2.2.18p18 eepro100 issues (packets per irq, shared irqs)

2.2.18p18 eepro100 issues (packets per irq, shared irqs)

[KJ Pickett]

First of all I would like to say that 2.2.18p18 does not have the 'card 
reports no resources' messages when I saturate the 100mbit network anymore.

But I was wondering about two things:
I have a 815EE board, which has an onboard eepro100 and a pci eepro100.  
They share an irq, no matter if I'm using intels e100.o driver or the stock 
linux one.  For performance reasons, can I make them each have a different 
irq?  Doing it from ifconfig gives me a notsupported error, with either 
driver.

Also: VMSTAT output, comparing stock linux driver with intel.  Heres the 
stock linux:

1  0  0      0  31836  67632  13240   0   0     0     0 24015 47215  20  15  
65
1  0  0      0  31836  67632  13240   0   0     0     1 24016 47016  21  16  
64

The card is recv'ng 24k packets per second, and seeing that many irqs.  The 
intel driver does 6 per irq, and vmstat from that driver (not posted, but I 
remember what it was) shows 6 times less irq/sec and cs/sec.  Even though, 
it appears the stock linux driver is supposed to do 20 things per irq?  I am 
no kernel hacker...can I get the stock linux driver to do multiple packets 
per irq with some config settings?

Thanks,

Karl Pickett
(Posting from hotmail since I cant use my companies email address)
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: 2.2.18p18 eepro100 issues (packets per irq, shared irqs)

[Alan Cox]

> They share an irq, no matter if I'm using intels e100.o driver or the stock 
> linux one.  For performance reasons, can I make them each have a different 
> irq?  Doing it from ifconfig gives me a notsupported error, with either 
> driver.

Under 2.2 no. Under 2.4 maybe

> no kernel hacker...can I get the stock linux driver to do multiple packets 
> per irq with some config settings?

We dont have sufficient docs to figure this out easily. Intel upload magic
numbers to the card microcode
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: 2.2.18p18 eepro100 issues (packets per irq, shared irqs)

[octave klaba]

> > They share an irq, no matter if I'm using intels e100.o driver or the stock
> > linux one.  For performance reasons, can I make them each have a different
> > irq?  Doing it from ifconfig gives me a notsupported error, with either
> > driver.
> 
> Under 2.2 no. Under 2.4 maybe

one question: using driver 1.11 of eepro, on a server with 1 
network carte only, we have something really dirty in dmesg
it writes IRQ 20 !? why all this folks ? ;) 
the good news there is no more eth0 problems :)

Octave

IRQ to pin mappings:
IRQ0 -> 2
IRQ1 -> 1
IRQ3 -> 3
IRQ4 -> 4
IRQ5 -> 5
IRQ6 -> 6
IRQ7 -> 7
IRQ8 -> 8
IRQ9 -> 9
IRQ14 -> 14
IRQ16 -> 0
IRQ20 -> 4
IRQ24 -> 8
IRQ25 -> 9


eth0: OEM i82557/i82558 10/100 Ethernet at 0xf0002000, 00:E0:18:02:15:02, IRQ 20.
  Board assembly 668081-002, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.
  General self-test: passed.
  Serial sub-system self-test: passed.
  Internal registers self-test: passed.
  ROM checksum self-test: passed (0x04f4518b).
  Receiver lock-up workaround activated.
eepro100.c:v1.11 7/19/2000 Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/eepro100.html
eth1: OEM i82557/i82558 10/100 Ethernet at 0xf0004000, 00:E0:18:02:15:02, IRQ 20.
[...]
eth7: OEM i82557/i82558 10/100 Ethernet at 0xf0010000, 00:E0:18:02:15:02, IRQ 20.
  Board assembly 668081-002, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.
  General self-test: passed.
  Serial sub-system self-test: passed.
  Internal registers self-test: passed.
  ROM checksum self-test: passed (0x04f4518b).
  Receiver lock-up workaround activated.
eepro100.c:v1.11 7/19/2000 Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/eepro100.html
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/