From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003 Description of problem: ksoftirqd_CPU0 using > 90% of CPU time. Drive by going to a page with lots of images in mozilla, using 802-11b pcmcia wireless card with orinoco_cs driver. Reproducible on kernel 2.4.18-24.8.0 on my hardware. Not drivable at all (so far) on kernel 2.4.18-14. Hardware is Dell Inspiron 8200, using Netgear 802-11b wireless card in first pcmcia slot with orinoco_cs driver as device "eth1" Software for 2.4.18-14 is Redhat 8.0 distribution with no tweaks or new drivers but ntfs file system module added). 2.4.18-24.8.0 is update done on 7-Feb-03 (no ntfs file system module if it matters...) Version-Release number of selected component (if applicable): How reproducible: Sometimes Steps to Reproduce: 1.with hardware and software described above use mozilla to open pages with many images (gardening catalogs in this case). 2.after loading a number of such pages, ksoftirqd_CPU0 will start to take > 85%, often > 90% of CPU time. 3.System remains unusable until rebooted. Additional info:
This affects my dual xeon 2.4 Ghz system as well. I have the latest kerner from RH (as of 2/17/03.....non-beta). My system is a Dell 530 with 4 gb of ram and two 120gb ide disks. I use a ps2 keyboard, a ps2 mouse, and a few usb peripherals. I also see memory use climbing as it sits still (is there a memory leak in the kernel or the gterm app?). It makes typing and mouse use impossible! For me, no browser is necessary...just a telnet terminal for (gterm) is all that is needed.It occurs randomely. My machine is typically up non-stop. About once every month I need to reboot because of this bug. Sometimes it happens two days after a reboot. Sometimes two weeks.... See bug 80279 too. This is likely the same thing. It would be great if RH addressed these two bugs as one bug and fixed the problem or at least came up with a "patch" (some solution for us). I moved to linux for stability. Even XP doesn't do this kind of stuff. Please help! Jeff
Two new comments: (1) I _can_ drive bug 83789 under 2.4.18.24 after all, but requires much larger transfer over wireless: 5M .rpm file being downloaded from ftp.research.bell-labs.com managed to hang wireless 4 times in a row this morning in 2.4.18.24. (2) See http://www.uwsg.iu.edu/hypermail/linux/net/0211.3/0020.html Suresh Singh Keisam claims that he had similar problem with linux-2.4.20-rc3 kernel, and problem was fixed by updating to orinoco-0.13beta1 driver. I do not have experience with reinstalling drivers (that is why I get pre-packaged linux), but please check it out and see if this fixes problem. -- my wireless card uses the orinoco driver, so this seems likely to be the problem.
I just got the orinoco-0.13b driver from http://ozlabs.org/people/dgibson/dldwd/orinoco-0.13b/, installed as per their instructions (including using the -DMODVERSIONS -include $(KERNEL_SRC)/include/linux/modversions.h flags as recommended in README.orinoco). Rebooted, and was able to download the file that I had consistently been unable to download previously: ftp://ftp.research.bell-labs.com/dist/smlnj/release/110.0.7/RPMS/smlnj-110.0.7-4.i386.rpm This certainly does not constitute a thorough test of the fix, but gives some indication that the orinoco driver packaged with the Red Hat 8.0 system _may_ have been at fault.
The above is very interesting...but I am running on a land line 100BaseT dhcp set-up. While orinoco drivers "might" be loaded by the kernel, they should not be in use....am I missing something? My box is a Dell Precision Workstation 530. Thanks.
Had identical problem on a recent Dell 8200 with DLink 650 (ie Prism2.0) wireless card, running stock RH8.0 with 2.4.18-26.8.0 kernel. ksoftirqd would consistently bog down the machine whenever a X connection using wireless was up, even with minimal traffic. Using the updated orinoco driver as described above _fixed_ the problem. I used the most recent one available at the time, orinoco-0.13c. Ivo
I have had similar problems in RedHat 9 with kernel-2.4.20-9. I am trying orinoco-0.13c from http://ozlabs.org/people/dgibson/dldwd/ now. This is the first indication in the log that something was wrong: Apr 17 18:16:24 zion kernel: eth1: orinoco_reset failed in orinoco_pci_open()<3>eth1: error -110 reading Rx descriptor. Frame dropped. Apr 17 18:16:24 zion kernel: eth1: Error -110 writing Tx descriptor to BAP Apr 17 18:16:25 zion last message repeated 13 times Apr 17 18:16:25 zion kernel: hermes @ MEM 0xd0957000: Timeout waiting for command completion. Apr 17 18:16:25 zion kernel: hermes @ MEM 0xd0957000: Error -16 issuing command. Apr 17 18:16:25 zion kernel: eth1: Error -110 writing Tx descriptor to BAP Apr 17 18:16:25 zion last message repeated 30 times Apr 17 18:16:25 zion kernel: hermes @ MEM 0xd0957000: Error -16 issuing command. Apr 17 18:16:25 zion kernel: hermes @ MEM 0xd0957000: Error -16 issuing command. Apr 17 18:16:25 zion kernel: eth1: Error -110 writing Tx descriptor to BAP Apr 17 18:16:26 zion last message repeated 1102 times This was repeated until I shutdown the interface and unloaded the hermes, orinoco and orinoco_pci modules. (When I previously used RedHat 7.3 and 8 on the same machine, I used the linux_wlan driver [http://www.linux-wlan.com/linux-wlan/] and it never did anything like this.)
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/