121472 - hangs, does not refresh

Status: RESOLVED WORKSFORME

(SeaMonkey :: General, defect)

Description

[phil]

Another "keep it up for days" issue.

I had two mozilla windows open, for a day or three. I went to sleep.
I came back, and the windows would not refresh. It was like I had suspended
mozilla... except that it was not suspended.

a partial output of "truss" on solaris:

read(7, "\b04FAA3 697DC92 %\0\0\0".., 32)       = 32
read(7, "\n03FAA3B4 7 @03\0 | `DF".., 32)       = 32
read(7, 0x080469CC, 32)                         Err#11 EAGAIN
door_return(0x00000000, 0, 0x00000000, 0) (sleeping...)
signotifywait()                 (sleeping...)
poll(0xDE4C0A80, 1, 35000)      (sleeping...)
poll(0x080467E0, 1, -1)         (sleeping...)
lwp_sema_wait(0xDD450E3C)       (sleeping...)
lwp_sema_wait(0xDF2BA370)       (sleeping...)
lwp_cond_wait(0xDF2BFEC8, 0xDF2BFED8, 0xDD714D40) (sleeping...)
poll(0xDE4C0A80, 1, 35000)                      = 0
door_return(0x00000000, 0, 0x00000000, 0) (sleeping...)
signotifywait()                 (sleeping...)
poll(0xDE4C0A80, 1, 35000)      (sleeping...)
poll(0x080467E0, 1, -1)         (sleeping...)
lwp_sema_wait(0xDD450E3C)       (sleeping...)
lwp_sema_wait(0xDF2BA370)       (sleeping...)
lwp_cond_wait(0xDF2BFEC8, 0xDF2BFED8, 0xDD714D40) (sleeping...)

It keeps looping around the poll and sema_wait calls.


this is the binary for solaris x86, moz 0.9.7 running on solaris 8.
Built by:  Paul Pietromonaco
e-mail:    [e-mail]
Build ID
2001122311

Comment 1

[phil]

This may be related to 121048, which I also filed.

I just experienced the "wont refresh", but this time, I noticed a
BadRequest error.

X Error of failed request:  BadRequest (invalid request code or no such operation)
  Major opcode of failed request:  0 ()
  Serial number of failed request:  3414565
  Current serial number in output stream:  3414564

This was not after leaving it alone, this time;
 I have been using fvwm to actively flick between mozilla, and other virtual
areas. (I always use fvwm)

Comment 2

[Asa Dotzler [:asa]]

gissburn, timeless,  can  either of  you  two  confirm this?

Comment 3

[timeless]

reporter: is the poll of a unix domain socket or an ip socket? (my guess is 
the former.)

doors are fun things, but i never had time to learn about them.

Comment 4

[Roland Mainz]

The (IMHO innocent) door used may be for the connection to "nscd" (Name Service
Caching Deamon).
"truss" is not the best tool here, better is a tool like /usr/proc/bin/pstack
which provides a snapshot of all thread stacks.

Reporter:
1. Did you apply a recommended patch cluster for Solaris 8 in the last month
(just asking, I do not want to hunt an already fixed OS bug... =:-) Some people
are still using Solaris 8 FCS (First Customer Ship) without any patches... ;-( )
...
2. Can you "try again" and provide a dump via /usr/proc/bin/pstack, please ?

Comment 5

[Donnie Cranford]

I can confirm that this is not being seen on any of my Solaris builds.
I am running Solaris 8 SPARC 10/01 w latest recommended patch update. 
Also Solaris 9 EA 

I am using the Forte 7 EA compiler for both.

Comment 6

[Roland Mainz]

dcran:
How long did your Zilla stay open ?

Comment 7

[Donnie Cranford]

Gisburn

I compiled moz on friday afternoon before I left work on both platforms and left 
it open (I forgot to close it) Checked it this morning and it worked just fine.
So this would have been the Friday afternoon trunk about 3:00 p.m. CST

Dcran

Comment 8

[Roland Mainz]

Mhhh, OK... this is IMHO long enougth... :)

Let's wait whether the reporter has applied the recommended patches or not...

Comment 9

[phil]

A few additional details:

I'm running solaris 8 MU5 (Solaris 8 s28_38shwp2 INTEL)

I was using Xsun at the time. so be sure to check against that.
Additionally, it may be a gcc issue, since I believe the binary was compiled
with gcc?

Plus, I personally dont think that "bring it up friday, come back monday" is a
very good test. I would say grab the binary, and use it regularly for a WEEK,
would be a fair test.

if I see the problem again, i will give a ptrace output. However, right now, I
dont run into it much, 'cause I'm busy porting utah-glx to xfree. lots of
reboots, and different X server.

Comment 10

[Stuart Parmenter]

I started to see this on Solaris 8 (sparc) immediatly after installing the
latest recommended cluster (1/25/02).  I'm not sure how to uninstall the
cluster, but once I figure that out, I will see if that fixes it.

Comment 11

[Andrew Hagen]

I don't think SunOS runs on the PC. Changing OS to Solaris. Does this problem
only occur with Solaris on the PC?

Comment 12

[Jim Crumley]

Does this problem still occur? If so, please re-open this bug.  Marking works
for me for now.