Re: [PATCH for 4.7] xen: Replace alloc_vcpu_guest_context() with vmalloc()

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 21/08/15 18:55, Konrad Rzeszutek Wilk wrote:
> On Fri, Aug 21, 2015 at 06:51:46PM +0100, Andrew Cooper wrote:
>> This essentially reverts c/s 2037f2adb "x86: introduce
>> alloc_vcpu_guest_context()", including the newer arm bits, but achieves
>> the same end goal by using the newer vmalloc() infrastructure.
> Could you explain what this fixes? Or perhaps with an explanation
> of why this will make Xen [ ]better; [ ] faster [ ] magical.
>
> :-)

Ain't the diffstat enough to qualify for [x]better ;) ?

>
>
> Thanks.

It is relevant to a patch of Rogers which I am reviewing from the no-DM
series.  I was writing in reply to that, but this can also do.

alloc_vcpu_guest_context() was introduced long before vmalloc(), and
attempts to make the same end result using per-pcpu fixmap entries,
which scale by the compile-time NR_CPUS.

This patch causes a net reduction in compiled size for each arch (small
for ARM, larger for x86) and removes a scalability limit for compiling
with large numbers of cpus.

I think I can guess what you are going to ask me to do, given this
explanation.

~Andrew