]> git.openfabrics.org - ~shefty/rdma-win.git/commit
[MTHCA] Added bugfixes and changes from Linux Gen2 code.
authorleonidk <leonidk@ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86>
Sun, 2 Apr 2006 19:39:37 +0000 (19:39 +0000)
committerleonidk <leonidk@ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86>
Sun, 2 Apr 2006 19:39:37 +0000 (19:39 +0000)
commitb741ac77b1bc353032e0e3a942372e0ee1cec01d
treed6ee97978227a81ebe092dd85c2f864a72919e95
parentba3d06476c8120ed5abd58c09eb485697ef54a5f
[MTHCA] Added bugfixes and changes from Linux Gen2 code.
[MTHCA] (LinuxGen2-rev5521) Add support to IBV_SEND_FENCE in post_send.
[MTHCA] (LinuxGen2-rev5520) Add support to IBV_SEND_FENCE in post_send.
[MTHCA] (LinuxGen2-rev5476) Have mthca's create_srq method return the actual capacity of the SRQ that gets created.  Also update comments in <rdma/ib_verbs.h> to clarify that this is what is expected from ib_create_srq().
[MTHCA] (LinuxGen2-rev5293) Generate SQ drained events when requested
[MTHCA] (LinuxGen2-rev5369) When debugging is enabled, the mthca_QUERY_DEV_LIM() firmware command
function prints out some of the device limits that it queries.
However the debugging prints happen before all of the fields are
extracted from the firmware response, so some of the values that get
printed are uninitialized junk.  Move the prints to the end of the
function to fix this.
[MTHCA] (LinuxGen2-rev5275) Fixing stuff for FMR, not ported yet
[MTHCA] (LinuxGen2-rev5273) Adding stuff for FMR, not ported yet
[MTHCA] (LinuxGen2-rev5244) Cleaning: Remove useless "extern" from function declarations.
[MTHCA] (LinuxGen2-rev5243) Cleaning: remove might_sleep()
[MTHCA] (LinuxGen2-rev5242) Make functions that never fail return void
[MTHCA] (LinuxGen2-rev5223) We have run into the following problem: if a task receives a signal
while in the process of e.g. destroying a resource (which could be
because the relevant file was closed) mthca could bail out from trying
to take a command interface semaphore without performing the
appropriate command to tell hardware that the resource is being
destroyed.As a result we see messages like
 ib_mthca 0000:04:00.0: HW2SW_CQ failed (-4)
[MTHCA] (LinuxGen2-rev5221) There are some cards around that have UAR (user access region) size different from 8 MB.  Relax our sanity check to make sure that the PCI BAR is big enough to access the UAR size reported by the device firmware instead.
[MTHCA] (LinuxGen2-rev5220) Fix typo in overflow test in mthca receive posting.

git-svn-id: svn://openib.tc.cornell.edu/gen1@271 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86
19 files changed:
trunk/hw/mthca/kernel/SOURCES
trunk/hw/mthca/kernel/ib_verbs.h
trunk/hw/mthca/kernel/mt_fmr_pool.c [new file with mode: 0644]
trunk/hw/mthca/kernel/mt_fmr_pool.h [new file with mode: 0644]
trunk/hw/mthca/kernel/mt_l2w.h
trunk/hw/mthca/kernel/mt_uverbs.h [deleted file]
trunk/hw/mthca/kernel/mthca_cmd.c
trunk/hw/mthca/kernel/mthca_cq.c
trunk/hw/mthca/kernel/mthca_dev.h
trunk/hw/mthca/kernel/mthca_eq.c
trunk/hw/mthca/kernel/mthca_main.c
trunk/hw/mthca/kernel/mthca_mcg.c
trunk/hw/mthca/kernel/mthca_mr.c
trunk/hw/mthca/kernel/mthca_pd.c
trunk/hw/mthca/kernel/mthca_qp.c
trunk/hw/mthca/kernel/mthca_srq.c
trunk/hw/mthca/user/driver.h
trunk/hw/mthca/user/mlnx_uvp.h
trunk/hw/mthca/user/mlnx_uvp_qp.c