Fork of ipxe; additional commands and features https://git.openslx.org/openslx-ng/ipxe.git/atom/src/net/infiniband?h=v1.0.0 2009-11-16T22:14:36+00:00 2009-11-16T22:14:36+00:00 Michael Brown 2009-11-13T19:02:04+00:00 urn:sha1:58b6794c111d8a8984d50bae10f720561c5c07f7 There is no such thing as a non-existent partition. 2009-11-16T22:13:44+00:00 Michael Brown 2009-11-16T20:34:54+00:00 urn:sha1:228ac9d0184006d22f12d5daecd0a03d81cd029e 2009-11-16T22:13:25+00:00 Michael Brown 2009-11-16T20:18:03+00:00 urn:sha1:e7018228fab855677c9256ead2871c9e354d4df2 IBA section 14.2.5.2 states that "the contents of the NodeDescription attribute are the same for all ports on a node". Satisfy this by using the HCA GUID rather than the port GUID to form the node description string. 2009-10-16T22:03:47+00:00 Michael Brown 2009-10-16T17:20:55+00:00 urn:sha1:224ef7f48300e3499275eb022ad05defb163245d 2009-08-10T21:31:55+00:00 Michael Brown 2009-08-09T20:00:48+00:00 urn:sha1:0ff5c456cb1ad38c3300a214799478101a3540a6 There is diagnostic value in being able to disambiguate between the various reasons why an IB CM has rejected a connection attempt. In particular, reason 8 "invalid service ID" can be used to identify an incorrect SRP service_id root-path component, and reason 28 "consumer reject" corresponds to a genuine SRP login rejection IU, which can be passed up to the SRP layer. For rejection reasons other than "consumer reject", we should not pass through the private data, since it is most likely generated by the CM without any protocol-specific knowledge. 2009-08-10T21:30:06+00:00 Michael Brown 2009-08-10T10:32:13+00:00 urn:sha1:a0d337912eee5ff0d72d0d85d83be88672d49004 Generate errors within individual MAD transaction consumers such as ib_pathrec.c and ib_mcast.c, rather than within ib_mi.c. This allows for more meaningful error messages to eventually be displayed to the user. 2009-08-10T21:27:33+00:00 Michael Brown 2009-07-17T21:11:42+00:00 urn:sha1:0c30dc6bc5f26e7011ddfcda34d7adac653464cf SRP is the SCSI RDMA Protocol. It allows for a method of SAN booting whereby the target is responsible for reading and writing data using Remote DMA directly to the initiator's memory. The software initiator merely sends and receives SCSI commands; it never has to touch the actual data. 2009-08-10T21:23:28+00:00 Michael Brown 2009-08-09T10:05:21+00:00 urn:sha1:4be11f523c57a22d18a7369fbe9ec6d9f1c88022 SRP over Infiniband uses a protocol whereby data is sent via a combination of the CM private data fields and the RC queue pair itself. This seems sufficiently generic that it's worth having available as a separate protocol. 2009-08-09T00:31:07+00:00 Michael Brown 2009-08-09T00:25:38+00:00 urn:sha1:46073f12391e5a467fbf5d5853ea91333fcce843 We will terminate our transaction as soon as we receive the first CM REP, since that provides all the state that we need. However, the peer may resend the REP if it didn't see our RTU, and if we don't respond with another RTU we risk being disconnected. (This protocol appears not to handle retries gracefully.) Fix by adding a management agent that will listen for these duplicate REPs and send back an RTU. 2009-08-08T22:56:28+00:00 Michael Brown 2009-08-06T00:18:38+00:00 urn:sha1:34bfc04e4c2ea74a025994fc37c83d640942a651