On Fri, Sep 11, 2009 at 9:50 PM, Chris Worley worl...@gmail.com wrote:
I'm thinking the reason you can't repeat the test may be latency
related. We get ~50usecs average latency (on small block sizes),
which can't be achieved using regular SSD's (and rotating drives are
nowhere close). Maybe
On Wed, Sep 16, 2009 at 2:44 PM, Bart Van Assche
bart.vanass...@gmail.com wrote:
I noticed that there are six SRP patches (bug fixes) present in OFED
1.4.1 that are not present in mainstream Linux kernels up to and
including version 2.6.30. Do you know whether it is documented
anywhere which
On Wed, Sep 16, 2009 at 9:41 PM, Chris Worley worl...@gmail.com wrote:
On Wed, Sep 16, 2009 at 12:15 PM, Vladislav Bolkhovitin v...@vlnb.net wrote:
Chris Worley, on 09/16/2009 12:51 AM wrote:
On Tue, Sep 15, 2009 at 11:10 AM, Vladislav Bolkhovitin v...@vlnb.net
wrote:
[ ... ]
[
Hello,
While /etc/init.d/openibd runs the following kernel messages were generated:
mstmread:2578 freeing invalid memtype fe60-fe61
mstmread:2578 freeing invalid memtype fe61-fe70
mstmread:2581 freeing invalid memtype fe60-fe61
mstmread:2581 freeing invalid memtype
On Wed, Oct 28, 2009 at 7:47 PM, Chris Worley worl...@gmail.com wrote:
It appears that SRP tries to coalesce and fragment initiator I/O
requests into 64KB packets, as that looks to be the size requested
to/from the device on the target side (and the I/O scheduler is
disabled on the target).
On Thu, Oct 29, 2009 at 12:10 PM, Gheorghe Ardelean
ardel...@ww.uni-erlangen.de wrote:
I must say first that I am new to Infiniband so maybe I am just making some
stupid mistakes. I have installed OFED-1.5-rc1 on OpenSUSE 11.1 (we need
gcc4.3). The installer expects libstdc++4.2 but it is
On Fri, Nov 6, 2009 at 6:28 PM, Arend Dittmer
aditt...@penguincomputing.com wrote:
Please find attached the gzip'ed /var/log/messages.
This log clearly show the login and logout actions from the different
initiators. I couldn't find anything unusual in the posted log file
however. Around which
On Fri, Nov 6, 2009 at 3:54 AM, Philip Pokorny
ppoko...@penguincomputing.com wrote:
On Thu, Nov 5, 2009 at 2:27 PM, Vladislav Bolkhovitin v...@vlnb.net wrote:
Philip Pokorny, on 11/05/2009 11:51 AM wrote:
Chris Worley asked that we post this to scst and linux-rdma lists for
On Fri, Nov 20, 2009 at 5:16 PM, frank zago fz...@systemfabricworks.com wrote:
It seems ib_post_send() is implemented slightly differently in the various
hardware drivers (as in kernel 2.6.31). Here are the differences I've noticed
regarding the bad_wr parameter.
amso1100/c2_qp.c :
On Fri, Nov 20, 2009 at 9:08 PM, Sean Hefty sean.he...@intel.com wrote:
mlx4/qp.c: mlx4_ib_post_send()
* when passing a list containing more than one item to
mlx4_ib_post_send(), and sending the second or later item fails (e.g.
because of QP overflow), the preceding items are sent anyway. This
On Sat, Nov 21, 2009 at 9:01 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Sat, Nov 21, 2009 at 12:17:32PM +0100, Bart Van Assche wrote:
ib_post_send() has to request a completion notification for each WR,
which has a negative performance impact. My opinion
Single-sentence grammar fix for Documentation/infiniband/ipoib.txt.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
diff --git a/Documentation/infiniband/ipoib.txt
b/Documentation/infiniband/ipoib.txt
index 6d40f00..64eeb55 100644
--- a/Documentation
On Mon, Sep 28, 2009 at 5:27 PM, Roland Dreier rdre...@cisco.com wrote:
If an SRP target processes SRP I/O slow enough, the SRP initiator locks up.
