Very frequent crashes on startup in Alpha Linux

VERIFIED WORKSFORME

Status

SeaMonkey
Build Config
--
critical
VERIFIED WORKSFORME
18 years ago
14 years ago

People

(Reporter: mach1, Assigned: msw)

Tracking

({crash})

Trunk
DEC
Linux
crash

Firefox Tracking Flags

(Not tracked)

Details

Attachments

(2 attachments)

(Reporter)

Description

18 years ago
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.2.16-3 alpha; en-US; 0.8.1) Gecko/20010310
BuildID:    0000000000

Frequently crashes on startup (95% of attempts) with Segmentation Fault on RH
Linux 6.2 (kernel 2.2.16-3) for Alpha. Seems to be dependent on what has been
run before and what initialization tasks are performed (e.g. component
registration).

Reproducible: Sometimes
Steps to Reproduce:
1.Just attempt to start (e.q. ./mozilla in mozilla/dist/bin directory).
2.
3.

Actual Results:  Browser performs its startup and fails with a core dump. The
failures occur at different points in the startup process, but once a failure
occurs it seems that immediate re-attempts to start the browser fail at the same
point. Removing component.reg often helps to get the browser started.

Expected Results:  The browser should start and present an initial window.

This problem is exhibited after a complete download of the source and build of
the browser, but also occurs when starting after performing other operations and
programs. Immediate execution of the browser following the build consistently
fails. The problem has occured on a version of the sources downloaded at 1:00 pm
MST on 10-MAR-2001, but similar problems have existed since early December of
2000. Prior to that time the problem did not exist. The problem seems to be
getting worse over time, but this is only my impression--I have not objective
evidence of this.

In the most recent build I was able to get gdb to read the various symbol
information (without exhausting memory, generating a gdb internal error, or
crashing the operating system) and give a stack trace, but I don't know how to
get it pasted in this field. If it is useful, I can provide it if someone can
tell me how to do so.
(Reporter)

Comment 1

18 years ago
Created attachment 27396 [details]
Stack trace of typical failure
(Reporter)

Comment 2

18 years ago
Created attachment 27397 [details]
Stack trace of failure immediately following previous one

Comment 3

18 years ago
um. perhaps networking: file?
Assignee: asa → dougt
Component: Browser-General → Networking: File
Keywords: crash
QA Contact: doronr → tever

Comment 4

17 years ago
Reporter still a problem in the latest nightlies?
(Reporter)

Comment 5

17 years ago
I have downloaded the complete sources and re-built the browser each week since
my original report--the latest download was today (07-APR-2001). All of these
attempts to build the browser have resulted in a browser that will not run long
enough to display a window and all of the previous methods of getting the
browser to start (e.g. deleting component.reg) have failed. All attempts to run
gdb have failed with gdb internal errors.

Comment 6

17 years ago
After talking to Asa on IRC he says mark this NEW so I am :)
Assignee: dougt → cls
Status: UNCONFIRMED → NEW
Component: Networking: File → Build Config
Ever confirmed: true
QA Contact: tever → granrose

Comment 7

17 years ago
Reassigning to msw as he claims the EV4 box netscape has is too old to upgrade
to RH6.x.  
Assignee: cls → msw

Comment 8

17 years ago
The 0.8.1 is (somewhat) runnable at RH 7.0/rawhide Linux/Alpha system,
presuming I use blizzards specs file from his RedHat site.

However there are possible problems lurking in the corners including:
- version of gcc (perhaps egcs ? a.k.a. kgcc)
- Need to use blizzards collection of support patches (?)
- 64-bit pointer issues where code stores said pointer into some 32 bit space,
and ...

Most cases of the last are in hash functions, which should know better, than
to cast a pointer into 32-bit quantity -- for hashing that is ok, but via:
    (PR32Uint)(long)ptrvar
or something like that to silence the compiler.  Now there is just too much
clutter to find the real problem cases.

Comment 9

17 years ago
Is this still a problem for anyone? (probably not?)
0.9.7+ compiles/runs fine for me (Linux 2.4.9-12 Alpha)

Comment 10

17 years ago
I don't know if this is the same thing, but I just installed Mozilla 0.9.9 and I
get a segmentation fault on at least one page. I'm running alphalinux (RH 7.1 on
a 164LX) and I get a segmentation fault not on startup, but when viewing this
page: http://www.sunspot.net.I can attach a stack trace although I'm new to
this, and I'm not sure if I did the right thing or not.

I was previously using 0.8.1 and that viewed that page no problem. 
(Reporter)

Comment 11

17 years ago
This may actually by the problem reported as bug 127455. I don't know if the
patch developed for that bug is present in the version of the browser (0.9.9)
being used. I built the browser from scratch on 16-MAR-2002 and it seems to
handle the indicated page just fine.

Comment 12

17 years ago
the patch for bug 127455 did not make it into 0.9.9, so if you compile 
from 0.9.9 sources, that bug will show up.  However, the RedHat RPMS (currently 
available as RawHide) do contain the patch for that bug.

Comment 13

16 years ago
mach1: so you're not seeing *this* bug anymore, right (Mozilla starts up ok)?
(Reporter)

Comment 14

16 years ago
That's right, the browser starts up consistently and runs fine.

Comment 15

16 years ago
excellent!

marking WORKSFORME
Status: NEW → RESOLVED
Last Resolved: 16 years ago
Resolution: --- → WORKSFORME

Comment 16

16 years ago
verifying worksforme.
Status: RESOLVED → VERIFIED
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.