A debate has started in the Linux kernel mailing list about the ide-scsi module. The cdrecord guys want it up while Linus points out that the module is “bad” and it imposes a number of problems to users. Our take: I’ve been hit by most of Linus’ points: from the ‘silly CD numbering’ confusion, to the need to have two LILO entries to load ide-scsi or not in order to either burn a CD or be able to watch DVD.
I guess I’ve just been lucky because I simply never had issues with it. I’ve been able to burn cd’s without problem and watch DVD’s without any drama.
Reading the article and from what I’ve read in the past its clear that IDE burners should work very well with 2.6. It also sounds like ide-scsi and the older “semi-flawed-but-working” version of cdrecord are going the way of the dodo bird. The one really important app that needed ide-scsi for cdburning(an obviously important function) has been patched and works fine now. The only caveat and this is all in theory are ide-tape devices which has been pointed out are not definitely affected by this.
So basically no need to freak out. To quote from the article.
“The bottom line seems to be that since cdrecord has been patched to handle IDE device names as well as the old SCSI naming scheme, and most applications for those devices (like K3B) use cdrecord to access them, there will be only a few broken by the move to 2.6. If and when that happens, someone will step up to fix it.”
Hope this spells things out for those of you who have no intention of reading the article and just FUDing about Linux in general.
Btw it looks like what Eugeina is reffering to is if you own a combo DVD/CDRW drive. Feel free to correct me if I’m wrong.
People who use seperate DVD and CDRW drives usually don’t have that problem but either way its good to see its been fixed.
I’m glad to see it going.
I never did understand the need to pretend your IDE devices were really SCSI devices. It smelled of a kludge, but if that was what it took, that’s what I did. Seemed rather silly, with most devices these days being IDE.
I’m looking forward to 2.6. I just hope it doesn’t require a double digit revision before it’s truely stable (i.e. 2.4.xx).
Just my $0.02 (Canadian, before taxes)
someone247356
I have never had to turn off ide-scsi to watch a dvd.
Normally you can indeed specify hdd=ide-scsi and still have hdc as your DVD-player while your burner is availlable through scsi.
I think the problem starts with DVD-burners.
Or combo drives.
those burns problems are sure as hell annoying, I have had many problems on my system, could not burn more than one CD per session, had to reboot if I wanted to burn a second CD (under 2.6). Finally downgraded my kernel to stable 2.4.23 and haven`t had a problem since.
good thing Linus can keep its end, otherwise it would be havoc.
Not on my laptop. I use a cdrw/DVD-combo and I have never had any problem of any kind (related to above).
Would someone be so kind as to list the things that need to be done to get cd-writing/dvd-writing/dvd-reading to work properly in the 2.6 kernel.
Also what would be needed to set up cd to cd copying/writing.
Thanx alot ! ! !
Funny, I was just ripping ide-scsi the other day with a friend. Besides the issues proposed by this article, there’s a huge problem affecting users of IDE devices connected via the USB bus.
I recently purchased a USB 2.0 external enclosure for some of my IDE drives. However, it wasn’t until I started playing with it that I realized that ide-scsi was necessary for interoperating with the usb-storage drivers. Forcing a scsi interface on these devices means I lose the ability to tweak IDE settings with hdparm (32-bit I/O, DMA, etc). Has anyone considered this?
-fp
I don’t know about dvd-writing, but I done cd-writing/dvd-reading under the 2.6 kernel, and must say I don’t think you have to do anything special, just make sure you have the right modules/build in drivers. IMO ide-scsi was just a work around for cd burners. Now that they are supported natively by linux, who needs it?
I had the impression that the Gnome CD Coaster guy is developing a lib for burning CD’s, independent of cdrecord… is that right? And has anybody tested it?
Victor.
a interesting point in the article is:
Linux is the worst OS I am aware of if you compare SCSI transport implementations. Every even year, a new completely disjunct new kernel interface appears.
And unfortunately, in the linux world, this is oh so true, every time a developer turns his head, someone goes and changes the api for the lib or the kernel used.
How bloody hard is it developing a api for each major branch, and then, in the NEXT major branch you can do a new api, but be backwardscompatible, and then remove the api in the next major version after that. No, its not hard.
well, the new kernel is hella stable in development, I haven’t had but one system lock on 2.6.0-test series, none under test11. I highly recommend download the source and compiling.