INFO: task fio:6389 blocked for more than 120 seconds.
echo 0 /proc/sys/kernel/hung_task_timeout_secs disables this message.
fio
On Sat, Dec 5, 2009 at 10:49 PM, Roland Dreier rdre...@cisco.com wrote:
I'm not sure that the non-error path for ordinary SCSI commands is
that trivial. If my interpretation of the SRP initiator source code is
correct, the statements complete(target-done) and
Adds documentation to the libsrp source code.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: James E.J. Bottomley james.bottom...@suse.de
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
diff --git a/drivers/scsi/libsrp.c b/drivers/scsi/libsrp.c
index 9ad38e8..bd3229e 100644
Adds documentation to the libsrp source code.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: James E.J. Bottomley james.bottom...@suse.de
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Cc: Randy Dunlap randy.dun...@oracle.com
This is the second version of this patch. While
Adds documentation to the libsrp source code.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: James E.J. Bottomley james.bottom...@suse.de
Cc: FUJITA Tomonori fujita.tomon...@lab.ntt.co.jp
Acked-by: Randy Dunlap randy.dun...@oracle.com
This is the third version of this patch
Hello Roland,
First of all, my best wishes for the new year. I hope that you will be
able to continue supporting the RDMA community in 2010.
I have a question about the SRP initiator available in the Linux
kernel. While consulting the SRP spec (r16a) I noticed that all
multi-byte integer fields
observing the initiator
state easier.
See also http://bugzilla.kernel.org/show_bug.cgi?id=14235
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Reported-by: Chris Worley worl...@gmail.com
Cc: Roland Dreier rola...@cisco.com
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
b/drivers/infiniband
On Sat, Jan 2, 2010 at 5:05 PM, Chris Worley worl...@gmail.com wrote:
Would this fix effect the Window's SRP initiator too?
I'm not familiar with the WinOF codebase. But at first sight it looks
like the WinOF SRP initiator doesn't support processing SRP_CRED_REQ
information units either, which
On Mon, Jan 4, 2010 at 2:34 AM, David Dillow d...@thedillows.org wrote:
On Sat, 2010-01-02 at 13:19 +0100, Bart Van Assche wrote:
This patch fixes this issue by adding support for SRP_CRED_REQ information
units in the SRP initiator. Additionally, this patch makes the per-session
variable
On Wed, Jan 6, 2010 at 10:40 PM, Roland Dreier rdre...@cisco.com wrote:
Is that regular kernel coding practice, to run away with the work
someone else did and to claim authorship ? As far as I know this is
considered as impolite.
I don't see with proper credit for Bart as claiming
On Thu, Jan 7, 2010 at 8:57 AM, Roland Dreier rdre...@cisco.com wrote:
It was only recently that I had the time to analyze this issue in
depth and that I added SRP_CRED_REQ support in SCST (checked in that
support about one week ago). Before SRP_CRED_REQ support was added in
SCST, the
On Thu, Jan 7, 2010 at 9:03 AM, Bart Van Assche
bart.vanass...@gmail.com wrote:
On Thu, Jan 7, 2010 at 8:57 AM, Roland Dreier rdre...@cisco.com wrote:
It was only recently that I had the time to analyze this issue in
depth and that I added SRP_CRED_REQ support in SCST (checked
On Fri, Jan 8, 2010 at 11:39 PM, Chris Worley worl...@gmail.com wrote:
On Fri, Jan 8, 2010 at 3:17 PM, David Dillow d...@thedillows.org wrote:
On Fri, 2010-01-08 at 14:40 -0700, Chris Worley wrote:
On Wed, Jan 6, 2010 at 6:57 PM, David Dillow d...@thedillows.org wrote:
On Wed, 2010-01-06 at
On Sat, Jan 9, 2010 at 6:16 PM, David Dillow d...@thedillows.org wrote:
On Sat, 2010-01-09 at 14:05 +0100, Bart Van Assche wrote:
The SRP spec says that the target must specify the maximum message
size in the SRP_LOGIN_RSP information unit. The largest value one can
set
On Thu, Jan 7, 2010 at 1:16 AM, Chris Worley worl...@gmail.com wrote:
3) In my iSCSI (tgt) results using the HCA as a 10G interface (not
IPoIB, but mlnx4_en), comparing this to the results of using the same
HCA as IB under SRP, I get much better results with SRP when
benchmarking the raw
On Sun, Jan 10, 2010 at 7:41 AM, Roland Dreier rdre...@cisco.com wrote:
The patch I posted is really fixing the original bug. The problem was
that neither the SRP target nor the SRP initiator had support for
SRP_CRED_REQ. Support for SRP_CRED_REQ has to be added to both
software
On Tue, Jan 12, 2010 at 6:13 PM, Roland Dreier rdre...@cisco.com wrote:
Note: I do not know of any SRP targets that implement approach (2). As
far as I know all SRP targets use approach (1).
