[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
| Title: | DIGITAL UNIX (FORMERLY KNOWN AS DEC OSF/1) |
| Notice: | Welcome to the Digital UNIX Conference |
| Moderator: | SMURF::DENHAM |
|
| Created: | Thu Mar 16 1995 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 10068 |
| Total number of notes: | 35879 |
8819.0. "Cannot debug multi-threaded programs" by TAEC::BRONOEL () Thu Feb 13 1997 07:05
Hi,
We encounter a new strange problem with our TeMIP executables
on both V3.2C and V3.2G platforms : we cannot debug anymore
our programs.
Symptoms are :
# dbx mcc_domain_fm
dbx version 3.11.8
Type 'help' for help.
(dbx) r
Interrupt
signal Interrupt at >*[__accept, 0x3ff8051a688] beq r19,
0x3ff8051a6a0
(dbx) tlist
thread 0x0 signal Interrupt at >*[__accept, 0x3ff8051a688] beq r19,
0x3ff8051a6a0
(dbx) tstack
Thread 0x0:
> 0 __accept(0x3ff80560760, 0x11fffeb18, 0x3ff81299914, 0x3ffc072d3f0,
0x1400036ec) [0x3ff8051a688]
1 mcc_accept(0x4, 0x11ffff498, 0x11ffff490, 0x100000000, 0x3ff81266210)
[0x3ff8129992c]
2 im__rpc_server(0x120016fc0, 0x3ff00000018, 0x10, 0x0, 0x0) [0x3ff81266320]
3 mcc_start_server(0x10, 0x0, 0x0, 0x1200396c0, 0x120016fe4) [0x3ff81299554]
4 main(0x2, 0x0, 0x120016b38, 0x355a0a3, 0x100000000) [0x120016fe0]
Note that for example this executable has been compiled with -g
option, has not been stripped and is working correctly. In addition
is is really multi-threaded as shown by ps command.
However it is now impossible to switch from main thread to
another using dbx. Note also that, as usual, ladebug behaviour
is worst : after a long startup it cannot be interrupted and
generally after a while produce a core dump.
Our OS used versions are : V3.2C and V3.2G.
Note also that this is the same problem either /proc is
mounted or not.
Hope you can help us.
Thanks,
Jean-Marc.
| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 8819.1 | | SMURF::DENHAM | Digital UNIX Kernel | Thu Feb 13 1997 09:16 | 6 |
| Well, just to make one thing clear -- if /proc isn't mounted,
you'll never see more than one thread in the task, in spite
of how many ps may show you. Why? Becuase the ptrace interface
used in the absense of /proc doesn't have the slightest clue
about what a thread is. Ptrace is extremely primitive. Unlike
like /proc, it is also reasonably maintainable....
|
| 8819.2 | | COL01::LINNARTZ | | Sun Feb 16 1997 14:25 | 5 |
| debugging also behaves diffrent for SUID programs. It's not
nice, but perform debugging as the SUID user with the application
having NON-SUID protection.
Pit
|
| 8819.3 | | COL01::LINNARTZ | | Mon Feb 17 1997 09:50 | 15 |
| and two more possibilities, i've observed from time to time.
If you're manipulating signalmask for SIGTRAP, the debugger
won't stop at your breakpoint. The ladebug conference
mentioned that this should be fixed with V4.0b, but the
partner I'm working with didn't give me a positigve feedback
yet.
Even harder to figure out is the ladebug behaviour when using
tcsh. I've observed a couple of times different behaviour
(not stoping at Bp's, etc) while working on the same executable.
in brief, after they switched from tcsh to csh the problems
dissapeared. Being honest, I diden't have the time to further
analyse the actu7al cause.
Pit
|