Home | Docs | Forums | Lists | Bugs | Planet | Store | GMN | Get Gentoo!
Not eligible to see or edit group visibility for this bug.
View Bug Activity | Format For Printing | XML | Clone This Bug
Trying to use a live CD to boot the system so I can install gentoo on a dell D410. The system has a external CD DRive in either the "mediabase" or via USB on a special connector for extra power (both are USB protocol according to XP) The CD boots then loads the following modules: - ehci-hcd - usb-storage - uhci-hcd - ohci-hcd - ieee1394 - ohci1394 - sbp2 Next it attemps to mount CD at: - /newroot/dev/cdroms/* - /newroot/dev/ide/cd/* - /newroot/dev/sr0 and fails with the following message: "Could not find CD to boot, something else needed"" "mount: mounting /newroot/dev on /dev failed: no duch file or directory" does keymap thing then fails with: "loadkmap: /dev/cv/0: No such file or directory" "umount: /dev: Invalid argument Determining root device... "The root block device is unspecifed or not detected" "Please specify a device to boot, or "shell" for a shell..." boot() :: Try pointing it to /dev/hda3 where the live cd is coppied onto a fat32 partition or to /dev/sr0 where the CD drive is and it kernel panics. using shell I can mount /dev/sr0 and find the files on the live CD located there Basically Help :) The Dell D410 is an intel Sonama laptop (centrino 2) with a 1.86Ghz CPU and 1GB RAM, the graphics is the built in intel 915GM. Have tried with acpi on and off, noapic, doscsi and gentoo-nofb and all die in this way. Other live CD's die in a similar way, but a ubuntu install CD works, completes and sends me on my way to a working ubuntu install. Any help would be welcome as I actually like running gentoo. Reproducible: Always Steps to Reproduce: 1. Boot Live CD 2. Wait 3. Failure as described Actual Results: Bootup fails Expected Results: Take me to a console so I can install Gentoo Dell D410 laptop specs can be found from www.dell.co.uk under the notbooks -> latitude section. The live CD is 2004.3 although I have also tried a ubuntu live CD & Knoppix all of which fail in a similar way.
Did you boot with "doscsi"?
I tried it without doscsi initially, however after an initial failure I tried it with doscsi and that failed in the same way. Mark
Any ideas would be appreciated please.
An addition to this, I have now tried other machines booting off USB CD drives and they fail to boot using a wide variety of live CD's including the Gentoo Boot CD's This is not restricted to Dell laptops and affects other brands also.
*** Bug 86953 has been marked as a duplicate of this bug. ***
*** Bug 88333 has been marked as a duplicate of this bug. ***
I also have the same problem on a IBM X40 mod. 2386H4G, Intel pentium M processor 1.40 GHz and a USB BenQ Slim Combo DW-224E cd drive. I'm completely new to Linux but i would like to start with Gentoo, so please help me to solve this problem!
After further testing I can conclude that many other distro's live CD's do exactly the same. I am now fairly certain that this has something to do with the way live CD's boot up when the main drive is a USB CD Device. It would appear that this is a common problem not linked to just gentoo. Hope this helps somebody. As an aside I found a ubuntu or similar full installer CD worked, so I put ubuntu on my laptop with that in to a minimal configuration, then did a full gentoo install from in that. Have latest gentoo now running, using only UDEV (devfs never touched the system) and all is happy and fine :)
there seems to be a workaround: http://forums.gentoo.org/viewtopic-t-320660.html so it's a problem with udev not making the right /dev entrys
Using "/dev/ram0" instead of "/dev/sr0" the program starts booting perfectly...the only problem is that it doesn't let you choose the keymaps! It directly choose the default keyboard setting ("us/41") without giving you the possibility to enter the verbose mode... Anybody knows how to enter the verbose mode and choose the proper keyboard after the program has started and i'm logged in to the "Live cd" as "root" ?? Maybe this is a stupid question, but since i'm completely new to the world of Linux any help would be greatly appreciated! Thank you very much!
