Text 474, 200 rader
Skriven 2007-02-10 16:26:32 av Alan Hess
     Kommentar till en text av Mike Luther
Ärende: pci.exe output
======================
Whilst masticating on <Feb 09 07>, Mike Luther (1:117/3001)
wrote to Alan Hess:

Howdy, Mike,

This discussion is also going on in the forums at www.os2world.com - I don't
know if you visit that site, but if you don't, it's a good one - as
comp.os.os2.misc, and I know you read the newsgroups.  OS/2 users remain very
helpful to nimrods like me, and it's appreciated.

ML> Does your motherboard BIOS support hard assignment of the IRQ's 
ML> to a given PCI slot as part of its capabilities?  If it does then 

I don't know, to be honest.  With my limited knowledge, I try to play with bios
as little as possible.  I was able to disable video irq, which was a suggestion
I got - next time my friend is here and tries the Radeon again, perhaps that
will solve the problem.  There's also a setting to disable USB irq, and I think
a setting to let the OS control USB.  I currently have no USB hardware, and I'd
probably get a USB 2.0 card if I get a USB device (I'm thinking of getting an
external USB hard drive to use for backups, perhaps disk images from DFSEE - I
have a Quantum tape drive, and use Fastback to back up to it, but Fastback is
limited to 2 gig partitions.  I'm not knocking Fastback, as it's crash recovery
disks have saved my butt several times, most recently after the Ubuntu install
disaster that killed everything on my master drive.)

ML> you may be able to force a different IRQ for an item in a given 
ML> PCI slot on your motherboard.  Ordinarily, if memory is correct, 
ML> your mouse ought to be normally looking at IRQ12.  Thus the 
ML> devices sharing IRQ 11 ought not to be a conflict issue there.

It's a PS/2 mouse.  IIRC, you're correct that it's IRQ12.

ML> In line with this, your below looks like an SBLive 5.1 PCI audio 
ML> card.

AH>>  Bus 0 (PCI), Device Number 9, Device Function 0
AH>>  Vendor 1102h Creative Labs
AH>>  Device 0002h EMU10K1 Audio Chipset (SB Live!)

AH>>  System IRQ 11, INT# A

ML> If so, then either of two possible 'solutions' for IRQ sharing on 
ML> it might help you.  First is the above.  I try to work with the 
ML> SBLive 5.1 card as an IRQ10 operation if possible.  If you cannot 
ML> 'force' assign that to this card in the slot it is in, you may be 
ML> able to move the card to another slot.  Usually, moving a PCI 
ML> card to another slot tends to coax a motherboard to grant it a 
ML> different IRQ.  If you can one way or another do this, you may 
ML> help get a possible conflict for the audio card driver out of 
ML> harm's way.  The audio card IRQ conflict deal can really hurt 
ML> things, especially if it is in conflict with a network card.

You're correct, it's an SBLive card.  With the Matrox video card in, I have no
problems that I can see - the trackball works, and Airboot happily appears. 
With the Radeon in, Airboot reports a drive problem, and halts the system.  If
I uninstall Airboot (boot from a CD, go to the Airboot directory on the hard
drive, and do the uninstall), then boot without Airboot, the boot proceeds
(though the eCS boot blob whizzes by - there's no way I'd be able to do, say,
an alt-F2), but the mouse pointer doesn't move.  eCS is much more difficult to
navigate without a mouse!  The SNAP driver recognizes the new card, and the
desktop is 1024x768, as it is with the Matrox card.

ML> The same thinking goes for the SCSI card .. if it is a card.  If 
ML> it is an on-board SCSI controller and this is the boot drive you 
ML> might note.

It is a card.

AH>>  Bus 0 (PCI), Device Number 11, Device Function 0
AH>>  Vendor 1022h Advanced Micro Devices (AMD)
AH>>  Device 2020h AM53C974 SCSI Bridge (AKA Qlogic 
AH>> Fast!SCSI Basic, Tekram DC-390)
AH>>  Command 0087h (I/O Access, Memory Access, BusMaster, Wait 
AH>> Cycles) Status 0200h (Medium Timing)
AH>>  Revision 10h, Header Type 00h, Bus Latency Timer 20h
AH>>  Minimum Bus Grant 04h, Maximum Bus Latency 28h
AH>>  Self test 00h (Self test not supported)
AH>>  PCI Class Storage, type SCSI
AH>>  Address 0 is an I/O Port : D800h..D87Fh
AH>>  System IRQ 11, INT# A
AH>>  Expansion ROM of 64 KiB decoded by this card, currently 
AH>> disabled

ML> I know nothing about the Tekram product.  But for the Adaptec PCI 
ML> SCSI cards, the raw on-board Adaptec BIOS when accessed by their 
ML> <CTRL A> at boot time card setup force effort can be specifically 
ML> set to a different IRQ than 'normal' if needed.  Getting this 
ML> card off IRQ11 by the first technique, or swapping it in a 
ML> different PCI slot, or .. by forcing the IRQ to other than IRQ11, 
ML> but not IRQ10 if that is what you want on the SBLive audion card 
ML> will usually help get rid of this error vector.

