On Fri, Jun 14, 2013 at 10:53:24PM +, Jeff Squyres (jsquyres) wrote:
> On Jun 12, 2013, at 5:47 PM, Jason Gunthorpe
> wrote:
>
> > Someone has to finish the ummunotify rewrite Roland
> > started. Realistically MPI is going to be the only user, can someone
> > from the MPI world do this?
>
>
On Jun 12, 2013, at 5:47 PM, Jason Gunthorpe
wrote:
> Someone has to finish the ummunotify rewrite Roland
> started. Realistically MPI is going to be the only user, can someone
> from the MPI world do this?
1. I tried to ask what needed to be done at the beginning of this thread and
didn't get
On Jun 12, 2013, at 5:17 PM, Jason Gunthorpe
wrote:
> Yes, it can, via MAP_FIXED. There are lots of fun tricks you can play
> using that.
You're missing the point.
Normal users (i.e., MPI users) don't do that. They call malloc() and they get
what they get.
The whole point of upper-layer AP
On Wed, Jun 12, 2013 at 09:18:34PM +, Jeff Squyres (jsquyres) wrote:
> > Well, it creates a mess in another sense, because now you've lost
> > context. When your MPI goes to do a 1byte send the kernel may well
> > prefetch a few megabytes of page tables, whereas an implementation in
> > usersp
On Jun 10, 2013, at 1:26 PM, Jason Gunthorpe
wrote:
>> I agree that pushing all registration issues out of the application
>> and (somewhere) into the verbs stack would be a nice solution.
>
> Well, it creates a mess in another sense, because now you've lost
> context. When your MPI goes to do
On Wed, Jun 12, 2013 at 09:10:57PM +, Jeff Squyres (jsquyres) wrote:
> > Another way to look at it is "specify IO access permissions" for
> > address space ranges. This could be useful to implement a buffer
> > pool to be used for a specific MR only, yet still map/unmap memory
> > within this
On Jun 10, 2013, at 11:56 AM, Liran Liss wrote:
>> "Register all address space" is the moral equivalent of not having userspace
>> registration, so let's talk about it in those terms. Specifically, there's
>> a subtle
>> difference between:
>>
>> a) telling verbs to register (0...2^64)
>> --
On Mon, Jun 10, 2013 at 02:49:24PM +, Jeff Squyres (jsquyres) wrote:
> On Jun 7, 2013, at 4:57 PM, Jason Gunthorpe
> wrote:
>
> >> We talked about this at the MPI Forum this week; it doesn't seem
> >> like ODP fixes any MPI problems.
> >
> > ODP without 'register all address space' changes
Raindel
> Subject: Re: Status of "ummunot" branch?
>
> On Jun 7, 2013, at 4:57 PM, Jason Gunthorpe
> wrote:
>
> >> We talked about this at the MPI Forum this week; it doesn't seem like
> >> ODP fixes any MPI problems.
> >
> > ODP without
On Jun 7, 2013, at 4:57 PM, Jason Gunthorpe
wrote:
>> We talked about this at the MPI Forum this week; it doesn't seem
>> like ODP fixes any MPI problems.
>
> ODP without 'register all address space' changes the nature of the
> problem, and fixes only one problem.
I agree that pushing all regi
gai Eran; Or Gerlitz; linux-rdma@vger.kernel.org; Shachar Raindel
Subject: Re: Status of "ummunot" branch?
On Fri, Jun 07, 2013 at 10:59:43PM +, Jeff Squyres (jsquyres) wrote:
> > I don't think this covers other memory regions, like those added via mmap,
> > right?
&
On Fri, Jun 07, 2013 at 10:59:43PM +, Jeff Squyres (jsquyres) wrote:
> > I don't think this covers other memory regions, like those added via mmap,
> > right?
>
> We talked about this at the MPI Forum this week; it doesn't seem
> like ODP fixes any MPI problems.
ODP without 'register all a
On Jun 6, 2013, at 4:33 PM, Jeff Squyres (jsquyres) wrote:
> I don't think this covers other memory regions, like those added via mmap,
> right?
We talked about this at the MPI Forum this week; it doesn't seem like ODP fixes
any MPI problems.
1. MPI still has to have a memory registration ca
On Jun 5, 2013, at 10:52 PM, Haggai Eran wrote:
>> Haggai: A verb to resize a registration would probably be a helpful
>> step. MPI could maintain one registration that covers the sbrk
>> region and one registration that covers the heap, much easier than
>> searching tables and things.
