linux install ends at "partition check"

Hello, I am new on this site and have not been able to read all threads relied to my problem, so be indulgent with me please!

I tried to install linux (REdHat 7.3) on a new hard drive these past days, and got a problem since the installation stops at the "partition check" level : it does not tells me about any error. It just stops, and doesn't go further!!!

For purpose, here is my hardware : I've just got a new motherboard (but still very old P1) : it is based on a "ALI Aladdin IV+" chipset (pentium 1). I could let it work with a K6-2/550AGR installed at 6x75Mhz, that is 450Mhz, with 128MB of RAM, and a 10GB Generic hd (that appeared to be bad these days since some bad clusters have been found with scandisk tool). So I bought a new hd Seagate 120GB, UDMA100, with 8Mo of cache, and tried to install linux on it.

My bios (amibios file) is very old and doesn't support >32GB hard drives. So the only way for me is, either to use the 32GB clip jumper, or to set the "not installed" parameter in bios at the location Primary/Secondary Master/Slave Drive.

Here is what I have experimented : boot with "bootable cdrom" as well as "floppy disk" (in order to switch through CDRom), and always got the same problem stop at "partition check"!! I tried master and slave positions, "set in bios" with 32GB clip jumper, nothing goes on. I had some windows partitions on this hard rive. I erased all of them, but the problem holds on!!

Please!! Help me!!!!

Bye, Xround

PS : I am not sure but I think I tried the hd on another older chipset and I could obtain the recognization of my both drives!! May that trick could confirm a chipset problem???

The kernel can't get a partition table reading off of the BIOS. As far as I can tell there's no real way around that using the BIOS you have. If the BIOS has a flash for greater then 32Gb drives, by all means, flash the BIOS. If not, there's always cheap ATA100/133 cards out there.... which are probably a good idea anyway as the board you have must have IDE controllers limited to ATA33, so you can triple your data throughput cheap.

The BIOS on the cards will report the drive geometry onto the kernel correctly and the card can usually handle 250Gb+

Watch out for what card you buy if you want to stick with rh 7.3, which is rather old, make certain not to get a Highpoint newer then a 370 or a Promise card newer then the 20269.

Cheers,

Finegan

Well, thanks a lot finegan, for your advise.

As I think I have the latest bios for my old M575 (supporting indeed only dma33 mode), I can't get an upgrade form now on. Under windows, I have tried a seagate utility that let me see my hd at its full size.
I do wonder if linux could speak with this utility. Will try that possibility, and if not I will then try to buy a card you suggested. I am not sure so can you confirm ? Do I have to purchase for Older models than the Highpoint 370 or Promise 20269 ??

Bye, X.

The seagate utility is 100% windows, Maxtor had one too to trick older bioses. Under Linux there's very little that can be done to get around an old buggy bios then to buy your way around the problem, I've had to do it in the past... heck, I've got 3 ATA100 cards. But honestly, the speed bump is worth it, its amazing how an old p2 233 with 128Mb of ram will zing along when the harddrive pull rate is not the data bottleneck. It also makes swapping somewhat bearable.

RH 7.3 released with kernel 2.4.18 that only has support for the Highpoint series up to the 370 and the Promise's up to the 20269. If you wanted to go with RH 9, you could use just about any ATA card you could find, kernel 2.4.20 added support for a buttload of ATA controllers. Offhand, avoid the Promise TX2 card and aside from that you should be fine. Honestly, the cheapest ATA100 card on the market will probably be more then enough for the job and well old enough for the kernel to be happy with.

Cheers,

Finegan

finigan

I have a Maxtor IDE controller with a Promise 20267 chipset. This chipset will not work when two disks are attached to the two IDE ports and try to run in parallel. Twelve letters and emails to Maxtor produced a magnificent run around. I finally gave up on it when it wouldn't work with kernel 2.4.22-ac4.

I have been searching for a replacement, paying meticulous attention to the specs of the prospective PCI IDE controllers. My current favorite is the HighPoint Rocket 100 2-channel ATA/100 IDE PCI Host Controller Adapter Card. I will have to order this by mail order which makes returning it more difficult if it doesn't work. Apparantly I have to install a kernel driver to make the HighPoint card work. I downloaded the HighPoint zip file twice (SuSE 8.1 version). The first time the download file was 1.6M. The second time the file was 5.4M. Neither download would unzip. I downloaded the tar.gz and unpacked that. Compiling the driver seems to be fairly straightforward so I think that I can order the card with confidence that I can get it to work.

"Honestly, the cheapest ATA100 card on the market will probably be more then enough for the job and well old enough for the kernel to be happy with."

I don't have the problem of an old kernel. Are there any ATA 100 cards that I can buy that will work correctly with two drives attached (no RAID) which I can use with the IDE support as shipped by kernel.org? Actually the chipset name would be a better answer. I have found boxes of the same brand IDE controller card side by side on the shelf with different chipsets.


