On Mon, 27 Dec 2004 21:50:47 -0800
Roland Dreier <[EMAIL PROTECTED]> wrote:
> > "David" == David S Miller <[EMAIL PROTECTED]> writes:
>
> David> Send it all over.
>
> OK, you asked for it... here's our latest tree, which should
> incorporate all the feedback I've seen.
W00t :-) All app
By, the way does anyone know what IPD is supposed to be used when the
injection port's rate is not a multiple of the rate for the path? For
example, suppose an 8X DDR port is sending to a 12X path -- 12 doesn't
divide 16 evenly.
I would guess that the correct thing to do is round up, so 8X DDR
->
Add OpenIB maintainers information to MAINTAINERS.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- linux-bk.orig/MAINTAINERS 2004-12-27 21:47:44.140300651 -0800
+++ linux-bk/MAINTAINERS2004-12-27 21:48:28.966702428 -0800
@@ -1081,6 +1081,17 @@
L: [EMAIL PROTECTED]
S: Mai
Add files to Documentation/infiniband that describe the tree under
/sys/class/infiniband, the IPoIB driver and the userspace MAD access driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/Documentation/infiniband/ipoib.txt 200
Add the 0x1b ioctl magic number used by ib_umad module to
Documentation/ioctl-number.txt.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- linux-bk.orig/Documentation/ioctl-number.txt2004-12-27
21:47:59.407053483 -0800
+++ linux-bk/Documentation/ioctl-number.txt 2004-12-27 21:48
Add a driver that provides a character special device for each
InfiniBand port. This device allows userspace to send and receive
MADs via write() and read() (with some control operations implemented
as ioctls).
All operations are 32/64 clean and have been tested with 32-bit
userspace running on a
Add functions for handling IPoIB multicast and multiple partitions.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/ulp/ipoib/ipoib_multicast.c 2004-12-27
21:48:27.157968669 -0800
@@ -0,0 +1,981 @@
+/*
+ *
Add ipv6_ib_mc_map() to convert IPv6 multicast addresses to IPoIB
hardware addresses, and add support for autoconfiguration for devices
with type ARPHRD_INFINIBAND.
The mapping for multicast addresses is described in
http://www.ietf.org/internet-drafts/draft-ietf-ipoib-ip-over-infiniband-08.txt
Add ip_ib_mc_map() to convert IPv4 multicast addresses to IPoIB
hardware addresses. Also add so INFINIBAND_ALEN
has a home.
The mapping for multicast addresses is described in
http://www.ietf.org/internet-drafts/draft-ietf-ipoib-ip-over-infiniband-08.txt
Signed-off-by: Roland Dreier <[EMAIL P
Add MAD (management datagram) code for Mellanox HCA driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_mad.c2004-12-27
21:48:24.331384733 -0800
@@ -0,0 +1,320 @@
+/*
+ * Copyright (c) 20
Add code for remaining InfiniBand objects (address vectors, multicast
groups, memory regions and protection domains)
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_av.c 2004-12-27
21:48:23.8
Add device initializaton code for Mellanox HCA driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_profile.c2004-12-27
21:48:23.120562962 -0800
@@ -0,0 +1,226 @@
+/*
+ * Copyright (c)
Add event queue code for Mellanox HCA driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_eq.c 2004-12-27
21:48:22.766615062 -0800
@@ -0,0 +1,690 @@
+/*
+ * Copyright (c) 2004 Topspin Com
Add firmware command processing code for Mellanox HCA driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_cmd.c2004-12-27
21:48:22.369673490 -0800
@@ -0,0 +1,1573 @@
+/*
+ * Copyright (c)
Add midlayer interface code for Mellanox HCA driver.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_provider.c 2004-12-27
21:48:22.043721469 -0800
@@ -0,0 +1,627 @@
+/*
+ * Copyright (c) 2
Add a low-level driver for Mellanox MT23108 and MT25208 HCAs. The
MT25208 is only fully supported when in MT23108 compatibility mode;
only the very beginnings of support for native MT25208 mode (required
for HCAs without local memory) is present.
(As a side note, I believe this driver would be th
Add support for sending queries to the SA (Subnet Administration). In
particular the PathRecord and MCMember (multicast group member) used
by the IP-over-InfiniBand driver are implemented.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- linux-bk.orig/drivers/infiniband/core/Makefile 2
Add MAD layer SMI (Subnet Management Interface) code.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/core/smi.c 2004-12-27 21:48:20.566938847
-0800
@@ -0,0 +1,234 @@
+/*
+ * Copyright (c) 2004 Mellanox Te
Add MAD layer private implementation headers.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/core/agent.h2004-12-27 21:48:20.224989180
-0800
@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) 2004 Mellanox Technologie
Add public headers for handling InfiniBand MADs (management
datagrams), including sending and receiving MADs as well as passing
MADs on to local agents.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0 +
+++ linux-bk/drivers/infiniband/include/ib
Add the appropriate lines to drivers/Kconfig and drivers/Makefile so
that the kernel configuration and build systems know about drivers/infiniband.
