Hello Roland,
This patch series addresses a few issues I noticed after having built
libibverbs from the source tree available at kernel.org. Feedback is welcome.
Bart.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
Tell automake that the libibverbs package does not follow the GNU standards.
This change makes it possible to use 'autoreconf' for the libibverbs package.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
configure.in |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git
Avoid that the code in src/memory.c triggers a compiler warning when compiled
on a 32-bit system.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
src/memory.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/src/memory.c b/src/memory.c
index faa43f3..9f13774
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
.gitignore |5 +
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/.gitignore b/.gitignore
index 78effef..a0a13a6 100644
--- a/.gitignore
+++ b/.gitignore
@@ -15,3 +15,8 @@ Makefile
config.status
stamp-h1
libtool
name for the same functionality. This
variable is deprecated; we suggest using AM_CPPFLAGS and per-target
_CPPFLAGS instead.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
Makefile.am |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/Makefile.am b/Makefile.am
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
src/memory.c |2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/src/memory.c b/src/memory.c
index 9f13774..7d97e55 100644
--- a/src/memory.c
+++ b/src/memory.c
@@ -269,6 +269,7 @@ static void __mm_rotate_left(struct
There is no need to invoke the different autotools individually - autoreconf
already does that.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
autogen.sh |7 +--
1 files changed, 1 insertions(+), 6 deletions(-)
diff --git a/autogen.sh b/autogen.sh
index fd47839..f6ff02c 100755
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+ /* heuristic sliding window algorithm to keep
+sender from overrunning receiver queues */
+ if (qp_type(qp) == IB_QPT_RC) {
+ if (psn_compare(qp-req.psn, qp-comp.psn +
+
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+int rxe_arbiter(void *arg)
+{
+ int err;
+ unsigned long flags;
+ struct rxe_dev *rxe = (struct rxe_dev *)arg;
+ struct sk_buff *skb;
+ struct list_head *qpl;
+ struct rxe_qp *qp;
+
+ /*
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
Bring up to date with the current version of the IBTA spec.
- add new opcodes for RC and RD
- add new groups of opcodes for CN and XRC
Signed-off-by: Bob Pearson rpear...@systemfabricworks.com
---
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static inline int pkey_match(u16 key1, u16 key2)
+{
+ return (((key1 0x7fff) != 0)
+ ((key1 0x7fff) == (key2 0x7fff))
+ ((key1 0x8000) || (key2 0x8000))) ? 1 : 0;
+}
Shouldn't the
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+ rxe_qp.c:char *rxe_qp_state_name[] = {
One more comment about rxe_verbs.h: why is the above declaration
present in this header file ? As far as I can see it's only used in
rxe_qp.c and not in any other source file.
Bart.
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static int rxe_query_port(struct ib_device *dev,
+ u8 port_num, struct ib_port_attr *attr)
+{
+ struct rxe_dev *rxe = to_rdev(dev);
+ struct rxe_port *port;
+
+ if (unlikely(port_num
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+ for (i = 0; i ARRAY_SIZE(rxe_dev_attributes); ++i) {
+ err = device_create_file(dev-dev, rxe_dev_attributes[i]);
+ if (err) {
+ pr_warn(device_create_file failed,
+
On Mon, Aug 15, 2011 at 4:45 PM, Greg KH gre...@suse.de wrote:
