firefox-44.0.1

[Fernando de Oliveira]

Hopefully, again, problem with system cairo fixed ???)

​https://ftp.mozilla.org/pub/firefox/releases/44.0.1/source/firefox-44.0.1.source.tar.xz

​https://ftp.mozilla.org/pub/firefox/releases/44.0.1/SHA512SUMS

1e6c2cc4525bfb2362bac1338b721e1021b5f6f794a9db1bf794bfd26d62a7f5ea2887af569e04c9ecf81c9da6fc9f52ba14d238de7c8274e5fb8d1f729b2e2f source/firefox-44.0.1.source.tar.xz

​https://ftp.mozilla.org/pub/firefox/releases/44.0.1/SHA512SUMS.asc

​https://www.mozilla.org/en-US/security/known-vulnerabilities/firefox/

Fixed in Firefox 44.0.1

None (at least not available yet)

​https://www.mozilla.org/en-US/firefox/44.0.1/releasenotes/

Release Notes

44.0.1
Firefox Release

February 8, 2016

Version 44.0.1, first offered to Release channel users on February 8,
2016

Fixed

  • Fix issue which could lead to the removal of stored passwords under
    certain circumstances (1242176)
  • Allows spaces in cookie names (1244505)
  • Fix WebSockets when used in a Service Worker context (1243942)
  • Disable opus/vorbis audio with H.264 (1245696)
  • Require NSS 3.21 (1244069)
  • Ship the Gecko SDK (1243740)
  • Fix for graphics startup crash (GNU/Linux) (1222171)
  • Fix a crash in cache networking (1244076)

Congrats! You’re using the latest version of Firefox

[Fernando de Oliveira]

Fixed at r16928.

[Fernando de Oliveira]

Added system-cairo back. Please, test.

Ken is the best tester.

Thanks

[ken@…]

I am dubious that this fixes the same problem - they talk of a crash at startup, but my memory says that crashes happened at random. I was originally going to continue using system cairo except on the BLFS-7.8 release, but I think I stopped doing that in December because I was fed up with the crashes.

Will give it a go on my second test box, but not sure how long I will be able test it. Too much else going on (I want to do a fresh build with unrelated things, and at some point I will have to power down everything to try a new KVM switch).

[ken@…]

Replying to ken@…:

I am dubious that this fixes the same problem - they talk of a crash at startup, but my memory says that crashes happened at random. I was originally going to continue using system cairo except on the BLFS-7.8 release, but I think I stopped doing that in December because I was fed up with the crashes.

Will give it a go on my second test box, but not sure how long I will be able test it.

Well, it has been up for 5 hours, and continually streaming youtube for almost the whole time (sometimes in a tab, at other times full screen), and I have had up to 15 tabs open, of which one was slashdot, on autorefresh : I suspect that one of the crashes I saw in the past, while watching youtube, was because of a slashdot update in the background.

So, this appears to do the job, I think I might put it on a gtk+-2 box (I was not going to bother with 44.0.1).

[Fernando de Oliveira]

Thanks for testing. Here it is not crashing, either. If everything goes well, we can change the comments in mozconfig.

I was suspicious that fix would solve our problems, after the comments (c1-c5), that they had broken it:

And this is a regression from bug 1209812, which removed gfxImageFormat::A1, which causes gfxImageFormatToCairoFormat to no longer work because gfxImageFormat and _cairo_format no longer match up.

​https://bugzilla.mozilla.org/show_bug.cgi?id=1222171#c1

Comment 1 [[https://secure.gravatar.com]] Markus Stange [:mstange] (away
until Feb 15th) 2015-11-05 13:20:51 PST

And this is a regression from bug 1209812, which removed
gfxImageFormat::A1, which causes gfxImageFormatToCairoFormat to no
longer work because gfxImageFormat and _cairo_format no longer match up.

Comment 2 Nicholas Nethercote [:njn] 2015-11-05 13:37:50 PST

This one's my fault.

In part 1 of bug 1209812 I made the conversions between gfxImageFormat
and cairo_format_t more explicit -- they now use a macro instead of just
casts, which is good.

But the macros still assume that the values in gfxImageFormat line up
with the equivalent values in cairo_format_t. And then in part 2 I
removed gfxImageFormat::A1, which breaks the assumption.

Should be an easy fix -- just make the values line up again by making
them explicit. Patch coming up soon.

Comment 3 Nicholas Nethercote [:njn] 2015-11-05 13:44:42 PST

Created attachment 8683873 [details] [diff] [review]
Re-establish equivalence between gfxImageFormat and cairo_format_t

Markus, can you test this? Thank you.

Comment 4 [[https://secure.gravatar.com]] Markus Stange [:mstange] (away
until Feb 15th) 2015-11-05 14:09:15 PST

Comment on attachment 8683873 [details] [diff] [review]
Re-establish equivalence between gfxImageFormat and cairo_format_t

Review of attachment 8683873 [details] [diff] [review]:
-----------------------------------------------------------------

Sure.

Comment 5 [[https://secure.gravatar.com]] Markus Stange [:mstange] (away
until Feb 15th) 2015-11-05 14:21:26 PST

It works!

[Fernando de Oliveira]

Bad news. Writing this after a crash. Will revert to not use system cairo :-(