---------------------------
Steve Stites

The kernel.org kernels have drivers that usually only support the cards as straight ata devices. I think the rocketraid is just another hpt 37x chip, covered by 2.4.21+ and thereabouts. The odd thing is what you say about the 20267 card, I've got one of those, but I've never set up two drives on it master/slave. Its part of a software RAID 5, never hiccups... makes me wonder if something else was wrong... a lot of those Promise Bioses were buggy... same with early highpoint cards, like the hpt370 and hpt370a.

Cheers,

Finegan

"The kernel.org kernels have drivers that usually only support the cards as straight ata devices. I think the rocketraid is just another hpt 37x chip, covered by 2.4.21+ and
thereabouts. "

Thanks. I found the support for HPT37x in my 2.4.22-ac4 xconfig. That is what I will use (I am not using RAID).

"The odd thing is what you say about the 20267 card, I've got one of those, but I've never set up two drives on it master/slave."

I haven't tried the master/slave combination either. I am setting up a disk configuration for speed where I can have 4 devices running in parallel. With a master/slave pair, only one can do I/O at a time. So I want to set up three disks and a CD-RW, each attached by its own cable to its own controller port. I want the following configuration:
SiS IDE0 /dev/hda - Maxtor disk
SiS IDE1 /dev/hdd - CD-RW
Maxtor IDE2 /dev/hde - Maxtor disk
Maxtor IDE3 /dev/hdg - Seagate disk

I cannot get /dev/hde and /dev/hdg to run in parallel. Whenever I try Linux crashes and the disk file systems and/or partition table are kaput. Somtimes fsck can fix it, sometimes not.

Some relevent diagnostics are:
I set up two Maxtor disks on /dev/hde and /dev/hdg. I run the stand alone Maxtor diagnostics against each disk in turn and there are no errors. I cannot run the diagnostics simultaneously against both disks, which I reported to Maxtor as a needed improvement in their diagnostics.

I can then boot Linux and set up the two disks with fdisk and mkfs. Of course these programs only access one disk at a time. When I run massive I/O to both /dev/hde and /dev/hdg (such as a backup from one disk to the other) I crash within 15 seconds max.

Maxtor denies that there is any problem with their card. I think that what they have done is set up the Windows driver so that it serializes access to the two IDE ports. Maxtor would not admit this because it means that their card will only run at half of the potential maximum rated speed.

This need for serialization could be caused by either the Promise 20267 chip having a serial bottleneck or by the Maxtor card trying to have one printed circuit serve both IDE ports. I held the Maxtor card up to the light and examined the circuitry for about 10 minutes with my admittedly poorly calibrated eyeballs. I came to the conclusion that the printed circuit was entirely duplicated and the problem was in the Promise 20267 chipset.

I searched the kernel development boards and found several developers grumbling that the Promise 2026x was buggy and that Promise refused to release the specs. The developers were trying to create work arounds for the Promise bugs by trial and error. I started with kernel 2.4.18 and over time worked my way through kernel 2.4.22-ac4 and there was no successful workaround.

"The odd thing is what you say about the 20267 card, I've got one of those, but I've never set up two drives on it master/slave. Its part of a software RAID 5, never hiccups... makes me wonder if something else was wrong."

It is possible that my conclusion about the Promise 20267 being buggy is wrong and the problem is really in the Maxtor printed circuit or even the Maxtor BIOS. But if the problem was the BIOS code then I would have expected Maxtor to release a BIOS upgrade.

Do you have a Promise 20267 chipset on your farm that will do I/O simultaneously to both ports?

"a lot of those Promise Bioses were buggy... same with early highpoint cards, like the hpt370 and hpt370a."

If I buy a HPT370 am I going to hit the same problem?


---------------------
Steve Stites

No, you seem to have delved into this a lot deeper then I have. I have an HPT370 onboard my old abit kt7a board, athy 1.2, I get the full throughput on writes from the dvd on ide3 to the 40gig on ide2, 45MB/sec thereabouts which makes sense for the external write speed of the HD and the read speed of the admitedly crap dvd. I've never poked around the lkml to find out which of these drivers were coded peak and poke, I guess it makes sense now as to why it took so long for an entire series of promise controllers to make it into 2.4.x

I've got a trio of 5400rpm ata100 maxtors spread accross a PDC20267 and a 20265:

