Mouse driver dies -- Solaris x86 2.5.1

Mouse driver dies -- Solaris x86 2.5.1

Post by Andrew Gabri » Wed, 09 Jul 1997 04:00:00





Quote:>This is driving me crazy: Solaris x86 2.5 patched to the hilt (all
>recommended patches, anyway) -- the mouse driver dies after just a few
>minutes in OpenWindows or CDE, leaving me with ... a terminal if I'm
>lucky.

>Any suggestions? Thanks in advance!

Well, it would probably help to know what sort of interface the
mouse uses.

One possibility is a mouse on COM1 and something on COM3 trying
to use the same IRQ (with hardware which doesn't support IRQ
sharing).

--

Consultant Software Engineer

 
 
 

Mouse driver dies -- Solaris x86 2.5.1

Post by Andrew Gabri » Sat, 12 Jul 1997 04:00:00




Quote:

>The only clue I have was this console message (which only
>happened once in 28 iterations):

>    NOTICE: asy0: silo overflow

>This makes me think that the dumb serial ports on this board
>are too slow to handle things when the CPU is busy with other

Yes.

Quote:>processing (like creating a new window).

No, it is interrupt level processing which is too slow.

Quote:

>Anybody have any other ideas (other than "just get a real Sun",
>which believe me is on my short list)? Can you recommend a good

Actually, this is one area where PCs (with their buffered 16550s)
are usually better than Sparcs (with their Zilog async/sync i/o
chip). My old 486 33MHz had no problem running the serial port
at 57.6k with Solaris x86, with no silo overflows (or ring buffer
overflows) ever.

Quote:>motherboard (Pentium 133 or so) with onboard serial ports that
>works with Solaris x86?

Maybe you don't have a 16550, but a 16450 or older (unbuffered)
UART, or you are using DU7 (which forgot to turn the buffer on).
I think all current motherboards now come with 16550's and have
done so for at least a year or longer now.

Another possibility is you have another device driver installed
which disables interrupts too long.

--

Consultant Software Engineer

 
 
 

1. X86 v2.6 & EtherExpress Pro/10+ die die

I have been banging my head against the wall on this one. Here are the
details. It's Sun's fault :-)

Solaris X86 v2.6 5/98 with DU3
Asus TX97-L Mainboard, with 64MB of RAM
ATI PCI Video, ASUS SC-200 PCI SCSI, Intel EtherExpress Pro/10+ PCI (82557
- network controller chip)

Solaris will see the card, but it thinks it's a Pro/100B, therefore it
assigns it the iprb driver. The problem is this card/computer can't been
seen on the network, or see the network.

To make sure it wasn't a hardware/configuration issue I ran PicoBSD
(FreeBSD on a floppy). from PicoBSD I was able to ping my Win95 box, and
Win95 was able to ping PicoBSD.

All the references on SunSolve, or anything from Sun say the
82557 is  a Pro/100B, but in the case it's a Pro/10+. I when to the Intel
web site and was able to ID the card as a Pro/10+ PCI (it support
full-duplex, I think thats the plus feature)

So it's a driver issue with Solaris v2.6 - anyone have any ideas, tips, or
anything that can help ? Is Sun assuming that all 82557s are Pro/100B.
Also dmesg | grep ether shows

PCI [blah blah] ether (unknown) [mac address]

                       ^------- Should this not be 10T or 10Mbps ???

Please Help
Thank You
 --

BTW: Yes I made sure that each PCI device had a unique IRQ.

2. dwarf stars histogram again

3. Solaris 2.4 x86, Mach 64 Video Driver, and Token Ring driver conflict??

4. Configuring system as a non networked standalone.?

5. asy driver in Driver update 9 for Solaris x86 2.4 broken?

6. PCNFS problems

7. in.named dies on Solaris 2.1 x86

8. pkgadd

9. Does a SPARC Solaris driver run on x86 Solaris?

10. Solaris x86 v7 NE2000 driver in Solaris 2.5.1

11. Solaris 8 X patches (was: New Solaris/x86 driver download page)

12. Mouse Systems PC Mouse 3D/6D Bus Mouse driver?

13. Evaluation pointers: x86-Solaris vs x86-NT vs SPARC-Solaris