[EMAIL PROTECTED] wrote on 09.10.2007 22:19:17:
I didn't see a response to my earlier email about the other uses of
min_t(int, x, INT_MAX) so I fixed it up myself and added this to my
tree. I don't have a working setup to test yet so please let me know
if you see anything wrong with this:
cheapxpsoftware . com
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
HR == Hal Rosenstock writes:
HR Here's a patch to fix this for OFED 1.3:
While you're there, can you change pointer dereference with memcpy,
e.g.:
memcpy (capmask, pc+2, sizeof(capmask));
capmask = ntohs(capmask);
Those pointer dereferenes are royal pain on ia64 unless
I'm not sure what you mean. During the 2.6.23 cycle I've been sending
any patches that potentially could conflict with the net-2.6 tree to
you and Jeff so that you can merge them upstream via your tree. Or do
you mean Jeff should become the maintainer of drivers/infiniband??
Not
On Fri, 12 Oct 2007 09:08:58 -0700
Brandeburg, Jesse [EMAIL PROTECTED] wrote:
Andi Kleen wrote:
When the hw TX queue gains space, the driver self-batches packets
from the sw queue to the hw queue.
I don't really see the advantage over the qdisc in that scheme.
It's certainly not
While working on this I observed that for mthca max_srq_sge
returned by ib_query_device() is not equal to max_sge returned
by ib_query_srq(). Why is that?
Not sure. I'll take a look. What are the two values that you get?
I get 28 and 16. This is on InfiniBand: Mellanox Technologies
related to this comment, does Linux have a lockless (using atomics)
singly linked list element? That would be very useful in a driver hot
path.
No; it doesn't. At least not a portable one.
Besides they tend to be not faster anyways because e.g. cmpxchg tends
to be as slow as an explicit
Use RCU? or write a generic version and get it reviewed. You really
want someone with knowledge of all the possible barrier impacts to
review it.
I guess he was thinking of using cmpxchg; but we don't support this
in portable code.
RCU is not really suitable for this because it assume
I'd like to patch the OFED-1.2.5 source file ib_srp.h (or use the modified
source file) and rebuild the source RPM (whichever one ib_srp.h comes
from) and
the OFED 1.2.5 distribution package. Just to make things nice and neat for
local use. The goal is to have a local OFED distribution package
cheapxpsoftware . com
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
infiniband-diags/perfquery.c: Fix issues when checking
PerfMgt:ClassPortInfo.CapabilityMask
1. bit 9, if we're counting from 0, will have mask of 0x200,
not 0x100. mask of 0x100 will be for counter aggregation according
to IBA 1.2.
2. If capmask is 16 bit big-endian word, then we're
Linus Torvalds wrote:
Oh, and obviously, the NAPI changes may well have resulted in a merge that
had no actual *conflicts* in it, but whether the end result works or not
(and whether any IB drivers need updating due to the NAPI changes), I
cannot tell. I've pushed out my tree, so people who
cheapxpsoftware . com
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
This email was generated automatically, please do not reply
git_url: git://git.openfabrics.org/ofed_1_3/linux-2.6.git
git_branch: ofed_kernel
Common build parameters: --with-ipoib-mod --with-sdp-mod --with-srp-mod
--with-user_mad-mod --with-user_access-mod --with-mthca-mod --with-mlx4-mod
In OFED 1.2 perfquery attempts to check if a port supports extended
counters:
} else {
/* Should ClassPortInfo be implemented in libibmad ? */
pc2 = (uint16_t *)pc[2]; /* CapabilityMask */
cap_mask = *pc2;
if (!(cap_mask 0x100))
This is a draft patch to address the following bug:
https://bugs.openfabrics.org/show_bug.cgi?id=728
Might be nice to include a description with the patch, so everyone
doesn't have to go figure out this bug report (the issue I guess is
that ehca doesn't support enough SG entries to handle 16
Doug Ledford wrote:
On Thu, 2007-10-11 at 18:51 -0700, Sean Hefty wrote:
I'm more referring to when you call rdma_bind_addr to bind to your
device before you call rdma_connect. In that instance, your address
isn't for the eventual destination, but just to bind you to your local
rdma device.
