PFE 0.32.56 signal support segfaults on Linux 2.4.18 
Author Message
 PFE 0.32.56 signal support segfaults on Linux 2.4.18

Subject pretty much says it all.  The stock "exceptn.4th" file will
crash under Linux 2.4.18.  Here's where it crashes:

' SIG-TERM-HDL SIGTERM SIGNAL .
SIGTERM RAISE   ( <-- it crashes here! )
0 SIGTERM SIGNAL ' SIG-TERM-HDL = Y

Basically, instead of a SIGTERM, I get a SIGSEG instead!  :D

--
Samuel A. Falvo II



Fri, 21 Jan 2005 04:29:02 GMT  
 PFE 0.32.56 signal support segfaults on Linux 2.4.18
Es schrieb "Samuel A. Falvo II":

Quote:

> Subject pretty much says it all.  The stock "exceptn.4th" file will
> crash under Linux 2.4.18.  Here's where it crashes:

> ' SIG-TERM-HDL SIGTERM SIGNAL .
> SIGTERM RAISE   ( <-- it crashes here! )
> 0 SIGTERM SIGNAL ' SIG-TERM-HDL = Y

> Basically, instead of a SIGTERM, I get a SIGSEG instead!  :D

I noticed that signal handling has been problematic for some time,
there are some interactions with the setjmp code. It needs to be
tracked down at some point but I had no time to go about it as of
last. The Tek work is on embedded platforms that does not need any
signal handling. A bad fix would be to comment out the offending
section but I'd not like that either.

The current README file says: ...Some checks in the
test/ directory are not autochecked with `make check` due
to problems in the signal/raise mechanics...
- and actually I know some platforms where the signal/raise
mechanisms never works even that the system says it does. Well,
on linux it should be working as the operating system does
implement it correctly (but without useful extensions that can
be on other unix systems).

have fun, guido



Fri, 21 Jan 2005 08:52:14 GMT  
 PFE 0.32.56 signal support segfaults on Linux 2.4.18
On 4 Aug 2002 13:29:02 -0700

Quote:
> Subject pretty much says it all.  The stock "exceptn.4th" file will
> crash under Linux 2.4.18.  Here's where it crashes:

        ^^^^^^^^^^^^^^^^^^
there might be a syscalls/libc problem. some of those calls were changed
to different arguments passing conventions (e.g. mmap) or, just behave
differently (e.g. mount) or, were in-consistently renamed - or, falsely
described in the rsp. man-pages.

Quote:

> ' SIG-TERM-HDL SIGTERM SIGNAL .
> SIGTERM RAISE   ( <-- it crashes here! )
> 0 SIGTERM SIGNAL ' SIG-TERM-HDL = Y

> Basically, instead of a SIGTERM, I get a SIGSEG instead!  :D

> --
> Samuel A. Falvo II

--
   >>> pse, reply to : clf -at- lxhp -dot- in-berlin -dot- de <<<
Linux,Assembly,Forth: http://www.lxhp.in-berlin.de/index-lx.shtml en/de


Fri, 21 Jan 2005 05:00:46 GMT  
 
 [ 3 post ] 

 Relevant Pages 

1. error Logout 56, Help with

2. Runcode 56 returned by run

3. Error 56 returned by Runcode

4. LOGOUT Already Active (56) - why ?

5. LOGOUT Already Active (56)

6. UDP read error code 56

7. -56 THROW is QUIT ?

8. Problems Integrating PASM386 into FPC3.56

9. Basic FPC3.56 Float Question

10. multiplier with 56 bit

11. Help: bitwise operations on a 56 bit word

 

 
Powered by phpBB® Forum Software