Closed Bug 328319 Opened 20 years ago Closed 15 years ago

System briefly hangs when loading Digg

Categories

(Firefox :: General, defect)

Product:
Firefox
Component:
General
Version:
1.5.0.x Branch
Platform:
x86
Windows XP
Type:
defect
Priority:
Not set
Severity:
major

Tracking

()

Status:
RESOLVED INCOMPLETE

People

(Reporter: kevin.breit, Unassigned)

References

(
URL
)

Details

(Keywords: hang)

User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322; .NET CLR 1.0.3705; InfoPath.1) Build Identifier: Whenever I load digg on my Ubuntu Linux box with Firefox 1.5.0.1, my system hangs for a second or two while the page is rendered. This is on a 900mhz computer with 512MB of RAM. Reproducible: Always
I can confirm this. Everytime I go to digg.com, my system hangs for a couple of seconds during page rendering... I tried with -safe-mode & with clean profile and it's the same. Currently, I am using: Build identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.1) Gecko/2006120814 Firefox/2.0.0.1 However, this used to happen with 1.x as well. What's different with 2.x is that now when digg.com /is/ loaded, firefox will behave very slow to requests and slow down the rest of the system from time to time. Though this might be because of changes to the site's code... I am using: model name : Intel(R) Celeron(R) CPU 2.80GHz cpu MHz : 2800.573 cache size : 128 KB OS: Ubuntu 6.06 (not using its packaging of Firefox) PS to bug reporter: please change the "OS" section to Linux. thanks
tried with latest nightlies of 2.0.2pre and 3 (of course with brand new profiles on both); both have the same problem...
(RAM: 1GB) (pls change OS to Linux)
Another confirm; Ubuntu Edgy version of Firefox 2.0 on Dual Athlon with 512MB RAM. It's not jsut the load, but also a redraw is taking a few seconds with the window being entirely unresponsive to interaction during that time.
(In reply to comment #4) > but also a redraw is taking a few seconds with the > window being entirely unresponsive to interaction during that time. > same for me...
booted to Windows and there were no problems there. This definitely is a Linux problem.
I've got 3 computers running Ubuntu Edgy with FF 2.0 as well. Digg is slow especially when expanding the hidden comments that have been dugg down. On windows, its a very snappy, smooth scroll.
I see this on Windows XP with most Digg stories that have a decent amount of comments. Such as: http://digg.com/tech_news/Kevin_Rose_dressed_up_as_Spock_Batboy_on_G4
I would like to add that I've experimented with this a lot and it appears to be the combination of visiting a site that uses transparent PNGs stacked on top of each other with CSS while not have 3D graphics card acceleration. Many Linux users do not have 3D graphics card acceleration due to the fact that closed-source graphics drivers are not installed by default (and the open-source graphics drivers capable of 3D acceleration are either alpha quality or do not support all cards). The point is, this bug is not specific to only Linux users. However, Linux users are much more likely to experience the problem due to the current graphics drivers situation. Basically the work of interpreting two or more stacked alpha channels is being pushed onto the processor rather than the graphics card, essentially maxing the proc out and bringing the entire system to a halt. I've experienced a 100% CPU usage just scrolling up and down on http://www.linuxactionshow.com . Proposed solutions: A) have an option within about:config to toggle off the display of transparent PNGs (this is only a temporary fix for those who even know what "about:config" is). An extension could be developed as well, but again, this is a poor solution. Not everyone knows what an extension is, nor do they actually know why web pages appear to be slow after installing Linux on their computer. B) rethink the way Fx renders stacked transparent PNGs C) detect when a user does not have 3D acceleration. If it is determined that no acceleration is present, then handle transparent PNGs differently Obviously (C) looks like the best choice. This is a VERY serious bug, and as such should have top priority. Fx cannot leave Linux users in the dust. This includes future One Laptop Per Child users, who will have no 3D acceleration whatsoever. test pages: http://www.digg.com/submit (continue the submission process to the point where you need to enter a description of the story. CPU slowdown ensues) http://www.linuxactionshow.com
Hi Brent, That is interesting - however I'm confused how this applies to my system; For reference I'm running Ubuntu/Edgy with a Radeon 7200 on a Dual Athlon MP system (Tyan S2460). Now the 3D driver does work; however it isn't exactly a new card. However, the 3D driver on this occasionally dies badly - and I've never had such a crash through Firefox; which makes me think that Firefox (fortunately) isn't even trying to use 3D drivers. Dave
