Looking more closely at a 2.2.18 bootup tonight I see that apm stuff
appears in dmesg before dmi_scan does (I added "#define DUMP_DMI" in
dmi_scan.c).
If this implies that apm is initialised *before* the dmi_scan then there
is potentially a problem with buggy BIOSen that oops instead of reporting
power status, given:
* passing the boot-parameter apm=debug causes a power status report as apm
is initialised
* dmi_scan is identifying at least one machine that oopses when the power
status report BIOS call is invoked.
... so with a buggy BIOS and "apm=debug" it's vital that the dmi_scan is
completed *before* apm is initialised - which is not what I am seeing with
2.2.18 (nor can I see where the order of such things is set). The test
would be to boot a buggy Dell with "apm=debug" (and "#define DUMP_DMI" in
dmi_scan.c - but if my theory is right it won't get that far) and watch
for smoke leakage.
I think it's obvious that this would all go away if apm is a module (as it
can be with 2.4).
Regards,
Neale.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/