Re: 2.6.39.1 immediately reboots/resets on EFI system

From: Maarten Lankhorst
Date: Mon Jun 13 2011 - 13:52:52 EST


Op 13-06-11 18:47, Matthew Garrett schreef:
> Ok, let's just try this in a somewhat more brute force way. Can you try
> this patch on top of current git, and give me the full dmesg and
> /proc/iomem ?
>
> diff --git a/arch/x86/platform/efi/efi.c b/arch/x86/platform/efi/efi.c
> index 0d3a4fa..ffbe47c 100644
> --- a/arch/x86/platform/efi/efi.c
> +++ b/arch/x86/platform/efi/efi.c
> @@ -312,11 +312,35 @@ void __init efi_reserve_boot_services(void)
> efi_memory_desc_t *md = p;
> unsigned long long start = md->phys_addr;
> unsigned long long size = md->num_pages << EFI_PAGE_SHIFT;
> + unsigned long long end = start + size;
>
> - if (md->type != EFI_BOOT_SERVICES_CODE &&
> - md->type != EFI_BOOT_SERVICES_DATA)
> + if (md->type != EFI_BOOT_SERVICES_CODE)
> continue;
>
> + /*
> + * Trying to map EFI code over the zero page is just asking
> + * for trouble. The good news is that it's almost certainly
> + * only there for graphics card bringup at POST time, so
> + * we should be able to get away with not having it there
> + */
> +
> + if (start == 0) {
> + md->num_pages = 0;
> + continue;
> + }
> +
> + /*
> + * We probably also want to avoid mapping code over the
> + * kernel...
> + */
> +
> + if ((start <= virt_to_phys(_etext)-1 && end >= virt_to_phys(_text)) ||
> + (start <= virt_to_phys(_edata)-1 && end >= virt_to_phys(_etext)) ||
_etext -> _sdata perhaps?
> + (start <= virt_to_phys(&__bss_stop)-1 && end >= virt_to_phys(&__bss_start))) {
> + md->num_pages = 0;
> + continue;
> + }
> +
> memblock_x86_reserve_range(start, start + size, "EFI Boot");
> }
> }
Will test though. It might also be appropriate to put a WARN when that condition happens.

~Maarten
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/