Re: FreeBSD-8-BETA2 on MacBookPro5.5 (regression issue)

From: Eygene Ryabinkin <rea-fbsd_at_codelabs.ru>
Date: Wed, 22 Jul 2009 21:16:39 +0400
Wed, Jul 22, 2009 at 12:15:39PM -0400, Jung-uk Kim wrote:
> > The thing is that the IRQ for SCI at my MacBook is 9, so it may be
> > really related to the keyboard issues.  May be SCI interrupt is
> > posted after ACPI is enabled and this confuses something.  I'll try
> > to look at how APIC stuff is routed, but not today.
> 
> That's quite possible scenario.  FYI, all new nVidia chipset Mac's
> have almost identical ACPI tables.  The SMM port is at 0x52e while old
> Intel Macs were at 0xb2, AFAIK.  When we do AcpiEnable(), ACPICA
> checks if the SCI_EN bit is set.  If not, it tries to enable it via
> writing ACPI_ENABLE value to SMI_CMD port from FADT.  In this case,
> that effectively ends up doing:
> 
> bus_space_write_1(ACPI_BUS_SPACE_IO, ACPI_BUS_HANDLE, 0x52e, 0xf0);
> 
> According to HPS, this freezes the system as if it doesn't come back
> from the SMI handler.

That's true: I see this too.  SMI_CMD is really 0x52e and prior to
calling AcpiHwSetMode() ACPI mode is legacy.

Jung-uk, what is the rationale for enabling the SCI mode and to install
the interrupt handler only afterwards?  Will it break something if the
interrupt handler will be installed prior to going to the ACPI native
mode switch?
-- 
Eygene
 _                ___       _.--.   #
 \`.|\..----...-'`   `-._.-'_.-'`   #  Remember that it is hard
 /  ' `         ,       __.--'      #  to read the on-line manual
 )/' _/     \   `-_,   /            #  while single-stepping the kernel.
 `-'" `"\_  ,_.-;_.-\_ ',  fsc/as   #
     _.-'_./   {_.'   ; /           #    -- FreeBSD Developers handbook
    {_.-``-'         {_/            #
Received on Wed Jul 22 2009 - 15:16:43 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:52 UTC