The Tekram card also seems to be accessible that way - I forget the key combo,
but it appears at every bootup.

ML> The NIC IRQ seems out of the way at this point...

AH>>  Bus 0 (PCI), Device Number 13, Device Function 0
AH>>  Vendor 8086h Intel Corporation
AH>>  Device 1229h 82559 Fast Ethernet Multifunction 
AH>> PCI/CardBus Controller (C-step)

AH>>  System IRQ 5, INT# A

ML> The IDE IRQ's seem normal at this point.,,

AH>>  Bus 0 (PCI), Device Number 17, Device Function 1
AH>>  Vendor 1106h VIA Technologies Inc
AH>>  Device 0571h VT82xxxx EIDE Controller (All VIA Chipsets)
AH>>    Primary   Channel is at I/O Port 01F0h and IRQ 14
AH>>    Secondary Channel is at I/O Port 0170h and IRQ 15

AH>>  Subsystem ID 05711106h VT82Cxxxx EIDE Controller (All
AH>> VIA Chipsets) (Generic ID)

AH>>  System IRQ (disabled), INT# A

ML> The USB controllers may be at issue in that they seem to be an 
ML> on-board chipset.  If so, doing anything to it is going to likely 
ML> by an issue ...

AH>>  Bus 0 (PCI), Device Number 17, Device Function 2
AH>>  Vendor 1106h VIA Technologies Inc
AH>>  Device 3038h VT8233A USB Controller

AH>>  System IRQ 11, INT# D

AH>>  Bus 0 (PCI), Device Number 17, Device Function 3
AH>>  Vendor 1106h VIA Technologies Inc
AH>>  Device 3038h VT8233A USB Controller

AH>>  System IRQ 11, INT# D

As I said, the USB IRQ can be disableed.

ML> And your PCI slot Matox video card is already on IRQ10.  So 
ML> obviously you don't want it on the same IRQ as your SBLive.  Thus 
ML> ..

AH>>  Bus 1 (AGP), Device Number 0, Device Function 0
AH>>  Vendor 102Bh Matrox Graphics Inc
AH>>  Device 0525h MGA-G400/450 Chipset

AH>>  System IRQ 10, INT# A

There should be no video IRQ now, if I run pci.exe again.  Hold on...  Yup,
video IRQ now reported as disabled.

ML> Looking at life in the fast lane, your choices for things boil 
ML> down to the following.

ML> IRQ 0  - Non Mask Interrup (reserved).
ML> IRQ 1  - Keyboard.
ML> IRQ 2  - Bootstrapped to IRQ 9 thus paired with it - whatever.
ML> IRQ 3  - COM2 normally - If not in use can be used otherwise.
ML> IRQ 4  - COM2 normally - If not in use can be used otherwise.
ML> IRQ 5  - In use for the NIC which ought to be fine.
ML> IRQ 6  - Floppy controller.
ML> IRQ 7  - LPT1 printer use normally.
ML> IRQ 8  - Real Time Clock (reserved).
ML> IRQ 9  - Looks available in your case .. I usually run SCSI here,
ML> IRQ 10 - In use for the VIA video chipset .. usually OK.
ML> IRQ 11 - *CAN* be shared for USB if device driver written 
ML> properly. IRQ 12 - Mouse (reserved - normally unless serial 
ML> mouse). IRQ 13 - Numeric Co-processor (reserved).
ML> IRQ 14 - Normally for IDE controller #1 as you report.
ML> IRQ 15 - Normally for IDE controller #2 as you report.

ML> If you are not using the second IDE controller and can free that 
ML> IRQ 15 up with your BIOS setup, that will make one IRQ free for 
ML> something else. We see that IRQ then available at times for the 
ML> SCSI controller, or for the video card.  Which would free up the 
ML> IRQ 10 you are using for the Matrox video card for other use.  
ML> That is what I see used at times for the SBLive 5.1 card on many 
ML> boxes.  Which, with the SCSI controller on IRQ 9 makes that all 
ML> go away from the IRQ 11 gambit.

I'm only using the first IDE controller.  With the video IRQ disabled, is it
necessary to change the SBLive IRQ?

ML> And .. if you are using a SERIAL mouse, SOMETIMES the OS/2 
ML> operating system cannot 'find' it unless you specifically comment 
ML> the MOUSE driver in CONFIG.SYS to tell it to look for a serial 
ML> mouse.  With some mice but not others, as I have seen.  Forcing 
ML> that with some mice on a serial port can stop this in some 
ML> systems.  But usually you wind up able to boot to the Desktop but 
ML> just get no mouse working at all.

I'm not using any serial ports right now, save for virtual ones (Vmodem - one
to telnet out, and one for the bbs.)  The serial ports are still enabled in
bios, and in the sio.sys line in config.sys.  I still have one USR modem
connected, but it's turned off, as I haven't needed it (I access the web via
cable modem.)  I could disable Com2 in the bios and config.sys, if necessary.

Thanks for the ideas.  *adh*

-->> Sleep well; OS/2's still awake! ;)

--- Msged/2 6.0.1
 * Origin: tncbbs.no-ip.com - Try the CROSSFIRE echo - all welcome (1:261/1000)