Re: [Xen-devel] [PATCH 5/8] xen/vmap: allow vmap() to be called during early boot

[David Woodhouse <dwmw2@xxxxxxxxxxxxx>]

On Tue, 2020-02-04 at 11:06 +0000, David Woodhouse wrote:
> 
> I still don't really get it. What if the zone boundary is at MFN 0x300?
> 
> What prevents the buddy allocator from merging a range a 0x200-0x2FF
> with another from 0x300-0x3FF, creating a single range 0x200-0x400
> which crosses nodes?
> 
> The MFN0 trick only works if all zone boundaries must be at an address
> which is 2ⁿ, doesn't it? Is that always true?


As yes, it is. And you called it out explicitly. Sorry.

I think I was conflating NUMA nodes, with the allocator zones.

On the plus side, this is a whole lot easier for the boot allocator to
cope with. It's easy enough to avoid handing out any high-order
allocations from the boot allocator which would cross a *zone*
boundary. It's crossing *node* boundaries which I was thinking of,
which would be harder before we've even parsed the SRAT...

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel