]> git.openfabrics.org - ~shefty/rdma-win.git/commitdiff
rename
authorU-AMR\MSHEFTY <MSHEFTY@mshefty-MOBL2.amr.corp.intel.com>
Wed, 6 Jan 2010 16:50:41 +0000 (08:50 -0800)
committerU-AMR\MSHEFTY <MSHEFTY@mshefty-MOBL2.amr.corp.intel.com>
Wed, 6 Jan 2010 16:50:41 +0000 (08:50 -0800)
meta
patches/old-queue_mads [new file with mode: 0644]
patches/queue_mads [deleted file]

diff --git a/meta b/meta
index 4d29eaeb2ab560397e389d4f066016a08933844b..2d307ea7d9efa4ab095b7888796b865b7b13af13 100644 (file)
--- a/meta
+++ b/meta
@@ -1,5 +1,5 @@
 Version: 1
-Previous: 815c34644aa17ec9daf6cd3de69e94fed9e31a11
+Previous: ef5ac51c0b30cab7a725fd38fa7f975c412e93dc
 Head: d6d2302f2d258fa2c07e55a2e124a9bdd75367a0
 Applied:
   rm-build: d6d2302f2d258fa2c07e55a2e124a9bdd75367a0
@@ -7,6 +7,6 @@ Unapplied:
   old-bld-32: bd1bd1d366e3927b7646da33e243213a5d59c235
   old-apphang: 7a6f7ff02c6035e54fc262414eb9484ea98018b9
   old-cm_send_cb: 70dceed7518c8429e4c9e0646c7ec035f1cf53cb
-  queue_mads: a9c572c34ae693764a33e87173125d3065a6d1c5
+  old-queue_mads: a9c572c34ae693764a33e87173125d3065a6d1c5
   epdisc: 4c931f6af5e37a502f7a10b78a902747f9738623
 Hidden:
diff --git a/patches/old-queue_mads b/patches/old-queue_mads
new file mode 100644 (file)
index 0000000..1350933
--- /dev/null
@@ -0,0 +1,28 @@
+Bottom: 71d8cf1ad32d88b92c8abddfc3b3206fb073519e
+Top:    71d8cf1ad32d88b92c8abddfc3b3206fb073519e
+Author: U-AMR\MSHEFTY <MSHEFTY@mshefty-MOBL2.amr.corp.intel.com>
+Date:   2009-12-17 14:31:44 -0800
+
+winverbs/endpoint: queue connection requests
+
+Winverbs models listening for connections and retrieving
+connection requests similar to receive queue processing.  If
+a GetRequest is not waiting when a connection
+request arrives, the request is dropped and must be retried.
+
+For applications that serialize their retrieval of connection
+requests, this can result in a lot of dropped requests that
+need to be retried.  As a result, the connection rate to
+such applications suffers dramatically.
+
+To support these applications, queue the connection
+request in the kernel until a GetRequest call comes down
+to retrieve it.  This greatly improves the connection rate for
+these applications.
+
+Signed-off-by: Sean Hefty <sean.hefty@intel.com>
+
+
+---
+
+
diff --git a/patches/queue_mads b/patches/queue_mads
deleted file mode 100644 (file)
index 1350933..0000000
+++ /dev/null
@@ -1,28 +0,0 @@
-Bottom: 71d8cf1ad32d88b92c8abddfc3b3206fb073519e
-Top:    71d8cf1ad32d88b92c8abddfc3b3206fb073519e
-Author: U-AMR\MSHEFTY <MSHEFTY@mshefty-MOBL2.amr.corp.intel.com>
-Date:   2009-12-17 14:31:44 -0800
-
-winverbs/endpoint: queue connection requests
-
-Winverbs models listening for connections and retrieving
-connection requests similar to receive queue processing.  If
-a GetRequest is not waiting when a connection
-request arrives, the request is dropped and must be retried.
-
-For applications that serialize their retrieval of connection
-requests, this can result in a lot of dropped requests that
-need to be retried.  As a result, the connection rate to
-such applications suffers dramatically.
-
-To support these applications, queue the connection
-request in the kernel until a GetRequest call comes down
-to retrieve it.  This greatly improves the connection rate for
-these applications.
-
-Signed-off-by: Sean Hefty <sean.hefty@intel.com>
-
-
----
-
-