This change is being checked into the XFS tree first, because that
group seems to have the most problems with the i386 kdb backtrace. The
backtrace heuristics have been tweaked so it now does two passes
looking for the next return address. The first pass only considers
code that can be validated as branching to the current function. Iff
the first pass fails, then do a second pass allowing for code that does
branch by register. Such code cannot be validated and must be assumed
to be valid, sometimes resulting in false positives.
If you still have backtrace problems on i386, capture the output from
these kdb commands and send it to kaos@xxxxxxxx
set LINES 2000
bt
set KDBDEBUG 0x19
bt
Date: Sun Jun 22 23:01:23 PDT 2003
Workarea: sherman.melbourne.sgi.com:/build/kaos/2.4.x-xfs
The following file(s) were checked into:
bonnie.engr.sgi.com:/isms/slinx/2.4.x-xfs
Modid: 2.4.x-xfs:slinx:151810a
linux/arch/i386/kdb/kdbasupport.c - 1.27
|