But no other SRP targets send SRP_CRED_REQ -- and that includes all
native IB storage arrays that
On Wed, Jan 13, 2010 at 12:24 AM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
[ ... ]
Also, I couldn't tell for sure from a cursory examination of the
patch, but the initiator must be designed to not stall processing the
RQ dependent on the SQ or the credit level, when using a
On Tue, Jan 12, 2010 at 11:57 PM, Roland Dreier rdre...@cisco.com wrote:
Having separate buffer sets for command receives and response sends
might seem a sensible way to design a target. But calling
ib_post_recv() before sending a response back to the initiator is not
acceptable
On Wed, Jan 13, 2010 at 12:24 AM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
I'd actually think that the decline caused by exhausting credit on
the initiator side is far worse than ib_post_recv.. If ib_post_recv
really is noticable then maybe just force loading the RQ after XX
On Wed, Jan 13, 2010 at 7:16 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Wed, Jan 13, 2010 at 08:23:27AM +0100, Bart Van Assche wrote:
On Wed, Jan 13, 2010 at 12:24 AM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
[ ... ]
Also, I couldn't tell for sure
On Wed, Jan 20, 2010 at 8:16 PM, Roland Dreier rdre...@cisco.com wrote:
- * may also be |ed in to request a hint about missed events, as
+ * may also be passed in to request a hint about missed events, as
I don't think this is a typo -- the comment is trying to say that the
report
.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
b/drivers/infiniband/ulp/srp/ib_srp.c
index 54c8fe2..f259ab2 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
On Wed, Feb 3, 2010 at 4:30 PM, Or Gerlitz ogerl...@voltaire.com wrote:
The following patch set removes some in efficiencies in the iser data path
through simplification and reducing the amount of code, using less atomic
operations, avoiding TX interrupts, moving to iscsi passthrough mode,
On Wed, Feb 3, 2010 at 5:30 AM, Amerigo Wang amw...@redhat.com wrote:
This patch introduces /proc/sys/net/ipv4/ip_local_reserved_ports,
it can be used like ip_local_port_range, but this is used to
reserve ports for third-party applications which use fixed
port numbers within
On Mon, Jan 11, 2010 at 7:44 PM, Vladislav Bolkhovitin v...@vlnb.net wrote:
[ ... ]
SRP initiator seems to be not too well optimized for the best performance.
ISER initiator is noticeably better in this area.
(replying to an e-mail of one month ago)
I'm not sure the above statement makes
of srp_create_target_ib(). I'm afraid that
it's too late to add an acked-by to that patch ? Anyway:
Acked-by: Bart Van Assche bart.vanass...@gmail.com
Bart.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo
On Wed, Mar 3, 2010 at 9:23 PM, Vladislav Bolkhovitin v...@vlnb.net wrote:
Bart Van Assche, on 03/01/2010 11:38 PM wrote:
On Mon, Mar 1, 2010 at 9:12 PM, Vladislav Bolkhovitin v...@vlnb.net
mailto:v...@vlnb.net wrote:
[ ... ]
It's good if my impression was wrong. But you've got
On Wed, Mar 31, 2010 at 8:18 PM, Roland Dreier rdre...@cisco.com wrote:
The SRP initiator works fine without this patch, and this patch slows
down the SRP initiator.
I do agree that there are no missed callbacks without this patch, but I
don't see how it would slow things down... it
On Wed, Mar 31, 2010 at 8:57 PM, Roland Dreier rdre...@cisco.com wrote:
Processing the flag IB_CQ_REPORT_MISSED_EVENTS causes several IB
drivers to lock and unlock a spinlock. Full-speed SRP I/O can cause
this operation to be invoked more than 100.000 times a second, so the
flag
On Wed, Apr 7, 2010 at 11:39 AM, Dan Carpenter erro...@gmail.com wrote:
The intent here was to check the mfrpl-mapped_page_list allocation.