http://digg.com/apple/Apple_Announces_iPhone_2 (warning, 1,459 comments!) Here is a better example of the problem I have. The browser will become unresponsive for a while and then come back. When it does, I have no problems scrolling. On http://www.linuxactionshow.com/ , I get slow scrolling, due to the stacked transparent pngs. I think the total hanging while loading many-commented digg stories, and the slow scrolling w/ transparent pngs, are different bugs.
Now I wouldn't immediately point this at 3D acceleration issues. I encounter this on my Thinkpad r50p with the fglrx proprietary drivers installed and functioning.
Dr. Gilbert: I am not too familiar with the capabilities/performance of fglrx in terms of how it renders multiple stacked alpha channels, but perhaps my calling it "3D acceleration" is conflating 3D graphics with alpha transparency. Maybe the problem is not with the lack of 3D acceleration per se, but rather most systems without functioning 3D graphics tend to also have issues with rendering alpha channels. Maybe it is the case that the fglrx driver handles 3D fine but chokes on alpha rendering. For what it's worth, testing with both Windows XP and many distributions of Linux, four different computers, all with nvidia graphics cards, in all cases the problem was reproducible and installing the proprietary "nvidia" driver rather than the open-source "nv" driver fixed the problem immediately.
http://digg.com/apple/Apple_Announces_iPhone_2 was a torture, at times it hanged the computer... I can confirm the slow scrolling with http://www.linuxactionshow.com/ as well. as per 3D, I believe I have 3D working on an Ati Mobility Radeon 9000 with the open source ati drivers. But that whole 3D thing is very confusing to me so I'm not sure...
...and now I just tested a machine with the open-source "radeon" driver. Fx renders transparent PNGs fine with "radeon", but completely chokes with the open-source "ati" driver. I haven't tested out fglrx, however. Just to note: the goal of "radeon" is to achieve 3D acceleration while "ati" is 2D only.
towsuno2003@gmail.com: Are you using the "radeon" driver? I confirm slow-ish scrolling as well, but it is categorically different from the slowdown I get when using "ati". I believe the slight slowdown you may get from visiting linuxactionshow.com with the "radeon" driver can be attributed to it's limited functionality--it is very early in development.
I just tested http://digg.com/apple/Apple_Announces_iPhone_2 on a Linux machine using proprietary "nvidia" driver and confirm slowdown of a different type. I agree with pile0nades that there seems to be two separate issues here. The digg page with 1400+ comments did not ever make Fx unresponsive on my system--it just sucked up a lot of system resources. This does seem halfway reasonable given that there are a whole lot of tiny avatars to render. I think this kind of problem affects all users of Fx on all platforms. I also think that these cases are somewhat more rare. But the problem with rendering a simple website that makes use of transparent PNGs is absolutely *not* reasonable. I'm talking about the kind of problem you get when visiting linuxactionshow.com, for example. It's not reasonable because it is a very simple website design, and it is also very common. Incidentally, digg.com also makes use of transparent PNGs in a similar way. [I believe this is unrelated to the problem of visiting a digg page with 1400+ comments]
(In reply to comment #16) > towsuno2003@gmail.com: > > Are you using the "radeon" driver? I'm using "ati" in xorg.conf... "radeon" results in random crashes...
PS. Sorry for the bug spam.. I just found a bug report at Ubuntu tracker as well: https://launchpad.net/firefox/+bug/81858 (I'm using upstream product)
So, I've got the 'ati' driver in the config, but reading the man page for it, it looks like it loads one of radeon/r128/etc depending on what you have - so I believe it's running Radeon. Just watching on top, the linuxactionshow.com page scrolling up and down shows most of the CPU is in Xorg; for the iPhone digg page I see it at ~100% on Xorg but then I see it bouncing to 100% on the firefox process for a while and back to Xorg. What's the right way to profile this - I guess we could attack Xorg/firefox with gprof or oprofile; but is there anything higher level to see what is going on in firefox or what primitives the X server is taking its time on?
Dr. Gilbert: I would hazard a guess that you are actually using the ati driver. How do 3D applications run? Have you tried running any 3D apps? I know on Debian you have to explicitly change your xorg.conf to use the radeon driver or else it will use the ati driver instead. On my ati machine, I'm using the radeon driver, which fixes the problem perfectly. When I switch back to the "ati" driver, it gets dog slow.