On Mon, Aug 15, 2011 at 04:33:14PM +0200, Bart Van Assche wrote:
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+ for (i = 0; i ARRAY_SIZE(rxe_dev_attributes); ++i) {
+ err
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+char *rxe_qp_state_name[] = {
+ [QP_STATE_RESET]= RESET,
+ [QP_STATE_INIT] = INIT,
+ [QP_STATE_READY]= READY,
+ [QP_STATE_DRAIN]= DRAIN,
+ [QP_STATE_DRAINED] =
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static ssize_t rxe_show_skb_num(struct device *device,
+ struct device_attribute *attr, char *buf)
+{
+ struct rxe_dev *rxe = container_of(device, struct rxe_dev,
+
On Mon, Aug 15, 2011 at 6:15 PM, Bob Pearson
rpear...@systemfabricworks.com wrote:
I want to eventually complete the IB transport parts of the driver although
they are not used by RoCE. They would be useful for a soft IB implementation
with say an FPGA based MAC/PHY. The long term goal is to
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+int rxe_fast_comp = 2;
+module_param_named(fast_comp, rxe_fast_comp, int, 0644);
+MODULE_PARM_DESC(fast_comp, fast path call to completer
+ (0=no, 1=no int context, 2=any context));
+
+int rxe_fast_resp =
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+enum rxe_mtu {
+ RXE_MTU_INVALID = 0,
+ RXE_MTU_256 = 1,
+ RXE_MTU_512 = 2,
+ RXE_MTU_1024= 3,
+ RXE_MTU_2048= 4,
+ RXE_MTU_4096= 5,
+ RXE_MTU_8192= 6,
+};
Do the
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static u64 rxe_dma_map_single(struct ib_device *dev,
+ void *cpu_addr, size_t size,
+ enum dma_data_direction direction)
+{
+
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+int rxe_cq_chk_attr(struct rxe_dev *rxe, struct rxe_cq *cq,
+ int cqe, int comp_vector, struct ib_udata *udata)
+{
+ int count;
+
+ if (cqe = 0) {
+ pr_warn(cqe(%d) = 0\n, cqe);
+
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static char *resp_state_name[] = {
+ [RESPST_NONE] = NONE,
Shouldn't that be const char * const instead of char * ?
+/* rxe_recv calls here to add a request packet to the input queue */
On Sat, Aug 20, 2011 at 4:47 PM, Bart Van Assche bvanass...@acm.org wrote:
On Mon, Aug 15, 2011 at 6:15 PM, Bob Pearson
rpear...@systemfabricworks.com wrote:
I want to eventually complete the IB transport parts of the driver although
they are not used by RoCE. They would be useful for a soft
On Thu, Sep 1, 2011 at 7:20 AM, Bob Pearson
rpear...@systemfabricworks.com wrote:
From: bvanass...@acm.org
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static int rxe_query_port(struct ib_device *dev,
+ u8 port_num, struct ib_port_attr *attr)
On Thu, Sep 1, 2011 at 7:20 AM, Bob Pearson
rpear...@systemfabricworks.com wrote:
From: bvanass...@acm.org
On Sun, Jul 24, 2011 at 9:43 PM, rpear...@systemfabricworks.com wrote:
+static struct ib_ah *rxe_create_ah(struct ib_pd *ibpd, struct
ib_ah_attr
*attr)
+{
+ int err;
+
On Thu, Sep 1, 2011 at 5:15 AM, Bob Pearson
rpear...@systemfabricworks.com wrote:
On Sat, Aug 20, 2011 at 4:47 PM, Bart Van Assche bvanass...@acm.org
More in detail, my comments with regard to multicast support in ib_rxe
are:
1. Using ipv6_eth_mc_map() for mapping multicast GIDs seems
On Mon, Sep 19, 2011 at 11:20 PM, Hefty, Sean sean.he...@intel.com wrote:
It would be easier to maintain opensm on windows if it truly shared the same
code base. For now, I'd just like to start with a common ib_types.h file.
(There are currently hundreds, if not thousands, of lines that
On Tue, Sep 20, 2011 at 7:45 PM, Ira Weiny wei...@llnl.gov wrote:
+#ifdef __cplusplus
+extern C {
+#endif
+
#include stdio.h
#include infiniband/mad.h
#include iba/ib_types.h
It is bad style to surround #include directives with 'extern C'.
Bart.
