Closed Bug 102457 Opened 23 years ago Closed 23 years ago

OS/2 - Crash on Japanese-only web site in UCONV.DLL

Categories

(Core :: Internationalization, defect)

x86
OS/2
defect
Not set
critical

Tracking

()

VERIFIED FIXED

People

(Reporter: rockowallaby, Assigned: mkaply)

References

()

Details

(Keywords: crash, intl, stackwanted)

Attachments

(4 files)

I don't THINK this has been reported yet (I did a search).  I dislike giving
duplicate bug reports as much as you all dislike getting them!

When going to the URL listed ( http://www.genki.co.jp ), it crashes.  Here's the
POPUPLOG.OS2 output:
------------------------------------------------------------

09-30-2001  18:39:24  SYS3175  PID 0052  TID 0001  Slot 007a
F:\WARPZILLA\BIN\MOZILLA.EXE
c0000005
1fd80a54
P1=00000001  P2=00000034  P3=XXXXXXXX  P4=XXXXXXXX
EAX=006d0072  EBX=00f86530  ECX=00000000  EDX=0000003c
ESI=006d0082  EDI=00f86534
DS=0053  DSACC=f0f3  DSLIM=ffffffff
ES=0053  ESACC=f0f3  ESLIM=ffffffff
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1fd80a54  CSACC=f0df  CSLIM=ffffffff
SS:ESP=0053:0012b748  SSACC=f0f3  SSLIM=ffffffff
EBP=0012b798  FLG=00012206

UCONV.DLL 0001:00000a54

I have only the en-us language set, and my default character coding is set to
blank (nothing selected).  I have the default US Region content pack installed.
The web site given is Japanese only, as I found out, if that makes a difference.
Whoops!  I'm on Build ID 2001091812  and I'm using the standard US English
version of OS/2 (Fixpack 15) and SDD Beta 43
Hmm.

UCONV.DLL is a system DLL.

I don't see the trap with the OS/2 convenience pack.

Do you see the trap as soon as you go to the page?
Yes.  I can just begin to see some text or graphics (not sure which - it's
rather fast), when it closes abrubtly with the error.  Is there any other
information about my system you need?

Can anyone else try and duplicate this?  I have a fairly standard system I'd
like to think...
Switching qa contact to mkaply for now since we don't have the test environment. 
 Please re-assign further as you see is appropriate.  Thanks.
Keywords: intl
QA Contact: andreasb → mkaply
Need a stacktrace
Severity: normal → critical
Keywords: crash, stackwanted
Jan, can you please give me your Os/2 information (video driver, fixpak level, etc.)

Thank you
I could reproduce this problem by taking special procedure as follows:
1) Run mozilla as usual - works fine.
2) Close  mozilla.
3) Move d:\language\codepage\ibm*  to some other directory.
4) Run mozilla again - get trap as below.
------------------------------------------------------------
11-01-2001  18:25:15  SYS3175  PID 006f  TID 0001  Slot 0090
F:\WARPZILLA-NIGHTLY\BIN\MOZILLA.EXE
c0000005
1f8d0a50
P1=00000001  P2=00000034  P3=XXXXXXXX  P4=XXXXXXXX  
EAX=00690072  EBX=01d1f9a0  ECX=00000000  EDX=0000003c
ESI=00690082  EDI=01d1f9a4  
DS=0053  DSACC=d0f3  DSLIM=3fffffff  
ES=0053  ESACC=d0f3  ESLIM=3fffffff  
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1f8d0a50  CSACC=d0df  CSLIM=3fffffff
SS:ESP=0053:0012bd70  SSACC=d0f3  SSLIM=3fffffff
EBP=0012bdc0  FLG=00212206

UCONV.DLL 0001:00000a50
------------------------------------------------------------
So I don't think this is mozilla's bug but system failure or misconfiguration
around codepage/unicode.
This is very helpful.

we can probably at least prevent the trap.

I will see what I can find out.
To the reporter - rocko@ao.net

Can you check in your x:\language\codepage\ directory and see if you have a file 
IBM943?

Can you post a list of all the IBM* named files in that directory?

I think I know what this problem is.

Thanks
Directory listing of codepages is attached. No ibm943 in the directory nor does
this file exist in any Warp 4 archive i have on my disk:?
What is your OS/2 version?

The issue is probably that we shouldn't be using 943 if it is not available. 

We have other choices of Japanese codepages.

There are two bugs here that we are going to fix.

1. We should be checking errors when Uconv fails so that we don't crash.

2. We should use the right Japanese codepage for the system.
OK, uploaded my complete listing of that directory.  I don't have IBM943 either.

Oh, the video card driver is the latest SDD/2 Beta, so it uses GRADD.
Attached patch Fix for problemSplinter Review
There were a few issues at work here.

The core problem was that we were using UniQueryUconvObject 
without checking whether or not UniCreateUconvObject 
succeeded. Os/2 does bad things when you do that.

I've added code to check all the return codes from the 
Unicode calls.

In addition, I've changed our default Japanese code page 
from 943 to 932 and added code similar to the code we use 
to check Chinese codepages.

We use 943 if PM supports it AND if we can create a 
converter for it, otherwise we use 932.

Hopefully this should take care of the issues.
Status: UNCONFIRMED → ASSIGNED
Ever confirmed: true
I appreciate everyone's help in getting this one nailed.

Marking fixed.
Status: ASSIGNED → RESOLVED
Closed: 23 years ago
Resolution: --- → FIXED
Verifying.

got this note from one of the people that saw this problem:

Excellent! 
Just loaded the latest build and checked edom.com.tw site, which works
fine:)


Thanks,
jan lambert
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.