Just installed the latest ATI fglrx driver, direct rendering is on, all is well, but google spreadsheets still crawls compared to performance on Windows. It's not totally debilitating, but noticeable. On a side note, http://www.linuxactionshow.com/ suffers only a smidgen of lag. While traversing a google spreadsheet: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 4388 root 15 0 84264 56m 7388 S 52.6 2.8 0:49.68 Xorg 5506 subwolf 16 0 176m 77m 23m R 46.1 3.8 0:38.06 swiftfox-bin Hardware: Athlon XP 3000+, 2Gb DDR400, Radeon 9600 Clone (9650) Software: Kubuntu Edgy 6.10 Edgy, SwiftFox V2.0.0.1-AthlonXP
There may well be issues with 3d acceleration, but the issue I'm experiencing with Google Spreadhseets and Ubuntu, which is at least in symptoms identical to the issues experienced by others above, is categorically and provably not a 3d acceleration issue. Under Ubuntu, with unknown 3D acceleration support, Google Spreadsheets crawls horribly. However, if I start up VMWare Server and boot Windows 2000, Firefox and Google Spreadsheets inside that VMWare (which has no 3D acceleration support, at least in my version) is fast as you like. Far, far faster than using Firefox/Google Spreadsheets natively in X. This may well be a separate issue from that with digg.com, of course. It also may be the case that, if others are not experiencing such a slowdown with 3D acceleration enabled, it may be that the 3D acceleration is mitigating the effect of a bug; for example, if something is causing excessive overdraw, the 3D acceleration may render the bug unnoticable, and this is masking the underlying problem. I'm behind the times... I had no idea hardware compositing even existed under Linux or Windows... Regards, -- C.
> with Google Spreadhseets and Ubuntu if you open a new report on that, please let us know :)
> if you open a new report on that, please let us know :) No need, as it seems there already is one: https://bugzilla.mozilla.org/show_bug.cgi?id=346382 ...which refers to this: http://www.suseforums.net/index.php?s=8497382120bc18e7f628ad55b18babcf&showtopic=29910 wherein the commenter already tried the precise diagnostic procedure I did. That's what I get for skimping on my research, apologies for the duplication. -- C.
http://digg.com/tech_deals/The_Moron_Who_Sued_Archive_org_Intends_to_Sue_Us_Next hangs the computer, let alone the browser... probably due to too many comments?
http://digg.com/tech_deals/The_Moron_Who_Sued_Archive_org_Intends_to_Sue_Us_Next does not kill my browser, does make it think hard though. See comment #22 for system specs.
got the same problem on a mac (os X 10.3) using: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8.1.3) Gecko/2007030919 Firefox/2.0.0.3 system slows down for a few seconds, but then everything works well again.
Hi, I am using Ubuntu 7.04 with Firefox from Mozilla site, and my graphic card is ATI 9250. I am having the same issue on digg.com, I found a workarround that may help some people. here is what I did. I have installed Adblock Plus extension for Firefox then, I visited digg.com and right click on comment picture and block it using Adblock Plus (It's a PNG picture which may be in blue or gray according depending on the rating the person has recieved). or simply after installing Adblock Plus, add this two filters to the list of Adblock Plus: http://www.digg.com/img/comment-1.png http://www.digg.com/img/comment-2.png I hope that was helpful
I had to block all the images and the speed became significantly better. I tried png, gif, and jpg each one by one but without blocking all, it didn't work. of course, this doesn't solve the problem... The filters I used with adblock plus were: http://digg.com/img/*.png http://digg.com/img/*.gif http://digg.com/img/*.jpg is anyone caring for this bug?
http://digg.com/tech_news/Digg_This:_09-f9-11-02-9d-74-e3-5b-d8-41-56-c5-63-56-88-c0 Here's a better story to test with. Over 1500 comments and almost 40,000 diggs. I'm thinking this is probably an HTML parser bug. With huge digg pages it's simply overloaded and locks up the UI until it's done.
(In reply to comment #31) [snip] > I'm thinking this is probably an HTML parser bug. With huge digg pages it's > simply overloaded and locks up the UI until it's done. > Then there are two separate bugs in here - otherwise why would blocking all the images from the site work (as in comments #29 & #30?
This bug was reported on Firefox 2.x or older, which is no longer supported and will not be receiving any more updates. I strongly suggest that you update to Firefox 3.6.3 or later, update your plugins (flash, adobe, etc.), and retest in a new profile. If you still see the issue with the updated Firefox, please post here. Otherwise, please close as RESOLVED > WORKSFORME http://www.mozilla.com http://support.mozilla.com/kb/Managing+profiles http://support.mozilla.com/kb/Safe+mode
Severity: normal → major
Keywords: hang
Version: unspecified → 1.5.0.x Branch
See Also: → https://launchpad.net/bugs/81858
No reply, INCOMPLETE. Please retest with Firefox 3.6.x or later and a new profile (http://support.mozilla.com/kb/Managing+profiles). If you continue to see this issue with the newest firefox and a new profile, then please comment on this bug.
Status: UNCONFIRMED → RESOLVED
Closed: 15 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.