--
To unsubscribe from this list: send
On Thu, Sep 22, 2011 at 8:29 PM, Hefty, Sean sean.he...@intel.com wrote:
Signed-off-by: Sean Hefty sean.he...@intel.com
---
libibnetdisc/src/ibnetdisc.c | 2 +-
src/ibportstate.c | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
mode change 100644 = 100755
On Tue, Oct 4, 2011 at 3:53 PM, Hal Rosenstock h...@dev.mellanox.co.il wrote:
+int ib_mtu_is_valid(IN const int mtu)
+{
+ if (mtu IB_MIN_MTU || mtu IB_MAX_MTU)
+ return 0;
+ return 1;
+}
Not that it matters, but the above three lines can also be written as
Tell automake explicitly that the libmlx4 package is not a GNU package. This
change makes it possible to regenerate the configure script via autoreconf
instead of only via autogen.sh.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
configure.in |2 +-
1 files changed, 1 insertions
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
include/infiniband/verbs.h |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/include/infiniband/verbs.h b/include/infiniband/verbs.h
index ad9d524..656abb8 100644
--- a/include/infiniband/verbs.h
+++ b/include
Avoid that link_local_gid() triggers the following compiler warning:
dereferencing type-punned pointer will break strict-aliasing rules
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
src/verbs.c |8 +++-
1 files changed, 3 insertions(+), 5 deletions(-)
diff --git a/src/verbs.c b
On Tue, Oct 11, 2011 at 7:39 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Tue, Oct 11, 2011 at 09:02:41AM -0500, Christoph Lameter wrote:
Has XRC support not been merged? How can I build the OFED libraries
against Linux 3.1? I'd really like to get rid of the OFED kernel tree
On Fri, Oct 14, 2011 at 3:48 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This patch adds the kernel module ib_srpt SCSI RDMA Protocol (SRP) target
implementation conforming to the SRP r16a specification for the mainline
drivers/target infrastructure as requested by Christoph for an
On Wed, Oct 19, 2011 at 7:38 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
Btw, Roland asked me to take a look at the current set of module
parameters for ib_srpt, and see which could be removed and/or converted
to per SPR target endpoint configfs attributes.. So on that note, a few
On Wed, Oct 19, 2011 at 8:35 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Wed, 2011-10-19 at 19:56 +0200, Bart Van Assche wrote:
Not sure what you mean here. Can you be more specific..?
I'm referring to this code:
sprintf(sport-port_guid, 0x%04x%04x
On Wed, Oct 19, 2011 at 9:47 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Wed, Oct 19, 2011 at 12:09:01PM -0700, Nicholas A. Bellinger wrote:
Are sport-gid.global.interface_id going to be different for each struct
srpt_port..?
Infiniband GID suffixes (interface_id) are not
On Wed, Oct 19, 2011 at 9:09 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
Ok, but just to verify.. This change will still allow us to
differentiate between different struct srpt_port entries from the
perspective of configfs, correct..?
Yes.
Also, this is the same format (minus the
On Thu, Oct 20, 2011 at 6:51 PM, Roland Dreier rol...@purestorage.com wrote:
On Wed, Oct 19, 2011 at 11:09 AM, Bart Van Assche bvanass...@acm.org wrote:
*) use_port_guid_in_session_name: This appears to be a legacy compat
item, can this be safetly removed for mainline..?
As far as I know
Register transport attributes after the attribute array has been
set up instead of before. The current code is racy because there
is no guarantee that the CPU examining the attribute container
will see all values written to the container.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc
On Mon, Oct 24, 2011 at 7:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This patch removes the legacy use_port_guid_in_session_name module parameter
that is no longer required in modern ib_srpt code.