*** Bug 89402 has been marked as a duplicate of this bug. ***
*** Bug 89580 has been marked as a duplicate of this bug. ***
*** Bug 93790 has been marked as a duplicate of this bug. ***
On my Dell Latitude X300 this bug is completely reproducible on CDs 2005-0 and 2004-3. 2004-2 works perfectly MOST of the time (when I tried it this afternoon it worked twice and failed once -- but on the failure I hit F2 very early in the sequence).
This should be at least tested on next 2005.1 Release. USB CD-ROM now shipped with many small-sized Notebook and tablets PC. I think some system may need some delay for USB device to initialize. May the /linuxrc scripts should wait for a few seconds for these USB devices initialized and probed correctly.
Great... do you have a USB CDROM that you can ship us to test with? Anyway, using both "doscsi" and "scandelay" should work. There is a test ISO available at your local mirror in /experimental/x86 that you can use to test this. Report back to here. Since we don't have the hardware, we can't test it. When we ask for testers, we never get a response, so if the hardware doesn't work on a given release, look around at your fellow USB CDROM users, especially ones subscribed to gentoo-dev, and blame them for not stepping up to take the few minutes to boot test the latest test CD sets.
Just tried the minimal x86 install 2005.1_pre1 with a DVD-Writer attached to a USB to IDE adaptor with 'doscsi' and 'scandelay' passed to the kernel. The bug is still present and presents in excatly the same way. The device doesn't appear in /dev. When I run udevstart at the prompt the device appears and is mountable. I notice that the DVD-Writer device is present as /dev/sr0 rather than /dev/sgX or /dev/sdX. Also, /dev/sr0 isn't in the list of devices checked for the install media. I'm more than happy to do any other testing that is required.
Yo! Genkernel monkeys! Earn those bananas! *grin*
please upgrade to genkernel-3.2.1 or better as that does check for /dev/sr. Let us know if this fixes the issue for you.
Just to clarify, the problem occurs when booting pre-release (in the case of 2005.1_pre1) and release (in the case of 2005.0) live CDs downloaded from a gentoo mirror on a system that isn't running gentoo. Couple of points: 1. When you say "upgrade to genkernel-3.2.1" do you mean upgrade and build a new live CD? Is there a profile availible for the pre-release 2005.1? 2. There may still be a problem if the device isn't present until udevstart is run manually.
Appologies for not offering to help earlier, but I have moved house and was without DSL for a month :( Anyway if you need testing do say as I am more than willing to try any pre release versions of a minimal live CD.
Eric: please assign to the release group and not to me directly... The current _pre1 pre-release for x86 was *not* built with this version of genkernel, so it will not have the proper fix. We are working on a new pre-release version as we speak.
It looks like udev is not running while probing for USB-CDROMs After enter shell (after failed finding cdroot). There are no /dev/scd0 or /dev/sr/0. I try to run 'udevstart' manually but udev still not running. Other things. There are no sr_mod in /lib/modules. I think it built-in in the kernel. Most or SCSI Devices uncluding HDD and CDROM still need sg ... But 'sg' built as modules while sr_mod or sd_mod are built-in in the kernel ?.
I found that lxnay's Reiser4 livecd can boot from USB-CDROM. I am trying to looks what it done at startup.
You absolutely do not need SCSI generic for disk or CD access, only for "generic" devices such as cameras and scanners. We'll let you know as soon as we have something to test.
It may not need. I don't know CD-Writes need it at boot time or not ?.
I didn't need to modprobe anything else to get my DVD-Writer mountable, I only needed to run udevstart to get the device node created. Unless the other modules are loaded after the search for the install media and before the emergency shell is presented I don't think more modules are needed. G.
