suite | Test | Platform | Options | Absolute values (old vs new) |
|---|---|---|---|---|---|
| 0.35% | installer size | osx-cross | 81,233,242.25 -> 80,946,299.17 |
For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=30757
|
||
Comment 6•3 years ago
|
||
| bugherder | ||
|
|
Assignee | |
Comment 7•3 years ago
|
||
Clearing the needinfo. This has been fixed and now pushed again --- will watch to make sure it sticks this time
|
||
Comment 8•3 years ago
|
||
The changes caused an improvement:
== Change summary for alert #30830 (as of Mon, 09 Aug 2021 03:16:53 GMT) ==
Improvements:
| Ratio | Suite | Test | Platform | Options | Absolute values (old vs new) |
|---|---|---|---|---|---|
| 2% | Base Content Heap Unclassified | linux1804-64-shippable-qr | 2,116,227.67 -> 2,065,195.33 | ||
| 2% | Base Content Heap Unclassified | linux1804-64-shippable-qr | 2,113,749.83 -> 2,065,067.00 |
For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=30830
|
|
Assignee | |
Comment 9•3 years ago
|
||
Yup, we are replacing compilers for wasm sandboxed code in this change. This differences are expected. I'm not sure what info is being requested by the "needinfo". Could you please clarify?
|
|
||
Comment 10•3 years ago
|
||
(In reply to Shravan Narayan from comment #9)
Yup, we are replacing compilers for wasm sandboxed code in this change. This differences are expected. I'm not sure what info is being requested by the "needinfo". Could you please clarify?
No needinfo needed. Thanks.
|
||
Updated•3 years ago
|
|
||
Comment 11•3 years ago
|
||
I'm wondering whether the upstreaming plan is still moving forward? It seems that the fork has been diverging from the upstream for quite a bit (50 commits / 3 months behind now). I saw there was a giant PR submitted two months ago, and the plan discussed there was that someone would be submitting separate PRs, but I saw no PR from that person to the upstream yet.
I'm raising this because I'm concerned about further divergence making the changes no longer mergeable and then the implementation turns unmaintained eventually similar to what happened to the RLBox Lucet fork.
Description
•