The patch looks fine to me but the description could be improved: the
On Mon, Oct 24, 2011 at 7:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
+static ssize_t srpt_tpg_attrib_store_srp_max_rsp_size(
+ struct se_portal_group *se_tpg,
+ const char *page,
+ size_t count)
+{
+ struct srpt_port *sport = container_of(se_tpg, struct
On Mon, Oct 24, 2011 at 7:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
+ snprintf(sport-port_guid, sizeof(sport-port_guid),
+ 0x%016llx,
+ be64_to_cpu(sport-gid.global.interface_id));
If I
On Mon, Oct 24, 2011 at 9:49 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-10-24 at 21:44 +0200, Bart Van Assche wrote:
On Mon, Oct 24, 2011 at 7:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
+static ssize_t srpt_tpg_attrib_store_srp_max_rsp_size
On Mon, Oct 24, 2011 at 10:05 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-10-24 at 21:58 +0200, Bart Van Assche wrote:
On Mon, Oct 24, 2011 at 9:49 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-10-24 at 21:44 +0200, Bart Van Assche wrote:
On Mon
On Mon, Oct 24, 2011 at 9:49 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-10-24 at 21:44 +0200, Bart Van Assche wrote:
Also, isn't kstrto*() preferred over strict_strto*() ?
Not AFAIK.
This is what I found in the description of commit 33ee3b2:
kstrto*: converting
On Mon, Oct 24, 2011 at 8:33 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
It would have been nice if you bothered to mention these in any of the
pre merge window reviews, but given your delay responsed on these items
they will have to be something we'll fix post merge.
Sorry for that
On Mon, Oct 24, 2011 at 11:01 PM, Roland Dreier rol...@purestorage.com wrote:
On Mon, Oct 24, 2011 at 11:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
- Handle IB completion timeouts. Although the InfiniBand Architecture
Manual specifies that a HCA must generate an error completion
On Mon, Oct 24, 2011 at 10:29 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
@@ -114,6 +114,7 @@ enum {
MIN_SRPT_SRQ_SIZE = 4,
DEFAULT_SRPT_SRQ_SIZE = 4095,
MAX_SRPT_SRQ_SIZE = 65535,
+ MAX_SRPT_RDMA_SIZE = 256U,
Hey btw, what should the proper
On Tue, Oct 25, 2011 at 1:17 PM, Roland Dreier rol...@purestorage.com wrote:
On Mon, Oct 24, 2011 at 11:07 PM, Bart Van Assche bvanass...@acm.org wrote:
As far as I know every HCA supported by Linux does implement this
correctly. Which class did you have in mind as not doing
On Mon, Oct 24, 2011 at 7:57 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
+static int srpt_write_pending(struct se_cmd *se_cmd)
+{
+ struct srpt_rdma_ch *ch;
+ struct srpt_send_ioctx *ioctx;
+ enum srpt_command_state new_state;
+ enum rdma_ch_state ch_state;
On Wed, Oct 26, 2011 at 12:04 AM, Roland Dreier rol...@purestorage.com wrote:
Sorry, but now I confused about what the bug is. You have a QP
associated with an SRQ, and you transition the QP to error. At some
point you get a last WQE received event for that QP (which means
all receive
On Mon, Oct 24, 2011 at 10:25 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-10-24 at 21:57 +0200, Bart Van Assche wrote:
On Mon, Oct 24, 2011 at 7:33 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
+ snprintf(sport-port_guid, sizeof(sport-port_guid
On Wed, Oct 26, 2011 at 10:06 PM, Roland Dreier rol...@purestorage.com wrote:
On Wed, Oct 26, 2011 at 11:05 AM, Bart Van Assche bvanass...@acm.org wrote:
Can I conclude from your reply that the last WQE event refers to the
SRQ only and that it does not provide any information about the send
On Mon, Oct 31, 2011 at 10:33 AM, James Bottomley
james.bottom...@hansenpartnership.com wrote:
On Fri, 2011-10-21 at 18:57 +0200, Bart Van Assche wrote:
Register transport attributes after the attribute array has been
set up instead of before. The current code is racy because
On Fri, Nov 4, 2011 at 9:10 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This is the PULL request for an initial merge of the ib_srpt driver
using mainline target infrastructure into v3.2-rc1.
In case anyone is interested, the most important unaddressed comments
for this version of
On Sat, Nov 5, 2011 at 7:04 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Sat, 2011-11-05 at 08:37 +0100, Bart Van Assche wrote:
On Fri, Nov 4, 2011 at 9:10 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This is the PULL request for an initial merge of the ib_srpt driver
srpt_device-srq_size.
