From: Andrew Cooper <andrew.cooper3@citrix.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: xen-devel <xen-devel@lists.xenproject.org>,
Dario Faggioli <dario.faggioli@citrix.com>,
Aravind Gopalakrishnan <aravind.gopalakrishnan@amd.com>,
suravee.suthikulpanit@amd.com
Subject: Re: [PATCH 5/5] AMD IOMMU: widen NUMA nodes to be allocated from
Date: Fri, 06 Mar 2015 12:15:31 +0000 [thread overview]
Message-ID: <54F99A63.6040200@citrix.com> (raw)
In-Reply-To: <54F96A440200007800066D57@mail.emea.novell.com>
On 06/03/2015 07:50, Jan Beulich wrote:
>>>> On 05.03.15 at 18:30, <andrew.cooper3@citrix.com> wrote:
>> On 26/02/15 13:56, Jan Beulich wrote:
>>> --- a/xen/include/asm-x86/hvm/svm/amd-iommu-proto.h
>>> +++ b/xen/include/asm-x86/hvm/svm/amd-iommu-proto.h
>>> @@ -158,12 +158,12 @@ static inline unsigned long region_to_pa
>>> return (PAGE_ALIGN(addr + size) - (addr & PAGE_MASK)) >> PAGE_SHIFT;
>>> }
>>>
>>> -static inline struct page_info* alloc_amd_iommu_pgtable(void)
>>> +static inline struct page_info *alloc_amd_iommu_pgtable(struct domain *d)
>>> {
>>> struct page_info *pg;
>>> void *vaddr;
>>>
>>> - pg = alloc_domheap_page(NULL, 0);
>>> + pg = alloc_domheap_page(d, MEMF_no_owner);
>> Same comment as with the VT-d side of things. This should be based on
>> the proximity information of the IOMMU, not of the owning domain.
> I think I buy this argument on the VT-d side (under the assumption
> that there's going to be at least one IOMMU per node), but I'm not
> sure here: The most modern AMD box I have has just a single
> IOMMU for 4 nodes it reports.
It is not possible for an IOMMU to cover multiple NUMA nodes worth of
IO, because of the position it has to sit relative to the IO root ports
and QPI/HT links.
In AMD systems, the IOMMUs lives in the northbridges, meaning one per
numa node (as it is the northbridges which contain the hypertransport links)
The BIOS/firmware will only report IOMMUs from northbridges which have
IO connected to their IO hypertransport link (most systems in the wild
have all IO hanging off one or two Numa nodes). On the other hand, I
have an AMD system with 8 IOMMUs in use.
In Intel systems, there is one IOMMU for each socket (to cover the
on-chip root ports and GPU if applicable) and one IOMMU in the IOH/PCH
(depending on generation) to cover the legacy IO.
In all cases, the IOMMUs are local to a single NUMA node, and would
benefit from having the control pages and pagetables allocated in local RAM.
~Andrew
next prev parent reply other threads:[~2015-03-06 12:15 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-02-26 13:44 [PATCH 0/5] (not just)x86/Dom0: NUMA related adjustments Jan Beulich
2015-02-26 13:52 ` [PATCH 1/5] x86: allow specifying the NUMA nodes Dom0 should run on Jan Beulich
2015-02-26 17:14 ` Dario Faggioli
2015-02-27 8:46 ` Jan Beulich
2015-02-27 10:04 ` Dario Faggioli
2015-02-27 10:50 ` Jan Beulich
2015-02-27 14:54 ` Dario Faggioli
2015-02-27 15:04 ` Jan Beulich
2015-03-03 10:51 ` Jan Beulich
2015-03-04 10:18 ` Dario Faggioli
2015-03-06 9:11 ` Jan Beulich
2015-03-06 10:46 ` Dario Faggioli
2015-03-06 11:33 ` Dario Faggioli
2015-03-06 13:26 ` Jan Beulich
2015-03-06 11:49 ` Jan Beulich
2015-03-03 9:59 ` Ian Campbell
2015-03-05 16:11 ` Andrew Cooper
2015-03-05 16:43 ` Jan Beulich
2015-03-05 17:27 ` Andrew Cooper
2015-03-06 9:19 ` [PATCH 1/5 v2] " Jan Beulich
2015-03-06 10:41 ` Dario Faggioli
2015-03-06 16:05 ` Andrew Cooper
2015-02-26 13:53 ` [PATCH 2/5] allow domain heap allocations to specify more than one NUMA node Jan Beulich
2015-02-27 11:34 ` Dario Faggioli
2015-03-02 17:12 ` Ian Campbell
2015-03-03 7:59 ` Jan Beulich
2015-03-05 16:18 ` Andrew Cooper
2015-02-26 13:54 ` [PATCH 3/5] x86: widen NUMA nodes to be allocated from Jan Beulich
2015-02-27 13:27 ` Dario Faggioli
2015-02-27 13:36 ` Jan Beulich
2015-02-27 14:11 ` Dario Faggioli
2015-02-27 13:38 ` Julien Grall
2015-02-27 13:55 ` Dario Faggioli
2015-02-27 13:58 ` Jan Beulich
2015-02-27 13:46 ` Ian Campbell
2015-02-27 14:00 ` Dario Faggioli
2015-02-27 14:03 ` Jan Beulich
2015-03-05 16:39 ` Andrew Cooper
2015-02-26 13:55 ` [PATCH 4/5] VT-d: " Jan Beulich
2015-03-05 17:08 ` Andrew Cooper
2015-03-09 3:07 ` Tian, Kevin
2015-02-26 13:56 ` [PATCH 5/5] AMD IOMMU: " Jan Beulich
2015-03-05 17:30 ` Andrew Cooper
2015-03-06 7:50 ` Jan Beulich
2015-03-06 12:15 ` Andrew Cooper [this message]
2015-03-09 15:42 ` Suravee Suthikulanit
2015-03-09 17:26 ` Andrew Cooper
2015-03-09 19:02 ` Suravee Suthikulanit
2015-03-10 7:35 ` Jan Beulich
2015-03-10 13:55 ` Boris Ostrovsky
2015-02-27 10:04 ` [PATCH 0/5] (not just)x86/Dom0: NUMA related adjustments Dario Faggioli
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=54F99A63.6040200@citrix.com \
--to=andrew.cooper3@citrix.com \
--cc=JBeulich@suse.com \
--cc=aravind.gopalakrishnan@amd.com \
--cc=dario.faggioli@citrix.com \
--cc=suravee.suthikulpanit@amd.com \
--cc=xen-devel@lists.xenproject.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.