From ROBB F! on Mon, 27 Mar 2000
I have a problem w/ the graphics on my intel system. They don't work.
I installed Redhat Linux onto my CA-810, but the only thing I can do is getinto nongraphix mode. Now, I can get into root, but am not very fimiliar w/ workig there and would very much like to be.
I didn't know what a CA-810 when I read this question. However, a quick trip to Google's Linux pages (http://www.google.com/linux) and the search string: "ca-810 intel video" did net me the following links:
Linux Today: How to Build Your Own 1U Rack Mount Server and
- Save a Bundle
- CA810 Motherboard Product Documentation
That told me that the CA810 is a line of integrated motherboards by Intel. Some of them have integrated video and ethernet. They seem to be primarily used for 1U (one rack unit: 1.75 in.) rackmount and slim case ("pizza box") systems.
Then I went back and changed my search query to "ca810 intel video" to find this link:
- Slashdot | Ask Slashdot | Cheap Rackmount Enclosures/Systems?
Then I used "find in page" (the "/" key in lynx) on the term "video" to find that this board includes a comment to the effect that "the video is the Intel 82810" and that the SVGA X server is appropriate for that video chipset.
I tried to cross-check that at the XFree86 web site (http://www.xfree86.org) which maintains a list of supported chipsets (http://www.xfree86.org/3.3.5/README3.html). However, I didn't find the 82810 listed there, and the only Intel video chipset listed was the i710.
So I went back to Google and tried a few searches on variations of "82810" with words like: intel, xfree86, x windows, xwindows, eventually finding references to this as the "Camino" chipset and to some discussions on the Linux kernel list (kernel panics and "oopsen" from overheating issues), and a reference to an XFCom_i810 server in S.u.S.E. version 6.3 (back in December of 1999).
I also found a kernel patch updating the list of PCI id codes with a set of 82810 chipsets.
I also found another "Ask Slashdot" thread that referred to this as compatible with the XFree86 SVGA (generic) driver. Finally I followed up by looking at the S.u.S.E. site and searching through their list of X servers (supported video chipsets).
S.u.S.E. has often offered support for video chipsets before they were available in the XFree86 sources. They've contributed back to XFree86 wherever possible. However, some of these were done with Precision Insite and some of them involved NDAs (non-disclosure agreements) with the video chipset manufacturers. Sometimes Linux and X programmers will sign NDAs in order to produce freely distributable but closed source support for some hardware.
Anyway, that was about 10 minutes of searching at Google. If I really needed to I could probably dig up more at Yahoo! (http://www.yahoo.com), Deja News (http://www.deja.com/usenet) or Alta Vista (http://www.altavista.com).
(In fact, I did go a bit further. From Deja I found that the unique thing about this chipset is that it uses main system memory for the video framebuffer (presumably by using the PCI chipset to reserve a block of motherboard/main RAM for its use. That's a clever trick, and potentially very handy for servers where you don't usually want to waste memory on graphics at all, but certain legacy operating systems, like MS Windows NT, can't be installed or operated with using a GUI. That and another reference to an old kernel issue involving the hardware clock support).
O.K. So this particular motherboard has had some issues. It is not well supported, but it is apparently quite inexpensive, thus making it popular. Naturally many people have gone out and bought them and found that Linux might have some problems with them.
It sounds like the main kernel issue (the hardware clock support) was fixed back in December; so a recent kernel will fix that. Linux doesn't require a GUI for any normal server operations, so most of the core users and target for this product aren't affected by that chipset. Some people have gotten it working fine under the SVGA XFree86 driver, and none of them reported any special steps (like CMOS settings) as necessary to make that work. Some others have reported that a S.u.S.E. XFCom_* server/driver works for them.
I've gone into considerable detail about how I did these searches so that you, and my other readers and see how that's done. Remember that, at first reading your question, I'd never heard of this Intel "Camino" CA810 motherboard with it's "82810/82810-DC100 Graphics and Memory Controller Hub." With only about 20 minutes of research and another 20 minutes of typing in this buffer I could sound like an expert on the bloody thing.
