firefox-41.0.2

[Fernando de Oliveira]

​https://ftp.mozilla.org/pub/firefox/releases/41.0.2/source/firefox-41.0.1.source.tar.xz

​https://ftp.mozilla.org/pub/firefox/releases/41.0.2/MD5SUMS

d71f0f761c51aeae03e367001afc9f8d

​https://ftp.mozilla.org/pub/firefox/releases/41.0.2/MD5SUMS.asc

Now 41.0.2, fixes Mozilla Foundation Security Advisory 2015-115

"Security researcher Abdulrahman Alqabandi reported that the fetch() API did not correctly implement the Cross-Origin Resource Sharing (CORS) specification, allowing a malicious page to access private data from other origins. Mozilla developer Ben Kelly independently reported the same issue. CVE-2015-7184"

Following 41.0.1 items NOT changed. In practice, that CVE is still marked as "reserved"

​https://www.mozilla.org/en-US/security/known-vulnerabilities/firefox/#firefox41.0.1

Security Advisories for Firefox

Sorted by Impact key level, not original time stamp order.

None

​https://www.mozilla.org/en-US/firefox/41.0.1/releasenotes/

Release Notes

Firefox Notes
Version 41.0.1, first offered to Release channel users on September 30, 2015
View notes for:

    Desktop
    Android

What’s New

    New

  • Reference: Release notes for Firefox 41.0

    Fixed

  • Fix a startup crash related to Yandex toolbar and Adblock Plus
    (1209124)
  • Fix potential hangs with Flash plugins (1185639)
  • Fix a regression in the bookmark creation (1206376)
  • Fix a startup crash with some Intel Media Accelerator 3150 graphic
    cards (1207665)
  • Fix a graphic crash, occurring occasionally on Facebook (1178601)

[ken@…]

Changing priority to 'high' since this is now a vulnerability fix.

[ken@…]

Fixed at r16533, I do not understand why my -j1 elapsed time was so much less than for 41.0 (on the same machine, both running 7.8).

[Fernando de Oliveira]

If what I write in the following is true, just add "using parallelism=N", where N is found below. No need to rebuild, for statistics.

In the past it happened to me, and after a couple of releases, I discovered that it was not honoring mozconfig and automatically setting the flag: MOZ_MAKE_FLAGS.

I find its value with

$ grep 'make -j' firefox-41.0-simulation-2015.09.30-19h08m32s.log
make -j4 -C /tmp/mozilla-release/firefox-build-dir
make -j4 -C /tmp/mozilla-release/firefox-build-dir install

That time, I discovered that no matter what I had in mozconfig, it gave never -j1, then introduced:

MOZ_MAKE_FLAGS='-jN' make -f client.mk

MOZ_MAKE_FLAGS='-jN' make -f client.mk install INSTALL_SDK=

where N is an integer, 1 ≤ N ≤ threads/cpu.

Or define globally MOZ_MAKE_FLAGS for the build, with export ... unset may work.

If it is the same problem, hopefully, flag is still named MOZ_MAKE_FLAGS and solution above may work.

To be sure, no need for a complete build. Sufficient start the build until four or five lines after making splines.

[bdubbs@…]

Ken, Are you holding off closing this ticket for timing issues? I have this:

Laptop, parallelism=8
1133.2 Elapsed Time -  firefox-41.0.2.source
SBU=11.153

Development system, parallelism=12
972.6 Elapsed Time -  firefox-41.0.2.source
SBU=10.458

I have some changes I want to make to FF and friends for the new freetype. I also took #6998 so I can do all at the same time. If you want, I can close this when I'm done.

[ken@…]

Didn't realise I'd forgotten to close it! I'm happy with the current SBU.

Interesting that going from 8 to 12 CPUs makes an insignificant difference to the SBU. I know that when I build with -j 4 and it eventually starts to do the linking everything grinds to a halt (in one case, literally - my desktop on a 4GB machine was unusable for a couple of minutes, and the clock didn't get updated during that), but perhaps the current code is not very parallelizable.