1132896 - Firefox hangs in print or print preview by using float and clear if content overflows to 2 pages

Status: RESOLVED WORKSFORME

(Core :: Layout: Floats, defect)

Description

[Volker]

Attachment: Firefox_hangs.jpg

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:35.0) Gecko/20100101 Firefox/35.0
Build ID: 20150122214805

Steps to reproduce:

Tried to print or print preview the webpage http://www.pollin.de/shop/banner/rauchmel.htm with Firefox Windows Version on Windows 8.1 Pro


Actual results:

Firefox hangs and must be killed


Expected results:

Print dialog or print preview window should appear

Comment 1

[Matthias Versen [:Matti]]

I can confirm a hang with Firefox35 and Firefox38 on windows7
The attached stacktrace could be wrong because I had some issues

Comment 2

[Matthias Versen [:Matti]]

Attachment: Windbg stacktrace

Comment 3

[Robert Utasi [:hunboy]]

Attachment: most reduced hang testcase

I also can confirm this hang, reduced testcase attached which one hangs during print-preview and print.

shortly the issue: 2pages, float and clear causes hang.

Comment 4

[Volker]

I can confirm, your testpage also causes the hang. Thanks a lot! Great that you reduced the page to the essential problem. Now only someone is missing who is able to fix the bug. :-)

Comment 5

[Robert Utasi [:hunboy]]

finished with regression window:

http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=557165a66267&tochange=88d3c5bde0ba

See Bug 652178 (assumed)

regression based on my reduced testcase

David, pls take a look at it

also I can confirm the bug under linux.

Comment 6

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

(I'm treating this request as low priority right now, given that it's a regression that took nearly 4 years to find.)

Comment 7

[Robert Utasi [:hunboy]]

Attachment: another reduced hang testcase2

another reduced hang testcase2 with another regression window for testcase2. looks this is a cummulated bug:

http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=bbefb7bcb41e&tochange=26ee1b556bd9

shortly: 2 or more pages + float + some margin or padding.
Similar than the first testcase, however that went wrong later.

Now we have 2 testcases with 2 regression windows

Comment 8

[Volker]

Attachment: How IE handles the mentioned webpage

Comment 9

[Volker]

@dbaron: I don't think it took 4 years to find that bug. I encountered hangs quite often before when printing. But I didn't mind and used for printing Internet Explorer instead. But meanwhile I thought it's time to say something so that this bug can be fixed. ;-)

I LOVE firefox in common but I really hate it's printing capabilities! They are very poor compared to that of the Internet Explorer. And I think it's time for firefox to close that gap to IE.

Just for comparison I added an screenshot hot IE treats this webpage (https://bugzilla.mozilla.org/attachment.cgi?id=8565925).

Comment 10

[Robert Utasi [:hunboy]]

David,

I just tried to find dupes for it and noticed several hangs were reported and almost all of them are affected in a some way with padding or margin as well.

Probably this means the fixing point is not there where the regression-windows show, those are only a points where the original bug shows up.

(Note, when you remove the margin or padding rules from the testcases, the bug hides.)

Currently we have 2 reduced testcases for hanging. I don't have time-capacity for reducing all of reported hanging webpages. So 2 reduced testcases are enough or try to find more?

Volker,

I'm not sure if the IE behavior is standard. <img> is a replaced element. Use diplay:block on the image. But this is not relevant to the hang here. This should be an other bugreport.

Comment 11

[Daniel Bodea [:danibodea]]

I have attempted to confirm this old bug by attempting to print and print preview the page in comment 0 on a Windows 7, on the latest Nightly, but no hang could be performed. Then I tried to reproduce it in the old Nightly v35.0a1, but I could not reproduce it either (it should probably be mentioned that the page's content is not rendered correctly on this old version).

I have succeeded in reproducing a hang (Firefox not responding) with the reduced test case from comment 3 and comment 7 in the old Nightly v35.0a1, but it does not reproduce in the latest Niglhy v87.0a1.

While searching for the change responsible for the fix, I have observed that it appears it was somewhere in v40. Results are unclear due to the diversity of behaviors around those old versions.

Considering the above information, I believe that we can close this bug as WORKSFORME.