>
> That'
On 05/06/2013 22:05, Jason Gunthorpe wrote:
> On Wed, Jun 05, 2013 at 06:45:13PM +, Jeff Squyres (jsquyres) wrote:
>> What happens if you:
>>
>> a = malloc(N * page_size);
>> ibv_reg_mr(..., a, N * page_size, ...);
>> free(a);
>> // incoming RDMA arrives targeted at buffer a
>
> Haggai should
On Jun 5, 2013, at 12:05 PM, Jason Gunthorpe
wrote:
>> It does seem quite odd, abstractly speaking, that a registration
>> would survive a free/re-malloc (which is arguably a "different"
>> buffer).
>
> Not at all: the purpose of the registration is to allow access via
> RDMA to a portion of th
On Wed, Jun 05, 2013 at 06:45:13PM +, Jeff Squyres (jsquyres) wrote:
> Hum. I was under the impression that with today's code (i.e., not ODP), if
> you
>
> a = malloc(N);
> ibv_reg_mr(..., a, N, ...);
> free(a);
>
> (assuming that the memory actually left the process at free)
>
> Then the
On Jun 5, 2013, at 11:18 AM, Jason Gunthorpe
wrote:
>> Are you saying that the 2nd malloc will magically be registered
>> (with the new physical address)?
>
> Yes, that is the whole point.
Interesting.
> ODP fundamentally fixes the *bug* where the HCA's view of process
> memory can become inc
On Wed, Jun 05, 2013 at 06:10:11PM +, Jeff Squyres (jsquyres) wrote:
> On Jun 5, 2013, at 10:14 AM, Jason Gunthorpe
> wrote:
>
> >> a = malloc(x);// a gets (va=0x100, pa=0x12345) back from malloc
> >> MPI_Send(a, ...); // MPI registers 0x100 for len=x, and saves (0x100,x) in
> >> reg ca
On Jun 5, 2013, at 10:14 AM, Jason Gunthorpe
wrote:
>> a = malloc(x);// a gets (va=0x100, pa=0x12345) back from malloc
>> MPI_Send(a, ...); // MPI registers 0x100 for len=x, and saves (0x100,x) in
>> reg cache
>> free(a);
>> a = malloc(x);// a gets (va=0x100, pa=0x98765) back from mallo
On Wed, Jun 05, 2013 at 04:53:48PM +, Jeff Squyres (jsquyres) wrote:
> On Jun 5, 2013, at 6:39 AM, Haggai Eran wrote:
>
> > Perhaps I'm missing something, but I believe ODP deals with the first
> > two problems in the list (slide 8), even if it doesn't solve them
> > completely.
>
> Unfortun
On Jun 5, 2013, at 6:39 AM, Haggai Eran wrote:
> Perhaps I'm missing something, but I believe ODP deals with the first
> two problems in the list (slide 8), even if it doesn't solve them
> completely.
Unfortunately, it does not. If we could register(0 ... 2^64) and never have to
worry about re
On 05/06/2013 15:45, Jeff Squyres (jsquyres) wrote:
> On Jun 5, 2013, at 12:14 AM, Haggai Eran wrote:
>
>>> Hmm; I'm confused. How does this fix the
>>> MPI-needs-to-intercept-freed-memory problem?
>> Well, there is no problem if an application frees registered memory (in
>> an on-demand paging
On Jun 5, 2013, at 12:14 AM, Haggai Eran wrote:
>> Hmm; I'm confused. How does this fix the
>> MPI-needs-to-intercept-freed-memory problem?
> Well, there is no problem if an application frees registered memory (in
> an on-demand paging memory region) and that memory is returned to the
> OS. The
On 04/06/2013 23:13, Jeff Squyres (jsquyres) wrote:
> On Jun 4, 2013, at 4:50 AM, Haggai Eran wrote:
>
>>> Does this mean that an MPI implementation still has to register memory upon
>>> usage, and maintain its own registered memory cache?
>> Yes. However, since registration doesn't pin memory,
On 04/06/2013 20:04, Jason Gunthorpe wrote:
> Thus, I assume, on-demand allows pages that are 'absent' in the larger
> page table to generate faults to the CPU?
