[ofw] [ANNOUNCE] winOFED 2.3 RC3 available for download
Chris Worley
worleys at gmail.com
Tue Oct 5 08:51:53 PDT 2010
Are there any testing procedures for the Windows SRP initiator? If
so, where can they be found?
Is there a specific revision of SCST the SRP developers use for
testing? If so, which?
Thanks,
Chris
On Fri, Oct 1, 2010 at 3:06 PM, Smith, Stan <stan.smith at intel.com> wrote:
>
> OFED for Windows Release 2.3 RC3 available.
>
> Download @ http://www.openfabrics.org/downloads/Windows/v2.3_RC3
>
> Please address comments and concerns to https://bugs.openfabrics.org and/or the
> Windows OpenFabrics email list ofw at lists.openfabrics.org
>
> Commits since RC2 @ svn.2933
> ----------------------------
>
> Revision: 2952
> Author: stansmith
> Date: 8:50:52 AM, Friday, October 01, 2010
> Message:
> [IPoIB_NDIS6_CM] initialize IBAL pointer in Endpoint struct @ EP creation;
> initialize only in one place as IBAL ptr does not change over life of EP.
> ----
> Modified : /gen1/branches/WOF2-3/ulp/ipoib_NDIS6_CM/kernel/ipoib_endpoint.cpp
> Modified : /gen1/branches/WOF2-3/ulp/ipoib_NDIS6_CM/kernel/ipoib_endpoint.h
> Modified : /gen1/branches/WOF2-3/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2949
> Author: tzachid
> Date: 10:26:21 AM, Tuesday, September 28, 2010
> Message:
> [core] Replace memory allocator with a memory allocator that works natively
> with 0 bytes allocations.
> ----
> Modified : /gen1/trunk/core/complib/kernel/cl_memory_osd.c
> Modified : /gen1/trunk/hw/mthca/kernel/mt_memory.h
>
> Revision: 2948
> Author: leonidk
> Date: 12:32:11 PM, Monday, September 27, 2010
> Message:
> [HCA] Prevent stack corruption
>
> In the case where umv_buf::command is FALSE,
> the else control segment is taken and a stack variable's address is stored
> by INIT_UDATA, to be written later in the call to alloc_pd.
> The stack variable then goes out of scope, so the call to alloc_pd could
> corrupt the stack.
>
> The fix uses the status local variable as temporary storage,
> as it is unused until after the call to alloc_pd.
>
> Signed-off-by: Fab Tillier <ftillier at microsoft.com>
> ----
> Modified : /gen1/trunk/hw/mlx4/kernel/hca/pd.c
>
> Revision: 2947
> Author: shefty
> Date: 10:13:53 AM, Wednesday, September 22, 2010
> Message:
> librdmacm: validate user private data size
>
> Check private data size to avoid
> overrunning the winverbs data buffer.
>
> Signed-off-by: Sean Hefty <sean.hefty at intel.com>
> ----
> Modified : /gen1/trunk/ulp/librdmacm/src/cma.cpp
>
> Revision: 2946
> Author: xalex
> Date: 2:40:52 AM, Wednesday, September 22, 2010
> Message:
> [IPoIB_NDIS6_CM] stop leaking spin lock
>
> Signed-off by: Smith Stan
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2945
> Author: xalex
> Date: 2:28:26 AM, Wednesday, September 22, 2010
> Message:
> [[IPoIB_NDIS6_CM] Spacing and tabulation - part II; no functionality changes
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_driver.cpp
>
> Revision: 2944
> Author: xalex
> Date: 2:26:18 AM, Wednesday, September 22, 2010
> Message:
> [[IPoIB_NDIS6_CM] Spacing and tabulation; no functionality changes
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2943
> Author: xalex
> Date: 9:07:09 AM, Tuesday, September 21, 2010
> Message:
> [[IPoIB_NDIS6_CM] LSO bug fix
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2942
> Author: xalex
> Date: 9:05:11 AM, Tuesday, September 21, 2010
> Message:
> [IPoIB_NDIS6_CM] Handling memory access violation at destroy when __port_init fails
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2941
> Author: xalex
> Date: 8:19:38 AM, Tuesday, September 21, 2010
> Message:
> [IPoIB_NDIS6_CM] [HW]
> 4K MTU support
> ----
> Modified : /gen1/trunk/hw/mlx4/kernel/bus/net/port.c
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_driver.cpp
>
> Revision: 2940
> Author: xalex
> Date: 8:15:20 AM, Tuesday, September 21, 2010
> Message:
> [IPoIB] [IPoIB_NDIS6_CM]
> Guids supported by several new vendors can't
> be distinguished only by first 2 bytes.
>
> This patch adds 3-rd byte of GUID to the translation table
> ----
> Modified : /gen1/trunk/ulp/ipoib/kernel/ipoib_xfr_mgr.c
> Modified : /gen1/trunk/ulp/ipoib/kernel/ipoib_xfr_mgr.h
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_xfr_mgr.cpp
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_xfr_mgr.h
>
> Revision: 2939
> Author: xalex
> Date: 7:47:23 AM, Tuesday, September 21, 2010
> Message:
> [IPoIB_NDIS6_CM]
> Fix at statistic calculation
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_port.cpp
>
> Revision: 2938
> Author: tzachid
> Date: 7:00:36 AM, Tuesday, September 21, 2010
> Message:
> [wsdinstall] Add support for user access control to the WSD installer.
> ----
> Modified : /gen1/trunk/tools/wsdinstall/user/SOURCES
>
> Revision: 2937
> Author: tzachid
> Date: 6:59:20 AM, Tuesday, September 21, 2010
> Message:
> [perf_utils] Add more information in error printing.
> ----
> Modified : /gen1/trunk/tools/perftests/user/perf_utils.c
>
> Revision: 2936
> Author: xalex
> Date: 5:42:46 AM, Tuesday, September 21, 2010
> Message:
> [Tools]
> Change Peak BW calculation when the difference between post
> and completion is abnormal (avoid 1#J printings)
> Merged from MLNX_WinOF 5894
> ----
> Modified : /gen1/trunk/tools/perftests/user/read_bw/read_bw.c
> Modified : /gen1/trunk/tools/perftests/user/send_bw/send_bw.c
> Modified : /gen1/trunk/tools/perftests/user/write_bw/write_bw.c
>
> Revision: 2935
> Author: xalex
> Date: 5:16:38 AM, Tuesday, September 21, 2010
> Message:
> [IPoIB_NDIS6_CM]
> Fixing a possible race when we get PORT_REMOVE
> before PORT_DOWN
> Merged from MLNX_WinOF_2.1.25168
> ----
> Modified : /gen1/trunk/ulp/ipoib_NDIS6_CM/kernel/ipoib_adapter.cpp
>
> Revision: 2934
> Author: tzachid
> Date: 5:18:34 AM, Sunday, September 19, 2010
> Message:
> [complib] Allow the timer to work with windows 2003. (mlnx 6543)
> ----
> Modified : /gen1/trunk/core/complib/user/cl_timer.c
>
>
> OFED for windows Release Summary
> --------------------------------
>
> 1) The winOFED 2.3 release is based on openib-windows source svn revision
> (branches\WOF2-3 svn.2953.
>
> Last OFED release (2.2) based on svn.2739.
>
> 2) New Features
>
> uDAT / uDAPL 2.0.30 code base:
> 32-bit providers available.
> UCM (Unreliable Datagram Connection Manager) DAPL provider available.
>
> OpenSM upgraded to version 3.3.6 (see %windir%\temp\osm.log for SM details).
>
> OFED (Open Fabrics Enterprise Distribution) verbs API documented in manual.htm
>
>
> 2) Bug fixes
>
> IB Core
> cl_timer race conditions fixed
> ibal memory leaks fixed.
>
> IPoIB
> DHCP is now working correctly in larger HPC Edition clusters.
> DHCP correctly inter-operating with Linux.
> DPC threads replaced with IO work queue threads to allow system scheduling fairness.
>
> WinVerbs
> Overlap structures initialized correctly.
>
> WinMAD
> ReadFile/WriteFile return values checked.
>
> OFED umad
> MAD receive timeouts handled correctly.
>
> NetworkDirect providers: IBAL & winverbs
> Memory leaks fixed.
>
> OFED Installer
> Winverbs, winmad, dat/dapl .dlls installed into %windir%\system32\
> For 64-bit architectures, 32-bit .dll versions are installed into %windir%\SysWow64\
>
>
> **** Known Issues ****
>
> HCA drivers are reclassified as BOOT_START drivers to support booting from
> an SRP IOC. An unfortunate side-effect is when OFED 2.3 is installed
> on a system for the first time, after installation a mandatory reboot is
> Required by Windows.
>
> After the OFED.msi file has started installation, an errant
> "Welcome to the Found New Hardware Wizard" window 'may' popup; XP & Svr 2003.
>
> Just ignore or 'cancel' the errant FNHW popup window in order to proceed with
> the installation. XP requires a cancel, for WLH & WNET, the notifier will
> disappear on their own.
>
> You do need to answer 'Yes' or 'Continue' to popup windows which refer to
> Non-Logo'ed (aka, Non-WHQL'ed) drivers being installed.
>
> If the install appears to hang, look around for popup windows requesting input
> which are covered by other windows.
> Such is the case on Server 2008 initial install - Answer 'yes' to always trust
> the OpenFabrics Alliance as a SW publisher.
>
>
> Please:
> Read the Release_notes.htm file!
>
> make 'sure' your HCA firmware is recent:
> vstat.exe displays HCA firmware version & PSID.
> flint.exe (found in the Mellanox firmware tools package) displays PSID.
>
> Thank you,
>
> OFED for Windows developers.
> _______________________________________________
> ofw mailing list
> ofw at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
>
More information about the ofw
mailing list