@@ -885,6 +901,19 @@ struct ib_send_wr {
u32 rkey;
struct ib_mw_bind_info bind_info;
} bind_mw;
+ struct {
+ struct ib_sig_attrs*sig_attrs;
+ struct ib_mr
applied - thanks!
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
applied - thanks!
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
applied - thanks!
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Subject: [PATCH librdmacm] examples: Remove unneeded include of header file
when using rocket.h
as it's included in rsocket.h
My preference is to include the file explicitly.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to
I would not call these workarounds a real fix, but they should point
out the problems which I am trying to solve.
Thanks for the update. I haven't had the time to investigate this, but did
want to at least acknowledge that this hasn't gotten lost.
- Sean
--
To unsubscribe from this list:
which kind of system is used by rsockets to address the remote host?
Is using IPoIB ?
rsockets is built over the RDMA CM. The rdma cm uses ipoib indirectly when
connecting over IB, by mapping the IP addresses to GIDs.
rsockets can also support using native IB addresses, though this is a new
We are using the RoCE ethertype in the MAC frames, and that is all of the
similarity between usNIC protocol and RoCE. L3 is not the GRH, but rather
a usNIC L3 header. Customers can distinguish between RoCE traffic and
usNIC traffic by looking at the version of the L3 frame. It is set to 6
The cma_acquire_dev function was changed by commit 3c86aa70bf67
to use find_gid_port because multiport devices might have
either IB or IBoE formatted gids. The old function assumed that
all ports on the same device used the same GID format. However,
when it was changed to use find_gid_port,
I think there are lots of #defines and enums replicated in both
nes_netlink.h and c4iw_netlink.h. For example, the IWPM_* defines and
enums. It seems like we could put all this in a common header file to
be included by all iwarp providers? Say include/rdma/iw_portmap.h or
something.
with 2 parallel connection i'm able to reach rate speed with iperf,
the same speed archived with rstream.
Is iperf affected by IPoIB MTU size when used with librspreload.so ?
Not directly. The ipoib mtu is usually set based on the mtu of the IB link.
The latter does affect rsocket
Another strange issue:
$ sudo LD_PRELOAD=/usr/local/lib/rsocket/librspreload.so iperf -c
172.17.0.2
Client connecting to 172.17.0.2, TCP port 5001
TCP window size: 128 KByte (default)
Increasing the window size may improve the
i've connected just one port between two hosts.
Ports is detected properly as 20Gb/s (4x DDR) but i'm unable to reach
speed over 5Gbit/s:
It's possible that this is falling back to using normal TCP sockets.
Can you run the rstream test program to verify that you can get faster than 5
Gbps?
Perhaps it would be best to have ULP's ask for 0 and have providers return
their = 0 max, which would implement this as the man page suggests?
Based on performance tests, increasing the maximum inline can result in a
significant decrease in overall bandwidth. So, I don't believe that we want
Based on performance tests, increasing the maximum inline can result in a
significant decrease in overall bandwidth. So, I don't believe that we
want
providers to increase max_inline automatically to what could be
supported.
So why would providers return a max_inline value that
2013/8/28 Hefty, Sean sean.he...@intel.com:
Can you explain your environment more? The performance seems low.
Ubuntu 13.04 Server on both nodes.
node1:
$ cat /proc/cpuinfo | grep 'model name'
model name : Intel(R) Xeon(R) CPU E5-2603 0 @ 1.80GHz
$ cat /proc/cpuinfo | grep 'model
Ubuntu 13.04 Server on both nodes.
node1:
$ cat /proc/cpuinfo | grep 'model name'
model name : Intel(R) Xeon(R) CPU E5-2603 0 @ 1.80GHz
If you can provide your PCIe information and the results from running the
perftest tools (rdma_bw), that could help as well.
