KT333, IO-APIC, Promise Fasttrak, Initrd
Hey there,
this is gonna be a long one, please bear with me. :-)
I've got several problems/questions. My slackware 8.1 iso-image can't install so I tried making my own install flop (I've made several kernels before since 2.2.4 for various machines) which I think is made ok (pretty sure of it), but I've got some problems with it.
First my setup:
MSI KT3 Turbo2-R mainboard
AthlonXP 2000+
512MB 333MHz DDR
MSI GF4-MX460-VTP
Realtek 8139
Promise Fasttrak Lite IDE RAID (onboard see
http://www.msi.com.tw/program/products/mainboard/mbd/pro_mbd_detail.php?UID=
341&MODEL=MS-6380E)
The promise fasttrak controller has 4 disks... I just use it as a fancy IDE controller. This controller doesn't have a jumper/BIOS setting for putting it in IDE mode so I tricked it. Attached the harddisks one by one so arrays were created of one disk each. This works under WinXP. Most of the partitions were created on my previous system with PIIX intel controllers and are read without probs under XP.
This goes for all my tests... made a dir /tmp/lindisk/<name>.i copied the .config from /usr/src/linux to the <name.i>/config and copied the System.map and gzipped it in the <name.i>. Put the makedisk sh script and the 1440k.img.gz (compressed formatted floppy image for mount usage) to the /tmp/lindisk. Then made kernels and copied bzImage to the <name.i> and ran ./makedisk <name.i> <name.i>/bzImage. Next I made the floppy disk using dd if=/tmp/<name.i> of=/dev/fd0u1440 (the docs say to use zcat, but I found with some defected floppy disks it didn't give errors whilst dd does). The floppies boot so it should be fine :-)
Now the problems start. I've used 2.4.19 and 2.4.20-pre11 (the latest at this moment)
On 2.4.19 I compiled with APIC but without IO-APIC. It gave messages about spurious IRQ's (7.. nothing is connected to the parallel port tho', also there is no parallel port support in the kernels cause I wanted to keep them as small as possible). The 2.4.20pre is compiled with IO-APIC and didn't give those messages.
It gives the message i should contact Vojtech so I did for the via chipset.
The RAID controller comes up with both hd[e-h] and d<x>p<y>'s. Can I use the hd[e-h]'s? since I have 1 disk arrays?
Tried booting into my old partitions with the hdg3 but came up with IO errors. I'm guessing it has to do with the south bridge not being supported or the ext 2 partitions having trouble with being on the raid controller.
I used the slackware install.[1-5] from the current tree (downloaded yesterday). After loading the first disk it nicely mentions it has to load 5 disks, asks for the 2nd I put it in and it's done reading in 0.1secs and asks for 3, it does that for all the remaining 4 disks (that means, it actually reads the first disk but only asks for the rest but doesn't actually read them). Then it boots.... Comes up with message that the dir #41 is invalid which probably is because the other 4 disks aren't loaded.
Hoping someone can tell me more or has some experimental patches. I have reasonable computer knowledge and am willing to help were I can. Can't really program yet but I'm learning C++ (the kernel looks really tough tho' :-()
Found something in the (unofficial) archives that's probably similar:
quote:
Quoted 1 line · Benjamin LaHaise
Quoted 1 line
Quoted 1 line
APIC makes perfect sense albeit rare. Single processor IO APICs are
very
Quoted 3 lines
Quoted 2 lines
Quoted 1 line
rare and are usually MP systems with only one processor.
I think most AMD Athlon boards have an IO APIC
I'd love to have it enabled in a distro kernel, but as Arjan pointed out,
it
Quoted 1 line
currently breaks some laptops if enabled.
Well, _not_ enabling IO-APIC on UP breaks my Athlon / KT333 at home; random freezes are the result, so I prefer to enable it...
So whichever default is chosen, someone is burned. I hate hardware.
--
Well nice weekend everyone and TIA
Quoted 4 lines · freaky
it in IDE mode so I tricked it. Attached the harddisks one by one so arrays were created of one disk each. This works under WinXP. Most of the partitions were created on my previous system with PIIX intel controllers and are read without probs under XP.
Thats still going to have strange raid blocks on it. However if you then partitioned the driver you should have blown it away with luck.
Quoted 2 lines
The RAID controller comes up with both hd[e-h] and d<x>p<y>'s. Can I use the hd[e-h]'s? since I have 1 disk arrays?
hde/f/g/h are the real disks as a normal controller would see them. The ataraid devices are interpreting a subset (the bits we know about) of the raid partitioning the promise does
Quoted 3 lines
Tried booting into my old partitions with the hdg3 but came up with IO errors. I'm guessing it has to do with the south bridge not being supported or the ext 2 partitions having trouble with being on the raid controller.
Need more details. Exact error messages matter here
Quoted 2 lines · freaky
Quoted 1 line
it in IDE mode so I tricked it. Attached the harddisks one by one so
arrays
Quoted 2 lines
Quoted 2 lines
were created of one disk each. This works under WinXP. Most of the partitions were created on my previous system with PIIX intel
controllers
Quoted 4 lines
Quoted 1 line
and are read without probs under XP.
Thats still going to have strange raid blocks on it. However if you then partitioned the driver you should have blown it away with luck.
No problems under XP so far...what kind of blocks would that be? Only had time-outs on one of the disks when it was only slave (no master present forgot to attach it :/) but since the master is attached that's gone.... all my old data plays fine, mainly mp3's, games and movies.
I'll supply all of the info later, low on time now and it's late. Want the kernel config and such as well? BIOS setup?
TIA and regards
Quoted 4 lines · freaky
No problems under XP so far...what kind of blocks would that be? Only had time-outs on one of the disks when it was only slave (no master present forgot to attach it :/) but since the master is attached that's gone.... all my old data plays fine, mainly mp3's, games and movies.
The HPT and Promise raid cards add extra partition table type data of their own identifying each volume. Their drivers then read and honour that info.
Quoted 2 lines
I'll supply all of the info later, low on time now and it's late. Want the kernel config and such as well? BIOS setup?
If it looks useful include it 8)
Quoted 2 lines · freaky
Quoted 1 line
No problems under XP so far...what kind of blocks would that be? Only
had
Quoted 2 lines
Quoted 2 lines
time-outs on one of the disks when it was only slave (no master present forgot to attach it :/) but since the master is attached that's gone....
all
Quoted 5 lines
Quoted 1 line
my old data plays fine, mainly mp3's, games and movies.
The HPT and Promise raid cards add extra partition table type data of their own identifying each volume. Their drivers then read and honour that info.
So that would be data on the MBR, or partition table? Perhaps win doesn't have probs because it can handle to partitions types properly. MSI told me themselves the trick would work (so I guess it might support to run as IDE controller but to save money they didn't include the jumper thingy.... I'll try contacting promise on this one, I tried that once before with the time-outs but no replies. :-().
Quoted 1 line
Quoted 1 line
I'll supply all of the info later, low on time now and it's late. Want
the
Quoted 3 lines
Quoted 1 line
kernel config and such as well? BIOS setup?
If it looks useful include it 8)
Sure. But since I can't boot, err ok it boots but no shell :-( it won't be all that much
Quoted 7 lines
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Quoted 2 lines · freaky
So that would be data on the MBR, or partition table? Perhaps win doesn't have probs because it can handle to partitions types properly. MSI told me
No its seperate. The hpt/promise raid "borrows" part of the disk and hides it.
There's no data loss that I know off... Shouldn't I atleast have notices something if data was overwritten? Shouldn't the partition tables be corrupt then? (Both disk manager and linux fdisk see them correctly). Don't get me wrong :-) I'm not saying you're wrong, I wouldn't dare :-), I just find it strange I haven't noticed anything and I've been running this setup for several weeks now. Is there anyway I could see this block so I can see it actually did create it? Perhaps it's written on space unlikely to be used?
Booted into the slack rescue disk this morning (busybox) and tried email the log but the telnet can't handle pipes or something (it didn't work :-() so I'll guess I'll have to write the whole thing down.
fdisk did recognize all the partitions as they were on the PIIX controllers the disks come from (on the hd[e-h] devices) I was unable to mount anything tho' mount /dev/hdh3 /mnt/hd gave me a file doesn't exist (mnt/hd dir exists) support for the required filesystems is compiled in.
Also, the /dev/d<x>p<y> devices didn't exist, so I probably have to create them myself perhaps then I can mount. I'll check the files for the major/minors.
The dmesg in busy box returned more data then the kernel itself spewed during boot. I've never seen that before? Is it writing extra debuging info nowadays?
----- Original Message -----
From: "Alan Cox" <[email protected]>
To: "freaky" <[email protected]>
Cc: "Linux Kernel Mailing List" <[email protected]>
Sent: Saturday, October 26, 2002 4:06 PM
Subject: Re: KT333, IO-APIC, Promise Fasttrak, Initrd
Quoted 2 lines · freaky
Quoted 1 line
So that would be data on the MBR, or partition table? Perhaps win
doesn't
Quoted 1 line
Quoted 1 line
have probs because it can handle to partitions types properly. MSI told
me
Quoted 10 lines
No its seperate. The hpt/promise raid "borrows" part of the disk and hides it.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/