Testing installation on the most recent release, 2005.1, using an external USB DVD-ROM drive on a TabletPC. The instructions located at (http://forums.gentoo.org/viewtopic-t-320660.html) no longer work, as the live-cd is unable to find anything on /dev/ram0 or /dev/sr0. A new forum thread has been created to deal with this issue (http://forums.gentoo.org/viewtopic-t-369937.html). Recap, after attempting to boot off /dev/ram0: cat /proc/scsi/scsi cat /proc/scsi/usb-storage/0 [your number here may vary] cat /proc/sys/dev/cdrom/info [I found my drive name was "sr0"] ls -l /dev/sr0 [shows that the device doesn't exist] cat /proc/devices | grep sr [shows me the major node (?), mine is "11 sr"] mknod /dev/sr0 b 11 0 exit boot() :: /dev/ram0 >> Mounting root... mount: Mounting /dev/ram0 on /newroot failed: Invalid argument !! Could not mount specified ROOT, try again !! The root block device is unspecified or not detected. I can mount /dev/sr0 and succesfully list the (correct) contents of the .iso, but am unable to boot from this mount: boot() :: /newroot >> Mounting root... mkdir: Cannot create directory `/newroot/tmp/`: Read-only file system mknod: //newroot/dev/console: No such file or directory mknod: //newroot/dev/tty1: No such file or directory >> Booting (initramfs)...umount: /sys Invalid argument *: Failed to unmoun the initrd /sys! .. chroot: cannot execute /bin/sh: No such file or directory Kernel panic - not syncing: Attempted to kill init!
I meet the same problem with "Comment #28 From lieut_data". I'm using ibm x31 with nec 3520a dvd writer + ide2usb device. But with minimal livecd 2005.0, If I execute with "gentoo doscsi" and "udevstart" I could mount using /de/ram0. Do it doesn't work with 2005.1 Any hints will be appreciate. Thank you.
Managed to get the 2005.1 minimal cd to boot on a USB CDROM with the following process: 1. Boot with "doscsi scandelay debug" (note the 'debug') 2. At the debug prompt (NOT the emergency shell prompt): # ls /dev/sr0 ls: /dev/sr0: No such file or directory # cat /proc/sys/kernel/hotplug /sbin/hotplug # echo /sbin/udev > /proc/sys/kernel/hotplug # cat /proc/sys/kernel/hotplug /sbin/udev 3. unplug the cdrom then plug it back in. 4. At prompt again: # ls /dev/sr0 /dev/sr0 # exit 5. Boot continues and finds livecd at /dev/sr0. I've tried the same procedure with the the emergency shell, however it didn't work. The important parts are the 'echo /sbin/udev > /proc/sys/kernel/hotplug' and the replugging of the cdrom. The echo sets udev as the hotplug agent for the kernel and the replug causes the kernel to send a new hotplug event to udev which in turn sets up the missing device. One way I think this can be fixed permenatly is to add the echo line to the linuxrc right after /proc is mounted. I'm going to try and build my own version of the livecd to test this. Hope this helps! G.
Created an attachment (id=66098) [details] Stage 1 spec file used to build test CD.
Created an attachment (id=66099) [details] Stage 2 spec file used to build test CD.
Created an attachment (id=66100) [details] Patch for the linuxrc file.
I've managed to create a test CD (not quite the same as the offical 2005.1 release, couldn't find the spec files for it) with the attached patch applied to the linuxrc file in the initramfs. The CD boots in a USB drive with _no_ additional kernel options required. The first scan for the device fails but the second (after a short delay) finds the device on /dev/sr0 and continues to boot normally. I've also managed to boot this off a 128mb USB memory stick (appeared as /dev/sdb). All the patch does is set up /sbin/udev as the hotplug agent in the kernel just after /proc is mounted. I think it should really use /sbin/udevsend as the agent, but I wasn't sure that it is present in the initrd environment. As far as I know this shouldn't impact on any of the other aspects of the live cd. I hope this resolves the issue. I have tested this extensively and it may not be the best way to do it, however it does appear to work. Thanks, G.
> I have tested this extensively and it may not be > the best way to do it, however it does appear to work. This should read "I _haven't_ tested this extensively and it may not be the best way to do it, however it does appear to work." Freudian slip there..... G.