LKML is full of these arguments over decisions already made and implemented, take it with a grain of sand. Combo DVD/CD-RW drives work just as well whether they are ide-scsi emulated or they are accessed via the new interface. The old way (ide-scsi) is a slapped on hack that has far outserved it’s usefullness. The ongoing scsi subsystem rewrite is moving this now obsolete interface out into the cold where it belongs. The new ATAPI interface is going to be more friendly to ATAPI devices that are less friendly than normal. What this means is being able to turn on or off U/DMA access to misbehaving ATAPI devices without the bother of loading/unloading/reloading the ide cdrom driver and the ide-scsi driver. When the bugs are finally worked out of the new interface all of this spurious arguing will fall to the wayside.
I don’t know if any one has actually read the original documentation from the cdrecord authors, most people seem content to simply install their OS’s prebuilt binaries, but it generally comes across as arrogant and decidedly snide in regaurds to Linux and GNU software. Such extreme bias discounts their arguments in my opinion.
Just remove all the SCSI emulation junk from the kernel and remove the ide-scsi line from your boot loader, grub or lilo, and you are good to go.
Thankfully, I don’t use cdrecord for burning. Nautilus can burn data to CD. And for CD ripping, I use sound-juicer. None of which use cdrecord.
After which burn and ripp away. The speed performance is amazing!
Thankfully, I don’t use cdrecord for burning. Nautilus can burn data to CD. And for CD ripping, I use sound-juicer. None of which use cdrecord.
Hummm… the last time i took a look at the source of nautilus-cd-burner, it was using cdrecord.
Victor.
Howdy
Hmm this SCSI -> IDE debate seems to be rather strange, on the one hand we have guys who seem to want it as they doon`t want to write seperate IDE and SCSI handlers and then you have poor old Linus saying that the SCSI -> IDE thing is a bad idea.
Soooo why doesn`t the kernel abstract the two types (block devices?) and why doesn`t CDRecord use it? … this is a question not a flame i mean do you need to get right down to a low level with these devices or something ? (which seems horribly wrong in the firstplace).
I never did understand the need to pretend your IDE devices were really SCSI devices. It smelled of a kludge, but if that was what it took, that’s what I did. Seemed rather silly, with most devices these days being IDE.
Well, it is a kludge I guess, but not where you’d think. The ATAPI hardware standard is SCSI over IDE.
The USB mass storage devices are SCSI over USB.
In the case of IDE (If I understand correctly, and someones going to probably tell me I’m wrong here), it’s because IDE is only made to handle basic tasks along the lines of “Read here, write here.” whereas CD Writing requires much more complex instructions, as does the various types of USB devices.
My guess is that in both cases, it was easier to slap an existing, and known to work, SCSI interface on USB and IDE, than it was to bugger about re-engineering everything all over again. Especially given that SCSI is already known to work over a bunch of different links (ribbon cable, ethernet, TCP, IP, fiber, to name a few).
Hi no problem on my side: I have an LG CDRW/DVD combo drive.
I use XCDRoast and Xine and never had to mess up with ide-scsi, using ATAPI IDE for both things. Only cdrdao requires ide-scsi on my setup, because cdrdao does not like ATAPI IDE. Dvdrip requires cdrdao . . .
Personally what I think from a user point of view is ‘use scsi with scsi devices and IDE ATAPI for the others’, like already happens on consumer OSes.
my 2:)
– Regards
– Pasha
This whole kludge emulation thing has been kinda interesting. I sort of bypassed it at home by puttin my IDE burner into a firewire enclosure.
But at one point I was very interested in writing a program that would by itself resample and cut a file up into tracks on the fly while burning it.
When I looked at the cdrtools suite I saw bo diddly jack anything for any sort of API or library I could use to do something simple like: intialize device, get a buffer to write to, write in buffer and send buffer for write, et al. The absolute only way to do anything was to set things up and make a call directly to cdrecord to do it.
I sent an email asking why this of functionality was unavailable and pointed to the SANE project as a way that software designers dealt with a similar sort of device problem. I was less than impressed with the response I got. The resistence to any suggestions reminds me now of how the xfree86 folks are.
I will tend to side with Linus on this issue.
Bottom line, all these issues are a joke and have been for a very long time.
This should have been put to bed in a perm solution and I find it comical that its still being argued about. Pathetic.
AdmV
sorry Anonymous, but Victor is right.
-atom
emerge nautilus-cd-burner -p
These are the packages that I would merge, in order:
Calculating dependencies …done!
[ebuild N ] app-cdr/cdrtools-2.01_alpha14
[ebuild N ] dev-libs/libxml2-2.5.11
…removed other packages…
[ebuild N ] gnome-base/eel-2.4.0
[ebuild N ] gnome-extra/nautilus-cd-burner-0.5.3
I’ve been using linux for 2 or 3 years now, and I finally got around to making my CD-R work (under the 2.4 kernel). The main problem I had was dealing with the ide-scsi stuff. Not having to deal with that is actually one of the major reasons I’m on the 2.6 branch right now.
