[Openib-windows] [PATCH] ib_local_mad

[Yossi Leybovich]

> -----Original Message-----
> From: ftillier.sst at gmail.com [mailto:ftillier.sst at gmail.com] 
> On Behalf Of Fabian Tillier
> Sent: Tuesday, May 16, 2006 1:51 AM
> To: Yossi Leybovich
> Cc: openib-windows at openib.org
> Subject: Re: [Openib-windows] [PATCH] ib_local_mad
> 
> Hi Yossi,
> 
> On 5/15/06, Yossi Leybovich <sleybo at mellanox.co.il> wrote:
> > Fab
> >
> > This patch add the ib_local_mad to user mode.
> > The code allow only to issue Get MAD.
> > The patch also include some Performance management 
> definitions (ib_gmp
> > ,port_counters) that I added to ib_types.h I also change 
> the vstat so 
> > that "vstat -c" will print out the error counters of the ports.
> > We found this very useful in debugging big cluster , and hopefully 
> > this the first step on the way to add the ib port counters 
> to perfmon.
> 
> I made minor changes to the patch and committed in revision 351.
> 
> The first change I made was to add version checking between 
> user-mode and kernel-mode, so that changes in the IOCTL 
> interface would not result in undetermined behavior.  This 
> was done in revision 350.
> 
> I moved the enforcement of GET to the proxy, so that the 
> interface can't be abused.  Other than that, I changed 
> formatting a little.
> 
> I applied only the IBAL portions of this - the MTHCA edits 
> seemed unrelated.
> I didn't apply the vstat changes because all I ever see is 
> zeros coming out for the port counters.  Are these supposed 
> to work over the
> MT23108 driver?  There shouldn't be anything HCA driver 
> specific here, so I'm puzzled by the results I'm seeing.
> 
> Please let me know if I screwed something up when applying 
> the patch that causes the results to always be zero.  I'll 
> let you apply the vstat changes.
> 

I tested the patch on MTHCA and its look fine . 
I will try to check it on MT23108.

> Thanks,
> 
> - Fab
>