Reported-by: Bart Van Assche bvanass...@acm.org
Cc: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rol...@purestorage.com
Signed-off-by: Nicholas Bellinger n...@linux-iscsi.org
---
drivers/infiniband/ulp/srpt/ib_srpt.c | 2 +-
1 files changed, 1 insertions(+), 1
On Sun, Nov 6, 2011 at 12:15 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This patch converts the srp_max_req_size module parameter to R/W access
so that it's accessable when built as CONFIG_INFINIBAND_SRPT=y. This
accessible ?
includes adding srpt_device-max_req_size that is
On Sun, Nov 6, 2011 at 11:09 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
We have also discussed srpt_service_guid a bit before, which you
indicated needed to stay in current code as global scope, and presumably
should not change value after loading. Looking at the actual usage, one
of the completion processing thread.
This patch has been generated against the LIO tree of a few weeks
ago.
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
drivers/infiniband/ulp/srpt/ib_srpt.c | 48 +++--
drivers/infiniband/ulp/srpt/ib_srpt.h |8 ++---
2
On Wed, Nov 2, 2011 at 5:33 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
This v3 patch adds the kernel module ib_srpt SCSI RDMA Protocol (SRP) target
Hi Nick,
Sorry that it took so long before I noticed, but checkpatch complains
about the code present on the for-next-merge branch. Can
On Mon, Nov 7, 2011 at 9:46 PM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
On Mon, 2011-11-07 at 20:00 +0100, Bart Van Assche wrote:
On Mon, Nov 7, 2011 at 7:10 PM, Roland Dreier rol...@purestorage.com wrote:
On Sun, Nov 6, 2011 at 9:31 AM, Bart Van Assche bvanass...@acm.org wrote
On Tue, Nov 8, 2011 at 11:00 AM, Nicholas A. Bellinger
n...@linux-iscsi.org wrote:
Thanks for verifying this one. Pushed into lio-core and will squash
into a single patch for target-pending.
I've had a quick look at
state.
Change one occurrence of if (e) ... else WARN_ON(!e) into
if (e) ... else __WARN().
Signed-off-by: Bart Van Assche bvanass...@acm.org
---
drivers/infiniband/ulp/srpt/ib_srpt.c | 80 -
1 files changed, 19 insertions(+), 61 deletions(-)
diff --git a/drivers
On Thu, Nov 10, 2011 at 10:03 AM, Dotan Barak dot...@dev.mellanox.co.il wrote:
+This library is thread safe library and verbs can be called from every
thread in the process (the
Looks like a is missing between is and thread safe ?
+However, it is up to the user to stop working with a
On Thu, Nov 10, 2011 at 10:03 AM, Dotan Barak dot...@dev.mellanox.co.il wrote:
+This library is thread safe library and verbs can be called from every
thread in the process (the same resource can even be handled from different
threads, for example: ibv_poll_cq can be called from more than one
Since commit 6b7227866 the first argument passed to encode_wr_id()
is always of type enum srpt_opcode. Make this explicit by updating
the function declaration. Also insert a blank line between
function definitions as is done in the rest of the Linux kernel.
Signed-off-by: Bart Van Assche bvanass
Fix spelling in one error message and in several source code comments.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Nicholas Bellinger n...@linux-iscsi.org
Cc: Roland Dreier rol...@purestorage.com
Cc: Christoph Hellwig h...@infradead.org
---
drivers/infiniband/ulp/srpt/ib_srpt.c | 10
to __srpt_lookup_port() are
protected via the same spinlock that protects srpt_dev_list. Hence
the sdev-device pointer is never NULL nor invalid inside
__srpt_lookup_port(). That means that it is safe to remove the
sdev-device test in that function.