Now back to your question:
First of all I need help installing the required patch to enable the graphics, I hear it has something to do w the chipset. I think I have the right patch from Intel, (Maybe you can send me the download?) but I can't even get Linux to run the floppy from root. Computer illeterate??? I'd like to change that.
If you think you have the "right patch from Intel" why would you want me to take up my time and bandwidth hunting it down and sending it to you.
I presume that your reference to your floppy problems is to warn me that you have no idea how to do perfectly mundane tasks under Linux.
Indeed, your writing is so disjointed that I should probably insert a translation to see I understand it correctly:
"I don't know how to get X Windows (or other graphics software) working with Red Hat Linux on my Intel CA810 system. I can get into text mode. But I don't know what to do with that.
First, I need help installing the required patch. I've heard that it has something to do with my chipset. I think I have the right patch from Intel. But I don't know what to do with it. I don't even know how to access my floppy. (Like I said, I need to learn more about Linux, and writing).
I'm also looking for a good tutorial on using and administering a Linux system. I need to know how to work as root. About all I know so far is how to cd (change directories), and run ls to see my files. I've seen people do alot more, and I know that much more sophisticated interfaces are available. I would love to be more proficient with Linux."
Hmm. That flows much better. (Of course I realize that my own writing is not perfect. I leave the occasional typo, am prone to some run-on sentences, meander and digress incessantly, and exhibit a poor sense of paragraph structure. However, I do try to put enough into it that people can read through my prose without too much effort). Note that my "translation" is actually about 20% shorter than your message.
I still don't know what patch you think you got from Intel. Is it on a floppy? Where did you get it? What's the filename?
If you downloaded something from Intel's web site to a PC running MS-DOS or MS-Windows, and copied that out to a floppy, then you should be able access that under Linux using the 'mtools' commands, or the 'mount' command with one of the MS compatible filesystems (msdos, umsdos, vfat, or uvfat).
Linux supports a number of variations of msdos/vfat filesystems. The "msdos" is the most basic support for "FAT" (file allocation table) filesystems. The umsdos variant of that supports Linux meta data, long filenames, owner and group association and permissions, by maintaining a set of "hidden/system" files which are transparent to MS-DOS and the Linux "userspace." The vfat filesystem support Win'9x long filenames (the VFAT feature), and the uvfat variant of that hides Linux/UNIX metadata thereunder.
You're kernel might be built with any or all of these. Possibly these drivers are linked in. More likely your copy of Red Hat has them available as loadable kernel modules (under the /lib/modules directory tree).
Generally it is somewhat easier to access your MS-DOS floppies using the MTools package. So you can put your floppy in "drive A:" and just type the command 'mdir' to see a list of MS-DOS files. You can use the 'mcopy' command to copy files to or from the floppy. The mdel, mcd, mmd, mrd, mtype, mren, mmove, mattrib, mlabel commands all work pretty much like their MS-DOS counterparts (DEL, CD, MD, RD, TYPE, REN, MOVE, ATTRIB, and LABEL, etc). If you look closely you'll find that all of these "m*" commands under Linux are actually just links to the same file -- the 'mtools' program. Basically this one program is acting in different modes depending on the name under which you called it.
So the following would get a file named foo.zip into your /tmp directory:
mcopy a:/foo.zip /tmp
...NOTE: you should probably avoid trying to use wildcard patterns like *.TXT, with mtools until you understand UNIX "globbing" (UNIX' way of handling wildcards is different than the way MS-DOS does it). Also note that UNIX is case-sensitive so you should type any filenames you pass to mtools commands exactly as you see them in your 'mdir' and 'ls' commands.
It's also possible to "mount" an MS-DOS FAT for MS Win '9x VFAT filesystem into a Linux directory tree.
Under UNIX filesystems are not accessed through device names like A:, C:, etc. That's a feature that MS-DOS, and Windows NT borrow from the TOPS-10, TOPS-20, and VMS family of operating systems. (MS-DOS is a descendent of CP/M, which is basically a cousin to TOPS-10, an old OS for the PDP-10 family of computers).