We checked mfrpl-ibfrpl.page_list earlier.
Signed-off-by: Dan Carpenter erro...@gmail.com
diff --git a/drivers/infiniband/hw/mlx4/mr.c
(resending as plain text)
On Thu, Nov 5, 2009 at 10:51 AM, Philip Pokorny
ppoko...@penguincomputing.com wrote:
[ ... ]
Performance (when it's working) is generally good at 800MB/sec aggregate, but
we'd like to see better. It appeared we were getting 1.3GB/s at one point.
(replying to an
On Thu, Jul 8, 2010 at 2:33 AM, Ira Weiny wei...@llnl.gov wrote:
From 80eecc4046455999254fb312c4ba229b3a52d4c6 Mon Sep 17 00:00:00 2001
From: Ira Weiny wei...@llnl.gov
Date: Wed, 7 Jul 2010 17:35:34 -0700
Subject: [PATCH] ib_qib: Allow writes to the diag_counters to be able to
clear them
On Thu, Jul 8, 2010 at 8:04 PM, Ira Weiny wei...@llnl.gov wrote:
On Thu, 8 Jul 2010 10:37:26 -0700
Bart Van Assche bvanass...@acm.org wrote:
On Thu, Jul 8, 2010 at 2:33 AM, Ira Weiny wei...@llnl.gov wrote:
From 80eecc4046455999254fb312c4ba229b3a52d4c6 Mon Sep 17 00:00:00 2001
From: Ira
On Sat, Jul 10, 2010 at 2:56 AM, Ira Weiny wei...@llnl.gov wrote:
On Fri, 9 Jul 2010 12:33:14 -0700
Bart Van Assche bvanass...@acm.org wrote:
On Thu, Jul 8, 2010 at 8:04 PM, Ira Weiny wei...@llnl.gov wrote:
On Thu, 8 Jul 2010 10:37:26 -0700
Bart Van Assche bvanass...@acm.org wrote
On Mon, Jul 12, 2010 at 6:57 AM, Pradeep Satyanarayana
prade...@linux.vnet.ibm.com wrote:
I realize that the following patch:
https://patchwork.kernel.org/patch/97243/
is queued in your backlog of patches and unlikely that it will go into 2.6.35.
What are the chances that it will make it
On Mon, Jul 12, 2010 at 12:20 PM, Pradeep Satyanarayana
prade...@linux.vnet.ibm.com wrote:
Bart Van Assche wrote:
On Mon, Jul 12, 2010 at 6:57 AM, Pradeep Satyanarayana
prade...@linux.vnet.ibm.com wrote:
I realize that the following patch:
https://patchwork.kernel.org/patch/97243
__srp_get_tx_iu(). This is a
race condition which can result in incorrect manipulation of the
req_lim field. The patch below fixes this race condition by
converting all manipulations of req_lim into atomic operations.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola
One of the most common operations when using the verbs API is to
dequeue and process completions. For many applications, e.g. storage
protocols, processing completions in order is a correctness
requirement. Unfortunately with the current IB verbs API it is not
possible to process completions in
On Sun, Jul 25, 2010 at 8:36 PM, David Dillow d...@thedillows.org wrote:
On Sun, 2010-07-25 at 18:12 +0200, Bart Van Assche wrote:
In the current implementation of ib_srp the req_lim field of
struct srp_target_port can be manipulated in a non-atomic way by
more than one CPU at a time: one
On Mon, Jul 26, 2010 at 4:21 PM, Steve Wise sw...@opengridcomputing.com wrote:
On 07/25/2010 01:54 PM, Bart Van Assche wrote:
[ ... ]
The only way I know of to prevent out-of-order completion processing
with the current OFED verbs API is to protect the whole completion
processing loop
On Mon, Jul 26, 2010 at 9:22 PM, Roland Dreier rdre...@cisco.com wrote:
[ ... ]
Another approach is to just always run the completion processing for a
given CQ on a single CPU and avoid locking entirely. If you want more
CPUs to spread the work, just use multiple CQs and multiple event
Can I conclude from this that the polling loop (2) from
http://www.mail-archive.com/linux-rdma@vger.kernel.org/msg04907.html
won't trigger a race on a multiprocessor when using mlx4 hardware ?
