1271829 - Compile nss with SSE2 optimizations

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Ralph Giles (:rillian)]

We've declared we're dropping support for non-SSE x86 cpus as of Firefox 49. Update the build scripts to remove this restriction.

Comment 1

[Ralph Giles (:rillian)]

Attachment: Use SSE instructions on win32 nss

Comment 2

[Ralph Giles (:rillian)]

It looks like the clang issue in https://llvm.org/bugs/show_bug.cgi?id=24335 was specific to /arch:IA32, and in any case it was fixed 6 months ago, so it should be safe to remove the conditional.

Comment 3

[Mike Hommey [:glandium]]

Comment on attachment 8751007 [details] [diff] [review]
Use SSE instructions on win32 nss

Review of attachment 8751007 [details] [diff] [review]:
-----------------------------------------------------------------

What Firefox does shouldn't affect the defaults for NSS proper. We should pass pass our -arch parameter down from config/external/nss/Makefile.in, through XCFLAGS, until Firefox builds NSS with its build system (bug 1237872)

Comment 4

[Ralph Giles (:rillian)]

Just an update: I've mostly traced through the XCFLAGS dependencies. It looks like that will override properly if we change the Makefile.in to merge instead of clobbering them. I'll work on a patch tomorrow.

Comment 5

[Ralph Giles (:rillian)]

Attachment: Don't clobber XCFLAGS

Comment 6

[Ralph Giles (:rillian)]

Attachment: 0003-Bug-1271829-Propagate-MSVC-arch-switch-to-nss.-r-gla.patch

More like this?

Comment 7

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9f4983dfd881

Comment 8

[Phil Ringnalda (:philor)]

Backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/cbb9d4a22c5c20703b6fedf78c7b27ae7d62d897 - on your push, you got a maxtime failure, https://treeherder.mozilla.org/logviewer.html#?job_id=28046368&repo=mozilla-inbound, and then we started getting a whole lot of failures trying to remove mangle.exe (https://treeherder.mozilla.org/logviewer.html#?job_id=28071738&repo=mozilla-inbound, https://treeherder.mozilla.org/logviewer.html#?job_id=28071928&repo=mozilla-inbound, https://treeherder.mozilla.org/logviewer.html#?job_id=28085959&repo=mozilla-inbound) and pk12util.pdb (https://treeherder.mozilla.org/logviewer.html#?job_id=28072191&repo=mozilla-inbound), and there you were down below all those, having been foolish enough to touch building NSS.

Also, you put the wrong bug number in https://hg.mozilla.org/integration/mozilla-inbound/rev/d0ab0d508a248687825497a6a50ec9a685a7a5f9, incrementing this bug's number by one, so you'll want to fix that before relanding, should it turn out that my wild guess is wrong and we just keep failing even with you backed out.

Comment 9

[Ralph Giles (:rillian)]

Huh, worked on try. See https://treeherder.mozilla.org/#/jobs?repo=try&revision=1ff8bb1e7ce8 and https://treeherder.mozilla.org/#/jobs?repo=try&revision=b346b08202b0.

Comment 10

[Ralph Giles (:rillian)]

Trying again with a clobber.

Comment 11

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/040014a28b11
https://hg.mozilla.org/integration/mozilla-inbound/rev/b0c9aee8f295

Comment 12

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/040014a28b11
https://hg.mozilla.org/mozilla-central/rev/b0c9aee8f295