It looks like I've found a small bug in insight-gdb.
I've fixed it in on my PC but I would like to see it fixed in the main
branch.
I have insight+dejagnu-weekly-20000412.tar.bz2 and Cygwin CD1.0
I compiled it for --target=m68k-motorola-coff and for host i686-cygwin
Then I tried to connect remotely to my m68k-based SBC.
Insight crashed in gdb-related code.
Crash happened in m68k-tdep.c in delta68_in_sigtramp:
int delta68_in_sigtramp (pc, name)
CORE_ADDR pc;
char *name;
{
return strcmp (name, "_sigcode") == 0;
}
The reason -- pc was pointed to ROMmed RTOS rather than any program segment.
So name parameter was NULL. Function strcmp crashed insight then.
I've fixed the code and now it works fine.
BTW, gdb 4.18 works fine, 'cuz it does not have this function.
My solution:
int delta68_in_sigtramp (pc, name)
CORE_ADDR pc;
char *name;
{
if (name) /* Serge's fix here */
return strcmp (name, "_sigcode") == 0;
return 0; /* Serge's fix here */
}
the diff file, generated as
diff -Naur insight+dejagnu-20000412-old\gdb\m68k-tdep.c
insight+dejagnu-20000412-new\gdb\m68k-tdep.c > m68fix.diff
*************Begin*********************
--- insight+dejagnu-20000412-old\gdb\m68k-tdep.c Fri Mar 24 09:55:13 2000
+++ insight+dejagnu-20000412-new\gdb\m68k-tdep.c Fri Apr 21 12:24:55 2000
@@ -91,7 +91,9 @@
CORE_ADDR pc;
char *name;
{
- return strcmp (name, "_sigcode") == 0;
+ if (name)
+ return strcmp (name, "_sigcode") == 0;
+ return 0;
}
CORE_ADDR
************End*********************