Bart.
On Tue, Jul 27, 2010 at 11:14 AM, Eli Cohen e...@dev.mellanox.co.il wrote:
I don't think this
On Tue, Jul 27, 2010 at 6:50 PM, Roland Dreier rdre...@cisco.com wrote:
[ ... ]
From Documentation/infiniband/core_locking.txt:
The low-level driver is responsible for ensuring that multiple
completion event handlers for the same CQ are not called
simultaneously. The driver must
If an SRP target sends an invalid SRP_RSP information unit to the SRP
initiator this can cause a NULL pointer dereference on the initiator system.
This patch avoids such NULL pointer dereferences and makes sure the SRP
inititator keeps working.
Signed-off-by: Bart Van Assche bart.vanass
Reduces the number of source code lines in ib_srp.c by replacing an explicit
loop by a call to print_hex_dump().
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
b/drivers/infiniband/ulp/srp/ib_srp.c
index ed3f9eb..a072719 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++
as follows:
* Post all receive buffers after channel establishment.
* After a receive buffer has been consumed and processed, post it again.
A side effect is that the ib_post_recv() call is moved out of the SCSI command
processing path.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc
This series of four patches adds SRP_CRED_REQ support in ib_srp, which is a
feature defined by the SRP (draft) standard. Compared to an earlier patch for
implementing this functionality that I posted a few months ago I have done my
best to address all comments that were posted to this mailing
does not change the behavior of ib_srp in any way.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 28 +++-
drivers/infiniband/ulp/srp/ib_srp.h
-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 151 +++
drivers/infiniband/ulp/srp/ib_srp.h |7 ++
include/scsi/srp.h | 19
Makes sure that the SCSI mid-layer never tries to queue more than
(SRP request limit) - 1 SCSI commands. Improves performance for targets whose
request limit is less than or equal to SRP_SQ_REQ_SIZE (63) by reducing the
number of BUSY responses.
Signed-off-by: Bart Van Assche bart.vanass
Exports req_lim via sysfs, which is convenient for debugging purposes.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 14 ++
1 files changed, 14 insertions
On Mon, Aug 2, 2010 at 3:08 PM, Vladislav Bolkhovitin v...@vlnb.net wrote:
Bart Van Assche, on 08/02/2010 12:15 PM wrote:
SRP I/O with small block sizes causes a high CPU load. Processing IB
completions on the context of a kernel thread instead of in interrupt context
allows to process up
On Mon, Aug 2, 2010 at 8:36 PM, David Dillow d...@thedillows.org wrote:
On Mon, 2010-08-02 at 22:16 +0400, Vladislav Bolkhovitin wrote:
Bart Van Assche, on 08/02/2010 07:57 PM wrote:
block size number of IOPS IOPS IOPS
in bytes threads without
On Mon, Aug 2, 2010 at 11:23 PM, David Dillow d...@thedillows.org wrote:
On Mon, 2010-08-02 at 17:32 +0200, Bart Van Assche wrote:
[ ... ]
@@ -1051,7 +1052,7 @@ static struct srp_iu *__srp_get_tx_iu(struct
srp_target_port *target,
srp_send_completion(target-send_cq, target
On Mon, Aug 2, 2010 at 11:44 PM, David Dillow d...@thedillows.org wrote:
On Mon, 2010-08-02 at 17:32 +0200, Bart Van Assche wrote:
[ ... ]
There seems to be an accounting change from using tx_head/tx_tail to
using tx_req and tx_rsp. I'm not sure that really makes sense --
splitting
This series of four patches adds SRP_CRED_REQ support in ib_srp, which is a
feature defined by the SRP (draft) standard. These patches depend on the
IB/srp receive buffer handling robustness improvement patch posted a few
days ago.
Changes in v3 compared to v2:
- Corrected completion queue size
the size of
the IB send completion queue by one element.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
b/drivers/infiniband/ulp/srp/ib_srp.c
index bc2d512..1284bc3
of
allocated requests and responses.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 157 +-
drivers/infiniband/ulp/srp/ib_srp.h | 14
for targets
whose request limit is less than or equal to SRP_SQ_REQ_SIZE (63) by reducing
the number of BUSY responses reported by ib_srp to the SCSI mid-layer.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
Exports req_lim via sysfs, which is convenient for debugging purposes.
