Bug 46490
| Summary: | xl driver generates lots of interrupts with 3c905B-TX/5.0-RC1 | ||
|---|---|---|---|
| Product: | Base System | Reporter: | enderle <enderle> |
| Component: | kern | Assignee: | Andre Oppermann <andre> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | Unspecified | ||
| Hardware: | Any | ||
| OS: | Any | ||
|
Description
enderle
2002-12-23 10:00:16 UTC
Just wanted to state that i now did read the manpage and i did not find any information about any tuning parameters. Also, the man page states it belongs to FreeBSD 4.7 and was written 1998: http://www.freebsd.org/cgi/man.cgi?query=xl&apropos=0&sektion=0&manpath=FreeBSD+5.0-current&format=html Responsible Changed From-To: freebsd-bugs->silby I'll look into it. I just did the update to RC2, but it didn`t change anything.
I can supply more information now, though:
This is what top states:
CPU states: 52.3% user, 0.0% nice, 12.3% system, 31.5% interrupt, 3.8% idle
21 root -68 -187 0K 12K WAIT 3:39 34.03% 34.03% irq11: xl0 acpi0
Some info about network utilisation:
% netstat -in 1
input (Total) output
packets errs bytes packets errs bytes colls
778 0 539500 2477 0 3378314 0
794 0 566540 2412 0 3262760 0
2893 0 3763242 1898 0 2113040 0
715 0 432990 2491 0 3423266 0
^C
It's just moderate, not realy impressing.
% ifconfig xl0
xl0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
options=3<RXCSUM,TXCSUM>
inet 192.168.0.2 netmask 0xffffff00 broadcast 192.168.0.255
inet6 fe80::250:4ff:fe43:5797%xl0 prefixlen 64 scopeid 0x1
ether 00:50:04:43:57:97
media: Ethernet autoselect (100baseTX <full-duplex>)
status: active
% dmesg | grep xl0
xl0: <3Com 3c905B-TX Fast Etherlink XL> port 0xdc00-0xdc7f mem 0xda000000-0xda00007f irq 11 at device 10.0 on pci0
xl0: Ethernet address: 00:50:04:43:57:97
miibus0: <MII bus> on xl0
Just a wild guess about what could be the problem:
% dmesg | grep "irq 11"
xl0: <3Com 3c905B-TX Fast Etherlink XL> port 0xdc00-0xdc7f mem 0xda000000-0xda00007f irq 11 at device 10.0 on pci0
irq 11 doesn't seem to be used twice.
I tried to follow /usr/src/UPDATING and the section about 5.0-CURRENT's speed, but as it seems, RC2 already has the debugging flags mentioned there turned off.
Also, the xl driver can be tricked to *run away* - 100% cpu utilisation and hang up, but network reachability will coninue though.
I copied a big file over network while it was accidentialy deleted from some other network station.
Imidiatly the irq11: xl0 acpi0 used 100% cpu and the system was extremly sluggish, altough staying useable. The load raised to about 7 the dropped back to 5 and raising up to 7 again, while the system was absolutly idle.
I rebooted the system with shutdown -r now, but it did not shut down cleanly. init stated that not all processes did stop and later bufdeamon did not die. Also a lot of unwriten pages where dropped later (about 750).
The system did reboot and came up cleanly again.
I did not try to reproduce it.
The NFS server is running IRIX 6.5.18m.
Hope this helps or someone can point my nose at some stupid problem i overlooked! (maybe?)
Steven, do still have this xl driver interrupt problem with 5.2RC2 or -CURRENT? -- Andre State Changed From-To: open->feedback This PR is too old. Asking Originator if problem still persists. Responsible Changed From-To: silby->andre This PR is too old. Asking Originator if problem still persists. State Changed From-To: feedback->closed Originator reports problem disappeared in 5.1R. |
