_rtc_trap_handler

Question

Hi,
I am running into a problem when enabling runtime checking
from dbx on a huge 64-bit executable. This execuatble is linked to 5 shared libraries. On enabling runtime checking on dbx, I hit
upon the 8M limit . As recommended , I set the dbx environment
variable rtc_use_traps to "on" since my machine is a sun4u machine. But now even, on enabling all access checking, I get the
following error -
signal SEGV (no mapping at the fault address) in __rtc_trap_handler
I find that the stack shown at this point is not in my application code.
The other ways to get rid of the 8M problem as recommended by Sun are using the rtc_patch_area. Is it recommended that I use the rtc_patch_area even for a sun4u arch machine where the trap method is said to work . Any suggestions will be helpful.
Thanks.

Answer 1

/export/home/husam/release/lib/libstorage_s.so.1:backend.c stab #530 host:(0,211),The key to your problem is probably the stabs reading error.
How was backend.c compiled?
Normally dbx will load stabs information on demand as needed.
You can start dbx and use the "module" command to load
the stabs information right away without running the program or using RTC.
Try this:
(dbx) module backend.o
Does that cause the stabs warnings?
If you can "repair" or "normalize" the stabs information in that module,
the crash will probably go away.