Signed-off-by: Bart Van Assche bart.vanass...@gmail.com
Cc: Roland Dreier rola...@cisco.com
Acked-by: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 14 ++
1 files changed, 14
On Tue, Aug 3, 2010 at 5:24 PM, David Dillow d...@thedillows.org wrote:
On Tue, 2010-08-03 at 16:08 +0200, Bart Van Assche wrote:
The SRP (draft) standard specifies that an SRP initiator must never queue
more
than (SRP request limit) - 1 unanswered SRP_CMD information units. This patch
makes
On Wed, Aug 4, 2010 at 10:40 PM, Roland Dreier rdre...@cisco.com wrote:
I believe this is a wrong approach for this problem. You are
workarounding it, not solving, and introducing a bad side effect of
additional context switch per command, so increasing its processing
latency. It
On Tue, Jul 27, 2010 at 8:20 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Tue, Jul 27, 2010 at 08:03:25PM +0200, Bart Van Assche wrote:
As far as I know it is not possible for a HCA to tell whether or not a
CPU has finished executing the interrupt it triggered. So
On Sat, Aug 7, 2010 at 6:32 PM, Roland Dreier rdre...@cisco.com wrote:
Not sure that I follow the problem you're worried about. A given
tasklet can only be running on one CPU at any one time -- if an
interrupt occurs and reschedules the tasklet then it just runs again
when it exits.
Also
On Mon, Aug 9, 2010 at 1:51 AM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
[ ... ]
Further, the approach you outline in your follow on message for
blkio, has problems.. Look at how IPOIB does NAPI to see how
this must look.
1) ib_req_notify_cq must only be called if you are
On Tue, Aug 3, 2010 at 5:44 PM, David Dillow d...@thedillows.org wrote:
On Tue, 2010-08-03 at 17:26 +0200, Bart Van Assche wrote:
[ ... ]
I'm not sure it is a good idea to allow that all transmit buffers get
allocated for sending CMD_RSP information units and that none remain
for replying
This series of three patches adds SRP_CRED_REQ support in ib_srp, which is a
feature defined in the SRP (draft) standard.
Changes in v4 compared to v3:
- Dropped the fourth patch since it has been merged.
- Introduced the symbolic constant SRP_TSK_MGMT_RSV, which represents the
number of slots
,
increases the size of the IB send completion queue by one element and
reserves one transmit ring slot for SRP_TSK_MGMT requests.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 33
.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 222 +--
drivers/infiniband/ulp/srp/ib_srp.h |7 +-
include/scsi/srp.h | 14
for targets
whose request limit is less than or equal to
SRP_SQ_REQ_SIZE - SRP_TSK_MGMT_RSV (62) by reducing the number of BUSY
responses reported by ib_srp to the SCSI mid-layer.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d
On Tue, Aug 10, 2010 at 8:34 PM, Bart Van Assche bvanass...@acm.org wrote:
The SRP (draft) standard specifies that an SRP initiator must never queue more
Note: this patch still has the old (incorrect) patch description. Will
post a patch with an updated description soon.
Bart.
--
To unsubscribe
or equal to
SRP_SQ_REQ_SIZE - SRP_TSK_MGMT_RSV (62) by reducing the number of BUSY
responses reported by ib_srp to the SCSI mid-layer.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c
On Fri, Aug 13, 2010 at 9:55 PM, David Dillow d...@thedillows.org wrote:
On Tue, 2010-08-10 at 20:33 +0200, Bart Van Assche wrote:
Implements SRP_CRED_REQ, which is an information unit defined in the SRP
(draft) standard and that allows an SRP target to inform an SRP initiator
that
more
On Fri, Aug 13, 2010 at 8:12 PM, David Dillow d...@thedillows.org wrote:
On Tue, 2010-08-10 at 09:55 +0200, Bart Van Assche wrote:
On Tue, Aug 3, 2010 at 5:44 PM, David Dillow d...@thedillows.org wrote:
On Tue, 2010-08-03 at 17:26 +0200, Bart Van Assche wrote:
[ ... ]
I'm not sure
On Fri, Aug 13, 2010 at 8:24 PM, David Dillow d...@thedillows.org wrote:
+
+ rsv = (req_type == SRP_REQ_NORMAL) ? SRP_TSK_MGMT_RSV : 0;
srp_send_completion(target-send_cq, target);
- if (target-tx_head - target-tx_tail = SRP_SQ_SIZE)
+ if (target-tx_head -
On Sat, Aug 14, 2010 at 7:04 PM, David Dillow d...@thedillows.org wrote:
On Sat, 2010-08-14 at 09:35 +0200, Bart Van Assche wrote:
On Fri, Aug 13, 2010 at 8:24 PM, David Dillow d...@thedillows.org wrote:
+
+ rsv = (req_type == SRP_REQ_NORMAL) ? SRP_TSK_MGMT_RSV : 0
This series of three patches adds SRP_CRED_REQ ans SRP_AER_REQ support in
ib_srp. Both are information units defined in the SRP (draft) standard.