PDC20267: chipset revision 2
PDC20267: not 100% native mode: will probe irqs later
PDC20267: (U)DMA Burst Bit ENABLED Primary PCI Mode Secondary PCI Mode.
    ide2: BM-DMA at 0x8400-0x8407, BIOS settings: hde:DMA, hdf:DMA
    ide3: BM-DMA at 0x8408-0x840f, BIOS settings: hdg:DMA, hdh:pio
PDC20265: IDE controller on PCI bus 00 dev 88
PCI: Found IRQ 10 for device 00:11.0
PCI: Sharing IRQ 10 with 00:0b.0
PDC20265: chipset revision 2
PDC20265: not 100% native mode: will probe irqs later
PDC20265: (U)DMA Burst Bit ENABLED Primary PCI Mode Secondary PCI Mode.
    ide4: BM-DMA at 0x6800-0x6807, BIOS settings: hdi:DMA, hdj:DMA
    ide5: BM-DMA at 0x6808-0x680f, BIOS settings: hdk:DMA, hdl:pio
hdb: HITACHI CDR-8130, ATAPI CD/DVD-ROM drive
hde: Maxtor 4W080H6, ATA DISK drive
hdg: Maxtor 4R080L3, ATA DISK drive
hdi: Maxtor 4R080L2, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide2 at 0x9800-0x9807,0x9402 on irq 9
ide3 at 0x9000-0x9007,0x8802 on irq 9
ide4 at 0x8000-0x8007,0x7802 on irq 10

2.4.20 stock, xfs patch. I get more or less full speeds out of the drives, the hdparm only comes back with about 20MB/sec, but it is afterall software RAID5 over 3 crap drives.

I'm pretty certain the 20267 must be doing simultaneous I/O or I would end up with huge bottlenecks right? I guess it could be possible that I just don't demand enough from this crate to notice.

Cheers,

Finegan

"I'm pretty certain the 20267 must be doing simultaneous I/O or I would end up with huge bottlenecks right?"

This configuration (with 7200 rpm drives) consistently crashes on my system. Therefore my problem must be the Maxtor card circuitry, not the Promise 20267 chipset.

"I have an HPT370 onboard my old abit kt7a board, athy 1.2, I get the full throughput on writes from the dvd on ide3 to the 40gig on ide2"

Then there is a reasonable hope that a HighPoint PCI IDE card with a HPT370 chipset should work on my computer.

Thank you for doing all the work to gather up all this information for me.

-----------------------
Steve Stites

Here, I'm back again.

I don't understand all what you both speak about, but I think now form what I have understood that I can purchase a HPT370 provided I use a RH7.3 or greater with confidence. Am I wright ??

Thank you for your informations, Xround.

" I think now form what I have understood that I can purchase a HPT370 provided I use a RH7.3 or greater with confidence. Am I wright ??"

I think so. I have a HPT370 in the mail. I have already compiled a kernel 2.4.22 with HPT370 support. When I install it I'll post back how it works.

___________________________________
Be prepared. Create a LifeBoat CD.
http://users.rcn.com/srstites/LifeBo...home.page.html

Steve Stites

I had a similar problem on the machine I am posting this with when I tried a Libranet install. I solved the issue by partitioning the drive so that it had one ~ 30GB partition and a small swap partition, all within the first 32GB of the drive before I started the install. I told the install to use these partitions and it was able to successfully ignore the part of the drive the BIOS could not deal with.

Back again, I ve juste read that problems

Thanks for the advise Idaho, it is true I didn't wait much time (less than 5 minutes). I will try your trick. Also will try to configure an ext3 or 2 partition (don't remember now) and then will at the end try to purchase a card since what I read meens that i must take great care. Here is the reason of my fear!


finegan wrote :
"a lot of those Promise Bioses were buggy... same with early highpoint cards, like the hpt370 and hpt370a."

I have just read on a french site that the hpt370 was not much compatible with the seagate 40GB barracuda 7200rpm (same seires as mine escept mine is a 120GB), in the sense that it was very long to boot under windows, and also that it produced strange noise before booting under windows!
I think (if I understood) that some hard disks are not compatible with pci ide controler cards. Could it be this problem you (jailbait) had with your promise pci card ??
Are their compatibility tests between hard disk drives and existing pci ide controlers ??

Thanks to you all, Xround.

"I think (if I understood) that some hard disks are not compatible with pci ide controler cards. Could it be this problem you (jailbait) had with your promise pci card ??"

I was running a Maxtor IDE PCI card with a Promise 20267 IDE chipset. At first I had 2 Maxtor hard disks installed. While I am waiting for my HPT370 to arrive I am running with a single Maxtor drive on the Maxtor card. The Maxtor card runs fine with only one disk attached. It fails to work when I have two disks attached and I try to do a large copy from one disk to the other.

So incompatibility between the Maxtor card and the two Maxtor disks was not the problem. The problem was a design flaw that the Maxtor card would not run two disks in parallel.

___________________________________
Be prepared. Create a LifeBoat CD.
http://users.rcn.com/srstites/LifeBo...home.page.html

Steve Stites

First, sorry for my previous post that I began with a non finished sentence!

Well, I've just briefly read the specs of the promise Ultra100 TX2 (for about the quivalent of 60$), that seems to be compatible with the RH7.0, that should from thus be compatible with newer RG versions. Does anyone have comment or experience about this card ?
I do not absolutely want to install thez RH7.3 provided it doesn't weight the system. In this way, I can try to manage with a newer version, it is not a barrier for me. I prefer changing the RH version than a card that wouldn't be compatible with my present hardware!

Xround.