On Fri, 2007-10-12 at 12:09 +0530, Sumit Gaur - Sun Microsystem wrote:
Hi ,
Sean Hefty wrote:
There is no per thread demuxing. You would need two different mad agents
to do this with one looking at the SMI side and the other the GSI side.
I haven't looked at libibmad in terms of using this
HR == Hal Rosenstock writes:
HR infiniband-diags/perfquery.c: Fix issues when checking
HR PerfMgt:ClassPortInfo.CapabilityMask
Looks good. Thanks.
max
___
general mailing list
general@lists.openfabrics.org
On Fri, 2007-10-12 at 22:50 +1000, Max Matveev wrote:
HR == Hal Rosenstock writes:
HR Here's a patch to fix this for OFED 1.3:
While you're there, can you change pointer dereference with memcpy,
e.g.:
memcpy (capmask, pc+2, sizeof(capmask));
capmask =
I am getting this error with the current packaged releases:
# ibcheckerrors
perfquery: iberror: failed: perfquery
Error check on lid 2 (Topspin DDR-HCAe LX x8) port all: FAILED
perfquery: iberror: failed: perfquery
Error check on lid 1 (ibtest1 HCA-1) port all: FAILED
## Summary: 2 nodes
On Fri, 2007-10-12 at 13:58 -0700, Greg Kurtzer wrote:
# perfquery 1 -a
perfquery: iberror: failed: perfquery
# perfquery 2 -a
perfquery: iberror: failed: perfquery
That's what I thought was going on.
Thanks!
On Oct 12, 2007, at 1:56 PM, Hal Rosenstock wrote:
Greg,
On Fri,
While working on this I observed that for mthca max_srq_sge
returned by ib_query_device() is not equal to max_sge returned
by ib_query_srq(). Why is that?
Not sure. I'll take a look. What are the two values that you get?
I get 28 and 16. This is on InfiniBand: Mellanox
My setup is as follows:
I have 4 ports on a DDN, each port has 2 LUNs mapped to it.
I have 4 ports active on the SRP initiator machine. I want to do a one to
one mapping of SRP initiator ports to SRP target ports. All ports are
connected via a switch.
Here are the 4 ports as seen by SRP:
Please pull from:
git://git.openfabrics.org/~shefty/rdma-dev.git for-roland
This will pick up a couple of recent rdma_cm bug fixes.
drivers/infiniband/core/cma.c | 160 +-
1 files changed, 83 insertions(+), 77 deletions(-)
Sean Hefty (2):
On Thu, 11 Oct 2007 18:17:19 -0700 (PDT)
David Miller [EMAIL PROTECTED] wrote:
From: Roland Dreier [EMAIL PROTECTED]
Date: Thu, 11 Oct 2007 18:08:52 -0700
This will get the batch of changes queued up for the 2.6.24 merge
window (although I still have a few more things to merge later, once
Attn:Winner
Congratulations, The Foundazion Di Vittorio has chosen you by the board
of trustees as one of the final recipients ofa cash Grant/Donation for
your own personal,educational,and business Tocelebrate the 30th
anniversary 2007 program,We are giving outayearlydonation
of
cheapxpsoftware . com
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
OSM Simulation Regression Summary
[Generated mail - please do NOT reply]
OpenSM binary date = 2007-10-12
OpenSM git rev = Tue_Oct_2_22:28:56_2007
[d5c34ddc158599abff9f09a6cc6c8cad67745f0b]
ibutils git rev = Tue_Sep_4_17:57:34_2007
[4bf283f6a0d7c0264c3a1d2de92745e457585fdb]
Total=520
29 matches
Mail list logo