There was a period when Linux guys tried to keep in sync Linux and Windows source code. In that period ib_types.h was splitted in two: ib_types.h and ib_types_extended.h. The last one is used only in Windows. The first one has a common part for Linux and Windows and Linux-only part. At current moment the sync between Linux and Windows is lost. Linux Opensm saw a lot of improvements, also in the area of scalability. Unfortunately, noone can find time to merge these changes into Windows.
As to ib_types*.h files, they have to be kept in sync with the stack's ib_types.h file, because OpenSm works over IBAL should use the same structures. (Of course, not all stack changes are relevant for OpenSm, but it is simpler to duplicate the ib_types.h' changes into ib_types_extended.h than to think are they relevant or will be sometimes relevant to OpenSm). > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Fab Tillier > Sent: Friday, September 19, 2008 7:00 PM > To: Hal Rosenstock; Alex Naslednikov > Cc: [email protected]; Yevgeny Kliteynik > Subject: RE: ***SPAM*** Re: [ofw] [patch] [IPoIB] Adding > support to dellguid and user-define mask for guid generation > > >> Index: ulp/opensm/user/include/iba/ib_types.h > >> =================================================================== > >> --- ulp/opensm/user/include/iba/ib_types.h (revision 3193) > >> +++ ulp/opensm/user/include/iba/ib_types.h (working copy) > >> @@ -7928,6 +7928,7 @@ > >> IB_INVALID_GID, IB_INVALID_LID, IB_INVALID_GUID, + > >> IB_INVALID_GUID_MASK, IB_INVALID_CA_HANDLE, IB_INVALID_AV_HANDLE, > >> IB_INVALID_CQ_HANDLE, > > > > Are we keeping ib_types.h in sync with Linux ? I thought > this file was > > intended on being platform independent. Should this change be made > > there as well ? > > There's 2 ib_types.h files, one for the windows drivers, one > in OpenSM's own directory. Only the second needs to be kept > in sync with Linux. > > There's also ib_types_extended.h in OpenSM's directory, which > is a copy of ib_types.h. I don't quite understand what's > going on here, though... > > -Fab > _______________________________________________ > ofw mailing list > [email protected] > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw > _______________________________________________ ofw mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
