> Looks like the OFED installation is faulty. Can we fix this issue in the next release of OFED?
> For some reason, your compilation script is not taking directory > /usr/src/ofa_kernel/kernel_addons/backport/2.6.16_sles10_sp2/include > in the include path before the regular kernel includes. Are you referring to the OFED compilation script or our module's? This problem happens on RedHat EL 5.0 as well. On Tue, Jul 7, 2009 at 2:17 AM, Jack Morgenstein<[email protected]> wrote: > The missing function declarations are all found in header files under > directory > (on your system) /usr/src/ofa_kernel/kernel_addons/backport/2.6.16_sles10_sp2. > These header files must be taken before the regular kernel header files > (they "include_next" to the regular files, and supplement these files > with missing macros, static inline functions, etc.). > > For example, dma_map_single_attr() is found in file > /usr/src/ofa_kernel/kernel_addons/backport/2.6.16_sles10_sp2/include/linux/dma-mapping.h > > For some reason, your compilation script is not taking directory > /usr/src/ofa_kernel/kernel_addons/backport/2.6.16_sles10_sp2/include > in the include path before the regular kernel includes. > > -Jack > > On Monday 22 June 2009 07:57, Rajouri Jammu wrote: > >> Tziporet, >> > We (Mellanox) are running regression testing on SLES10 SP2 with OFED 1.4.1 >> without any problem >> >> I don't think it's a runtime issue. >> >> The problem: >> /usr/src/ofa_kernel/include/rdma/ib_verbs.h in OFED 1.4.1 is referring to >> symbols like "dma_map_single_attrs" etc, but when I looked for those symbols >> in /usr/src/linux-2.6.16.60-0.21 I did not find any references. >> >> Can you suggest a file that I should include that will resolve this issue? >> >> Thanks. >> >> On Sun, Jun 21, 2009 at 1:34 PM, Tziporet Koren <[email protected] >> > wrote: >> >> > Rajouri Jammu wrote: >> > >> >> Thanks for the response Evan. >> >> >> >> It appears to me that this is a problem in OFED 1.4.1 since the header >> >> file in /usr/src/ofa_kernel/include/rdma/ib_verbs.h is referring to data >> >> structures that are not present in the SLES 10 u 2 kernel. >> >> >> >> However, someone else indicated it was a problem in our code. >> >> >> >> Btw, my code compiled fine with OFED 1.3.1 on the same kernel. >> >> >> >> >> > We (Mellanox) are running regression testing on SLES10 SP2 with OFED 1.4.1 >> > without any problem >> > So I guess its something specific in your system, or order of include files >> > you use >> > >> > Tziporet >> > >> > _______________________________________________ > general mailing list > [email protected] > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general > > To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general > _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