--
To unsubscribe from this
While doing some tests, I've found that rdma_client failed
on my QLogic/Intel QLE7340 / QLE7342 HCA:
# rdma_client
rdma_client: start
rdma_post_send 22
rdma_client: end -1
I had a deeper look on the examples and found that max_inline_data was
returned as 0,
The man
thanks - applied
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
P1- addr_resolve()
select(cm_event_fd)
gets RDMA_CM_ADDR_RESOLVED
rout_resolve()
gets RDMA_CM_ROUTE_RESOLVE
connect():
P2: Gets Connect request
P1: - Dies
When process 1 dies after trying to establish a connection, this generates a
connection
I have added the patch and re-tested: I still encounter
hangs of my application. I am not quite sure whether the
I hit the same error on the shutdown because now I don't hit
the error always, but only every now and then.
I guess this is at least some progress... :/
WHen adding the patch to
Where is the documentation for this? Multiple people have referred to it, but
I don't see any mention of it in libibverbs.git.
This is an unmerged, yet to be accepted patch set. Extensions were added as
part of adding support for XRC.
Yishai Hadas posted v9 of the series on 8/1 - Add
Can you see if the patch below fixes the hang?
Signed-off-by: Sean Hefty sean.he...@intel.com
---
src/rsocket.c | 11 ++-
1 files changed, 10 insertions(+), 1 deletions(-)
diff --git a/src/rsocket.c b/src/rsocket.c
index d544dd0..e45b26d 100644
--- a/src/rsocket.c
+++ b/src/rsocket.c
Bumped the ABI version to 7 (the new verb will return -ENOSYS if
abi_verb is 7).
How does this break the ABI?
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
It doesn't *break* the ABI, but it does add a new downcall into the kernel.
That requires bumping the ABI version to 7, no?
No - adding a new command is fine. Older kernels will return ENOSYS if that
command is not supported. In that case, you can handle things like Jason
suggested.
- Sean
This breaks the libibverbs ABI. You can't modify ibv_context_ops because it
changes struct ibv_context.
Any suggestions on how one adds a new driver call without breaking ABI?
It could be built on the verbs extension mechanism.
Is it necessary to call into a provider library, versus
@@ -884,6 +884,13 @@ int main(int argc, char *argv[])
if (ctx.use_event)
ibv_ack_cq_events(ctx.recv_cq, num_cq_events);
+ /* Process should wait before closing its resources to make sure
+ * latest daemon's response sent via its target QP destined to an XSRQ
+
I am looking at a multithreaded application here, and I believe that
the race is between thread A calling the rpoll() for POLLIN event and
thread B calling the shutdown(SHUT_RDWR) for reading and writing of
the (r)socket almost immediately afterwards.
I modified a test program, and I can
The first question I would have is: why is the rpoll() split into
these two pieces? There must have been some reason to do a busy
loop on some local state information rather than just call the
real poll() directly.
As Scott mentioned in his email, this is done for performance reasons. The
I found a workaround for my (our) problem: in the librdmacm
code, rsocket.c, there is a global constant polling_time, which
is set to 10 microseconds at the moment.
I raise this to 1 - and all of a sudden things work nicely.
I am adding the linux-rdma list to CC so Sean might see
This is slightly off topic, but what is the underlying reason for having the
libibverbs config directory? Is it simply to support the library name being
different than the device name, or is there something more?
- Sean
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
I am currently testing rsockets in connection with ceph.
I am using LD_PRELOAD and the librspreload.so to force
the application (ceph) to use rsockets instead of regular
tcp/ip sockets.
All this works pretty well - until the point where an
established connection is shut down: this seems to
I don't believe that errno is supposed to be cleared by calls. It is only set
when an error occurs. This way, errno does not get lost. Is poll() returning
-1?
I begin to believe that this may be a more general
problem: it seems to me that errno is not always
initialized to 0 when the
and this is obviously missing the actual source code :P
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
It should be in network rather than host endian.
The TID is basically an abstract 64-bit handle. I don't see that it matters
what value is used, provided that the sender use it consistently.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to
There can be holes in pkey table although this is not usually the case
but since IBA spec allows for this, it should be handled.
Signed-off-by: Hal Rosenstock h...@mellanox.com
---
src/acm.c | 14 +++---
1 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/src/acm.c
+ case IBV_EVENT_CLIENT_REREGISTER:
+ if (dev-port[i].state == IBV_PORT_ACTIVE) {
+ acm_port_join(dev-port[i]);
+ acm_log(1, %s %d has reregistered\n,
+
applied
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
If a application is using AF_IB with a UD QP, but does not
provide any private data, we will end up accessing invalid
memory. Check for this case and handle it appropriately.
