I had trouble booting from an USB CD-ROM in an Intel BladeCenter. Please note that I'm fully aware of the bugs 74359 and 72055 but I think that my observation might help to solve the USB issue in bug 74359 without having to use "doscsi". I used Gentoo LiveCD 2004.3 x86 (minimal and universal -- same effect) and had the following effects: - initial booting worked - mounting cdrom didn't work When entering shell mode I realized that the USB cdrom's device /dev/sr0 was present. Mounting the cdrom handishly worked as well. However, in /linuxrc there's no delay between probing the modules (usb_storage among others) and trying to mount the cdrom. My guess is that the timespan between probing the usb_storage module and mounting the cdrom is too short when using udev (the device isn't there yet). I suspect this because when using the options gentoo devfs scandelay at the boot prompt, mounting the cdrom actually works without any problems. Possible solution: work the scandelay option into /linuxrc's sections for udev (same way as in the devfs sections). Christian Reproducible: Always Steps to Reproduce: 1. Boot from USB CD-ROM drive 2. 3.
Tim, what do you think?
<plasmaroo> I think what would be even better would be this: <plasmaroo> * If in CD-mode, and mounting fails; automatically scandelay <plasmaroo> * Try again in 10 seconds. <plasmaroo> * Be happy if it works. Otherwise, complain. I like this idea. Does this sound like it would accomplish the task, Christian?
Chris, that sounds like an excellent idea. This might even prevent potential problems with SCSI CD-ROMs that have similiar "timing problems" after a successful modprobe (device "pops up" a millisecond too late). Thumbs up for this fix from my side. :-) Would you like me to suggest a code snippet? I could try to come up with something tomorrow (I don't have the current linuxrc script + included scripts handy right now). Cheers, Christian
Created attachment 48990 [details, diff] Patch
Fixed in 3.1.0e; please reopen this bug if the problem still persists. Thanks!