On Wed, 26 Apr 2006, Christian Haul wrote: > Alan Stern wrote: > > On Tue, 25 Apr 2006, Christian Haul wrote: > >> This block repeats for every 2 seconds..... and top shows around CPU 40% hi > > > > I have trouble believing that a few milliseconds of activity occurring > > once every two seconds is enough to push your CPU usage to 40%. > > Well.... I've taken this after I have stopped hald for some ten minutes > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 2989 root 10 -5 0 0 0 S 0.0 0.0 3:57.77 usb-storage > > You see, the system spent 3:57.77 minutes in the usb-storage module, and > that with an uptime of 1:35 hrs.
About 1 part in 25? That would amount to 80 ms for each poll, if the polls occur at 2-second intervals. Sounds way too long. 5 ms or less would be about what I expect. It's hard to account for where the time was spent from the ps listing. It would help if you could provide another dmesg log with CONFIG_PRINTK_TIME set. > # uptime > 20:05:26 up 1:35, 4 users, load average: 1.69, 2.18, 2.23 > > idle system with hald running (for about 30 secs): > > top - 20:12:28 up 1:42, 5 users, load average: 2.67, 1.90, 2.00 > Tasks: 160 total, 2 running, 157 sleeping, 1 stopped, 0 zombie > Cpu(s): 9.6% us, 7.0% sy, 0.0% ni, 54.5% id, 5.3% wa, 22.9% hi, 0.7% si > Mem: 1037080k total, 1027028k used, 10052k free, 335928k buffers > Swap: 1245932k total, 12k used, 1245920k free, 128192k cached I can't really say what that means. > > They're right about the polling, but they're wrong about the kernel driver > > needing to be fixed. In fact nothing needs to be fixed as far as I can > > see, except possibly the statistics you get from top. Have you tried > > experimenting with a real load present, such as running a benchmark > > program? Try to find out how much hald and usb-storage really do slow > > down performance. > > Actually, this was where it all started. I experienced a load > 10 when > only browsing the web and the machine was more than sluggish. OK, this > is only an XP 1800+ with 1GB RAM, still, for web browsing a load of 10 > is incredible. I don't recall anyone else reporting this kind of trouble. It's possible that there's something funny going on with your card reader. > After removing usb-storage / stopping hald everything is nice and > smooth. Except for removable media auto-detection :-( > > Yes, I really do think something is broken here. Maybe it's the hardware. Alan Stern ------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users