[SOLVED] Possibile regresion on mozilla-firefox-32.0.3, for current (slow and processor hog, on heavy load)

Darth Vader · 09-30-2014, 03:37 PM

For some mysterious reason, the latest 32.0.3 for current behave like a slowwwww processor hog, eating a core solo, then I need to reverse quickly & temporary to previous version, mozilla-firefox-32.0-i486-1.txz.

Still, I do not find any un-synchronized mirror to get it. Any of you still have this package?

LATER:

So, finally, I can confirm that in my particular hardware (Phenom x4 9950, 8G DDR2 RAM), software (Slackware 32bit current) and habits (around 10 Firefox windows, every one with around 20 tabs open), Firefox 32.0.3 behave like a snail and eat an entire core processing power, while the previous version, 32.0, behave normal.

teckk · 09-30-2014, 03:44 PM

ftp://ftp.slackware.com/pub/slackwar...0-x86_64-1.txz
ftp://ftp.slackware.com/pub/slackwar...2.0-i486-1.txz

Darth Vader · 09-30-2014, 03:47 PM

Quote:

Originally Posted by teckk

ftp://ftp.slackware.com/pub/slackwar...0-x86_64-1.txz
ftp://ftp.slackware.com/pub/slackwar...2.0-i486-1.txz

Well, there, in ftp://ftp.slackware.com, is available, of course, the latest version, 32.0.3, BUT I need the previous build on current, 32.0. That from Thu Sep 4 19:43:25 UTC 2014

moisespedro · 09-30-2014, 04:35 PM

Can't you build it yourself?

rkelsen · 09-30-2014, 05:56 PM

Did you submit a bug report to PV with your findings? If not, why not?

dugan · 09-30-2014, 06:20 PM

I agree with rkelsen. You've tested the package (thank you) and your test has found a show-stopping regression (big thank-you). Please report your findings to Pat, so that he can fix the problem (possibly by reverting the slackware-current Firefox package to 32.0).

chrisretusn · 09-30-2014, 07:29 PM

Firefox 32.0.3 is working just fine for me on Slackware64-current. Perhaps it's a 32-bit issue.

Why not download the Firefox source tarball from ftp://ftp.mozilla.org/pub/mozilla.org/firefox/releases/ and compile it using the SlackBuild from the Slackware source. You will of course need the entire contents of slackware-current/source/xap/mozilla-firefox/ minus the 32.0.3 tarball and signature file.

Code:

firefox-32.0.3.source.tar.bz2
firefox-32.0.3.source.tar.bz2.asc
firefox.moz_plugin_path.diff.gz
gold/
mimeTypes.rdf.gz
mozilla-firefox-mimeTypes-fix.diff.gz
mozilla-firefox.SlackBuild*
mozilla-firefox.desktop
slack-desc

MadMaverick9 · 09-30-2014, 10:23 PM

http://www.mmnt.net/db/0/11/ftp.pies...slackware/xap/
http://www.mmnt.net/db/0/12/ftp.pies...slackware/xap/

And you can and should verify that this is an original build, by doing a "gpg --verify mozilla-firefox-32.0-i486-1.txz.asc".

Code:

bash $ gpg --verify mozilla-firefox-32.0-i486-1.txz.asc 
gpg: Signature made 2014-09-05T02:51:00 ICT using DSA key ID 40102233
gpg: Good signature from "Slackware Linux Project <security@slackware.com>"

Since this is not an official mirror as far as I can tell, I would advise not to skip the verification.

ftp://ftp.pieskovisko.sk/pub/slackwa...slackware/xap/

Darth Vader · 10-01-2014, 02:13 AM

Quote:

Originally Posted by MadMaverick9

http://www.mmnt.net/db/0/11/ftp.pies...slackware/xap/
http://www.mmnt.net/db/0/12/ftp.pies...slackware/xap/

And you can and should verify that this is an original build, by doing a "gpg --verify mozilla-firefox-32.0-i486-1.txz.asc".

Code:

bash $ gpg --verify mozilla-firefox-32.0-i486-1.txz.asc 
gpg: Signature made 2014-09-05T02:51:00 ICT using DSA key ID 40102233
gpg: Good signature from "Slackware Linux Project <security@slackware.com>"

Since this is not an official mirror as far as I can tell, I would advise not to skip the verification.

ftp://ftp.pieskovisko.sk/pub/slackwa...slackware/xap/

Big Thanks!

