Re: [Xen-devel] [PATCH RFC 34/35] arm : acpi workarounds for firmware/linux dependencies

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 06.02.15 at 15:38, <stefano.stabellini@xxxxxxxxxxxxx> wrote:
> On Fri, 6 Feb 2015, Parth Dixit wrote:
>> On 5 February 2015 at 23:18, Stefano Stabellini
>> <stefano.stabellini@xxxxxxxxxxxxx> wrote:
>> > On Wed, 4 Feb 2015, parth.dixit@xxxxxxxxxx wrote:
>> >> +#ifdef CONFIG_X86
>> >>  #define is_xmalloc_memory(ptr) ((unsigned long)(ptr) & (PAGE_SIZE - 1))
>> >> +#else
>> >> +#define is_xmalloc_memory(ptr) 1
>> >> +#endif
>> >
>> > Why?
>> this test is failing,leading to crash.
> 
> Actually I don't understand this check even on x86.
> 
> Jan, why are we assuming that xmalloc cannot return align pointers on
> x86?

Because all allocations coming through here are sub-page-size,
and such allocations are never providing page-aligned block of
memory. (I think it goes without saying that making this a
constant 1 on ARM is a no-go - if you have a need for larger
allocations, the macro still needs to distinguish both cases in
order to not break acpi_os_free_memory().)

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel