Can you give me your latest crash report IDs?

1. Enter about:crashes in the Firefox address bar and press Enter. A Submitted Crash Reports list will appear, similar to the one shown below.
2. Copy the 5 most recent Report IDs that start with bp- and paste them into your response here.

Hi,

"here" is the answer :

   bp-f9be41ea-c1e5-49bb-bb6c-49bc02140410
   bp-d5db916e-6238-44ec-8ee1-22b902140409
   bp-f81626fe-8309-4d88-a715-24d6a2140409
   bp-f058f58b-61ad-42a4-b973-850832140408
   bp-acb629c1-51ea-488a-801f-db7a02140408

Sometimes when there are several crashes consecutively I don't send comments

Best regards,

Trebly (Bernard TREMBLAY)

________________________________________

About difficulties to answer : I had problems to answer for two reasons

• Because I receive a notification by mail,
  • which content don't allows easily to answer from mail
  • the mail doesn't contents a link to the original thread
• Because there is no "unique reference of a thread" easily to find with search function

So an answer message, sent to "no-reply" had the interest to have been written. The process creates a misunderstanding ( which first appear not clear to me).

I receive the answer by mail to a message send on https://support.mozilla.org/... where ?

The mail box sender is : no-reply@mozilla.org -> normally this means that if you answer to this address your mail will be sent to garbage.

• Into the message there is
  • no address nor reference to the original message,
  • nor an url to find back the discussion thread

• You tell me :
  • Can you give me your latest crash report IDs?
  • Enter about:crashes in the Firefox address bar and press Enter. A Submitted Crash Reports list will appear, similar to the one shown below. Copy the 5 most recent Report IDs that start with bp- and paste them into your response here.
• But what means "here" into the mail context

I could not find back quickly the thread on support site :

• url inside : https://support.mozilla.org/questions?filter=unsolved not found the subject :
  • not found search box,
  • not found into my messages,
  • nor into https://support.mozilla.org/fr/questions/firefox?owner=all

Finally I could reach the thread :

• I had to go to my profile
• Found my message
• Click on the link (title of message)

Before I tried with "advanced search" without success (could be an option of "search").

The main reason was simply that I post my message on /en-US/ I often don't take care of the language... But the searches seems to be performed in one lonely language, not sure. With the full title copied from the mail there is no answer on both fr and en-US, why ?

Note : I recognize well "tikiwiki" as tools of this site on which CMS I am working right now. May be the search problem comes from the structure of the site(s) and search option on a multi-tiki ? This doesn't changes anything to the problem of content of the notification mail and the utility of a multi-sites unique reference of threads

Hi thank you for the feedback on the usability on the site. I have forwarded the feedback to the developers of the usability of the support forum. Thank you, we really appreciate the feedback and hope to add it to the improvements of the site.

The crash signatures:

std::vector<CanvasRenderingContext2D*>::iterator iter = std::find(DemotableContexts().begin(), DemotableContexts().end(), context);,

Looks like some of the comments on the other sites are experiencing this on a new tab or in a File Maker. I do not know much, but does this also occur on a fresh reinstall of Firefox 28? Esr is a different branch and the most current version is version 24. It is meant for enterprise environments and the next update will come in version 30 or 31 esr.

Could you please also give some insight on what you are doing right before the crash? A rendering of a 2d canvas is crashing but I do not know where to look for a regression unless I know if it also crashes in a fresh install of version 28. Please make a back up of the profile as well before doing this. You can launch Firefox in a new profile with a fresh installation with the cmd code in this article. Profile Manager - Create, remove or switch Firefox profiles

Thank you, looking forward to your reply.

Hi Trebly, Kadir followed up with me and there is a bug addressing this in the forum :-)

• Bug 994367 - Email to original poster is misleading in that it contains no obvious link back to the support forum.

Hi,

I had a system crash. This explain how long time had spread before this answer.

Thanks, using the function name "CanvasRenderingContext2D" (graphic dimensioning images which manipulates objects) I made a research into *.xpi.