Signed-off-by: Sean Hefty sean.he...@intel.com
Roland, hold off on this patch, but please push the other 2 for
thanks - applied all 4
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
This is still unclear.
From the README (
http://git.openfabrics.org/git?p=~shefty/librdmacm.git;a=blob;f=README;h=e1f222
740144ed8f7d8bc935ea6643355b077bcd;hb=refs/heads/master
) I can read:
Using multiple interfaces
The librdmacm does support multiple interfaces.
so, multiple
IPoIB is very slow, I prefere to try with rsockets and the preloader library.
Currently, rsockets cannot fail over between local HCA ports. This is an
implementation restriction resulting from no automatic path migration support.
- Sean
--
To unsubscribe from this list: send the line
thanks - applied
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Steve Wise sw...@opengridcomputing.com
This patch modifies the type of local_addr and remote_addr fields in struct
iw_cm_id from struct sockaddr_in to struct sockaddr_storage to hold IPv6 and
IPv4 addresses uniformly. It changes the references of local_addr and
remote_addr in
See Annex 8; DevMgt class version 2 rather than 1 is currently supported.
Could older devices still return version 1? If so the kernel should allow
DevMgt without RMPP, correct?
This check has been this way since 2.6.17. I think it's reasonable to say that
there aren't any devices using
See Annex 8; DevMgt class version 2 rather than 1 is currently
supported.
Could older devices still return version 1? If so the kernel should
allow DevMgt without RMPP, correct?
This check has been this way since 2.6.17. I think it's reasonable to say
that
there aren't
I am seeing build warnings in drivers/infiniband/core/cma.c starting with
v3.11-rc1. These can be reproduced with gcc 4.6.3.
Would you consider applying the following fix ?
A patch to fix this was submitted to the linux-rdma list last week.
Thanks.
- Sean
--
To unsubscribe from this
thanks - committed
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
+ kern_spec = kern_flow_attr + 1;
+ ib_spec = flow_attr + 1;
+ for (i = 0; i flow_attr-num_of_specs; i++) {
+ err = kern_spec_to_ib_spec(kern_spec, ib_spec);
+ if (err)
+ goto err_free;
+ flow_attr-size +=
+ ((struct
diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c
index dc3fd1e..9be6754 100644
--- a/drivers/infiniband/core/mad.c
+++ b/drivers/infiniband/core/mad.c
@@ -2663,6 +2663,7 @@ static int ib_mad_port_start(struct ib_mad_port_private
*port_priv)
int ret, i;
Building cma.o triggers this GCC warning:
drivers/infiniband/core/cma.c: In function ‘rdma_resolve_addr’:
drivers/infiniband/core/cma.c:465:23: warning: ‘port’ may be used
uninitialized in this function [-Wmaybe-uninitialized]
drivers/infiniband/core/cma.c:426:5: note: ‘port’ was
I hadn't looked at the kernel side yet; I was waiting for the userspace side
to
sort itself out first.
I think it makes sense to start with how user space can get the data. Without
eating up reserved fields, we're starting with 8 bit values.
Hmm. 16 bits is probably enough for the MTU
+ssize_t ib_uverbs_create_flow(struct ib_uverbs_file *file,
+ const char __user *buf, int in_len,
+ int out_len)
+{
+ struct ib_uverbs_create_flow cmd;
+ struct ib_uverbs_create_flow_resp resp;
+ struct ib_uobject
Jeff's patch doesn't break old binaries, old binaries, running with
normal IB MTUs work fine. The structure layouts all stay the same,
etc.
FWIW, I did a simple test to confirm this. I installed a stock git HEAD
libibverbs into $HOME/libibverbs-HEAD and a libibverbs with the MTU patch
A source change is completely unvaoidable. Supporting the new MTU
values requires updated source.
