On 2005-07-18 09:22 Sandy Drobic wrote:

Anders Norrbring wrote:

...

MegaRAID SCSI 320-1, BIOS: G119, F/W: 1L37.

I have the same Controller working fine out of the box with Suse 9.2.

...

I also tried SuSE 9.3 Pro in both 64/32-bit versions, with that I came to the selection screen, choose "Install" and the kernel loaded. After that the system stops with a blinking cursor in the upper left corner. Same problem with acpi=off.

With [Alt]-[F4] you should be able to see the error message up to that point. Have you tried to define an Array in the Controller Bios prior to installation?

Now I tried setting the HBA to emulate an I2O device, same problem.. I also looked closely to the driever loading reports, it says it cannot get adapter status.. Anders.

On 2005-07-18 10:45 Sandy Drobic wrote:

Anders Norrbring wrote:

...

...

...

MegaRAID SCSI 320-1, BIOS: G119, F/W: 1L37.

With [Alt]-[F4] you should be able to see the error message up to that point. Have you tried to define an Array in the Controller Bios prior to installation?

Now I tried setting the HBA to emulate an I2O device, same problem.. I also looked closely to the driever loading reports, it says it cannot get adapter status..

Could be a compatibility problem with the mainboard. Though it is strange that you can define Arrays but can't see the controller during installation.

Have you tried to integrate the Controller into an existing running installation, just in order to have all commands available to investigate the module loading? If you can't get it running with an installed system chances are you will not be able at all to get the hardware combination working.

Well, the installtion detects the HBA correctly, lists it at the correct PCI slot etc. When the module is loaded, it tried to get adapter status, and fails doing so. The exact error displayed is: aborting=2 cmd=2 . I have tried the HBA in 3 different systems, same shit on all of them, except for the Arima HDAMA board where the installation locks up before even going to the first selection screen where I choose Installation, Rescue, Boot installed system, Memtest, etc. I'll try to plug it into a Windows 2k server system, just to see what happens there... I've sent a report to LSI-Logic as well, maybe it's a RMA case. Anders.

On 2005-07-18 11:07 Sandy Drobic wrote:

Anders Norrbring wrote:

...

Well, the installtion detects the HBA correctly, lists it at the correct PCI slot etc. When the module is loaded, it tried to get adapter status, and fails doing so. The exact error displayed is: aborting=2 cmd=2 .

I have tried the HBA in 3 different systems, same shit on all of them, except for the Arima HDAMA board where the installation locks up before even going to the first selection screen where I choose Installation, Rescue, Boot installed system, Memtest, etc.

I'll try to plug it into a Windows 2k server system, just to see what happens there...

I've sent a report to LSI-Logic as well, maybe it's a RMA case.

Just to be thorough, have you tried to exchange the SCSI cable and the terminator? Defective terminators often have strange effects.

Oh yeah.. It doesn't even load without anything connected! Anders.

On 2005-07-18 18:03 Brad Dameron wrote:

On Mon, 2005-07-18 at 11:15 +0200, Anders Norrbring wrote:

...

On 2005-07-18 11:07 Sandy Drobic wrote:

...

Anders Norrbring wrote:

...

Well, the installtion detects the HBA correctly, lists it at the correct PCI slot etc. When the module is loaded, it tried to get adapter status, and fails doing so. The exact error displayed is: aborting=2 cmd=2 .

I have tried the HBA in 3 different systems, same shit on all of them, except for the Arima HDAMA board where the installation locks up before even going to the first selection screen where I choose Installation, Rescue, Boot installed system, Memtest, etc.

I'll try to plug it into a Windows 2k server system, just to see what happens there...

I've sent a report to LSI-Logic as well, maybe it's a RMA case.

Just to be thorough, have you tried to exchange the SCSI cable and the terminator? Defective terminators often have strange effects.

Oh yeah.. It doesn't even load without anything connected!

Anders.

Try booting with mem=2000M. There are known issues with installing certain versions of SuSe and these controllers. Once you get it installed update your kernel and edit your /boot/grum/menu.lst file and remove the mem statement. Should work fine. If this does not fix it you could very possibly have a bad controller. LSI is very good with warranties.

Nope, it's not that either.. The controller now works fine in both tested 64-bit PCI systems. However, it still doesn't answer to the "get status" command when seated in a 32-bit PCI slot. LSI suggests it's a driver problem, but as far as I can see, the driver that comes on the 32-bit SLES distribution disks should be supporting a 32-bit bus. I'm not sure though. Anders.

On 2005-07-18 19:13 L. Mark Stone wrote:

On Monday, July 18, 2005 12:39 pm, Anders Norrbring wrote:

...

However, it still doesn't answer to the "get status" command when seated in a 32-bit PCI slot.

What's the voltage in that slot? 3.3V or 5+? The controller may have a preference...

It's a 3.3 volts 32-bit slot, just as the HBA likes it.. However, the problem seems to be the Linux driver for the MegaRAID. Apparently it isn't aware of 32-bit slots, so the driver ASSUMES that the HBA is placed in a 64-bit slot. This is according to LSI support. The controller behaves just fine under Windows on the same hardware platform, and that driver is aware of the PCI slot differencies. I guess it would be nice with a patch/mod of the driver... Anders.

On 2005-07-19 11:48 Sandy Drobic wrote:

Anders Norrbring wrote:

...

It's a 3.3 volts 32-bit slot, just as the HBA likes it.. However, the problem seems to be the Linux driver for the MegaRAID. Apparently it isn't aware of 32-bit slots, so the driver ASSUMES that the HBA is placed in a 64-bit slot. This is according to LSI support. The controller behaves just fine under Windows on the same hardware platform, and that driver is aware of the PCI slot differencies.

I guess it would be nice with a patch/mod of the driver...

Is this an issue specific to Suse 9.3 or did it already occur in Suse 9.2 as well? I have 64-Bit Slots on my server mainboard and had no problems at all with Suse 9.2.

Well Sandy, then you're on the "good side" of it all... The driver is not aware of the 32-bit slots, so when you put the controller into a 32-bit, it fails since it presumes the controller is in a 64-bit slot. When you actually have it in a 64-bit slot, everything runs just fine. Anders.