Panic with latest sources.
Matthew Dillon
dillon at apollo.backplane.com
Thu Feb 12 11:17:52 PST 2004
:Ok here we go...
:Regression report (GENERIC kernel):
:
:1) 2004.02.06.00.00.00 - works
:2) 2004.02.08.00.00.00 - works
:3) 2004.02.10.00.00.00 - buildkernel failure
:4) 2004.02.10.12.00.00 - panic
:5) 2004.02.09.12.00.00 - works
:6) 2004.02.10.06.00.00 - works
:
:So...something committed (and which still exists) between 2004.02.10.06.00.00
:and 2004.02.10.12.00.00 results in the following sort of early panic.
All right, lets see. Not a whole lot changed between those dates.
Try undoing some of these recent changes, one at a time, and see if it
fixes the problem:
The double psm reset:
--- dev/misc/psm/psm.c 27 Aug 2003 10:35:18 -0000 1.8
+++ dev/misc/psm/psm.c 10 Feb 2004 15:56:38 -0000 1.9
@@ -1040,6 +1044,11 @@
if (verbose)
printf("psm%d: failed to reset the aux device.\n", unit);
endprobe(ENXIO);
+ } else if (!reset_aux_dev(sc->kbdc)) {
+ recover_from_error(sc->kbdc);
+ if (verbose >= 2)
+ printf("psm%d: failed to reset the aux device (2).\n",
+ unit);
}
}
The PCI unit secondary assignment:
--- bus/pci/pcisupport.c 15 Jan 2004 20:35:06 -0000 1.8
+++ bus/pci/pcisupport.c 10 Feb 2004 07:26:06 -0000 1.9
@@ -785,7 +795,7 @@
secondary = pci_get_secondarybus(dev);
if (secondary) {
- child = device_add_child(dev, "pci", -1);
+ child = device_add_child(dev, "pci", secondary);
*(int*) device_get_softc(dev) = secondary;
return bus_generic_attach(dev);
} else
The pccard support (in i386/conf/GENERIC) (get rid of 'card', 'pcic*',
'pccard', 'cardbus', and 'cbb').
--- i386/conf/GENERIC 27 Jan 2004 18:26:43 -0000 1.8
+++ i386/conf/GENERIC 10 Feb 2004 07:55:47 -0000 1.9
@@ -171,9 +171,9 @@
device apm0 at nexus? disable flags 0x20 # Advanced Power Management
# PCCARD (PCMCIA) support
-device card
-device pcic0 at isa? irq 0 port 0x3e0 iomem 0xd0000
-device pcic1 at isa? irq 0 port 0x3e2 iomem 0xd4000 disable
+device pccard
+device cardbus
+device cbb
# Serial (COM) ports
device sio0 at isa? port IO_COM1 flags 0x10 irq 4
I also updated the token code but that would have crashed in a
different place, and I updated a few malloc() calls in the NFS
code which should not have been touched that early in the boot
sequence.
Make sure you do complete kernel builds (do not use -DNOCLEAN)
when testing.
-Matt
Matthew Dillon
<dillon at xxxxxxxxxxxxx>
More information about the Kernel
mailing list