Re: progress on infinite DTLB-miss

On Tue, 15 Jun 2004 09:34:02 -0700
Chuck Silvers <[email protected]> wrote:

> I checked in some fixes this morning, both for the infinite DTLB-miss
> and for the SANITY() assert that was causing trouble for DEBUG
> kernels.
About the infinite error messages: When this return(0) is removed the
kernel jumps into DDB instead repeating the error messages ad infinitum.
Index: hppa/db_interface.c
===================================================================
RCS file: /cvsroot/src/sys/arch/hppa/hppa/db_interface.c,v
retrieving revision 1.3
diff -u -r1.3 db_interface.c
--- hppa/db_interface.c 28 Nov 2003 19:02:25 -0000      1.3
+++ hppa/db_interface.c 15 Jun 2004 20:50:28 -0000
@@ -174,7 +174,6 @@
                        db_error("Caught exception in DDB;
continuing...\n");
                        /* NOT REACHED */
                }
-               return (0);
        }
 
        /* XXX Should switch to kdb`s own stack here. */


> I haven't tried the boot-from-disk stuff yet, though.
I did a cvs update of my kernel source and: No fun.

biomask 00000068 netmask 00000078 ttymask 000000fe
scsibus0: waiting 2 seconds for devices to settle...
sd0 at scsibus0 target 2 lun 0: <DEC, DSP3210S, 441G> disk fixed
sd0: 2048 MB, 3045 cyl, 16 head, 86 sec, 512 bytes/sect x 4194685
sectors
sd0: sync (100.00ns offset 8), 8-bit (10.000MB/s) transfers
boot device: sd0
root on sd0a dumps on sd0b
panic: trap: uvm_fault(0x630c90, 30000000, 0, 2): 14
Stopped in pid 0.1 (swapper) at netbsd:cpu_Debugger+0x10:       ldo    
R'40(%r3
),%r30
db> t
cpu_Debugger(587d48,90000000,0,8) at netbsd:cpu_Debugger+0x10
panic(598b18,630c90,30000000,0) at netbsd:panic+0x1b8
panic(f,6a7a80,0,68) at netbsd:panic+0x148
trap(1490ca8,0,0,a) at netbsd:trap+0x8d8
netbsd:$tlb_missend+0x2a8:
db> 
-- 


tschüß,
       Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/