I don't really care one way or the other; I'll submit whatever patch people
want. :-)
But FWIW, I tend to believe the Doug/Jason position:
- MTU really needs to be a plain integer
Thanks - I pulled in these patches, but see below:
diff --git a/autogen.sh b/autogen.sh
index f433312..6c9233e 100755
--- a/autogen.sh
+++ b/autogen.sh
@@ -1,9 +1,4 @@
#! /bin/sh
set -x
-test -d ./config || mkdir ./config
Without the above line, the build fails. I added it back in.
1) Perhaps a better way to silence GCC is to drop port entirely, and
assign to id_priv-id.port_num directly. Would that be acceptable?
Yes, this would work.
- Sean
I've pulled in this series, with the changes mentioned in a separate email,
plus updates to the patches below (mostly renames of functions, variables,
enums, etc.). The updated patches are available from my ibacm.git tree in the
preload branch.
Please look over those changes and see if you
I looked them over and retested this. I have a few minor fixups to
follow shortly.
I've incorporated your changes into the patch set and updated the preload
branch.
The one thing I noticed is that the check for using this (address)
preload option without the previous (route) preload option
+static void acm_parse_hosts_file(struct acm_ep *ep)
+{
+ FILE *f;
+ char s[120];
+ char addr[INET6_ADDRSTRLEN], gid[INET6_ADDRSTRLEN];
+ uint8_t name[ACM_MAX_ADDRESS];
+ struct in6_addr ip_addr, ib_addr;
+ struct acm_dest *dest, *new_dest;
+ uint8_t addr_type;
The TCP/IP filters are broken into separate filters based in L4/L3. It
would
seem to
make sense if the IB filters were similarly divided into L2/L3/L4 filters.
IB and IPv6
could probably share the same filter definition.
IPv6 filters wasn't defined through this submission, but as I
Still for the initial set of patches that goes in I tend to just
remove the IB filter structure and define the different IB filters
along your proposal in a follow-up patches/es, OK?
That sounds fine to me.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a
Patches 1-9 clean up documentation and sample config files.
Rather than trying to keep the sample ibacm_addr/opts files in sync, I elected
to remove them from the git tree. There's really no use for them, and I don't
think they're included with the released package anyway.
Doing this drops
diff --git a/src/acm.c b/src/acm.c
index e956b09..a124e93 100644
--- a/src/acm.c
+++ b/src/acm.c
@@ -1277,7 +1277,7 @@ static void acm_process_acm_recv(struct acm_ep *ep,
struct ibv_wc *wc, struct ac
rec = (struct acm_resolve_rec *) mad-data;
acm_format_name(2, log_data, sizeof
To preload these caches, ibacm service is started with -p option
which can also take optional path_rec_file which defaults to
ACM_CONF_DIR/ibacm_path_records.dump
I would rather use the config file versus adding command line parameters.
So add a new preload option in the config file
The input to ib_create_flow is instance of struct ib_flow_attr which
contain few mandatory control elements and optional flow specs.
struct ib_flow_attr {
enum ib_flow_attr_type type;
u16 size;
u16 priority;
u8 num_of_specs;
u8 port;
Preloading of these caches is supported via a file which is
produced by OpenSM by the dump_pr plugin which contains
sufficient SA PathRecord information. Details on this
file format and configuring OpenSM for this are found in
dump_pr_notes.txt in dump_pr.
File format is specified in
raises hand I'm ready for a new project. I already had one in mind,
but it required some changes to libibverbs that are pretty deep changes.
Instead of doing that deep work in libibverbs, I can start here and
move on to my next project when this is up and running.
That would be
I used the name num_to_ibv_mtu because it is in the spirit of the other
enum-
to-int/int-to-enum function pair naming conventions:
int ibv_rate_to_mult(enum ibv_rate rate);
enum ibv_rate mult_to_ibv_rate(int mult);
int ibv_rate_to_mbps(enum ibv_rate rate);
enum ibv_rate
I would argue that this is because the libraries are so disjoint (that
librdmacm needs the deep internal knowledge it needs of libibverbs
indicates that maybe these two shouldn't be separate from each other for
example, or that maybe libibverbs should provide a unified connection
API to the
Subject: [PATCH] IB/core: Fix error return code in add_port()
From: Wei Yongjun yongjun_...@trendmicro.com.cn
Fix to return -ENOMEM in the alloc_group_attrs() error handling
Patch itself looks fine, but please change alloc_group_attrs() - add_port() in
the description.