Yes, that's correct.
> So how does lifetime work here?
>
> - Can you populate the larger page table as soon as registration
>happen
On Jun 4, 2013, at 4:50 AM, Haggai Eran wrote:
>> Does this mean that an MPI implementation still has to register memory upon
>> usage, and maintain its own registered memory cache?
> Yes. However, since registration doesn't pin memory, you can leave
> registered memory regions in the cache for
On Tue, Jun 04, 2013 at 02:50:33PM +0300, Haggai Eran wrote:
> Our HCAs use their own page tables, in addition to a TLB cache. A miss
> in the TLB cache that can be filled from the HCA's page tables will not
> cause an RNR NAK, since the HCA can fill it relatively fast without the
> help of the op
On 04/06/2013 13:56, Jeff Squyres (jsquyres) wrote:
> On Jun 4, 2013, at 2:54 AM, Haggai Eran wrote:
>
>> We wish to get there eventually. In our current implementation you still
>> have to register an on-demand memory region explicitly. The difference
>> between a regular memory region is that t
On Jun 4, 2013, at 2:54 AM, Haggai Eran wrote:
> We wish to get there eventually. In our current implementation you still
> have to register an on-demand memory region explicitly. The difference
> between a regular memory region is that the pages in the region aren't
> pinned.
Does this mean tha
On 04/06/2013 11:37, Or Gerlitz wrote:
> On 04/06/2013 04:24, Jeff Squyres (jsquyres) wrote:
>> On May 29, 2013, at 1:53 AM, Or Gerlitz wrote:
>>
>>> Have you looked on ODP? see
>>> https://www.openfabrics.org/resources/document-downloads/presentations/doc_download/568-on-demand-paging-for-user-sp
On 04/06/2013 04:24, Jeff Squyres (jsquyres) wrote:
On May 29, 2013, at 1:53 AM, Or Gerlitz wrote:
Have you looked on ODP? see
https://www.openfabrics.org/resources/document-downloads/presentations/doc_download/568-on-demand-paging-for-user-space-networking.html
Is the idea behind ODP that,
On May 29, 2013, at 1:53 AM, Or Gerlitz wrote:
> Have you looked on ODP? see
> https://www.openfabrics.org/resources/document-downloads/presentations/doc_download/568-on-demand-paging-for-user-space-networking.html
Is the idea behind ODP that, at the beginning of time, you register the entire
On May 30, 2013, at 1:09 AM, Or Gerlitz wrote:
>> Has this been run by the MPI implementor community?
>
> The team that works on this here isn't ready for submission, so community
> runs were not made yet
If this is a solution to an MPI problem, it would seem like a good idea to run
the speci
On 30/05/2013 01:56, Jeff Squyres (jsquyres) wrote:
On May 29, 2013, at 4:53 AM, Or Gerlitz wrote:
Have you looked on ODP? see
https://www.openfabrics.org/resources/document-downloads/presentations/doc_download/568-on-demand-paging-for-user-space-networking.html
Is this upstream?
No
Has
On May 29, 2013, at 4:53 AM, Or Gerlitz wrote:
> Have you looked on ODP? see
> https://www.openfabrics.org/resources/document-downloads/presentations/doc_download/568-on-demand-paging-for-user-space-networking.html
Is this upstream?
Has this been run by the MPI implementor community?
The limi
On Tue, May 28, 2013 at 8:51 PM, Jeff Squyres (jsquyres)
wrote:
> I ask because, as an MPI guy, I would *love* to see this stuff integrated
> into the kernel and libibverbs.
Hi Jeff,
Have you looked on ODP? see
https://www.openfabrics.org/resources/document-downloads/presentations/doc_downlo
On May 28, 2013, at 1:52 PM, Roland Dreier wrote:
> Haven't touched it in quite a while except to keep it building. Needs
> work to finish up.
What kinds of things still need to be done? (I don't know if we could work on
this or not; just asking to scope out what would need to be done at this
On Tue, May 28, 2013 at 10:51 AM, Jeff Squyres (jsquyres)
wrote:
> I see a ummunot branch on your kernel tree at git.kernel.org
> (https://git.kernel.org/cgit/linux/kernel/git/roland/infiniband.git/log/?h=ummunot).
>
> Just curious -- what's the status of this tree? I ask because, as an MPI
> g
39 matches
Mail list logo