[openib-general] [PATCH 14/14] IB/mthca: Bump version

It's about time for a version bump. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_dev.h |4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) --- linux.git.orig/drivers/infiniband/hw/mthca/mthca_dev.h 2005-06-23 13:03:09.283107

[openib-general] [PATCH 13/14] IB/mthca: Align FW command mailboxes to 4K

Future versions of Mellanox HCA firmware will require command mailboxes to be aligned to 4K. Support this by using a pci_pool to allocate all mailboxes. This has the added benefit of shrinking the source and text of mthca. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers

[openib-general] [PATCH 12/14] IB/mthca: Encapsulate command interface init

Encapsulate mthca command interface initialization/cleanup. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_cmd.c | 21 + linux.git/drivers/infiniband/hw/mthca/mthca_cmd.h |2 ++ linux.git/drivers/infiniband/hw/mthca/m

[openib-general] [PATCH 11/14] IB/mthca: Fix memory leak on error path

Free page_list buffer on error path of mthca_reg_phys_mr(). Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_provider.c |1 + 1 files changed, 1 insertion(+) --- linux.git.orig/drivers/infiniband/hw/mthca/mthca_provider.c 2005-06-23 13:03:

[openib-general] [PATCH 10/14] IB/mthca: Split off MTT allocation

Split allocation of MTT range from creation of MR. This will be useful for implementing shared memory regions and userspace verbs. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_dev.h |6 linux.git/drivers/infiniband/hw/mthca/mthca_mr

[openib-general] [PATCH 09/14] IB/mthca: Move mthca_is_memfree checks

Make mthca_table_put() and mthca_table_put_range() NOPs if the device is not mem-free, so that we don't have to have "if (mthca_is_memfree())" tests in the callers of these functions. This makes our code more readable and maintainable, and saves a couple dozen bytes of text in ib_mthca.ko as well.

[openib-general] [PATCH 08/14] IB/mthca: Fix memset size

Fix memset to use sizeof *props instead of just sizeof props. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_provider.c |2 +- 1 files changed, 1 insertion(+), 1 deletion(-) --- linux.git.orig/drivers/infiniband/hw/mthca/mthca_provider.c

[openib-general] [PATCH 06/14] IB/mthca: Set RDMA/atomic capabilities correctly

mthca apparently had the meanings of the max_rd_atomic and max_dest_rd_atomic QP attributes backwards. max_rd_atomic limits the maximum number of outstanding RDMA/atomic requests as an initiator (on a send queue), and max_dest_rd_atomic specifies the resources allocated to handle RMDA/atomic reque

[openib-general] [PATCH 07/14] IB/mthca: Enable unreliable connected transport

Add support for unreliable connected (UC) transport to mthca driver: - Add attributes for UC to modify QP table. - Add support for posting UC work requests. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_qp.c | 79 ++

[openib-general] [PATCH 04/14] IB/mthca: Use dma_alloc_coherent instead of pci_alloc_consistent

Switch all allocations of coherent memory from pci_alloc_consistent() to dma_alloc_coherent(), so that we can pass GFP_KERNEL. This should help when the system is low on memory. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_cq.c | 25 ++

[openib-general] [PATCH 05/14] IB/mthca: Set QP static rate correctly

Fix offset of static_rate in QP context. Pointed out by Dror Goldenberg. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_qp.c |2 +- 1 files changed, 1 insertion(+), 1 deletion(-) --- linux.git.orig/drivers/infiniband/hw/mthca/mthca_qp.c

[openib-general] [PATCH 01/14] IB/mthca: Add Sun copyright notice

From: Tom Duffy <[EMAIL PROTECTED]> Add Sun copyright to files modified by Tom Duffy. Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_av.c |1 + linux.git/drivers/infiniband/hw/mthca/mthca_c

[openib-general] [PATCH 00/14] IB/mthca: merge

Here is a series of patches to the mthca Mellanox HCA driver. Some of the patches are largish but there shouldn't be anything too major here. This is mostly to bring my tree closer to the main kernel and make the merge of direct userspace access smaller, since it will require plenty of review. T

[openib-general] [PATCH 03/14] IB/mthca: Clean up CQ debug

Clean up CQ debugging code: make dump_cqe print on one line, and only dump error CQ entries for local operation errors. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_cq.c | 39 +++-- 1 files changed, 18 insertions

