<feed xmlns='http://www.w3.org/2005/Atom'>
<title>kernel.git/drivers/infiniband/core, branch linux-2.6.21.y</title>
<subtitle>Hosts the 0x221E linux distro kernel.
</subtitle>
<id>https://git.0xinfinity.dev/distro/kernel.git/atom?h=linux-2.6.21.y</id>
<link rel='self' href='https://git.0xinfinity.dev/distro/kernel.git/atom?h=linux-2.6.21.y'/>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/'/>
<updated>2007-03-06T22:58:11Z</updated>
<entry>
<title>RDMA/ucma: Avoid sending reject if backlog is full</title>
<updated>2007-03-06T22:58:11Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-03-06T19:58:32Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=3492856e33da734501f2bdd8656cbfdf61f60330'/>
<id>urn:sha1:3492856e33da734501f2bdd8656cbfdf61f60330</id>
<content type='text'>
Change the returned error code to ENOMEM if the connection event
backlog is full.  This prevents the ib_cm from issuing a reject
on the connection, which can allow retries to succeed.

Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>RDMA/cma: Initialize rdma_bind_list in cma_alloc_any_port()</title>
<updated>2007-03-06T20:41:44Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-03-05T20:50:17Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=cb164b8c6a2bdf995c938e5f157d41465b18e5c3'/>
<id>urn:sha1:cb164b8c6a2bdf995c938e5f157d41465b18e5c3</id>
<content type='text'>
The struct rdma_bind_list fields for hlist are not being initialized,
resulting in a corrupted list.  Fix this by using kzalloc() to make
sure all pointers are NULL.

Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>RDMA/cma: Remove unused node_guid from cma_device structure</title>
<updated>2007-02-23T01:54:35Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-02-22T19:37:44Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=1836854f25b1bc63766bff06aeeb83d2a602b050'/>
<id>urn:sha1:1836854f25b1bc63766bff06aeeb83d2a602b050</id>
<content type='text'>
Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>IB/cm: Remove ca_guid from cm_device structure</title>
<updated>2007-02-23T01:54:33Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-02-22T19:37:38Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=e971b8cd19d39366b9fdc9eadafec988d785264d'/>
<id>urn:sha1:e971b8cd19d39366b9fdc9eadafec988d785264d</id>
<content type='text'>
The cm_device references an ib_device, which already contains the node_guid.

Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>RDMA/cma: Request reversible paths only</title>
<updated>2007-02-23T01:54:07Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-02-22T00:40:44Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=962063e64b0c55d270979fa0e4ae26daedac6282'/>
<id>urn:sha1:962063e64b0c55d270979fa0e4ae26daedac6282</id>
<content type='text'>
The rdma_cm requires that path records be reversible.  Set the
reversible bit when issuing an path record query.

Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>IB/core: Set hop limit in ib_init_ah_from_wc correctly</title>
<updated>2007-02-23T01:54:02Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-02-22T00:37:31Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=47645d8d25387c08bb3ccd84e1405c3776f21d24'/>
<id>urn:sha1:47645d8d25387c08bb3ccd84e1405c3776f21d24</id>
<content type='text'>
The hop_limit value in the ah_attr should be 0xFF, not the value read
from the received GRH (which should be 0).  See 13.5.4.4 in the 1.2 IB
spec.

Signed-off-by: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>IB/uverbs: Return correct error for invalid PD in register MR</title>
<updated>2007-02-22T21:16:51Z</updated>
<author>
<name>Roland Dreier</name>
</author>
<published>2007-02-22T21:16:51Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=aaf1aef55f50f53812871693692c7cbefcd57f39'/>
<id>urn:sha1:aaf1aef55f50f53812871693692c7cbefcd57f39</id>
<content type='text'>
If no matching PD is found in ib_uverbs_reg_mr(), then the function
jumps to err_release without setting the return value ret.  This means
that ret will hold the return value of the call to ib_umem_get() a few
lines earlier; if the function reaches the point where it looks for
the PD, we know that ib_umem_get() must have returned 0, so
ib_uverbs_reg_mr() ends up return 0 for a bad PD ID.  Fix this by
setting ret to -EINVAL before jumping to the exit path when no PD is
found.

Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>IB/core: Set static rate in ib_init_ah_from_path()</title>
<updated>2007-02-16T23:31:24Z</updated>
<author>
<name>Roland Dreier</name>
</author>
<published>2007-02-16T23:31:24Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=7084f8429c940bac856123ce6d3946638fe20364'/>
<id>urn:sha1:7084f8429c940bac856123ce6d3946638fe20364</id>
<content type='text'>
The static rate from the path record should be put into the address
vector -- a long time ago the rate in the address attributes needed to
be a relative rate, which required more munging, but now that the
conversion from absolute to relative is done in the low-level driver,
it's easy for ib_init_ah_from_path() to put the absolute rate in.

Cc: Jason Gunthorpe &lt;jgunthorpe@obsidianresearch.com&gt;
Cc: Sean Hefty &lt;sean.hefty@intel.com&gt;
Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>IB/core: Fix sparse warnings about shadowed declarations</title>
<updated>2007-02-16T22:41:14Z</updated>
<author>
<name>Roland Dreier</name>
</author>
<published>2007-02-16T22:41:14Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=38abaa63bf38dd5e086911e63fbea47ae34674ab'/>
<id>urn:sha1:38abaa63bf38dd5e086911e63fbea47ae34674ab</id>
<content type='text'>
Change a couple of variable names to avoid sparse warnings about
symbols being shadowed.

Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
<entry>
<title>RDMA/cma: Add multicast communication support</title>
<updated>2007-02-16T22:29:07Z</updated>
<author>
<name>Sean Hefty</name>
</author>
<published>2007-02-16T01:00:18Z</published>
<link rel='alternate' type='text/html' href='https://git.0xinfinity.dev/distro/kernel.git/commit/?id=c8f6a362bf3eb28ade6027b49bb160a336dd51c0'/>
<id>urn:sha1:c8f6a362bf3eb28ade6027b49bb160a336dd51c0</id>
<content type='text'>
Extend rdma_cm to support multicast communication.  Multicast support
is added to the existing RDMA_PS_UDP port space, as well as a new
RDMA_PS_IPOIB port space.  The latter port space allows joining the
multicast groups used by IPoIB, which enables offloading IPoIB traffic
to a separate QP.  The port space determines the signature used in the
MGID when joining the group.  The newly added RDMA_PS_IPOIB also
allows for unicast operations, similar to RDMA_PS_UDP.

Supporting the RDMA_PS_IPOIB requires changing how UD QPs are initialized,
since we can no longer assume that the qkey is constant.  This requires
saving the Q_Key to use when attaching to a device, so that it is
available when creating the QP.  The Q_Key information is exported to
the user through the existing rdma_init_qp_attr() interface.

Multicast support is also exported to userspace through the rdma_ucm.

Signed-off-by: Roland Dreier &lt;rolandd@cisco.com&gt;
</content>
</entry>
</feed>
