Grigoriy Strokin wrote: [] > I have tried to include anything SATA-related into the kernel as > built-in rather than as modules, but it still doesn't work. Either I've > missed something, or this initrd does more than just loading required > modules: since the disk is eventually seen as /dev/sda, I suppose that > initrd configures some kind of SCSI emulation in addition to loading > required modules. JFYI: All SATA-related stuff (and new PATA aka IDE, but based on libata) will be /dev/sd* (and /dev/sr* for cd-roms), not /dev/hd*. That to say -- If, on the same controller, you've both /dev/hd* and /dev/sd*, that means that some "parts" of the controller are served by [SP]ATA driver, and some by IDE driver. Which doesn't quite work right, and such an ability has been removed from recent 2.6.* kernels, simplifying and cleaning up the code *alot*. Note that this all has - almost - nothing to do with initrd/initramfs. Whichever modules you'll load (or compile into the kernel), those devices will be present. Order does matter as well - having both IDE and PATA modules (or built-in), you can have EITHER /dev/sd* OR /dev/hd* - depending on which module has been loaded (or driver probed for built-in drivers) first. The only way initrd is involved is that it has to load the modules (if not built-in) -- but that's all, it can't do more than loading wrt setting up - at least IDE/[SP]ATA - devices. /mjt
This archive was generated by hypermail 2.1.3 : Thu Mar 29 2007 - 00:43:55 PST