mp_machdep.c (was Re: [Fwd: Re: Bug reports requested - acpi])
rik at cronyx.ru
Wed Sep 22 00:56:07 PDT 2004
> John Baldwin wrote:
>> On Tuesday 21 September 2004 01:10 pm, Nate Lawson wrote:
>>> A quick "no" vote from me until this is really understood. I think the
>>> real problem is an interference between the pmap for the AP trampoline
>>> and the acpi wake code (sys/i386/acpica/acpi_wakeup.c). The address
>>> gave (0x9f000) is right before the base address we use for the wakeup
>>> code (0xa0000). As I woke up this morning, I was wondering if this
>>> could be the issue. An easy way to test is to disable the call to
>>> acpi_install_wakeup_handler() in sys/i386/acpica/acpi_machdep.c and see
>>> if this alone fixes the problem.
>>> If I'm wrong, feel free to commit your patch.
>>> P.S. Spaces instead of tabs in your diff.
>> Umm, 0xa0000 is the start of Video RAM, so I sure hope the ACPI wake
>> code doesn't try to write code into Video memory. The
>> pmap_invalidate_page is certainly needed.
> This call to alloc the memory seems to say 0xa0000 is the first valid
> start address:
> if (bus_dma_tag_create(/* parent */ NULL, /* alignment */ 2, 0,
> /* lowaddr below 1MB */ 0x9ffff,
> /* highaddr */ BUS_SPACE_MAXADDR, NULL, NULL,
> PAGE_SIZE, 1, PAGE_SIZE, 0, busdma_lock_mutex,
> &Giant, &acpi_waketag) != 0)
> In any case, you're right about needing to invalidate the mapping
> after creating it so feel free to commit. Do we also need this after
> pmap_enter() in acpi_sleep_machdep() (i386/acpica/acpi_wakeup.c)?
No, a quick view tells that pmap_enter() do this by it self.
In my case pmap_kenter ().
More information about the freebsd-current