Changes in v5 compared to v4:
- Added support for processing the SRP_AER_REQ information unit.
- Introduced a new symbolic name,
,
increases the size of the IB send completion queue by one element and
reserves one transmit ring slot for SRP_TSK_MGMT requests.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 30
to add more forward declarations.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c | 244 ++-
drivers/infiniband/ulp/srp/ib_srp.h |7 +-
include
or equal to
SRP_NORMAL_REQ_SQ_SIZE (62) by reducing the number of BUSY responses reported
by ib_srp to the SCSI mid-layer.
Signed-off-by: Bart Van Assche bvanass...@acm.org
Cc: Roland Dreier rola...@cisco.com
Cc: David Dillow d...@thedillows.org
---
drivers/infiniband/ulp/srp/ib_srp.c |9
On Thu, Aug 19, 2010 at 2:48 AM, David Dillow d...@thedillows.org wrote:
On Mon, 2010-08-16 at 20:55 +0200, Bart Van Assche wrote:
Implements SRP_CRED_REQ and SRP_AER_REQ, which are information units defined
in the SRP (draft) standard. Adds declarations for the SRP_CRED_REQ,
SRP_CRED_RSP
On Thu, Aug 19, 2010 at 12:52 PM, David Dillow d...@thedillows.org wrote:
On Thu, 2010-08-19 at 12:10 +0200, Bart Van Assche wrote:
On Thu, Aug 19, 2010 at 2:48 AM, David Dillow d...@thedillows.org wrote:
[ ... ]
If I do not receive any further feedback,
I will modify srp_handle_req
On Thu, Aug 19, 2010 at 12:44 PM, David Dillow d...@thedillows.org wrote:
On Thu, 2010-08-19 at 08:04 +0200, Bart Van Assche wrote:
On Thu, Aug 19, 2010 at 1:44 AM, David Dillow d...@thedillows.org wrote:
On Mon, 2010-08-16 at 20:54 +0200, Bart Van Assche wrote:
@@ -989,19 +989,21
On Thu, Aug 19, 2010 at 7:25 PM, David Dillow d...@thedillows.org wrote:
On Thu, 2010-08-19 at 18:53 +0200, Bart Van Assche wrote:
On Thu, Aug 19, 2010 at 12:44 PM, David Dillow d...@thedillows.org wrote:
Put the assignment on the same line as the declaration and I'll deal
with the name
On Fri, Aug 20, 2010 at 9:49 AM, Bernd Schubert
bs_li...@aakef.fastmail.fm wrote:
In ib_srp.c sg_tablesize is defined as 255. With that value we see lots of IO
requests of size 1020. As I already wrote on linux-scsi, that is really sub-
optimal for DDN storage, as lots of IO requests of size
On Sat, Aug 21, 2010 at 6:27 PM, David Dillow dillo...@ornl.gov wrote:
On Sat, 2010-08-21 at 13:14 +0200, Bart Van Assche wrote:
On Fri, Aug 20, 2010 at 9:49 AM, Bernd Schubert
bs_li...@aakef.fastmail.fm wrote:
In ib_srp.c sg_tablesize is defined as 255. With that value we see lots
On Sat, Aug 21, 2010 at 8:20 PM, Bernd Schubert
bs_li...@aakef.fastmail.fm wrote:
On Saturday, August 21, 2010, Bart Van Assche wrote:
[ ... ]
task attribute SIMPLE, so a target is allowed to process these
requests concurrently. For the ib_srpt target I see the following
results over a single
1 - 100 of 1130 matches
Mail list logo