I found two extensions using particularly this function (that I uses too in some developments). There are two extensions of the same author Piro ( piro_or):

• TreeStyleTab
• Multiple TabHandler

They are very well designed and useful.

about :

• I got before known to bug leading to crashes and the way to reproduce
• I tried to reach the author but the link provided by Mozilla (try to find where) is to a japanese site not updated (http://piro.sakura.ne.jp/latest/blosxom/mozilla/extensions/ 404 error), I finally today found Piro - TreStyleTab and Piro MultipleTabHandler

So now I will use time to look at. Before this I got a third crash with TreeStyleTab by redim width of the Tree frame (quite surely using CanvasRenderingContext2D).

Done :

• inactivate

Just after I got a system crash, and I could not work till today.

Result :

I had no more crashes (but not sufficient work to tell that they are the cause of the crashes). Previously I had inactivated them separately and got no result, but know they are both inactivated. It is not pleasant to work without but it is for test. I will report my suspicions on the site that I found.

My suspicions are that, during the load of a new page (most of crashes) there are reorganisations of both MultipleTabHandler and TreStyleTab tabs references display. Probably during these operations objects are freed but reused by another process which generates the crashes (the processes are complex because of the use of hover functions which can generated concurrent processes).

I will send complementary informations :

• if a get new crashes as the extensions are in-activated
• If I get informations from Piro site (forum or bugs handling)

It seem well that the problem is unlocked by your information, that I was not able to find, : "CanvasRenderingContext2D" is directly involved into these crashes.

Thanks

Trebly

Hi,

I got again an error from another extension : TabGroupManager. The crash can occur simply when a new tab is opened. The test that I have done :

• each time that I was trying to open http://www.jetbrains.com/phpstorm/web.../scope.html from another page of the summary I was getting a crash. Then I test this :
• open an empty tab
• paste the url

This functions normally.

So it is the creation of a new tab, which changes the horizontal bar of tabs while loading and filling the title which crashes FF.

I was getting the same with TreeStyle which opens a new tab into the tree (vertically), is it the same bug ? I have found right now the corresponding xpi for searching which function is used

Trebly

Did you try Firefox Safe Mode?

You can try to disable hardware acceleration in Firefox (you need to close and restart Firefox).

• Tools > Options > Advanced > General > Browsing: "Use hardware acceleration when available"

Start Firefox in Safe Mode to check if one of the extensions (Firefox/Tools > Add-ons > Extensions) or if hardware acceleration is causing the problem (switch to the DEFAULT theme: Firefox/Tools > Add-ons > Appearance).

• Do NOT click the Reset button on the Safe Mode start window.

• https://support.mozilla.org/kb/Safe+Mode
• https://support.mozilla.org/kb/Troubleshooting+extensions+and+themes

In safe mode I have no crash.

I just test the change of hardware acceleration off. I will report.

All test of activation (inactivation) of add-ons didn't give any significant change

Last week, I moved my working place and I had to re-install all equipment. This is not ended, problems with the box and router and firewalls.

I note that when a session is restored I have not the tabs in right place with tab-group-manager : some old tabs comes back and the tab-group are not set in right place if tabs have been eliminated from restore panel.

I have upgraded to 29. I get always the same amount or more crashes

In Firefox Safe mode these changes are effective:

• all extensions are disabled (about:addons)
• default theme is used (no persona)
• userChrome.css and userContent.css are ignored (chrome folder)
• default toolbar layout is used (file: localstore-safe.rdf)
• Javascript JIT compilers are disabled (prefs: javascript.options.*jit)
• hardware acceleration is disabled (Options > Advanced > General)

• plugins are not affected
• preferences are not affected

I don't understand why you repeat what I know very well about Safe Mode.

In safe mode the problem isn't present.

guig2 had explore some ways and ask smoe questions

The crash signatures:

std::vector<CanvasRenderingContext2D*>::iterator iter = std::find(DemotableContexts().begin(), DemotableContexts().end(), context);,

Seems to be always the same. But I don't know how to check the new traces. About the versions, I had used 17esr, then 24esr since December (if I remember well, but the crash which began in December was still there) then since end of January I changed to standard version 27, 28 and now 29

guigs2 asked too : Could you please also give some insight on what you are doing right before the crash? A rendering of a 2d canvas is crashing but I do not know where to look for a regression unless I know if it also crashes in a fresh install of version 28. Please make a back up of the profile as well before doing this. You can launch Firefox in a new profile with a fresh installation with the cmd code in this article. Use the Profile Manager to create and remove Firefox profiles.

