1522803 - Update Rust toolchain to stable 1.32 for the win aarch64 build

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Comment 1

[Benjamin Bouvier [:bbouvier] (inactive)]

Attachment: bump.patch

Comment 2

[Benjamin Bouvier [:bbouvier] (inactive)]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=9e1b83974d68c107269dd8c7519748df113ae8df as a proof it works!

Comment 3

[Nathan Froyd [:froydnj]]

Comment on attachment 9039064 [details] [diff] [review]
bump.patch

Review of attachment 9039064 [details] [diff] [review]:
-----------------------------------------------------------------

Thank you!

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/1bb6959a6990
Use a stable Rust version for the aarch64 Windows repack; r=froydnj

Comment 5

[(Away)]

I was under the impression that we were gonna update to a new rustc nightly for bug 1512519, as soon as https://github.com/rust-lang/rust/pull/57645 lands. Does that mess up your plans here? (In other words did you specifically need the stable branch?)

Comment 6

[Nathan Froyd [:froydnj]]

Rust nightlies are currently unusable due to bug 1521249; bbouvier tried nightly versions in bug 1522713 comment 8.

Comment 7

[(Away)]

(In reply to Nathan Froyd [:froydnj] from comment #6)

Rust nightlies are currently unusable due to bug 1521249; bbouvier tried
nightly versions in bug 1522713 comment 8.

That bug doesn't have a comment 8... wrong bug? Thanks for the pointer though, that saves me a surprise from the next time I update my nightly.

Comment 8

[Nathan Froyd [:froydnj]]

Ah, bug 1522173 comment 8, whoops.

Comment 9

[Cristina Coroiu [:ccoroiu]]

https://hg.mozilla.org/mozilla-central/rev/1bb6959a6990

Comment 10

[Ionuț Goldan [:igoldan]]

Either this bug or bug 1522173 brought some big build times improvements on Windows platforms:

== Change summary for alert #18929 (as of Fri, 25 Jan 2019 15:09:19 GMT) ==

Improvements:

696% sccache hit rate windows2012-64 asan opt taskcluster-c5.4xlarge 0.12 -> 0.94
654% sccache hit rate windows2012-32 debug taskcluster-c5.4xlarge 0.12 -> 0.92
636% sccache hit rate windows2012-64-noopt debug taskcluster-c5.4xlarge 0.12 -> 0.87
528% sccache hit rate windows2012-64 asan fuzzing-asan taskcluster-c5.4xlarge 0.15 -> 0.92
26% build times windows2012-32 debug taskcluster-c5.4xlarge 2,577.58 -> 1,915.17
23% build times windows2012-64-noopt debug taskcluster-c5.4xlarge 2,246.23 -> 1,731.57

For up to date results, see: https://treeherder.mozilla.org/perf.html#/alerts?id=18929

Comment 11

[Benjamin Bouvier [:bbouvier] (inactive)]

Could be a mix of both:

• updating the Rust compiler might improve compilation times of Rust in general
• the Cranelift bump in bug 1522173 may have addressed the build time regression noted in bug 1506511.

Ionuț, regarding bug 1506511, did we get back to the baseline values? If so, we could mark these two bugs as the cause for resolution. (We could also do try builds including only one of the two patches to see which patch in particular solved the regression, for completeness)

Comment 12

[Ionuț Goldan [:igoldan]]

(In reply to Benjamin Bouvier [:bbouvier] from comment #11)

Could be a mix of both:

• updating the Rust compiler might improve compilation times of Rust in general
• the Cranelift bump in bug 1522173 may have addressed the build time regression noted in bug 1506511.

Ionuț, regarding bug 1506511, did we get back to the baseline values? If so, we could mark these two bugs as the cause for resolution. (We could also do try builds including only one of the two patches to see which patch in particular solved the regression, for completeness)

This is kinda tricky to confirm. There seems to be a fix for some of the build times. Still, the plain builds haven't improved at all.

Comment 13

[(Away)]

I'm going to claim bug 1522771 for the sccache and build time improvements. :-)