Bug 221076 - SIGSEGV loop in uw_frame_state_for in Java code
Summary: SIGSEGV loop in uw_frame_state_for in Java code
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: gcc
Version: 6
Hardware: x86_64
OS: Linux
medium
high
Target Milestone: ---
Assignee: Jakub Jelinek
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2007-01-01 10:15 UTC by Robin Green
Modified: 2007-11-30 22:11 UTC (History)
1 user (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2007-01-01 15:51:28 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Backtrace from gdb (12.76 KB, text/plain)
2007-01-01 10:15 UTC, Robin Green
no flags Details

Description Robin Green 2007-01-01 10:15:23 UTC
Description of problem:
When I attempt to connect to a new Derby database (a pure Java database) in
OpenOffice.org Base, it crashes.

Running it in gdb shows uw_frame_state_for is generating a SIGSEGV when trying
to generate a stacktrace for a previous SIGSEGV, and this in turn causes
uw_frame_state_for to be called again, which again falls over, and so on ad
infinitum.

Version-Release number of selected component (if applicable):
gcc-4.1.1-30

How reproducible:
Always

Steps to Reproduce:
1. Download Derby from
http://apache.oss-mirror.org/db/derby/db-derby-10.2.2.0/db-derby-10.2.2.0-lib.tar.gz
 and untar it
2. Run ooffice
3. Go into Tools->Options->Java
4. Verify that the GNU GCJ runtime is selected
5. Click the Classpath button
6. Add db-derby-10.2.2.0-lib/lib/derby.jar to the Classpath
7. Exit the Options dialog
8. Create a new database of type JDBC, with the connection string
jdbc:derby:/tmp/testdb;create=true and the driver
org.apache.derby.jdbc.EmbeddedDriver (no username is required). Then test the
connection. This should succeed, and create a directory "/tmp/testdb" containing
the (empty) Derby database.
9. Click Finish.
10. Click on the Tables button at the top-left.
  
Actual results:
Crash - see attached backtrace

Expected results:
No crash

Additional info:
openoffice.org-base-2.0.4-5.5.3

Comment 1 Robin Green 2007-01-01 10:15:23 UTC
Created attachment 144603 [details]
Backtrace from gdb

Comment 2 Jakub Jelinek 2007-01-01 11:08:48 UTC
Which glibc you are using?

Comment 3 Robin Green 2007-01-01 14:34:43 UTC
glibc-2.5-3

also, I forgot about this:
kernel-2.6.19-1.2877.fc7

Comment 4 Robin Green 2007-01-01 15:51:28 UTC
After I remembered to put these at the very *start* of my gdb session:

handle SIGPWR nostop noprint
handle SIGXCPU nostop noprint

I no longer get the SIGSEGV loop and the corrupt stack. Those are irrelevant
artifacts. I'm closing this bug as NOTABUG and I'll refile a bug, probably
upstream, once I figure out what's really causing the crash.


Note You need to log in before you can comment on or make changes to this bug.