I have cleaned the current tabs and after update to 28. But I have not created a new profile (I had created several one to try an organization of the job to do, so I know well their management) completely clean. It is long to install add-ons, then work and check if the crash succeeds and get back the job done while testing.

Most of time the crashes occurs when loading a new page or when several page are automatically re-loaded from existing tabs (by tabGroupManager and TreeTabView). It is difficult to be be sure of the user action because there are several operations performed at same time.

Note : I have too others crashes of which reason is different : there are consecutive to the error message joined 1 and 2 : this occurs (seems to) because I save each 15 mn and sometimes I get the error of outofmemory (line 869) and a crash appears when a new try to save session occurs (any).

I got a crash while loading the first image joined. At restore session I got the image well in place (this as used probably : CanvasRenderingContext2D) . Material acceleration is off. This is repeated for second image. (note : I know the code which is coming from tikiwiki and generates the crash, quite sure)

May be we are near the solution.

Best regards

Trebly

You can try to disable hardware acceleration in Firefox (you need to close and restart Firefox).

• Tools > Options > Advanced > General > Browsing: "Use hardware acceleration when available"

• https://support.mozilla.org/kb/Troubleshooting+extensions+and+themes
• https://support.mozilla.org/kb/upgrade-graphics-drivers-use-hardware-acceleration

If it works in Safe Mode and in normal mode with all extensions (Firefox/Tools > Add-ons > Extensions) disabled then try to find which extension is causing it by enabling one extension at a time until the problem reappears.

Close and restart Firefox after each change via "Firefox > Exit" (Windows: Firefox/File > Exit; Mac: Firefox > Quit Firefox; Linux: Firefox/File > Quit)

I just tell " acceleration is off." sorry I was tired and tell "material" for "hardware" (because in French "hardware is "materiel" while "material" is in French "tissu").

You propose to do what I have done since four month without success. I have reported this into this thread several times.

I just tell that it seems that it is one lonely function which crashes "CanvasRenderingContext2D" (told by guigs2) and found by chance a way to repeat the crash when executing a well known code. Nevertheless today I could test the load of images without crash. It seems that the change of hardware acceleration had changed the behavior ( only after restart not requested ?)

This code is this user interface when you load an image "add images" and the crash happens when the image is displayed.

So there is no direct relation with any add-on nor with 'hardware acceleration". The lonely link with add-on(s) can be that they use the same function which crashes FF (js to resize image before display may be not well used).

You seem to answer without reading what I write or I had written before.

What you tell is repeatedly the same, something what I know and practice since long years. It is completely usefulness.

I think that "we" are near from the solution.

Don't worry and use time to think please.

Best regards

Trebly

Note : Generally I never translate my texts, I think and write directly in English but I make some mistakes

Is stack undefined because an antivirus or firewall is blocking the crash report uploader? I have not seen an undefined session.backup error before. I presume this happened when one of the addons was one and the other not. As you mentioned the two of them before (TreeStyleTab and Multiple TabHandler )

Is there a line in the session backup file that has "undefined"?

Hi,

About sessions backup crash (they don't crash FF) For session.backup it was the first time while the outofmemory is an old problem. It seems that there is no relation with the uploader. I think that there is a problem with the saved sessions stack array in particular conditions. It is always after a long session of work, consequently after many automatic backup. This time the two error seems to have occured at the same time (I was no there). It seems that the session manager first meets the outofmemory error. There is no backup without name, it has not been performed. After the outofmemory error a manual backup of session cannot be performed so, when this occurs, I just quit and restore from the previous saved session. Because of the number of crashes I make an automatic back up each 15 mn.

note about backup management : I make too a manual back-up when I have cleaned and re-organized the tabs. So when I get a crash I can restart with the better configuration of tabs. Previously I had a little backup stack (may be 12) One time the automatic backups each 15mn, as I had no crash during a while, lead to loose a previous important step that I forgot to backup permanently. So I think that an enhancement could be the ability to convert any automatic backup to a permanent one (without to have to save the current session, reload the old saved save it permanently and finally restart the first one) and an optional warning before the cleaning the oldest automatic backup. Currently by security I set the automatic backup to 100 (more than 24hrs and I could always find back a job done). I clean too using bookmarks and Zotero.

Trebly