Comment 47 for bug 532374

Revision history for this message
Colin Ian King (colin-king) wrote :

Unfortunately, it looks like a BIOS issue to me at this point.

Looking at Tom's logs I get:

[1266874884.830945] ACPI Error: Hardware did not change modes (20090903/hwacpi-144)
[1266874884.830954] ACPI Error: Could not transition to ACPI mode (20090903/evxfevnt-93)

On my T410 I get the same. This error occurs when the kernel ACPI driver attempts to transition to ACPI mode
by writing ACPI_ENABLE to the SMI_CMD port but fails. This is dependant on the FADT containing the correct
ACPI_ENABLE and/or ACPI_DISABLE values - I've disassembled the FADT and they are 0xf0 and 0xf1 respectively
which means they are defined according to ACPI 2.0.

One possibility of the transition to ACPI mode failing is that hardware is taking more than the 3 seconds allowed or that
the FADT contains the wrong values, the latter is very unlikely.

It is noteworthy that the BIOS should have disabled all GP events by the time the transition occurs - if it hasn't transitioned
correctly, perhaps these are still enabled causing a huge number of IRQ's on line 9 which causes the later error message:

irq 9: nobody cared (try booting with the "irqpoll" option)

this seems to happen when the IRQ handler detects more than 99900 unhandled IRQs.

I will debug the mode transition code to see if I can tease a little more out of this bug, but it looks like a firmware problem at this stage.