konflikt pri sdileni IRQ vice zarizenimi

[Petr Bezděk]

Zdravim a omlouvam se za delsi mail,

mam nasledujici problem - v serveru bezicim na 5.3-STABLE mam 3 diskove 
radice a 4 sitove karty:

Nov 21 21:05:07 ares kernel: atapci0: <Promise PDC20378 SATA150 
controller> port 0x9800-0x987f,0x9400-0x940f,0x9000-0x903f mem 
0xe9200000-0xe921ffff,0xe92a3000-0xe92a3fff irq 10 at device 4.0 on pci0
Nov 21 21:05:07 ares kernel: bge0: <Broadcom BCM5705 Gigabit Ethernet, 
ASIC rev. 0x3003> mem 0xe9280000-0xe928ffff irq 11 at device 5.0 on pci0
Nov 21 21:05:07 ares kernel: fxp0: <Intel 82550 Pro/100 Ethernet> port 
0x9c00-0x9c3f mem 0xe9260000-0xe927ffff,0xe92a0000-0xe92a0fff irq 11 at 
device 7.0 on pci0
Nov 21 21:05:07 ares kernel: fxp1: <Intel 82550 Pro/100 Ethernet> port 
0xa000-0xa03f mem 0xe9220000-0xe923ffff,0xe92a4000-0xe92a4fff irq 10 at 
device 8.0 on pci0
Nov 21 21:05:07 ares kernel: fxp2: <Intel 82559ER Pro/100 Ethernet> port 
0xa400-0xa43f mem 0xe9240000-0xe925ffff,0xe92a1000-0xe92a1fff irq 11 at 
device 9.0 on pci0
Nov 21 21:05:07 ares kernel: atapci1: <HighPoint HPT370 UDMA100 
controller> port 
0xb800-0xb8ff,0xb400-0xb403,0xb000-0xb007,0xac00-0xac03,0xa800-0xa807 
irq 11 at device 10.0 on pci0
Nov 21 21:05:07 ares kernel: atapci2: <VIA 6420 SATA150 controller> port 
0xd000-0xd0ff,0xcc00-0xcc0f,0xc800-0xc803,0xc400-0xc407,0xc000-0xc003,0xbc00-0xbc07 
irq 10 at device 15.0 on pci0
Nov 21 21:05:07 ares kernel: fdc0: <floppy drive controller> port 
0x3f7,0x3f2-0x3f5 irq 6 drq 2 on acpi0
Nov 21 21:05:07 ares kernel: sio0: <16550A-compatible COM port> port 
0x3f8-0x3ff irq 4 flags 0x10 on acpi0
Nov 21 21:05:07 ares kernel: sio1: <16550A-compatible COM port> port 
0x2f8-0x2ff irq 3 on acpi0
Nov 21 21:05:07 ares kernel: atkbd0: <AT Keyboard> irq 1 on atkbdc0
Nov 21 21:05:07 ares kernel: Interrupt storm detected on "irq10: fxp1 
atapci+"; throttling interrupt source

Problem je v te posledni radce vypisu - na irq10 je naveseno fxp1, 
atapci0 a atapci2. Radice atapci0 a atapci2 nevykazuji zadnou chybovost, 
ale fxp1 je nepouzitelne. Ma znacnou ztratovost - napr. ve vypise 
`netstat -w1 -Ifxp1` jde videt znacne mnozstvi chyb.

Na irq11 jsou navesena dokonce 4 zarizeni - bge0, fxp0, fxp2 a atapci1 a 
nastesti zadne problemy se neprojevuji.

Z vypisu `ps ax|grep irq` jde videt, ze jeste nejake irq jsou volne:
    11  ??  WL     0:00,00 [irq0: clk]
    12  ??  WL     0:00,00 [irq1: atkbd0]
    13  ??  WL     0:00,00 [irq3: sio1]
    14  ??  WL     0:00,00 [irq4: sio0]
    15  ??  WL     0:00,00 [irq5:]
    16  ??  WL     0:00,00 [irq6: fdc0]
    17  ??  WL     0:00,00 [irq7:]
    18  ??  WL     0:00,00 [irq8: rtc]
    19  ??  WL     0:00,00 [irq9: acpi0]
    20  ??  DL     1:57,76 [irq10: fxp1 atapci+]
    21  ??  RL     8:42,65 [irq11: bge0 fxp0++]
    22  ??  WL     0:00,00 [irq12:]
    23  ??  WL     0:00,00 [irq13:]
    24  ??  RL     0:13,13 [irq14: ata0]
    25  ??  RL     0:12,90 [irq15: ata1]

Na desce je jeste jeden PCI slot volny - je nejaka sance, ze kdyz 
poprehazim poradi karet v PCI slotech, ze se kartam priradi IRQ nejak 
rovnomerneji? V techto situacich se jeste radi zmenit nastaveni v BIOSu 
na opacnou hodnotu nez je nastaveno u PnP OS. Predpokladam ze upravou 
/boot/device.hints niceho nedocilim, to se tyka jen ISA karet - ja zadne 
nemam.

Nemate nahodou jeste nekdo nejakou radu, co provest, abych se daneho 
konfliktu zbavil? Server jiz bezi v ostrem provozu a tudiz nemam moc 
moznosti na pokusy. Na testovani plne osazane HW konfigurace bohuzel 
nebyl cas, nebot se komponenty predelavaly z doted beziciho serveru.
Ctvrta sitova karta je nutna pro beh site a ted sit bezi v dost 
degradovanem rezimu :(

Prosim, kazda dobra rada nad zlato :)

-- 
Petr Bezděk