[openib-general] [PATCH 02/14] IB/mthca: Clean up error messages

From: Bernhard Fischer <[EMAIL PROTECTED]> - Fix incorrect cut-n-paste in error messages. - Add missing newlines in error messages. - Use DRV_NAME instead of "ib_mthca" in a couple of places. Signed-off-by: Roland Dreier <[EMAIL PROTECTED]> --- linux.git/drivers/infiniband/hw/mthca/mthca_eq.c

[openib-general] Re: [PATCH] rdma_lat-09 and results

Roland> It seems like the best would be to have libmthca.a in Roland> $(libdir) and mthca.so in $(libdir)/infiniband. That way Roland> we make it easy to do static linking and avoid having two Roland> names for the same object. Michael> OK, but I dont know how to do that in au

Re: [openib-general] Asynchronous Event notification and registration

Kevin> People need to excuse all my annoying startup questions. I Kevin> just notices that the user space ib verbs lib doesn't have Kevin> any functions to register completion or event handlers. Is Kevin> the intent that the API will only support polling or are Kevin> the asyn

[openib-general] Re: [PATCH] kDAPL: remove dat_os_realloc()

What do you think about doing it this way? (see attached) On Thu, 23 Jun 2005, Tom Duffy wrote: tduffy> There was only one user of the dat_os_realloc() function. Roll it into tduffy> the base function. tduffy> tduffy> Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> tduffy> tduffy> Index: linux-k

[openib-general] Asynchronous Event notification and registration

People need to excuse all my annoying startup questions. I just notices that the user space ib verbs lib doesn't have any functions to register completion or event handlers. Is the intent that the API will only support polling or are the asynchronous event handling stuff in somebodies branch

[openib-general] Re: [PATCH] kDAPL: cstyle dat-provider/dapl_ring_buffer_util.c

Committed in revision 2705. On Thu, 23 Jun 2005, Tom Duffy wrote: tduffy> Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> tduffy> tduffy> Index: linux-kernel-work/dat-provider/dapl_ring_buffer_util.c tduffy> === tduffy> --- linux-kern

Re: [openib-general] [PATCH] kDAPL: remove dapl_os_realloc()

On 6/23/05, Tom Duffy <[EMAIL PROTECTED]> wrote: > There was only one user of the dat_os_realloc() function. Oops. That should of course read *dapl*_os_realloc(). -tduffy ___ openib-general mailing list openib-general@openib.org http://openib.org/mailm

[openib-general] [PATCH] kDAPL: remove dat_os_realloc()

There was only one user of the dat_os_realloc() function. Roll it into the base function. Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> Index: linux-kernel/dat-provider/dapl_evd.c === --- linux-kernel/dat-provider/dapl_evd.c(

[openib-general] [PATCH] kDAPL: cstyle dat-provider/dapl_ring_buffer_util.c

Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> Index: linux-kernel-work/dat-provider/dapl_ring_buffer_util.c === --- linux-kernel-work/dat-provider/dapl_ring_buffer_util.c (revision 2704) +++ linux-kernel-work/dat-provider/dapl_rin

Re: [openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

At 05:09 PM 6/23/2005, Grant Grundler wrote: >On Thu, Jun 23, 2005 at 04:55:38PM -0400, James Lentini wrote: >> My argument in favor of retaining them is that dapl_evd_wc_to_event() >> will crash if the cookie NULL. A BUG_ON will detect this situation ... >The tombstone from the data page fault p

[openib-general] Re: [PATCH] rdma_lat-09 and results

Quoting r. Ronald G. Minnich : > Subject: Re: [PATCH] rdma_lat-09 and results > > > > On Fri, 24 Jun 2005, Michael S. Tsirkin wrote: > > > I had this impression that I can have a .so not being present on the > > slave at boot, and then dlopen could pull it across the network with > > some custo

[openib-general] Re: [PATCH] rdma_lat-09 and results

On Fri, 24 Jun 2005, Michael S. Tsirkin wrote: > So, if you want to run without nfs (or such), you basically need to link > the applications statically, is that right? the .so files you want have to be in /lib on the node, e.g. [EMAIL PROTECTED] ~]$ bpsh 0 ls /lib ld-2.3.3.so ld-linux.so.2 libd

[openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

Committed in revision 2704. On Thu, 23 Jun 2005, Tom Duffy wrote: tduffy> This patch removes dapl_os_assert(). In most cases, replacing with tduffy> BUG_ON(!). Some cases, I just called panic() where others I removed the tduffy> assert all together because the next line dereferences the pointe

[openib-general] Re: [PATCH] rdma_lat-09 and results

On Fri, 24 Jun 2005, Michael S. Tsirkin wrote: > I had this impression that I can have a .so not being present on the slave > at boot, and then dlopen could pull it across the network with some > custom protocol without going over NFS. > > And I was asking, if so, what other calls can do this b

[openib-general] Re: [PATCH] rdma_lat-09 and results

On Fri, 24 Jun 2005, Michael S. Tsirkin wrote: > I had this impression that I can have a .so not being present on the > slave at boot, and then dlopen could pull it across the network with > some custom protocol without going over NFS. not at present, at least on bproc. dlopen needs a path name

Re: [openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

Gotcha. I missed that. On Thu, 23 Jun 2005, Tom Duffy wrote: On Thu, 2005-06-23 at 16:55 -0400, James Lentini wrote: Why did you remove the asserts in dapl_ib_post_recv() and dapl_ib_post_send()? The code continues 6 lines down: < assert if cookie is null here > sg_list = ep->recv_

[openib-general] Re: [PATCH] rdma_lat-09 and results

Quoting r. Ronald G. Minnich : > Subject: Re: [PATCH] rdma_lat-09 and results > > > > On Thu, 23 Jun 2005, Michael S. Tsirkin wrote: > > > I know bproc supports dlopen from remote applications, but > > what about glob() on a remote directory? Does that work from a blade? > > I'm not totally s

Re: [openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

On Thu, 2005-06-23 at 16:55 -0400, James Lentini wrote: > Why did you remove the asserts in dapl_ib_post_recv() and > dapl_ib_post_send()? The code continues 6 lines down: < assert if cookie is null here > sg_list = ep->recv_iov; for (i = 0; i < num_segments; i++, sg_list++) {

[openib-general] Re: [PATCH] rdma_lat-09 and results

On Thu, 23 Jun 2005, Michael S. Tsirkin wrote: > I know bproc supports dlopen from remote applications, but > what about glob() on a remote directory? Does that work from a blade? I'm not totally sure what you're saying here but: if you have a .so and you need it on a bproc slave node, you c

Re: [openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

On Thu, Jun 23, 2005 at 04:55:38PM -0400, James Lentini wrote: > My argument in favor of retaining them is that dapl_evd_wc_to_event() > will crash if the cookie NULL. A BUG_ON will detect this situation > sooner rather than later and therefore make the problem easier to > diagnose. Did I miss s

[openib-general] Re: [PATCH] kDAPL: remove dapl_os_assert()

Why did you remove the asserts in dapl_ib_post_recv() and dapl_ib_post_send()? My argument in favor of retaining them is that dapl_evd_wc_to_event() will crash if the cookie NULL. A BUG_ON will detect this situation sooner rather than later and therefore make the problem easier to diagnose.

[openib-general] Re: [PATCH] rdma_lat-09 and results

Quoting r. Roland Dreier <[EMAIL PROTECTED]>: > Subject: Re: [PATCH] rdma_lat-09 and results > > Michael> At least for this usage, I think it would be best to > Michael> rename mthca.a (and mthca.so) to libmthca.a (and > Michael> libmthca.so), respectively, and put them in libpath, th

Re: [openib-general] NFS/RDMA/kDAPL

At 04:11 PM 6/23/2005, Tom Duffy wrote: >I am new to NFSoRDMA. Where do I start? Is there a good doc about >how to set it up? Consult your OpenSolaris documentation? :-) Seriously, there isn't a cookbook for Linux, because we're not there yet. But it's not too hard. I will plan to roll up the u

[openib-general] RE: [PATCH][kdapl] Integrate dapl_hca_alloc/dapl_hca_free to dapl _provider.c

hi James, I think that you are worng. There is no functionalty to this struct. it is only a struct that hold some data together. if we look into the files dapl_hca_util.[h.c] we will find 4 functions: 1) dapl_hca_alloc -> which is kmalloc and set to 4 data members of this struct 2) dapl_hca_fr