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- linux-bk.orig/drivers/Kconfig 2004-12-27 21:47:59.198084242 -0800
+++ linux-bk/drivers/Kc
Add public headers for core InfiniBand support. This can be thought
of as a midlayer that provides an abstraction between low-level
hardware drivers and upper level protocols (such as
IP-over-InfiniBand).
Signed-off-by: Roland Dreier <[EMAIL PROTECTED]>
--- /dev/null 1970-01-01 00:00:00.0
> "David" == David S Miller <[EMAIL PROTECTED]> writes:
David> Send it all over.
OK, you asked for it... here's our latest tree, which should
incorporate all the feedback I've seen.
(Individuals trimmed from CC list, since they probably don't want to
get all 24 patches over again)
Thank
Here's a patch that adds active_width and active_speed members to the
port attributes structure. This is one step towards having IPoIB set
the static rate of its address handles properly.
I also added a "rate" sysfs attribute for each port, which shows the
data rate. I don't have any 12X or DDR
On Mon, 27 Dec 2004 21:19:26 -0800
Roland Dreier <[EMAIL PROTECTED]> wrote:
> Dave, did you want to handle the entire merge of the whole IB stack,
> or just the net/ parts, which are pretty trivial and stand alone,
> since AF_INFINIBAND is already defined in the tree?
Send it all over.
__
David> If Roland can resubmit his patch queue to me with the fixes
David> folks have recommended to him, I can start merging this
David> stuff in.
Andrew has indicated to me that he has merged the whole InfiniBand
patch into -mm now.
Dave, did you want to handle the entire merge of th
On Mon, 20 Dec 2004 15:38:45 +0900 (JST)
YOSHIFUJI Hideaki / [EMAIL PROTECTED](B <[EMAIL PROTECTED]> wrote:
> In article <[EMAIL PROTECTED]> (at Sun, 19 Dec 2004 22:14:43 -0800), Roland
> Dreier <[EMAIL PROTECTED]> says:
>
> > The following series of patches is the latest version of the OpenIB
>
I've just committed skeleton versions of userspace/libibverbs and
userspace/libmthca. This is mostly to get early feedback on the
structure of the code, etc. -- what's there doesn't do much that's
useful. libibverbs is the device-independent part of userspace verbs,
and is what applications actua
Michael> Could you please remove openib/src/userspace/osm.old as a
Michael> first step? Its really painful to get two copies of
Michael> opensm each time. If you thjink you need it for
Michael> reference, you could always just move it somewhere ...
I agree osm.old should be delet
Hello!
Quoting r. shaharf ([EMAIL PROTECTED]) "RE: [openib-general] Problem with
userspace tree structure":
>
>
> > -Original Message-
> > From: Roland Dreier [mailto:[EMAIL PROTECTED]
> > Sent: Monday, December 27, 2004 6:37 PM
> > To: shaharf
> > Cc: openib-general@openib.org
> > Subje
Hi
all,
I
added very simple retries mechanism to the SM. It is not really checked,
because I don’t have a controlled way to loose packets.
It
is committed. Please let me know if it you encounter related problems (or
non-related problems ;-)
Shahar
Hello!
Quoting r. Roland Dreier ([EMAIL PROTECTED]) "Re: [openib-general] Problem with
userspace tree structure":
> We already have osm, diags, and tvflash which are all independent.
> I'll commit directories for libibvers and libmthca, which will be two
> more independent projects.
Could you add
shaharf> OK. How about the following tree?
That's OK, but why do we need a hierarchy? Why not just put every
package in a top-level directory like:
tvflash
management
libibverbs
libmthca
mpi
and so on.
- Roland
___
openib-gen
> -Original Message-
> From: Roland Dreier [mailto:[EMAIL PROTECTED]
> Sent: Monday, December 27, 2004 6:37 PM
> To: shaharf
> Cc: openib-general@openib.org
> Subject: Re: [openib-general] Problem with userspace tree structure
>
> We already have osm, diags, and tvflash which are all ind
shaharf> I am not sure everything in usermode is separate
shaharf> project. For example, osm, util, lib* do belong to the
shaharf> same package. The dependencies argument is for handling
shaharf> everything under the same root. You are correct about the
shaharf> remake problem,
>
> I think gnome is a pretty good parallel, because the contents of
> userspace/ are really separate projects, with different sets of
> developers and different maintainers. Having one main makefile is
> confusing for users and packagers, and the current scheme is also
> error prone, since it's
Ido> 1. We can divide the single CQ into two separate completion
Ido> queues: one for the RQ and the other for SQ. Then we can
Ido> change the CQ policy affiliated with the SQ into
Ido> IB_CQ_CONSUMER_REARM and in mainstream not arm the CQ. In
Ido> such case the poll_cq_tq wil
Hello
I have been investigating the performance of the IPoIB for a while and I
have discovered 3 interesting points
which I think are worth implementing in gen2.
1. We can divide the single CQ into two separate completion queues: one for
the RQ and the other for SQ.
Then we can change the CQ p
Tom,
I debugged the opensm a little, and it seems like a simple
opensm bug. Somehow the destination lid of the report target was not
found, and as there were no checks for that, the opensm faulted. I added
the missing checks. I don't know how this happened - i.e. why was the
dest port inval
39 matches
Mail list logo