I say burn ide-scsi like the witch it is!
scsi emulation enables more nice things like hyper-scsi.
where you can use scsi over standard raw ethernet(not tcp/ip). and that would be nice fore some cheap san
yeah i never did get the scsi thing either i just want to burn straight to ide
I’ve been using my DVD burner (LiteON 401S) and haven’t had a problem on 2.6-test10. I don’t see a reason to use ide-scsi.
I used ide-scsi for many years and it worked reasonably well. Last week I finally dumped ide-scsi. Now my computer aoutmounts (via supermount) and recognizes automatically both drives(dvd and cd-rw) and recognizies if the inserted disc is an audio CD, DVD or data CD. And this is under kernel 2.4.20.
Eliminating ide-scsi profoundly simplified my setup. No more stuff in GRUB, less stuff in FSTAB, less symlinks in /dev and less problems in general. cdrecord works fine with atapi burning using k3b-gawds gift to cd burners everywhere, one needs only the very,very latest version of cdrecord.
here is a snipit from my fstab:
/dev/dvd /mnt/dvd supermount ro,fs=udf,fs=iso9660,dev=/dev/dvd,–,user 0 0
/dev/cdrom /mnt/cdrom supermount ro,fs=iso9660,dev=/dev/cdroms/cdrom0,–,user 0 0
/dev/cdrw /mnt/cdrw supermount rw,fs=iso9660,dev=/dev/cdroms/cdrom1,–,use0 0
/dev/ide/host0/bus1/target1/lun0/cd /cdwriter auto ro,noauto,user,exec 0 0
Although this may not be new to some of you, for me it is a first- first time I have gotten the whole thing working like it should-correct permission, automounting, availability of DVD without do anything extra. Using gnome2.5 a drive icon on my desktop changes when I insert the drive, along with the text beneath it, I open the little spatial-nautilus browser window “computer” and There are all of my drives and partitions- having finally gotten my user permissions nailed down I can right click on any partition and mount it.
The reason I am so happy is this-If something goes amock with your permissions you can have one hell of a time trying to find the solution to the problem-there are so many files, and so many configuration files, scattered everywhere, one must hire a detective from scottland yard.
Some number of months ago this happened to me-I was constantly switching back and forth between the 2.6 and 2.4 kernel, and due to the configuration differences between the the two, I had to modify my local.start and /etc/fstab files accordingly and somehow ended up confused. My nightmare began when K3b announced that my drive wasn’t supported by cdrdao- well it was last week, why not now, so I spent hours trying to get cdrdao to recognize my (already previously) recognized drive. In the course of which I mucked around with permissions. The result utter and absolute frustration. Some number of weeks later I installed nautilus cdbrurner and tried it and voila- it worked with zero configuration. (Perhaps I got lucky and had the right combination of values in the various files, unbeknownst to me). Then last week I decided to dump ide-scsi. Now my headaches are over, at least for now…..
Although I love the 2.6 kernel, and am blown away by how much faster it is than the 2.4 kernel(all variations thereof), I can’t make the switch yet- two many things I use are dependent upon 2.4. The fritz capi module for my isdn card does not yet exist for 2.6- thus crippling my answering machine, which saves and emails me my messages in mp3 format, and my fax send/recieve, and my telephone logging app which records the name and numbers of those who called me which is coupled with a mysql telephone database. My HP ps-1320 usb printer/scanner is also not yet supported under 2.6- both the printer and the scanner work fine under linux, albeit I wish the max resoution of the device was better supported…And finally my IP masquerading via iptables is still not working right under 2.6 even though I am using iptables for 2.6-meaning I cannot ssh into my machine from elsewhere and my local network does not work. So I have to wit a little bit longer to really taste the fruits of 2.6…..
Funny, i do all the things you mention with ide-scsi 8o
But i am using a 2.4.22 and never used K3B.
from what i remember there is a tool on freebsd that can burn cds (on IDE burners) without the need of a kernel module or anything weird, so why does cdrecord work this way ?
I don’t know what this discussion is all about.
I am happily burning CDs right now without any ide-scsi module. I don’t why these consumer distros (SuSE, RH, Mandrake) load ide-scsi.
This works for me on Debian & Gentoo under a 2.4.22 kernel witout any problems:
cdrecord dev=ATAPI:n,n,n some.iso