Under all UNIX systems and clones, filesystems (peripheral storage devices) are "merged" into a single directory tree. For example, it is common to have a small root filesystem which stores the kernel, the minimum necessary utilities to bring up the rest of the system, and to troubleshoot and repair it, etc. The generally also have a larger filesystem which stores all of the major applications and the less fundamental and non-critical utilities that are commonly needed by users. That is usually mounted under the /usr directory. As networked filesystems became commonly supported under UNIX it was common to group quite a bit of the documentation and many other files unto a file server. That filesystem was "exported" to the others which normally mount it under /usr/share.
All of this is done using the mount command. It's also possible to detach a filesystem using the umount command. (I know, it should be "unmount" but whatever! UNIX programmers have a tradition of leaving one or two characters off of a command for function name --- though usually they pick on vowels).
So, you can usually mount your "C:" drive under Linux using a command like:
mount -t msdos /dev/hda1 /mnt/c
mount -t uvfat /dev/sdb2 /mnt/e
Here I'm using a common convention of mounting "temporary" and "removable" filesystems under /mnt. First you'd have to create your "mountpoint" directories using the Linux mkdir command (you can't use the 'MD' command from MS-DOS). You can name the mountpoint pretty much anything you want.
Once a filesystem is mounted you can access the files and directories under using all of the normal UNIX/Linux command names (ls, cp, mv, cat, etc).
To access your MS-DOS floppies under Linux you might create a use a command like:
mount -t msdos /dev/fd0 /mnt/floppy
Personally I prefer to access MS-DOS floppies using mtools, since it let's me get at my files with just one or two commands instead of having to mount it, get at them, and then remember to umount it when I'm done. (Actually I use whichever technique makes sense for what I'm doing. Usually I want to just get a file unto or off of my system. If I need to really work with a file, will mount it up).
So that gives you a brief run down on accessing floppies under Linux. Now, I have no idea what sort of file you MIGHT have (I'm not even sure if your comment on floppies has anything to do with the rest of your questions, or if you were just citing it as an example of your own unfamiliarity). Consequently I have no idea what you'd do with some file that you got from Intel.
Notice that I never came across any reference to any Intel drivers or support for Linux on these motherboards. Those, like most Linux and XFree86 drivers seem to have been done independently of the manufacturers.
So, how would you get X working on this system. Well, you might start by running the SuperProbe command. This might recognize your video chipset and tell you how much video memory it reports as available. That's a standard trick for looking detecting video cards. Of course we think we know which chipset this motherboard is using. However, there are a number of models of the CA810 motherboard out there. It's certainly possible that your's is different. (It's also tragically common for manufacturers to change chipsets and components in their motherboards --- often without changing the name or model number).
Anyway, running SuperProbe (which comes with XFree86) can also let us know if our chipset is NOT recognized. That might mean that we need a newer copy of XFree86. You don't way which version of Red Hat Linux you have installed there. It might be that you want a newer version, or that you want to download updated XFree86 RPMs.
RPMs are Red Hat Package Manager files. The RPM system is used by several Linux distributions, including Red Hat, Mandrake, S.u.S.E., Caldera, and TurboLinux. You can download RPMs for various distributions at the web sites of their respective distributors. I should note that some RPMs are not interchangeable among distributions. This is particularly likely for the "core" packages that are part of a distribution.
If you do have to upgrade your XFree86 RPMs you can probably download them Red Hat's FTP site (ftp://updates.redhat.com). You might manage to find it somewhere on their web site. As Red Hat has gotten larger and "more corporate" I've found their web site to be far less useful. Looking there for "updates" at this point seems to push for you to pay for their "priority update service" and seems to make it more difficult to actually get what you came for. I can understand it in some ways --- they need to make their money somehow. They still do make the updates freely available; you just have to know more about how to find them. So, use FTP in this case.
However, just downloading the one or two RPMs might not be enough. XFree86 is a complex program. It is the largest and most complex subsystem that comes with most Linux systems. It is dependent on many of system libraries. In other words you might have to upgrade many of the other packages on the system in order for a new XFree86 RPM to work.
It is often easiest to upgrade the whole system at once. If there is a newer version of Red Hat available, you might be better installing that. (You're definitely better off getting a new CD than downloading a whole ISO image. The few bucks you spend at Cheap Bytes (http://www.cheapbytes.com) or even for a whole new "Official Red Hat Linux" boxed CD set is well worth all the time and bandwidth you'll save.