Signed-off-by: Bart Van Assche bvanass...@acm.org
On Sun, Nov 13, 2011 at 12:29 PM, Dotan Barak dot...@dev.mellanox.co.il wrote:
On Sun, Nov 13, 2011 at 12:23 PM, Bart Van Assche bvanass...@acm.org
wrote:
Maybe the text in parentheses should be changed into a sentence on its
own, and maybe it's a good idea to mention that when polling
On Sun, Nov 13, 2011 at 5:18 AM, Bill Boas bb...@systemfabricworks.com wrote:
At SC we are running a video streaming demo across a 7000 mile fiber
distance using IB extension boxes from Bay Microsystems at 40 Gbps. The
video streaming app uses SRP to transmit the streams across this long
On Mon, Nov 14, 2011 at 10:43 PM, Or Gerlitz or.gerl...@gmail.com wrote:
On Sun, Nov 13, 2011 at 11:55 PM, Dave Dillow dillo...@ornl.gov wrote:
SRP uses RDMA, so you cannot use UC mode.
per the IB spec, RDMA write is supported for UC
Agreed. But an SRP target does not only issue RDMA write
On Mon, Nov 14, 2011 at 11:49 PM, Albert Chu ch...@llnl.gov wrote:
+ if (opt-vlarb_high)
+ free(opt-vlarb_high);
Those if-statements are superfluous - invoking free(NULL) is safe. See
e.g. http://pubs.opengroup.org/onlinepubs/009695399/functions/free.html.
Bart.
--
To
On Thu, Nov 17, 2011 at 3:34 AM, Randy Dunlap rdun...@xenotime.net wrote:
On 11/15/2011 06:33 PM, Stephen Rothwell wrote:
Hi all,
When CONFIG_BUG is not enabled:
next-2011-1116/drivers/infiniband/ulp/srpt/ib_srpt.c: In function
'srpt_cm_dreq_recv':
With CONFIG_BUG=n the __WARN() macro is not defined. Avoid that
building with CONFIG_BUG=n fails by replacing __WARN() with
WARN_ON(true). Also make sure that each such statement is preceeded
by an appropriate pr_err() statement.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Nicholas
On Fri, Nov 18, 2011 at 12:45 AM, Roland Dreier rol...@purestorage.com wrote:
On Thu, Nov 17, 2011 at 11:25 AM, Bart Van Assche bvanass...@acm.org wrote:
+ pr_err(%s[%d]: unexpected opcode %d, __func__, __LINE__,
+ opcode);
+ WARN_ON(true
With CONFIG_BUG=n the __WARN() macro is not defined. Avoid that
building with CONFIG_BUG=n fails by changing pr_err(...);
_WARN() into WARN(true, ...).
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Nicholas Bellinger n...@linux-iscsi.org
Cc: Roland Dreier rol...@purestorage.com
Cc
On Fri, Nov 18, 2011 at 9:10 PM, Roland Dreier rol...@purestorage.com wrote:
On Thu, Nov 17, 2011 at 11:52 PM, Bart Van Assche bvanass...@acm.org wrote:
As far as I can see with CONFIG_BUG=n WARN() is defined as follows:
#define WARN(condition, format
This patch series makes the ib_srp driver better suited for use in a H.A.
setup because:
- Switchover without triggering read or write errors become possible. Such
errors are bad because these can make a filesystem switch to read-only
mode.
- A ping mechanism has been added that allows to
Make the logging code more a little more brief by replacing
printk(KERN_WARNING PFX ...) by pr_warn(...) and by replacing
printk(KERN_ERR PFX ...) by pr_err(...). Remove one trailing
space to avoid a checkpatch warning.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo
Remove sysfs attributes before removing a target instead of
testing the target state in every sysfs attribute callback
method. Note: it is safe to invoke a sysfs attribute removal
method like device_remove_file() twice on the same attribute.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc
Currently the ib_srp driver allows to log in multiple times to the
same target via its sysfs interface. This leads to each target LUN
being imported multiple times at the initiator side, something that
should not be allowed.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo
an
error completion. The block layer timeout must be larger than
the transport layer timeout because otherwise it can happen
that an SRP response is received after the SCSI layer has
already killed a command.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo...@ornl.gov
Cc
If a user misconfigures the block layer timeout such that it is below
the InfiniBand RC timeout it can happen that an SRP reply arrives after
the SCSI error handler has already killed the associated SCSI command.
