Well if that is the case, I would suggest doing it the way that it has been done for the other drivers, as you suggested.
woody -----Original Message----- From: Steve Wise [mailto:sw...@opengridcomputing.com] Sent: Tuesday, March 03, 2009 11:03 AM To: Woodruff, Robert J Cc: Todd Rimmer; John Russo; tzipo...@dev.mellanox.co.il; OpenFabrics EWG Subject: Re: [ewg] shipping firmware with ofed Woodruff, Robert J wrote: > If this is the case, then why not just build the firmware into the driver > and submit it with the driver when the driver is sent upstream to kernel.org. > > I have seen firmware imbedded into source code before, in the form of > a header file that contains the firmware code in hex. Something like, > > unsigned long firmware[] = { > 0xfefefefe, > 0xdededede, > . > . > . > } > > I don't think thats the recommended approach in the upstream kernel. The kernel supports an infrastructure where the driver can request a firmware file to be opened and passed into the driver at device probe time. So I think the recommended approach is to have distros ship the firmware in /lib/firmware/*, and then have the driver request the firmware if it needs to download and run a new version using these standard Linux kernel services. This is how RHEL5.3 handles cxgb3 today. They ship the binary image. The driver loads it at device probe time. The Linux kernel service is request_firmware(). Steve. _______________________________________________ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