Anyway, then you try running XF86Setup (note the case) and/or xf86config. The former is a graphical menu-driven configuration program. The latter is sort of like a shell script --- it spits out lists of options on the screen and prompts for your information. Give it your best shot with the former. If that doesn't work, then try xf86config.
If neither of these works then its possible that someone might be able to get it working by manually tweaking your XF86Config file (usually stored in /etc/X11 these days). However, I can't describe how to do that here. (It's pretty hard to explain, and I prefer to avoid doing it when I can).
If all else fails you can try one of the commercial X server packages. XFree86 is the popular free and open source X server for Linux. There's a couple of commercial ones including the X-Inside package from Xig (http://www.xig.com) and Metro-X from Metrolink (http://www.metrolink.com). These often offer support for video chipsets that aren't supported by XFree86. That happens with hardware manufacturers refuse to release sufficient technical specifications for the open source community to write a driver. Sometimes these software vendors (Metrolink and Xig) donate drivers to the XFree project.
Remember that there might be a driver to support this at the S.u.S.E. site. You might down load that. It's possible that the S.u.S.E. XFCom_i810 driver will support your motherboard. That might install O.K. under Red Hat, or you might have to fuss with it. You might even want to just switch to the S.u.S.E. distribution in that case.
Obviously I could go on and on. You'll just have to try a few things. In many cases I'm suggesting an "easy" way even for situations where I might be able to get it working differently (and less expensively). Linux is very flexible. It's possible to do almost anything with it if you have the time, expertise and energy.
Secondly, Maybe you know a good tutorial in this respect. Dealing with, and getting around in root. I get th cd/thing (i.e. cd/usr gets you into usr and you can ls the stuff to see what's there) that's aboutwhere i am. I have seen alot more happen and know that total interface is possible. I would LOOOVE to be more profficient!!! HELP?!? -ROBB
You should start with the Linux Documentation Project's "Installation and Getting Started" (by Matt Welsh et al):
You can read the entire book online. In fact, like most of the LDP you might find that there are already copies installed under /usr/doc on your system (or on your Red Hat CDs). There are many other Guides, FAQs and HOWTOs at the LDP site: http://www.linuxdoc.org LDP is the free documentation project which is a major part of the whole Linux movement. In fact, the Linux Gazette, and this column is part of the LDP. That's one of hte reasons I spend so much time writing it.
There is also another book, "Running Linux" by Matt Welsh, Lar Kaufman, et al which shares some chapters with LIGS, but is a bit different in other respects. This latter book is published by O'Reilly & Associates and you can read more about it at: http://www.oreilly.com/catalog/runux3/index.html. Of course there are many other books on Linux out there. The only other one I really recommend for a beginner at this point, is Mark G. Sobell's "Practical Guide to Linux" (http://www.sobell.com/LINUX/linux.html).
I did a review of "Running Linux, 3rd edition" for Linuxcare (my employer) a few months ago. That's at:
Anyway, this has hopefully given you some ideas on how to do a search to find information about Linux. As you can see, most of it is available online and quite a bit comes with your system and is installed under /usr/doc.
From anonymous on Thu, 30 Mar 2000
Hmmm. One of the guys at work pointed out that I missed the boat on my first cut at this answer. I hadn't heard about the Intel CA810, but the guys in Linuxcare's hardware certification labs (*) have gone the distance with this chipset and XFree86 3.3.x
- ( Linuxcare Labs: http://www.linuxcare.com/labs)
So, if you're building a workstation around this system, and you want X support you'll want to check out Linuxcare's certification report on the Hewlett Packard Brio BA400 (which is built around this motherboard).
Unfortunately the details are a bit hairy. The involve grabbing the sources to XFree86 3.3.6, editing those to uncomment a section of C code, and getting a kernel patch from Intel's site and building a new kernel and a new X server. Hopefully with the recent release of XFree86 version 4 this sort of thing will become less problematic.
As more hardware vendors get certified then this sort of information should be easier to find in the future.
Here's the gory details of running XFree86 on the Intel "Camino" CA810: