|
DVD-R Installation history I got the file cdrecord-prodvd-1.11b19-i586-pc-linux-gnu from Joerg Schelling's cdrecorder site, under ProDVD (an ftp directory with binary files only). I followed the instructions in xcdroast and copied the file above over /usr/bin/cdrecorder, after first making a copy of that file called /usr/bin/cdrecorder-original. The new file kept the same file permissions as the old, but I had to manually modify the ownership with chown. I created a bash scrip /home/steen/.prodvd that looks like this:
The long line in the middle is the license key; I wrote the author for a personal license but he said he don't need one. This means that you can start the proDVD program with this command:
and xcdroast will still be the interface. It seems to be working fine
(I mastered a disk but didn't burn one). DVD-R: regionset In order to play DVDs on my new DVD-R, I had to do this on 21 June 2002:
Summary Both drives are now DVD -- one reader and one DVD-R. Whether CD or DVD doesn't seem to matter at all to Linux -- it's simply handled through hardware detection on boot. What matters is that the drives must be defined in fstab, including mount points in /mnt, and the /dev files must link to the sr1 or sr0 defined during boot. Lilo must pass boot parameters defining both drives as scsi devices. It's a little nightmare to set up the first time, when you don't know what you're doing, and trivial after that. Kernel 2.4.16p Summary After building a new kernel, complete with a patch for the 160GB drive, the CD Recorder no longer worked. I had no problems with the CD-ROM. Note that under SuSE's installation, the CD-Recorder works fine. Possible developments
Narrative Boot of Saturday 29 - Sunday 30 December 2001, after reinstallation of soundcard, which now works fine. You should list the kernel elements systematically. Now just some error messages: ACPI: System description tables not found Note that all drives are recognized at the outset: hda: Maxtor 91360D8, ATA DISK drive Really, this is a remarkably good boot, with an excellent kernel that makes the most out of the hardware in some detail. Linux is great for building systems stepwise, I mean if you get some things right, they stay right. Now, the CDRW drive is having some weird problems. The booting process recognizes it fine and assigns it to hdd. YAST2's hardware detection program finds it with no problem at device name /dev/hdd. It for some reason lists it as "notready: Yes", but sees that it's a CD-RW CRX100E on the IDE bus. Yet YAST is unable to mount it. If I try, I get the same result: mount /dev/hdd cdrecorder So there's a problem with hdd as a block device -- it could be that this requires a change in the kernel. The problem seems to be that /etc/fstab lacks the necessary definition! /dev/hdd /mnt/cdrecorder iso9660 noauto,ro,usr 0 0 What I had was this: /dev/cdrecorder /media/cdrecorder auto ro,noauto,user,exec 0 0 Now, it may be that you should just change the auto to iso9660 and leave the rest -- but I know the boot process labels it hdd. As for /media/cdrecorder, it does indeed exist. However, nothing works. When I try mount /dev/hdd I get a surprising result: mount: mount point /mnt/cdrecorder does not exist This suggests that the system recognizes that /dev/hdd is a cdrecorder and is about to use /mnt/cdrecorder to mount it. However, I think it's likely the problem is deeper -- mounting to /cdrecorder just gives "/dev/hdd is not a valid block device". Actually, now I know why it's looking for /mnt/cdrecorder -- that's what I just entered in /etc/fstab. So I changed that to /cdrecorder. Now I get simply the old error message, "/dev/hdd is not a valid block device." There's a problem in the kernel, most likely. It turns out it's been a common problem with RedHat 7.2. Here's one of their fixes: the problem was that the installer failed to include the following line in modules.conf: alias scsi_hostadapter ide-scsi Now, they all assume I'm using modules, which I'm not. But I do need the alias scsi_hostadapter for an "ide-scsi" -- and my kernel configuration says this: more kernel-2.4.16p.config | grep SCSI_* Now my guess is that I need the first one, though maybe not the others. Here's the guide for it: SCSI emulation support So it looks like I need three things I left out: * CONFIG_BLK_DEV_IDESCSI That should be the problem. At least you're getting better at this stuff now! Now note that in the working first installation, you simply had /dev/cdrecorder /media/cdrecorder auto ro,noauto,user,exec 0 0 You didn't have to specify the ISO9660 in fstab. Instead, I believe that is done when defining the mount point, or the device cdrecorder. It could be that the classification hdd stems from the lack of the pseudo-scsi support -- it should turn out differently with a new kernel. So I'll take fstab back to the line above. I'll try building a new kernel. I had to switch user to root -- make xconfig wouldn't let me use su. So remember to read the old configuration, and then find the new element. For a handy overview, see http://www.linuxdoc.org/LDP/solrhe/Securing-Optimizing-Linux-RH-Edition-v1.3/chap7sec80.html Basically, I reloaded the old values from /boot/kernel-2.4.16p.config and then found the CONFIG_BLK_DEV_IDESCSI in "IDE, ATA, and ATAPI block devices". I also found SCSI support and SCSI generic support in the SCSI section, and added all three. That should do it for tonight. Later, when the CD is working, you can remove the firewall, which is just for the modem, which I never use. Now, the remaking of the kernel went without a hitch -- while it was "make dep" I realized I should also at some point (no hurry) deal with the power-off problem and perhaps also remove the ipv6 (Internet 6) which I have no use for. Recall also that your ACPI is giving this error message -- this could mean you have to enable more if you want it, or leave it out altogether: ACPI: System description tables not found But the CDRW fix ran into opposition. Here are some of the messages: First, it finds the drive: hdd: CD-RW CRX100E, ATAPI CD/DVD-ROM drive Then, it installs the scsi-ide support -- this looks good: SCSI subsystem driver Revision: 1.00 In fact, this looks extremely promising -- it gets it all right. Then it runs into a very peculiar problem: VFS: Disk change detected on device ide1(22,64) Now, ide1 is the 15GB 7200rpm drive -- my main Linux drive. It's absurd to suggest there was a disk change on it. What is the "unsupported command in request quue (0)? This message repeats several times, culminating in isofs_read_super: bread failed, dev=16:40, iso_blknum=16, block=32 Now, I'd left a CD in the drive and it could be it was trying to read it and for some reason failed -- not too likely. It is conceivable that the first and fourth error message refer to hdc (my main Linux drive), while the second and third refer to the CDRW. Try once more -- and power down first. Well, the second boot produced none of these last error messages -- again, a perfectly clean boot. However, now the system has decided to place the CDRW at /dev/sr0 -- and it is of course "not a valid block device". Now, /etc/fstab hasn't changed -- so here the system has tried something new and decided that it should mount on sr0 -- which then doesn't work. It's peculiar I have such a problem with this -- maybe the redhat 7.2 problem really is the same one -- this could be a problem with the 2.4.16 kernel. I guess I could try swapping out the other CD ROM and use that instead? That would be one solution. The other would be what? Looking at dmesg, I'm getting "hdd: CD-RW CRX100E, ATAPI CD/DVD-ROM drive" and then SCSI subsystem driver Revision: 1.00 But wait: at the end you're getting just the same junk: VFS: Disk change detected on device ide1(22,64) Here's the identical problem from a 1999 Debian mailing list: VFS: Disk change detected on device ide1(22,0) Wow. So it's not likely kernel 2.4.16. Here's a howto: http://www.linuxdoc.org/HOWTO/CD-Writing-HOWTO.html -- here's what they say: ".. you have to load the compatibility driver ide-scsi. But this driver can only access your CD-Writer if no other driver has already done so. In other words, you have to tell the regular IDE driver to leave your CD-writer unrecognized, so the ide-scsi driver can grab it. hda = IDE bus/connector 0 master device The table above showsthe relation of device file names and the placing of devices on the IDE busses. The device file name representing your CD-Writer has to be passed to the driver in the Linux kernel. Example: hdb=ide-scsi. Such a setting should be added to lilo.conf or chos.conf if the driver is statically compiled into your kernel, which seems to be the most common setup. If you need to pass more than one parameter to the kernel, then seperate them with spaces (like shown in the chos example). The next two listings show example configurations containing more lines than just the relevant append-line. Please note the append- and cmdline-entries are image-specific (ie. don't add them immediatly at the top)." Now at least this clears up why the cdrecorder is referred to as hdd -- but is it in fact till the slave on the second drive? I think this is still correct, so this is not that helpful. Maybe you should just swap it out with the normal drive for the moment. Try that. Well, if you do that, you'd have to change Lilo -- and the whole thing was working so well at first! So here's a simpler solution: the zip drive had an extra jumper. I'll try to install both CD drives. If only one works, I can leave the other for the moment. Having said that I now read the manual for the CD-ROM drive and it said the second jumper -- the one I lost -- was only to disable UDMA, and thus of no consequence. So I put back the jumper for the Zip disk, just in case. Now, to be on the same side, I rebooted to the SuSE 2.4.10 kernel, with modules, which had worked for both CD drives before. Here's the result: First details on APIC (which works in this installation): Using local APIC timer interrupts. Second, APM seems to also be enabled: apm: BIOS version 1.2 Flags 0x07 (Driver version 1.14) It finds the four disks (but of course not the new one): hda: Maxtor 91360D8, ATA DISK drive It diagnoses the first correctly: hdb: ATAPI 48X CD-ROM drive, 128kB Cache, UDMA(33) Then you get this: SCSI subsystem driver Revision: 1.00 Now, the scsi_hostadapter -- can you compile, or have you compiled, that into the kernel? Not sure about this -- it could be what you need. Then you get: VFS: Mounted root (ext2 filesystem). Note that you get a problem iwth VFS on the other kernel: it won't mount and complains of "Disk change detected on device ide1(22,64)." This seems to be symptomatic of the problem. You then get: Attached scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0 This is what the other kernel is not managing to do. One more detail: eth0: no IPv6 routers present This likely means I can remove IPv6 from the other kernel. So, that's the story. It works perfectly in the other kernel. And it likely works because of the "SCSI host adapter emulation", which is handled by the module [scsi_hostadapter]. Well, it turns out that [scsi_hostadapter] is an optional name for a module -- it's hard to see that I really need it. There is some little tweak I need to do -- in fact, it could be that now that the other CD drive is in place, it will work again. Or at least one of them will work, which is fine for now! So try that. A final idea: when you do make dep after making your kernel selections, you could pipe that to a file and inspect it for warnings later. This might tell you about features that aren't going to be implemented anyway. After rebooting, I got these results. First, the four drives are found: hda: Maxtor 91360D8, ATA DISK drive Next, there is some gratifying recognition of the CD-ROM: hdb: ATAPI 48X CD-ROM drive, 128kB Cache, UDMA(33) This is identical to the SuSE kernel. Then you get as before: SCSI subsystem driver Revision: 1.00 You also get this one: VFS: Mounted root (ext2 filesystem) readonly. Note that you got that already before the SCSI subsystem detection in the SuSE kernel -- and not "readonly". On the other hand, the usb detection works much better in the new kernel -- likely because of improvements in the kernel itself, rather than in my configuration of it. In addition, of course, you get the huge drive and then firewire with the new kernel -- the stuff you really want to use. Question: are the drives working? Should I add a line in /etc/fstab? Do I need to define valid mount points? These could still be issues. I first get an error message: VFS: Could not mount the source medium. However, when I inserted the CD, it works fine! So at least you have a functioning CD -- you can figure out the details on the CD-RW later. As I write this it's 6:30 am. I've spent another night on Linux. I made things work, but at a huge price. Kernel 2.4.16m3 I still cannot figure out the CDRW problem. The disadvantage in going with SuSE is that they have their own non-standard configuration that's hard to penetrate. The guide is at http://www.linuxgazette.com/issue57/stoddard.html The brief point is that the CDRW should be fully detected during boot:
In my case, the first two lines do appear, but the last two do not (see kernel 2.4.16m3). The CD-ROM is found, the scsi driver is also loaded, but the two are never linked. It is possible that I need to edit /etc/modules.conf:
I replaced this line with:
It could be it's that simple -- yet this should only have to do with modules, so it doesn't make sense to me. Some claims and voices from various places:
On making nodes:
Then make links, in my case:
Names of the devices under devfs: CD-ROM will be something like /dev/scsi/host1/bus0/target0/lun0/cd
or So /dev/sr* (reading) and /dev/sg* (writing) are what you need. You obviously need SCSI generic and CD-ROM support in the kernel or as modules as well as SCSI-emulation compiled into the kernel or as a module.
Become root, setup devices:
Check the new SCSI settings:
Setup cdrecord's environment - edit /etc/default/cdrecord:
Input the right values, the fields are described in the manpage of cdrecord.
Alternatively, you may use this values as It's done! Insert a CD and try "cdrecord -v -toc" http://www.lugatgt.org/articles/cd_burning/ 1.2. Copying a CD
Or you could burn directly from CD to CD assuming you have two drives:
Lastly, you can use dd and produce an image and then burn that, like the cat example above. Kernel choices In a weak moment, I made the idescsi in the kernel into a module! So it must be called -- insmod -- and I have to know where it is and what it's called. Here's a diff from this and the previous kernel's configuration file:
So it looks like you didn't set DEV_SR, which you need to read! the DEV_SG is for writing. Kernel 2.4.16m4 Whatever I did between kernel 2.416m3 and m4 made the CDRW work!
As you can see, it ended up on sr0, and I now have no problem mounting it. Here is the diff file: I turned the scsi emulator from module to yes:
I also turned the scsi general from module to yes:
I turned on DEV_SD in m4 -- SD_EXTRA_DEVS=40 may also have been added:
I turned on DEV_SR in m4 and converted CHR_DEV_SG from module to yes:
I turned off devfs in m4:
xcdroast On 4 June 2002 I installed the following SuSE rpms for X-CD-Roast 0.98 a10 on cyberspace. I did it because the program supposedly can read DVDs and possibly rip them -- it may not be necessary, as I should likely use another program for DVD ripping, but the package is very well made and works great on gubbio. I got them here: http://www.xcdroast.org
The rpms for SuSE 7.3 are all here: I downloaded the rpms (according to the instructions) and issued
They installed nicely. Then I did the same for xcdroast:
There was some kind of conflict and I uninstalled the cdrtools and koncd and tried again. The second time is was like a hot knife through butter. I had to be root to run it the first time -- and my machine is tied up with transcoder! But still, this is fine. modules.conf I then messed around with /etc/modules.conf # # Used by i2c and lm_sensors (not currently working -- a /dev problem?) I also uncommented the following: # Creative SB Live! (ALSA will find it with the kernel's PCI support): I also uncommented these: #***************************************************************************** # # Uncomment this line too, you can setup your ALSA device permissions
here: Now, I might live to regret it -- indeed, I had to just comment it up
again immediately. I don't even recall the error messages, but there was
massive protest. June 2002 update Since early June 2002, I've been using xcdroast to burn CDs with no problems. On 14 June I finally figured out how to let user steen burn and not just root -- the trick is to allow hosts and not just users. The current setting is allow all users and hosts, which can hardly be a problem. Here is the output from a CD-burning session. Note that you're using cdrecord 1.11a19 (1.11b19 is now out), and that packet writing (see below) may be supported:
Early September 2002 On 5 September I started using the 2.4.19-ac4 kernel; this made no difference to my CD and DVD-R setup -- that is to say, I haven't actually tried writing, but it didn't seem to make any difference. On 9 September 2002, I installed a new DVD, which replaced the previous CD driver. I also did this: 1. /etc/fstab -- change the existing /media/dvd to dvd-r 2. /media -- change the existing /media/dvd to dvd-r 3. /dev -- change the existing /dev/dvd to dvd-r This worked. Hell if you don't know it, heaven if you do. Lilo On the CD front, I also added this to lilo.conf --
This works great -- see the 2.4.16m5 kernel log.
|
|
|
|
|
|
|
|