Genkernel: please reassign this back to me once you have this (or some other similar) patch in place in genkernel, as I want to leave this open until new media is available with this patch included
Who's ready to test a CD? http://dev.gentoo.org/~wolf31o2/test/install-x86-minimal-testing.iso Once I hear that this is working properly, I'll roll it up into a new LiveCD build, along with a new set of Minimal and Universal CD releases for the mirrors.
(In reply to comment #37) > Who's ready to test a CD? > > http://dev.gentoo.org/~wolf31o2/test/install-x86-minimal-testing.iso > > Once I hear that this is working properly, I'll roll it up into a new LiveCD > build, along with a new set of Minimal and Universal CD releases for the mirrors. it hangs with the following error message: /init: 211: Syntax error: "done" unexpected (expecting "fi")
Yeah... I noticed that myself this morning... I'm rolling a new CD now with that fixed...
> All the patch does is set up /sbin/udev as the hotplug agent in the kernel just > after /proc is mounted. I think it should really use /sbin/udevsend as the > agent, but I wasn't sure that it is present in the initrd environment. I've done a bit more digging and found that 'udevsend' is present in the initrd, however, the documentation for it suggests that 'udevd' would need to be running prior to adding it as the agent. I guess it would be possible to do this but I'm inclined to think this is a bit OTT and that 'udev' be left as the agent. G.
(In reply to comment #39) > Yeah... I noticed that myself this morning... I'm rolling a new CD now with that > fixed... I just pinched your testing2 .iso image, burned it and booted up. Short of it being in debug mode and me having to type exit, it worked sweet as anything on the Dell D410 apart from a few niggles with wireless. Anyway it booted just fine on the USB CD, so seems good.
The new CD works fine for me too. Ignore my last comment 'udevsend' is obviously fine without udevd :-) Did you have to change anything else other than put the echo line into runUdev? G.
Honestly, I didn't change it. One of our genkernel developers did. I just made a test CD. As far as I know, there was logic added to determine if udevsend was built or not, and to copy the binary into the initrd if it was built, or symlink it if it was not. Then the echo is done to proc. The debug issue is due to another patch I was testing at the same time, or at least I'm pretty sure it is. I'm investigating that a bit further.
Cool, cheers for the info. It definately looks like the best way to do it. G.
(In reply to comment #39) > Yeah... I noticed that myself this morning... I'm rolling a new CD now with that > fixed... Am going to test it also this evening...
One thing to note is that because the test CD is booting into debug mode, there is enough of a delay for the USB device to be picked up the first time round the scan loop. I imagine without the debug delay, the device will get picked up on the second iteration. However, this should be checked when the final release CD is built.... G.
(In reply to comment #45) > (In reply to comment #39) > > Yeah... I noticed that myself this morning... I'm rolling a new CD now with that > > fixed... > > Am going to test it also this evening... OK, I can confirm that the new CD boots successfully :) Waiting for a release with debugging-mode disabled.
confirmed it booted successfully. the new LiveCD will have its own version, 2005.1.1 or something?
The LiveCD, since it won't be based off 2005.1, will be dated. I also am *not* changing the normal install media at this time. I've built another -testing CD, but won't be able to upload it until later today. This CD should have debug mode turned off. Once we're sure it works, I'll actually be building a new Installer LiveCD. Give me a few days, as things are kinda crazy for me at work right now.
Try this ISO. It no longer hits debug mode. http://dev.gentoo.org/~wolf31o2/test/install-x86-minimal-testing3.iso
The ISO you provided picks up the CDROM (firewire) for me after the initial try.
install-x86-minimal-testing3.iso works for me on an HP Proliant DL140-G2 with an external USB CD drive where 2005.0 and 2005.1_pre1 live CDs did not. Thanks!
3rd ISO works for me too (both from pendrive and from CD-ROM). As suspected first search fails, but the usb drive gets picked up on the second. Thanks for the work you've put in fixing this. G.
solved it for me Thanks very much http://forums.gentoo.org/viewtopic.php?p=2676329#2676329
The genkernel part of this is now in CVS and in a released version of genkernel (3.3.6)... Reassigning to release team until a new release is made with these changes.
This is fixed with the 2005.1-r1 release media.