- Sean
--
To
Define AF_IB and sockaddr_ib to allow the rdma_cm to use native IB
addressing.
Signed-off-by: Sean Hefty sean.he...@intel.com
---
include/linux/socket.h |2 +
include/rdma/ib.h | 89
2 files changed, 91 insertions(+), 0
The concept of a libibverbs 2.0 has been NAK's by pretty much everyone
involved. This is why we are suffering with the complex extension
mechanism.
Are you saying that libibverbs must always always always be backwards
compatible, and there will never be an ABI break at any version in the
+
#ifdef HAVE_IBV_REGISTER_DRIVER
static __attribute__((constructor)) void mlx4_register_driver(void)
{
- ibv_register_driver(mlx4, mlx4_driver_init);
+ verbs_register_driver(mlx4, mlx4_driver_init);
+
}
#else
Shouldn't ibv_register_driver() need to be called in
enum verbs_context_mask {
VERBS_CONTEXT_XRCD = 1 0,
- VERBS_CONTEXT_RESERVED = 1 1
+ VERBS_CONTEXT_SRQ = 1 1,
+ VERBS_CONTEXT_RESERVED = 1 2
};
Why is _RESERVED being re-numbered here? That worries me..
For that matter, what is it for?
I called it
+uint32_t qp_num; /* QP number */
+void *qp_context; /* Associated context of the QP */
Would be nice to clarify what this value is too. I took a quick peak
and it wasn't clear to me. The XRC example does not use it.
Is it a user defined cookie? If so can we
enum verbs_context_mask {
VERBS_CONTEXT_XRCD = 1 0,
- VERBS_CONTEXT_RESERVED = 1 1
+ VERBS_CONTEXT_SRQ = 1 1,
+ VERBS_CONTEXT_RESERVED = 1 2
};
Would VERBS_CONTEXT_XSRQ or VERBS_CONTEXT_SRQ_EX be more clear?
The structure being extended is ibv_srq, so
Are you suggesting to do this in addition to ib_types.h or instead of ?
umad_sm.h seems to hold only a small subset of what's found in ib_types.h.
umad_sm.h was only recently added. I'm suggesting to put the definitions there
instead of within the opensm header file, so that the definitions
diff --git a/include/iba/ib_types.h b/include/iba/ib_types.h
index d5cf287..5c238ae 100644
--- a/include/iba/ib_types.h
+++ b/include/iba/ib_types.h
@@ -3,6 +3,7 @@
* Copyright (c) 2002-2011 Mellanox Technologies LTD. All rights reserved.
* Copyright (c) 1996-2003 Intel Corporation. All
[root@hpc-hn1 libibverbs-1.1.4]# ibv_xsrq_pingpong -d mlx4_0 192.168.174.52
local: LID 0001, QPN RECV 98004b SEND 18004c, PSN 5b6d99, SRQN 0042
remote: LID 0002, QPN RECV d4004a SEND 54004b, PSN d7ba7a, SRQN 0042
The same SRQN on both sides looks suspicious.
[root@hpc-cn2
So under some unknown conditions, many of the clients connection
attempts fail with RDMA_CM_EVENT_UNREACHABLE event and the status is
-ETIMEDOUT. Looking on the rdma-cm kernel code, I see that the only
location which generates this event is in cma_ib_handler when getting
IB_CM_REQ_ERROR (or
One thing seen in the nodes dmesg is a message from an old patch of
yours which exists in ofed1.5.3 but didn't hit (or wasn't accepted?)
upstream saying ib_cm: calculated mra timeout 67584 8192, decreasing
used timeout_ms does this provides any insight into the problem?
I don't
So is it intended that cooperating processes sharing an xrc domain will
choose one process to create the xrc qp, and the rest will open it?
yes - The QPN of the shared QP must be known between the cooperating processes.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
Do we have some public quoted usages/feedback for rsockets? I think
you've mentioned something during the panel at the Linux EU summit
last week but I am not sure...
