On Wed, 16 Jun 2004, dann frazier wrote: > here's an annotated diff between our x86_64 configs, which includes what my > plan is for the default - can you skim through and see if you have any > disagreements?
Sure. I've included my comments below. I can't think in short sentences, so I've not tried to put my comments as comments on your comments... ;) [...] > # > # ACPI Support > # > -# CONFIG_ACPI is not set > +CONFIG_ACPI=y ## y: i would think some machines would require > ACPI > CONFIG_ACPI_BOOT=y > +CONFIG_ACPI_BUS=y [...] Actually, I'm not convinced about this. ACPI is one of those lovely things that sometimes seems to be necessary, and other times seems to cause nothing but trouble. The only thing that I can speak to for sure is that we've had all kinds of trouble in the past with ACPI-enabled kernels losing track of interrupts for IDE disks, and causing really weird behavior with some network cards, rendering them useless. In those cases, disabling the use of ACPI was required in order to make the machine function properly. With that said, though, you might be right about including it in the kernel -- and then documenting that there are kernel command-line arguments which can be used in the pxelinux.cfg files to disable ACPI if it causes trouble. That way, the generic kernel will support those machines that *need* ACPI, and the instructions will provide insight on how to make it work on machines that *can't* use ACPI. (Oh, and I should mention the kernel args: 'acpi=off' and 'pci=noacpi' are the ones we've found to be quite useful, usually also in connection with 'noapic.') [...] > # > # SCSI support > # > -CONFIG_SCSI=m > +CONFIG_SCSI=y # m: scsi i'll leave modular since order matters > [...] > @@ -296,14 +310,24 @@ > # CONFIG_SCSI_AHA1542 is not set ## m: i'll leave lots of modules > enabled > # CONFIG_SCSI_AHA1740 is not set > # CONFIG_SCSI_AACRAID is not set > -# CONFIG_SCSI_AIC7XXX is not set > +CONFIG_SCSI_AIC7XXX=m > +CONFIG_AIC7XXX_CMDS_PER_DEVICE=253 [...] Agreed. I mostly left SCSI modules disabled because some of them didn't want to build correctly on my system (I'm not sure which ones anymore, and I don't remember why). But for a generic kernel, SCSI is *definitely* needed. [...] > -# CONFIG_HW_RANDOM is not set > +CONFIG_HW_RANDOM=y # why not... [...] Actually, I'll come right back and ask "why?" Is there a need for hardware random support during the image install? (NB: I don't really care -- as you pointed out, we're not trying to conserve space to make this fit on a floppy...) [...] > -CONFIG_DEVFS_FS=y # i'll have to play with this [...] Good luck. I couldn't get it to compile, and anytime I ran 'make oldconfig' to fix up any mistakes I made in the .config, this was removed, even if I listed it as 'y.' I've been made to understand that devfs might work right for the x86_64 in the 2.6 kernels, but I've not heard anyone claiming to have made it work in the 2.4 series. And I've heard *lots* of complaints about it not working in 2.4... Thanks, Dave ------------------------------------------------------- This SF.Net email is sponsored by The 2004 JavaOne(SM) Conference Learn from the experts at JavaOne(SM), Sun's Worldwide Java Developer Conference, June 28 - July 1 at the Moscone Center in San Francisco, CA REGISTER AND SAVE! http://java.sun.com/javaone/sf Priority Code NWMGYKND _______________________________________________ Sisuite-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/sisuite-devel