Re: [openib-general] Re: [PATCH][kdapl] Integrate dapl_hca_alloc/dapl_hca_free to dapl_provider.c

One more thing, if we decide to keep these files, I will rename them dapl_hca.[ch] for consistency. On Thu, 23 Jun 2005, James Lentini wrote: For consistency, I think we should keep the HCA object in its own file. However, I'm not sure we need an HCA object. Is there a better way to organi

[openib-general] Re: [PATCH][kdapl] Integrate dapl_hca_alloc/dapl_hca_free to dapl_provider.c

For consistency, I think we should keep the HCA object in its own file. However, I'm not sure we need an HCA object. Is there a better way to organize the data being stored in the dapl_hca structure? I have this feeling that we should merge all the structures that we create on a per-hca basi

[openib-general] [PATCH] kDAPL: remove dapl_os_assert()

This patch removes dapl_os_assert(). In most cases, replacing with BUG_ON(!). Some cases, I just called panic() where others I removed the assert all together because the next line dereferences the pointer anyways. Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> Index: linux-kernel/dat-provider/da

[openib-general] Re: [PATCH][kdapl] Integrate dapl_hca_link_ia/dapl_hca_unlink_ia into dapl_ia.c

Do we even need to link each IA onto an HCA? I don't see where in the code we use an HCA's ia_list. Is this just dead code? On Mon, 20 Jun 2005, Itamar Rabenstein wrote: itamar> Integrate dapl_hca_link_ia/dapl_hca_unlink_ia into dapl_ia.c itamar> (no need for functions that just call LIST_ADD a

[openib-general] Re: [PATCH] kDAPL: add a couple more static

Committed in revision 2703. On Thu, 23 Jun 2005, Tom Duffy wrote: tduffy> Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> tduffy> tduffy> Index: linux-kernel-static/dat-provider/dapl_cr.c tduffy> === tduffy> --- linux-kernel-static/da

Re: [openib-general] A new simple ulp (SPTS)

> Brief intro to NFS/RDMA work: > > Current client version is on Sourceforge, supporting various flavors of > 2.4. I'm preparing a new release for 2.6. > > > The client needs a patched version of Sunrpc in order to hook in as > an NFS transport. This pat

Re: [openib-general] NFS/RDMA/kDAPL

On 6/23/05, Talpey, Thomas <[EMAIL PROTECTED]> wrote: > Can the Linux OpenIB client connect to Solaris 10? If so, we might > consider using Sol10's NFS/RDMA server. If not, we'll have to use a > NetApp filer (which is fine by me but maybe hard for you), because > the CITI NFS/RDMA server is accepti

[openib-general] SDP: still getting sk_alloc() panic, any ideas?

I am still getting the panic when you try to connect to a machine and it is not listening (but has ib_sdp loaded): [EMAIL PROTECTED] ~]# --- [cut here ] - [please bite here ] - Kernel BUG at "/build1/tduffy/openib-work/linux-2.6.12-openib/in:352 invalid operand: [1] SM

[openib-general] [PATCH] kDAPL: add a couple more static

Signed-off-by: Tom Duffy <[EMAIL PROTECTED]> Index: linux-kernel-static/dat-provider/dapl_cr.c === --- linux-kernel-static/dat-provider/dapl_cr.c (revision 2695) +++ linux-kernel-static/dat-provider/dapl_cr.c (working copy) @@ -41,7

Re: [openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

Grant> If so, then gcc or the code gen backend should generate a Grant> warning about overflowing the encoding. This sounds like a Grant> toolchain bug. gcc does generate a warning for constant shifts that are too big, as in: bit = 1ull << 64; No diagnostic is generated when

[openib-general] Re: NFS/RDMA/kDAPL

On Thu, 2005-06-23 at 15:22 -0400, Talpey, Thomas wrote: > Can the Linux OpenIB client connect to Solaris 10? Yes. IPoIB works now. You can connect a Solaris 10 client to an OpenSM managed network. > If so, we might > consider using Sol10's NFS/RDMA server. OpenSolaris would be more accessible

[openib-general] NFS/RDMA/kDAPL

At 03:11 PM 6/23/2005, Tom Duffy wrote: >On Thu, 2005-06-23 at 13:54 -0400, Talpey, Thomas wrote: >> We look forward to running NFS/RDMA over OpenIB, when its kDAPL >> is ready. > >Let's get NFSoRDMA going sooner rather than later on James's kDAPL. I >think this will both be a good test case as we

Re: [openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

On Thu, 2005-06-23 at 12:41, Roland Dreier wrote: > Roland> It's odd that you saw the expession end up as 0. It seems > Roland> that on every system I have, 1ull << 64 is 0, and 0 - 1 == > Roland> -1 == 0x. So although it isn't correct C, > Roland> it should work.

Re: [openib-general] A new simple ulp (SPTS)

On Thu, 2005-06-23 at 13:54 -0400, Talpey, Thomas wrote: > We look forward to running NFS/RDMA over OpenIB, when its kDAPL > is ready. Let's get NFSoRDMA going sooner rather than later on James's kDAPL. I think this will both be a good test case as well as a vehicle to demonstrate the functionali

[openib-general] Re: Setting kdapl ep_state relative to calling disconnect

We should make them consistent. The state should be updated before the disconnect is sent. On Thu, 23 Jun 2005, Hal Rosenstock wrote: halr> Hi, halr> halr> In dapl_evd.c, there are cases where the ep_state is updated before halr> calling disconnect and others where this is done after. Should t

Re: [openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

On Thu, Jun 23, 2005 at 07:45:55PM +0300, Michael S. Tsirkin wrote: > It seems that compiler generates a single instruction for this shift. If so, then gcc or the code gen backend should generate a warning about overflowing the encoding. This sounds like a toolchain bug. > And CPU designers appa

RE: [openib-general] Re: [PATCH] cleanup dat provider registratio n

Making it a pointer would save memory, but it would open up some difficult questions. If it is a pointer, how is the underlying memory managed? This structure is returned by dat_registry_list_providers(). When would the registry be allowed to free this data? If we leave it as is, we will a

Re: [openib-general] [PATCH] kDAPL: use Grant's suggestions

Committed in revision 2691. On Thu, 23 Jun 2005, Tom Duffy wrote: tduffy> On Wed, 2005-06-22 at 23:20 -0700, Grant Grundler wrote: tduffy> > On Wed, Jun 22, 2005 at 10:17:09AM -0700, Tom Duffy wrote: tduffy> > > This patch removes the function dapl_os_panic() in favor of calling tduffy> > > pani

Re: [openib-general] RDMA write with immediate

On Thu, 2005-06-23 at 08:58 -0700, Roland Dreier wrote: > Josh> I get a completion with failed status 13. > > Josh> Again, this works just fine with IBV_WR_RDMA_WRITE. I'm > Josh> using 2.6.11.11 with SVN rev 2606. I'll update stuff and > Josh> try again. > > status 13 is "RNR r

Re: [openib-general] A new simple ulp (SPTS)

At 01:11 PM 6/23/2005, Jeff Carr wrote: >I didn't know there was a nfs/rmda module? > >[EMAIL PROTECTED]:/test/gen2# find . |grep -i nfs >[EMAIL PROTECTED]:/test/gen2# Brief intro to NFS/RDMA work: Current client version is on Sourceforge, supporting various flavors of 2.4. I'm preparing a new re

Re: [openib-general] RDMA write with immediate

On Thu, 2005-06-23 at 08:56 -0400, Hal Rosenstock wrote: > On Thu, 2005-06-23 at 08:47, Josh England wrote: > > Alright, I'm probably being retarded, but just in case: > > uverbs doesn't seem to work anymore. Simple pingpong uverbs test gives: > > > > Couldn't get context for mthca0 > > > > This

RE: [openib-general] mapping between IP address and device name

> From: Roland Dreier [mailto:[EMAIL PROTECTED] > Sent: Thursday, June 23, 2005 10:32 AM > > James> Perhaps a bit of motivation of how the GID->IP service can > James> be used is in order. > > James> kDAPL uses this feature to provide the passive side of a > James> connection with

[openib-general] [PATCH] kDAPL: use Grant's suggestions

On Wed, 2005-06-22 at 23:20 -0700, Grant Grundler wrote: > On Wed, Jun 22, 2005 at 10:17:09AM -0700, Tom Duffy wrote: > > This patch removes the function dapl_os_panic() in favor of calling > > panic() directly. > > just some nits... > > > if (!evd) > > - dapl_os_panic("NULL == cont

Re: [openib-general] mapping between IP address and device name

James> Perhaps a bit of motivation of how the GID->IP service can James> be used is in order. James> kDAPL uses this feature to provide the passive side of a James> connection with the IP address of the remote peer. kDAPL James> consumers can use this information as a weak auth

[openib-general] Re: [PATCH] ib_verbs.h ib_send_wr imm_data endianness

Quoting r. Roland Dreier <[EMAIL PROTECTED]>: > Subject: Re: [PATCH] ib_verbs.h ib_send_wr imm_data endianness > > Michael> Mark imm_data as big endian in ib_send_wr, same as we do > Michael> in ib_wc. > > Thanks, applied. > > Have you tried using -Wbitwise when building drivers/infiniba

[openib-general] Re: [PATCH] rdma_lat-09 and results

Michael> At least for this usage, I think it would be best to Michael> rename mthca.a (and mthca.so) to libmthca.a (and Michael> libmthca.so), respectively, and put them in libpath, then Michael> softlink libmthca.so to infiniband directory. Michael> This avoids the need for su

Re: [openib-general] mapping between IP address and device name

On Thu, 23 Jun 2005, Hal Rosenstock wrote: On Thu, 2005-06-23 at 11:47, Roland Dreier wrote: James> The IBAT API provides the two key services needed by kDAPL: James> the ability to obtain a route based on an IP address and James> the ability to map a GID to an IP address. Ja

Re: [openib-general] A new simple ulp (SPTS)

On 06/23/2005 09:52 AM, Talpey, Thomas wrote: > At 12:43 PM 6/22/2005, Jeff Carr wrote: > >>On 06/21/2005 12:50 PM, Roland Dreier wrote: >> >> >>>What happens if you try replacing the send_flags line with the one you >>>have commented out? >>> >>>+// send_wr.send_flags = IB_SEND_SIGNALED; >> >

[openib-general] Re: [PATCH] ib_verbs.h ib_send_wr imm_data endianness

Michael> Mark imm_data as big endian in ib_send_wr, same as we do Michael> in ib_wc. Thanks, applied. Have you tried using -Wbitwise when building drivers/infiniband? How bad is it? - R. ___ openib-general mailing list openib-general@openib.o

Re: [openib-general] A new simple ulp (SPTS)

At 12:43 PM 6/22/2005, Jeff Carr wrote: >On 06/21/2005 12:50 PM, Roland Dreier wrote: > >> What happens if you try replacing the send_flags line with the one you >> have commented out? >> >> +// send_wr.send_flags = IB_SEND_SIGNALED; > >Thanks, you are correct. IB_SEND_SIGNALED gives me the be

[openib-general] Re: [PATCH][kdapl] remove NULL check for input params

Committed in revision 2688 (minus the error Tom Duffy pointed out). On Wed, 22 Jun 2005, Itamar Rabenstein wrote: itamar> Apply This patch on top of the the previous patch set (5 patches) itamar> itamar> dapl_cr.c | 23 -- itamar> dapl_ep.c | 134

[openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

Quoting r. Roland Dreier <[EMAIL PROTECTED]>: > It's odd that you saw the expession end up as 0. It seems that on every > system I have, 1ull << 64 is 0, and 0 - 1 == -1 == 0x. > So although it isn't correct C, it should work. It seems that compiler generates a single instruction

[openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

On Thu, 2005-06-23 at 12:26, Roland Dreier wrote: > Hal> It fixes 64 bit sizes (at least on 32 bit machines). The > Hal> problem was with the mask computation when size_bits = 64. > > Hal> mask = cpu_to_be64(((1ull << desc[i].size_bits) - 1) << shift); > > Hal> yielded a mask of 0

Re: [openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

Roland> It's odd that you saw the expession end up as 0. It seems Roland> that on every system I have, 1ull << 64 is 0, and 0 - 1 == Roland> -1 == 0x. So although it isn't correct C, Roland> it should work. Oh, I see. A constant left shift of 64 is evaluated to 0

[openib-general] Re: [PATCH] rdma_lat-09 and results

Quoting r. Roland Dreier <[EMAIL PROTECTED]>: > Subject: Re: [PATCH] rdma_lat-09 and results > > Michael> Currently I want to be able to script linking mthca.a > Michael> statically. At least for this usage, I think it would be best to rename mthca.a (and mthca.so) to libmthca.a (and lib

[openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

Hal> It fixes 64 bit sizes (at least on 32 bit machines). The Hal> problem was with the mask computation when size_bits = 64. Hal> mask = cpu_to_be64(((1ull << desc[i].size_bits) - 1) << shift); Hal> yielded a mask of 0 (for size_bits 64 and shift of 0). The 32 Hal> bit mask c

[openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

On Thu, 2005-06-23 at 12:04, Roland Dreier wrote: > Hal> Fix ib_pack/unpack for 64 bits > > Can you be more specific about what this is fixing? It fixes 64 bit sizes (at least on 32 bit machines). The problem was with the mask computation when size_bits = 64. mask = cpu_to_be64(((1ull << des

Re: [openib-general] Re: page allocation failure with kdapl/mthca

On Thu, Jun 23, 2005 at 08:51:46AM -0700, Roland Dreier wrote: > Grant> On ia64 I expect both functions to land in the same bit of > Grant> code. For ZX1 platforms, that is sba_alloc_coherent(). > Grant> See arch/ia64/hp/common/sba_iommu.c since the behavior is > Grant> different d

[openib-general] Re: [PATCH] rdma_lat-09 and results

Michael> Currently I want to be able to script linking mthca.a Michael> statically. Michael> In the future I can imagine someone might want to script Michael> installing additional plugins into an existing system. I'm not sure this makes sense. libibverbs has a certain default pa

Re: [openib-general] mapping between IP address and device name

On Thu, 2005-06-23 at 11:47, Roland Dreier wrote: > James> The IBAT API provides the two key services needed by kDAPL: > James> the ability to obtain a route based on an IP address and > James> the ability to map a GID to an IP address. > > James> Is there agreement that an IB addr

[openib-general] Re: [PATCH] Fix ib_pack/unpack for 64 bits

Hal> Fix ib_pack/unpack for 64 bits Can you be more specific about what this is fixing? - R. ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org

Re: [openib-general] RDMA write with immediate

Josh> I get a completion with failed status 13. Josh> Again, this works just fine with IBV_WR_RDMA_WRITE. I'm Josh> using 2.6.11.11 with SVN rev 2606. I'll update stuff and Josh> try again. status 13 is "RNR retries exceeded." This means that the remote side did not have a rece

Re: [openib-general] Re: page allocation failure with kdapl/mthca

On Thu, Jun 23, 2005 at 06:45:21AM -0400, Hal Rosenstock wrote: > > On ia64 I expect both functions to land in the same bit of code. > > For ZX1 platforms, that is sba_alloc_coherent(). > > See arch/ia64/hp/common/sba_iommu.c since the behavior is > > different depending on CONFIG_NUMA setting. > >

Re: [openib-general] RDMA write with immediate

> # ll /dev/infiniband > crw-r--r--1 root root 231, 255 Jun 23 05:31 ucm > crw-rw-rw-1 root root 231, 128 Jun 23 05:31 uverbs0 > crw-rw-rw-1 root root 231, 129 Jun 23 05:31 uverbs1 If you're using a static /dev directory, you'll need to remake th

Re: [openib-general] Re: page allocation failure with kdapl/mthca

Roland> I just checked in a change that converts from using Roland> pci_alloc_consistent(), which is always GFP_ATOMIC, to Roland> dma_alloc_coherent(..., GFP_KERNEL). Grant> This arch specific. Grant> On ia64 I expect both functions to land in the same bit of Grant> code.

Re: [openib-general] mapping between IP address and device name

James> The IBAT API provides the two key services needed by kDAPL: James> the ability to obtain a route based on an IP address and James> the ability to map a GID to an IP address. James> Is there agreement that an IB address translation service James> must provide these two se

[openib-general] [PATCH] Fix ib_pack/unpack for 64 bits

Fix ib_pack/unpack for 64 bits Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]> Index: packer.c === -- packer.c (revision 2681) +++ packer.c(working copy) @@ -1,5 +1,6 @@ /* * Copyright (c) 2004 Topspin Corporation. All

[openib-general] Setting kdapl ep_state relative to calling disconnect

Hi, In dapl_evd.c, there are cases where the ep_state is updated before calling disconnect and others where this is done after. Should these be made consistent ? void dapl_evd_connection_callback(...) { ... case DAT_CONNECTION_EVENT_DISCONNECTED: ep->param.ep_state = DAT_E

[openib-general] Re: [PATCH] [kDAPL} Fix locking problem in some CM callback functions

Committed in revision 2686. On Wed, 22 Jun 2005, Hal Rosenstock wrote: halr> Fix locking problem in some CM callback functions halr> Since modify QP can sleep, dapl_ib_disconnect must be called without halr> holding spinlock. halr> halr> Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]> halr

Re: [openib-general] mapping between IP address and device name

On Tue, 21 Jun 2005, Hal Rosenstock wrote: On Mon, 2005-06-20 at 23:40, Roland Dreier wrote: There was some discussion a while ago about an "IB address translation" service proposed by the Voltaire crew, which would encapsulate some of this. However, we didn't make much progress towards a goo

RE: [openib-general] kdapl locking problem

In this instance, the code path getting us in trouble is: ib_modify_qp() calls mthca_modify_qp() calls mthca_MODIFY_QP() calls mthca_cmd_box() calls mthca_cmd_wait() calls wait_for_completion() calls schedule() but there are numerous code paths t

RE: [openib-general] Re: [PATCH] cleanup dat provider registratio n

i have a question: after we have cleaned up the dat_provider_info struct and now it is define as : struct dat_provider_info { char ia_name[DAT_NAME_MAX_LENGTH]; }; why cant we replace it with a simple char * it is only a string that the only thing we do with it is to store it and compar

Re: [openib-general] RDMA write with immediate

On Thu, 2005-06-23 at 08:47, Josh England wrote: > Alright, I'm probably being retarded, but just in case: > uverbs doesn't seem to work anymore. Simple pingpong uverbs test gives: > > Couldn't get context for mthca0 > > This is with rev 2685 on 2.6.12.1 under RHEL3. Fresh rebuild/install of >

Re: [openib-general] RDMA write with immediate

Alright, I'm probably being retarded, but just in case: uverbs doesn't seem to work anymore. Simple pingpong uverbs test gives: Couldn't get context for mthca0 This is with rev 2685 on 2.6.12.1 under RHEL3. Fresh rebuild/install of libibverbs, libmthca, and libibcm. Ports are ACTIVE and ipoib

Re: [openib-general] RDMA write with immediate

On Wed, 2005-06-22 at 19:02 -0700, Roland Dreier wrote: > Josh> I'm trying to figure out how to do an RDMA write with > Josh> immediate data. I've got a regular RDMA write working and > Josh> tried changing wr.opcode from IBV_WR_RDMA_WRITE to > Josh> IBV_WR_RDMA_WRITE_WITH_IMM and

Re: [openib-general] mapping between IP address and device name

Hi again Kevin, I'd like to go back to your original query to make sure I understand things. On Mon, 2005-06-20 at 22:23, Kevin Reilly wrote: > Maybe somebody could help me understand the proper way to map between an IP > address assigned to a port to the > "device name" and "port number" in the

Re: [openib-general] mapping between IP address and device name

On Wed, 2005-06-22 at 23:16, Kevin Reilly wrote: > > Hal and others, thanks for your responses. > > This kernel IBAT looks like it provides the function to map an IP address > of an network interface to the device name and port > number. Yes. > My reluctance is i don't think is accessible f

Re: [openib-general] Re: page allocation failure with kdapl/mthca

On Thu, 2005-06-23 at 02:42, Grant Grundler wrote: > On Wed, Jun 22, 2005 at 12:00:10PM -0700, Roland Dreier wrote: > > Roland> I just checked in a change that converts from using > > Roland> pci_alloc_consistent(), which is always GFP_ATOMIC, to > > Roland> dma_alloc_coherent(..., GFP_