Hi Sasha,
On Tue, Jun 1, 2010 at 11:32 AM, Sasha Khapyorsky sas...@voltaire.com wrote:
Hi Hal,
On 10:11 Wed 16 Dec , Hal Rosenstock wrote:
In order to better handle non responsive SMAs (when link is physically up
but the SMA does not respond), a rate based mechanism for SMPs is added
rather than just switches for common SystemImageGUID
Now consistent with original grouping.c from which this was ported
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
---
infiniband-diags/libibnetdisc/src/chassis.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git
when grouping by SystemImageGUID
Without doing this, find_chassisguid will always return without
finding the already discovered chassis so no grouping occurs
when it should
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
---
infiniband-diags/libibnetdisc/src/chassis.c |4
1
On 18:37 Tue 01 Jun , Ira Weiny wrote:
From: Ira Weiny wei...@llnl.gov
Date: Mon, 22 Dec 2008 11:03:59 -0800
Subject: [PATCH] opensm/osm_perfmgr.c: Remove unnecessary lock reference from
Performance Manager object
Signed-off-by: Ira Weiny wei...@llnl.gov
Applied. Thanks.
Sasha
--
On 09:04 Wed 02 Jun , Hal Rosenstock wrote:
when grouping by SystemImageGUID
Without doing this, find_chassisguid will always return without
finding the already discovered chassis so no grouping occurs
when it should
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Both
On 10:29 Wed 02 Jun , Yevgeny Kliteynik wrote:
Right now gcc allows you to do implicit casting in both ways, g++ doesn't.
OpenSM is written in C, not C++.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
On 06:58 Wed 02 Jun , Hal Rosenstock wrote:
I had started with an algorithm along these lines but evolved towards
the proposed one based on CPU utilization. An algorithm along the
lines of the above wastes CPU (when idling and other times) which
significantly impacts any other apps
On Wed, 2010-05-26 at 02:16 -0700, David Miller wrote:
From: Eli Cohen e...@dev.mellanox.co.il
Date: Wed, 26 May 2010 12:08:52 +0300
So if I understand you correctly, you think that I should not bother
to set a default value of 1. Each driver that cares about the value
of this field,
setlinebuf() is pretty intuitive to understand, compared to setvbuf().
I finally applied this; however in the end I decided to do
setvbuf(stdout, NULL, _IOLBF, 0);
instead of setlinebuf(), since in the past I've prefered more pedantic
stuff (eg posix_memalign instead of memalign) to
From: Ben Hutchings bhutchi...@solarflare.com
Date: Wed, 02 Jun 2010 17:42:29 +0100
On Wed, 2010-05-26 at 02:16 -0700, David Miller wrote:
From: Eli Cohen e...@dev.mellanox.co.il
Date: Wed, 26 May 2010 12:08:52 +0300
So if I understand you correctly, you think that I should not bother
to
On Wed, Jun 02, 2010 at 10:50:56AM -0500, Steve Wise wrote:
I have this library I'm writing that uses libibverbs. Somehow it seems
to be using the 1.0 version instead of the newer version (I'm guessing).
When I call ibv_create_cq() I get a seg fault and the stack looks like
this:
Make
Roland Dreier wrote:
I have this library I'm writing that uses libibverbs. Somehow it
seems to be using the 1.0 version instead of the newer version (I'm
guessing). When I call ibv_create_cq() I get a seg fault and the
stack looks like this:
#0 0x00319d80871d in
Fix warnings in windows build.
These are typecasts in several places, with a couple additional changes
needed to libibnetdisc. Windows uses ISO definitions instead of
POSIX definitions, so calls like _read, _write, etc. are used on
Windows instead of read, write. The result is that internal
qib has the code
void qib_disarm_piobufs_set(struct qib_devdata *dd, unsigned long *mask,
unsigned cnt)
{
struct qib_pportdata *ppd, *pppd[dd-num_pports];
it would probably be safer to avoid the variable length array pppd[] on
without patch:
1M memory region120usec
16M memory region 1970usec
with patch v2:
1M memory region172usec
16M memory region 2030usec
So if I read this correctly this patch introduces almost a 50% overhead
in the 1M case... and probably much worse (as a fraction) in say
if (status 0) {
-ep-com.cm_id-rem_ref(ep-com.cm_id);
+if (ep-com.cm_id)
+ep-com.cm_id-rem_ref(ep-com.cm_id);
Steve, does this make sense?
--
Roland Dreier rola...@cisco.com || For corporate legal information go to:
Sure. I will work on a patch.
On Wed, 2010-06-02 at 14:37 -0700, Roland Dreier wrote:
qib has the code
void qib_disarm_piobufs_set(struct qib_devdata *dd, unsigned long *mask,
unsigned cnt)
{
struct qib_pportdata *ppd,
Rather than use a variable size array allocation on the stack,
define a constant for the maximum array size possible.
Signed-off-by: Ralph Campbell ralph.campb...@qlogic.com
---
drivers/infiniband/hw/qib/qib.h|3 +++
drivers/infiniband/hw/qib/qib_tx.c |2 +-
2 files changed, 4
Roland Dreier wrote:
if (status 0) {
- ep-com.cm_id-rem_ref(ep-com.cm_id);
+ if (ep-com.cm_id)
+ ep-com.cm_id-rem_ref(ep-com.cm_id);
Steve, does this make sense?
I actually think that ep-com.cm_id is always valid when
connect_reply_upcall() is
I actually think that ep-com.cm_id is always valid when
connect_reply_upcall() is called. But I'd have to test it more. I'd
rather not add this change unless someone convinces me that there
actually is a path where the cm_id null...
Then would it make sense to change the code
This replace the PCI DMA state API (include/linux/pci-dma.h) with the
DMA equivalents since the PCI DMA state API will be obsolete.
No functional change.
For further information about the background:
http://marc.info/?l=linux-netdevm=127037540020276w=2
Signed-off-by: FUJITA Tomonori
21 matches
Mail list logo
