87076 - MathML-enabled Mozilla crashes on any page with MathML

Status: RESOLVED DUPLICATE of bug 86368

(Core :: MathML, defect)

Description

[brentmh]

My MathML-enabled Linux build 2001062103 is crashing on any page with MathML
content.

Here are the options I used to compile it:

ac_add_options --disable-tests
ac_add_options --disable-md
ac_add_options --disable-debug
ac_add_options --disable-dtd-debug
ac_add_options --enable-optimize="-O2"
ac_add_options --enable-strip-libs
ac_add_options --enable-cpp-rtti
ac_add_options --enable-cpp-exceptions
ac_add_options --enable-elf-dynstr-gc
ac_add_options --enable-x11-shm
ac_add_options --enable-mathml
ac_add_options --enable-svg
ac_add_options --with-extensions=default,irc,xmlextras
ac_add_options --enable-chrome-format=both

Comment 1

[rbs]

Is it the same problem as that reported in bug 86368?

Comment 2

[brentmh]

It's possible.  Mine doesn't generate a core file so I can't tell if it's dying
in the same place.  The symptom is the same though: browser crash.  Is there any
way this can be put on the priority list for 0.9.2?  MathML in Mozilla is
completely unusable until this is fixed.

Comment 3

[rbs]

Without a stack trace, it is very hard to assess the problem and commit on
a fix. Is this a recent problem? When there is a problem like that at the edge
of a milestone things get tricky.

Comment 4

[brentmh]

I understand.  Do you know of a way to force mozilla to dump core?  Is there
some other debugging I can do?

Comment 5

[rbs]

Cc:ing dbaron in case he has an answer.

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

You could run under gdb (if your gdb works reliably), or, easier, attach with
gdb right before you do what crashes.  So you would just run "./mozilla" from
one shell and then from another, right before you do the action that crashes, do
"./mozilla -g -d gdb" and then type "attach PID" into gdb where PID is the PID
of Mozilla's main thread.

Comment 7

[brentmh]

Attachment: stack trace

Comment 8

[brentmh]

I've attached the stack trace from the crash when running mozilla under gdb.  It
was long but the first few lines were.

#0  0x410f7700 in FLG_NONE ()
   from /net/mntb2/local/mozdaily/mozilla-daily/components/libgfx_gtk.so
#1  0x410d8c3e in nsFontGTKNormal::GetWidth ()
   from /net/mntb2/local/mozdaily/mozilla-daily/components/libgfx_gtk.so

Please let me know if there's anything else I can do to help test this

Comment 9

[brentmh]

I've attached a stack trace from the crash when running under gdb.  It's pretty
long, but the first few lines are:

#0  0x410f7700 in FLG_NONE ()
   from /net/mntb2/local/mozdaily/mozilla-daily/components/libgfx_gtk.so
#1  0x410d8c3e in nsFontGTKNormal::GetWidth ()
   from /net/mntb2/local/mozdaily/mozilla-daily/components/libgfx_gtk.so

Please let me know if there's anything else I can do to help debug this.

Comment 10

[rbs]

Does the patch that bstell attached on bug 86368 fix the problem?

Comment 11

[brentmh]

Yes!  That fixes it.  I guess it was the same bug after all.

Comment 12

[rbs]

*** This bug has been marked as a duplicate of 86368 ***