[ofa-general] Possible bug of inkernel OFED of RHEL5.3?

[Yossi Etigin]

Brian J. Murrell wrote:
> So, is this a confirmation that the OFED in question (in the RHEL 5.3
> kernel) is in fact broken?  Is the breakage serious enough that one
> should want to avoid using that version?
> 
> b.
> 

Liang's report is enough to show taht it indeed has a bug. 

Looks like the only affected case is setting an IPoIB interface to connected
mode via sysfs. This bug can be avoid if you avoid using connected mode and make
sure all  network/openib start scripts do not use connected mode. 

If you want to use connected mode, you'll probably want to upgrade the OFED version.

--Yossi