Most feedback I can think of has come via private emails or personal
interactions, especially specific details of various usage
My reading of the code (i.e. hw/mlx4/cq.c) is that the hardware cqe
owner_sr_opcode field contains MLX4_CQE_OPCODE_ERROR when there is an
error and therefore, the only way to recover what the opcode was is
through the wr_id you used when submitting the WR.
Is my reading of the code correct?
So if you were pushing these private conversations to linux-rdma, more
have been known on rsockets for the benefit of all... oh well. I think
you mentioned something re Intel HPC group, or I am wrong?
rsockets will continue to be supported by myself and Intel going forward. The
rsocket work
- Remove the pointer to the ibv_send_wr_ext pointer from the wr union
in ibv_send_wr, and put the ibv_send_wr (less the duplicated data structures)
after the union, with the usual comp_mask flag to indicate what is supported.
This would be our #1 preference, because of the performance
Previously we had agreed to use a new send_flag , IBV_SEND_EXTENDED, which
would indicate the presence of the extended data structure. With new
functionality, there may also be new extended opcodes, but this is not a
requirement.
I think that's fine.
--
To unsubscribe from this list: send
So are you are OK with this option ? There were 2 options, and this one is
our
preference.
Remove the pointer to the ibv_send_wr_ext pointer from the wr union
in ibv_send_wr, and put the ibv_send_wr (less the duplicated data
structures) after the union, with the usual comp_mask flag to
What would be the correct way to go for consumers, is that making sure
they destroy 1st all their IB
objects (PDs, MRs, CQs, QPs, etc) prior to destroying the last rdma_cm
id on a device removal event?
any other idea?
The user should free any objects that are associated with the device
So think on the case, e.g for iser where a ULP maintains a per-device
context where they stroe IB objects used for all connections of that
device (e.g PD, CQs, DMA-MR, etc) and per connection object where the
QP and RDMA-CM ID are storage.
The per-device context is created on demand
Conceptually, RSS/TSS are a set of send/receive work queues all
belonging to the same transport level address. There's no
parent-child relationship or needed pairing of send and receive
queues together in order to form a group.
This view makes sense to me as well. Can someone also
On Mon, Apr 22, 2013 at 7:46 PM, Or Gerlitz or.gerl...@gmail.com wrote:
Sean, Tzahi -- I understand now that there have been few talkings @
the OFA meeting re this patch set. So what's the way to move forward,
any concrete feedback that needs to be addressed here? This series is
hanging
This seems reasonable, but still concerns me a bit. The original
version was flat out wrong because you can't re-arrange any exposed enum
like this without requiring that all user space apps be recompiled.
This is especially true because ibv_mtu_enum_to_int is an inline
ib_mtu_enum_to_int()
Any thoughts on these two patches? They're pretty trivial, enable use with
modern versions of Autotools, and now feature the proper Signed-off-by line.
It may help if you identify the library this patch is against. :)
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
Was there a reason to define sm_key as an array of uint8_t's vs a be64_t?
I don't think the sm_key is 64-bit aligned, so you would need to pack the
structure and deal with misaligned accesses.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to
I have no issue with RSS/TSS. But the 'qp group' interface to using this
seems kludgy.
lets try to be more specific
On a node, this is multiple send/receive queues grouped together to form a
larger
construct. On the wire, this is a single QP - maybe? I'm still not clear
on
any feedback?
I have no issue with RSS/TSS. But the 'qp group' interface to using this seems
kludgy.
On a node, this is multiple send/receive queues grouped together to form a
larger construct. On the wire, this is a single QP - maybe? I'm still not
clear on that. From what's written,
Agreed, but changing to an int would seem to have some fairly serious
backwards
compatibility issues.
Why can't IB_MTU_1500 = 1500?
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
RDMA_NODE_VENDOR would be great, actually. Should I work up a patch for that?
I would prefer taking this approach and would be fine accepting such a change.
Roland, do you have an opinion on this?
- Sean
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a
301 - 400 of 1258 matches
Mail list logo