Then reverted the Firefox, and everything is OK.

So, finally, I can confirm that in my particular hardware (Phenom x4 9950, 8G DDR2 RAM), software (Slackware 32bit current) and habits (around 10 Firefox windows, every one with around 20 tabs open), Firefox 32.0.3 behave like a snail and eat an entire core processing power, while the previous version, 32.0, behave normal.

Darth Vader · 10-01-2014, 02:19 AM

Quote:

Originally Posted by dugan

I agree with rkelsen. You've tested the package (thank you) and your test has found a show-stopping regression (big thank-you). Please report your findings to Pat, so that he can fix the problem (possibly by reverting the slackware-current Firefox package to 32.0).

I can consider this thread as a bug report, knowing that P.V. and his merry guys lurks there...

ruario · 10-01-2014, 05:06 AM

You really should not downgrade and I think it highly unlikely that Pat will downgrade the official package. Firefox 32.0.3 was released specifically to fix the BERserk SSL Flaw (CVE-2014-1568, a vulnerability that could enable a digital signature forgery attack).

Intel Security General Manager Mike Fey has commented

Quote:

Given that certificates can be forged for any domain, this issue raises serious concerns around integrity and confidentiality as we traverse what we perceive to be secure websites

As you can see this is a really nasty issue and if it wasn't for Shellshock I suspect we would have had a lot more discussion about this here.

P.S. I should add that almost every release of all major browsers includes one or more security updates. It is therefore not recommended that you run older browsers

1337_powerslacker · 10-01-2014, 09:01 AM

Here's another idea about regressing to 32.0: There's a way to repackage the binary tarball provided by Mozilla, and I have been using it for months now with great success.

Note: Make a mozilla-firefox/ sub-directory and make sure you specify it when downloading before following these steps.

First, download the 32.0 tarball: http://ftp.mozilla.org/pub/mozilla.o...releases/32.0/

Then, go download the SlackBuild for Firefox from here: http://slackware.osuosl.org/slackwar...zilla-firefox/

Select 'Save File As' or 'Save link as' when right-clicking each file. Save in the directory you created.

Modify the version number in the SlackBuild, and make it executable, like so:

Code:

chmod +x mozilla-firefox.SlackBuild

You may want to change the makepkg line in the SlackBuild from '.tgz' to '.txz', but that's just my personal preference.

Hope this helps.

Regards,

Matt

gus3 · 10-01-2014, 11:14 AM

So it comes down to security vs. usability? Having worked in an InfoSec shop, I know the two are not mutually incompatible.

Here's the spec on my netbook:

ASUS EeePC 900
512M RAM
900MHz Celeron
External USB drive as boot

Launching Firefox with just one tab (email login) takes well over a minute, with CPU pegged at 100%, *even after the page is finished loading.* Keyboard input gets buffered by X, so I can have my entire username and passsword typed before any of the input shows up on-screen. And then it takes almost 3 minutes between hitting Enter and having my Inbox show on-screen.

During this slow-down, there is little to no paging I/O. The system isn't thrashing (1G swap, 995M free as I type this). It's just running the CPU entirely too much.

Something has gone seriously wrong. The previous version didn't behave at all like this.

bassmadrigal · 10-01-2014, 01:08 PM

Have the people who have been noticing the bug tried building it directly from Mozilla's tarball? It could help determine if it is just an issue with the slack pacakge, or something with Mozilla itself. That would dictate who needs to work on getting it resolved. If it's Mozilla, they would have to release an update before anything can be fixed, if it's Pat, he would just need to repackage it and then provide the new package.

Darth Vader · 10-01-2014, 03:48 PM

Quote:

Originally Posted by bassmadrigal

Have the people who have been noticing the bug tried building it directly from Mozilla's tarball? It could help determine if it is just an issue with the slack pacakge, or something with Mozilla itself. That would dictate who needs to work on getting it resolved. If it's Mozilla, they would have to release an update before anything can be fixed, if it's Pat, he would just need to repackage it and then provide the new package.

OK, I tested the binary shipped by Mozilla, and surprise! That build works OK, maybe slightly slower, but barely noticeable. Then, looks that is something wrong in the Slackware build...

I have no time to try a local build, but maybe is something wrong with the default toolkit, or maybe even Firefox do not like the default optimizations of our GCC? I noticed also, maybe after so long time, that the Mozilla build is i686, not i586... Maybe Firefox want a i686 build?