On Mon, 11 Jul 2005, vitko wrote: > > with -o sync or something similar. You might find it interesting to try > > doing the exact same write transfer under both 2.6.11 and 2.6.12, with > > usb-storage debugging turned on for both systems, and then compare the > > debugging logs. If the timings are so drastically different then I would > > expect the sequence of commands to be different as well. > > Please see attached dmesg outputs from 2.6.11 and 2.6.12.1 debug enabled > kernels, they contain only messages generated while writing file to flash > drive. Also please note that I got 1.4 MB/s write with usb-storage > debugging enabled (on "slow" kernel) instead of 135 kB/s with "normal slow" > kernel. Quite strange.
I don't understand that at all. If you rebuild the driver with debugging turned off, does the throughput decrease again? (That is, was this caused by the rebuild itself or by the debugging code?) The one obvious difference between the two logs is the lengths of the tranfers. With 2.6.11 the transfers were all 8192 bytes, and with 2.6.12 they were mostly 512 bytes with 2048 bytes occuring 1/4 of the time. This might be the result of a change in the filesystem layer. You could experiment with not using -o sync. Under 2.6.13-rc2 or later it may turn out to be just as reliable. Alan Stern ------------------------------------------------------- This SF.Net email is sponsored by the 'Do More With Dual!' webinar happening July 14 at 8am PDT/11am EDT. We invite you to explore the latest in dual core and dual graphics technology at this free one hour event hosted by HP, AMD, and NVIDIA. To register visit http://www.hp.com/go/dualwebinar _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
