347938 - (burrito-overflow) Cannot eat 10 burritos in 2 hours without vomiting

Status: RESOLVED WORKSFORME

Description

[J. Paul Reed [:preed]]

Firefox cannot handle eating 10 Taco Bell Burrito Supremes without vomiting.

Everything starts out rendering correctly, but as time progresses, the rendering slows down... and then about 90 minutes in, Firefox just barfs.

I think this is a problem with the front-end to back-end migration code.

Seth, can you take a look?

Comment 1

[(not reading, please use [e-mail] instead)]

yes, matt has been able to reproduce this bug.

matt, can I get a screen shot?

Comment 2

[J. Paul Reed [:preed]]

Created attachment 232807 [details]
Pre-crash

This was a screenshot from right before it just barfed...

Comment 3

[Majken Connor [:Kensie]]

can we get some more info on this?  how many burritos does firefox handle before it barfs?

Comment 4

[J. Paul Reed [:preed]]

(In reply to comment #3)
> can we get some more info on this?  how many burritos does firefox handle
> before it barfs?

Looks like it can handle about three bites shy of seven burritos before biffing it.

Comment 5

[Phil Ringnalda (:philor)]

Did the burritos talk back before it barfed? See http://kb.mozillazine.org/Talkback or http://en.wikipedia.org/wiki/Burping

Comment 6

[(not reading, please use [e-mail] instead)]

see http://www.netdareredux.com/images/ for some photos (before, after, and during.)

see also http://www.netdareredux.com/claudius/index.html

Comment 7

[J. Paul Reed [:preed]]

(In reply to comment #5)
> Did the burritos talk back before it barfed?

Hrm...

http://talkback-public.mozilla.org/search/start.jsp?search=1&searchby=comments&match=contains&searchfor=barf&vendor=MozillaOrg&product=All&platform=All&buildid=&sdate=&stime=&edate=&etime=&sortby=bbid&rlimit=500

Looks like this is causing lots of distress.

Comment 8

[Jo Hermans]

(In reply to comment #3)
> can we get some more info on this?  how many burritos does firefox handle
> before it barfs?
> 

I wonder what does PETA (www.peta.org) have to say about this. Force-feeding a poor animal :-)

Comment 9

[Joey Minta]

Is this in the test-suite yet?  This seems like the kinda thing that should be run on a regular basis.

Comment 10

[J. Paul Reed [:preed]]

Looks like firefox will barf on 4 Burger King Whoppers in < 3 hours, too.

Upping severity.

Comment 11

[Simon Howes]

Should we add this to the Tinderbox?

Comment 12

[Robert Strong [:rstrong] (use needinfo to contact me)]

I think you are premature in setting the severity to blocker... that really won't be known for at least a few hours.

Comment 13

[Majken Connor [:Kensie]]

Firefox shouldn't be trying to force 4 of the 10 whoppers through in the first 3 hours of 8 allotted.  Clearly smarter CPU usage handling is required.  Firefox should know it's using 100% CPU and queue the rest of the whoppers rather than locking up and crashing.

Comment 14

[Simon Howes]

Which developer should we assign to fix this bug?

Comment 15

[J. Paul Reed [:preed]]

Created attachment 232817 [details]
All smiles and sunshine... (a.k.a. you got your whopper on my burrito. no, you got your burrito on my whopper). 

Another screenshot: upon launch, everything looks good...

Comment 16

[Russell Howe]

WFM two days ago with homemade burritos, so it must be something specific to Taco Bell. Unfortunately, I don't have access to such a platform for testing here in London..

Comment 17

[Simon Howes]

(In reply to comment #15)
> Created an attachment (id=232817) [edit]
> All smiles and sunshine...
> 
> Another screenshot: upon launch, everything looks good...
> 

Nice Power Book / Mac Book Pro.

Comment 18

[Phil Ringnalda (:philor)]

(In reply to comment #14)
> Which developer should we assign to fix this bug?

Unfortunately, other than bz, who wrote mozFlushType.h, we don't have a lot of developer expertise in this area, and I strongly doubt he'd touch this problem with a ten foot pole.

Comment 19

[gabe]

gives everybody antacid pill and says there problem fixed

Comment 20

[Mano (::mano, needinfo? for any questions; not reading general bugmail)]

-> Autocomplete.

Comment 21

[Mark Mentovai]

As this bug causes the loss of data that (hopefully) cannot be recovered, it should be marked security.

Comment 22

[Phil Ringnalda (:philor)]

Nobody in their right mind would verify the outcome of this bug; changing qa contact accordingly.

Comment 23

[Alex Vincent [:WeirdAl]]

Note that the digestive system has been fixed on trunk, but due to frequent and repeated API changes to nsICalorieProcessor, no patch from trunk can land on the 1.8 branch.

Comment 24

[Justin Dolske [:Dolske]]

With a reduced testcase, my system quickly dumps core. However, I don't see any recognizable symbols... Best I can make out is that the function at 0xFEEDFACE overflowed a buffer with repeated instances of 0xDEADBEEF.

Comment 25

[Mike Beltzner [:beltzner, not reading bugmail]]

--> Gecko 1.9a1, as per comment 23. I think that this might be something for the reflow branch to handle, so cc'ing dbaron ...

Comment 26

[Jeff Walden [:Waldo] (I'm baaaaaaack...)]

I think the problem is with the ingredients in the ingested foods.

We now have demonstrable success at digesting an entire 21-piece* loaf of Wonder Classic bread in an hour, which contains a full 44% of a 2000-calorie diet's daily value of fiber.  Clearly high-speed fiber is the key to success, and adding some would speed things up immensely.  Over to Server Operations Projects to handle installation...

* mscott and sspitzer feel cheated out of the 22nd slice

Comment 27

[Marria Nazif]

*** Bug 350104 has been marked as a duplicate of this bug. ***

Comment 28

[J. Paul Reed [:preed]]

As sad as I am to do it, it looks like no one has been able to eat 10 burritos in 2 hours yet.

Resolving WONTFIX.

Reopen if you think you can prove us all wrong.

Comment 29

[chris hofmann]

interesting that someone put the dogfood keyword on this bug and didn't give it to sparky to test.  his version of fx3 can easily digest 10 burritos in about 20 minutes.  

Comment 30

[chris hofmann]

sparky can you work with sspitzer to get your patch in?

Comment 31

[chris hofmann]

performance after that 10th burrito isn't really great, but still functional

http://flickr.com/photos/41711850@N00/398221766/