Stefan Reinauer <[EMAIL PROTECTED]> writes: > Eric W. Biederman wrote: >> coreboot tracker <[EMAIL PROTECTED]> writes: >> >> >>> #110 Allow for per-device subsystem IDs >> >> What is the proper procedure for saying per-device subsystem IDs is >> a dumb idea. >> >> The subsystem IDs roughly identify the PCB a component sits on. So >> unless you have multiple pluggable boards in a system there should >> only be one subsystem vendor and one subsystem device id. > Is this the official definition? Where does this originate from?
A Paraphrased version of the official definition. It comes from the PCI SIG as I recall. I haven't rechecked this since I added support years ago. > In practice the subsystem vendor IDs are quite arbitrary and definitely not > the > same for all PCI devices per mainboard on any of the mainboards we have here. > In > fact, what I have seen quite often is that every PCI device has it's > device/vendor ID set as subsystem IDs. Which is probably a default, giving you no board information. Quite likely a bug. The subsystem vendor is not supposed to be the ASIC vendor. > The subsystem information is used by lspci to correctly identify some > devices. Yes. You need to know which board they are on, to identify them. > Plus, more importantly, we use it in flashrom to identify > mainboards. Using a unique device ID per board will require us to have two > sets > of information, one for legacy bios and one for coreboot. That sounds weird. > While using per-board subsystem ids sounds the right thing as per your > reasoning, I vote to not enforce that behavior, but make it configurable. I can understand overriding it for compatibility, or in other weird cases. The default should still be global to the mainboard. Eric -- coreboot mailing list [email protected] http://www.coreboot.org/mailman/listinfo/coreboot
