Hi,
Here's an updated version of the mspec driver. I have tried to address
pretty much all the issues reported by Christoph. I left out most of
the comments, but specific details on a few items below.
Patch at the end - I hope it's ready for the next -mm release now.
Cheers,
Jes
> "Christo
On Tue, Apr 12, 2005 at 10:50:08AM -0400, Jes Sorensen wrote:
> +config MSPEC
> + tristate "Special Memory support"
> + select GENERIC_ALLOCATOR
should depend on IA64_GENERIC || IA64_SGI_SN2 because it's using sn2
functions like bte_copy
> +#define BTE_ZERO_BLOCK(_maddr, _len) \
> + b
> "Andrew" == Andrew Morton <[EMAIL PROTECTED]> writes:
Andrew> Does this compile without CONFIG_EFI?
Andrew> (It seems that ia64 Kconfig tries to turn on EFI always, but I
Andrew> thing allnoconfig will turn it off)
Nope it doesn't. I don't think the ia64 port can operate without EFI
at all
> +/*
> + * Walk the EFI memory map to pull out leftover pages in the lower
> + * memory regions which do not end up in the regular memory map and
> + * stick them into the uncached allocator
> + */
> +static void __init
> +mspec_walk_efi_memmap_uc (void)
> +{
> + void *efi_map_start, *efi_map_
[EMAIL PROTECTED] (Jes Sorensen) wrote:
>
> + getpage:
> +/*
> + * Is this really correct?
> + */
> +page = alloc_pages(GFP_USER, 0);
> +spin_unlock(&vdata->lock);
> +return page;
> +
sleeping allocation inside a spinlock.
-
To unsubscribe from this list: send the li
[EMAIL PROTECTED] (Jes Sorensen) wrote:
>
> + if (atomic_dec(&vdata->refcnt) == 0) {
atomic_dec() normally returns void. ia64's returns int, which is a bit
risky for cross-arch developemnt.
atomic_dec_and_test() would be more conventional.
-
To unsubscribe from this list: send the line "unsu
6 matches
Mail list logo
