Description of problem:

After getting into troubles with 2.6.12-1.1372_FC3smp (it does not work
as described in bug 163407 and bug 163437) I tried to boot one of machines
with 2.6.12-1.1372_FC3.  This did boot, very slowly, but I got flooded with
errors like that:

    ACPI-0686: *** Error: acpi_ev_gpe_dispatch: No handler or method for
GPE[20], disabling event
    ACPI-0686: *** Error: acpi_ev_gpe_dispatch: No handler or method for
GPE[21], disabling event
.....

and all of this seems to slow down a machine to a crawl.

Looking at logs these errors start to be recorded immediately after syslogd
is active and they come is a rate about forty per minute.  Indices for "GPE"
contnuously cycle in 0x20-0x3F range.  An output from dmesg was useless
because even before /var/log/dmesg was written it was already overrun by
error lines like the above.

Attached are dmesg output for a working 2.6.11-1.35_FC3smp kernel, an
output from dmidecode for the board, an output for 'lspci -tv' and 
whatever is in /proc/acpi/dsdt.  An output from that does not differ
between 2.6.11-1.35_FC3smp and 2.6.12-1.1372_FC3 kernels.

I am afraid that my access to this machine is somewhat limited and pretty
soon for quite a while I will be not able to get to it at all.

Version-Release number of selected component (if applicable):
2.6.12-1.1372_FC3

Created attachment 116888 [details]
dmidecode data for the board in question

Created attachment 116889 [details]
'lscpi -tv ' output

Created attachment 116890 [details]
compressed output from 'cat /proc/acpi/dsdt

I should add that the same 2.6.12-1.1372_FC3 kernel on different boards does
not show any acpi problems; but none of other boards where I could try that
happens to be SMP.

I booted the machine in question with 2.6.12-1.1376_FC3smp (booting with
2.6.12-1.1372_FC3smp was not ever possible, even with an updated mkinitrd)
and so far it does not show any ill-effects.

An output from dmesg is very close to the one from 2.6.11-1.35_FC3smp.

ok, closing based on comment #6