Avoid that late replies cause a kernel crash.
Signed-off-by: Bart Van Assche bvanass
Help the CPU branch predictor by rearranging two if-statements such
that the most likely code is in the if-part instead of the else-part.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo...@ornl.gov
Cc: Roland Dreier rol...@purestorage.com
---
drivers/infiniband/ulp/srp
Factor out common code into a new function.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo...@ornl.gov
Cc: Roland Dreier rol...@purestorage.com
---
drivers/infiniband/ulp/srp/ib_srp.c | 19 +++
1 files changed, 11 insertions(+), 8 deletions(-)
diff
Document the srp_transport sysfs attributes according to the rules specified
in Documentation/ABI/README.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Cc: Brian King brk...@linux.vnet.ibm.com
---
Documentation/ABI/stable/sysfs-transport-srp
Register transport attributes after the attribute array has been
set up instead of before.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Cc: Brian King brk...@linux.vnet.ibm.com
Cc: David Dillow dillo...@ornl.gov
Cc: Roland Dreier rol
that macro when adding new
attributes.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Cc: Brian King brk...@linux.vnet.ibm.com
Cc: David Dillow dillo...@ornl.gov
Cc: Roland Dreier rol...@purestorage.com
---
drivers/scsi/scsi_transport_srp.c | 26
a reference to a SCSI device can be deleted via the
sysfs SCSI host delete attribute.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David Dillow dillo...@ornl.gov
Cc: Roland Dreier rol...@purestorage.com
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Cc: Brian King brk...@linux.vnet.ibm.com
Make it possible to disconnect via sysfs the IB RC connection used by the
SRP protocol to communicate with a target.
Let the SRP transport layer create a sysfs delete attribute for
initiator drivers that support this functionality.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: David
On Fri, Dec 2, 2011 at 1:04 AM, Hefty, Sean sean.he...@intel.com wrote:
- What should we do with modules like SDP that are not in kernel?
Either remove them or carry them forward as experimental features.
Wat I expect is that reworking the SDP implementation such that it can
be included
On Fri, Dec 2, 2011 at 7:12 PM, Christoph Lameter c...@linux.com wrote:
What were the issues that prevented the merging of the SDP
implementation?
At least AF_INET_SDP - there might have been other issues. See e.g.
http://lkml.org/lkml/2006/3/6/70.
Bart.
--
To unsubscribe from this list: send
Hi,
Has anyone had a look recently at the sparse endianness complaints
about the code under drivers/infiniband ? I haven't analyzed these
messages closely yet - these might indicate real bugs.
$ make C=2 CF=-D__CHECK_ENDIAN__ M=drivers/infiniband modules
[ ... ]
CHECK
On Tue, Dec 6, 2011 at 10:15 PM, Hefty, Sean sean.he...@intel.com wrote:
Fix errors reported by sparse in the rdma core stack. Note that these
are real bugs, but don't affect any existing code to the best of my
knowledge. The mlid issue would only affect kernel users of
rdma_join_multicast
On Fri, Dec 2, 2011 at 12:32 AM, David Dillow dillo...@ornl.gov wrote:
On Thu, 2011-12-01 at 16:35 -0500, chas williams - CONTRACTOR wrote:
perhaps this should use the unlikely() macro instead:
if (unlikely(wc.status)) {
...
}
If it needs to change at all, this is
On Thu, Dec 15, 2011 at 8:37 PM, David Dillow dillo...@ornl.gov wrote:
+static int srp_slave_alloc(struct scsi_device *sdev)
+{
+ struct Scsi_Host *shost = sdev-host;
+ struct srp_target_port *target = host_to_target(shost);
+
+ if (!WARN_ON(target-rq_tmo_jiffies == 0))
+
201 - 300 of 1130 matches
Mail list logo
