This patch-set implements per device dirty page throttling. Which should solve the problem we currently have with one device hogging the dirty limit.
Preliminary testing shows good results: mem=128M time (dd if=/dev/zero of=/mnt/<dev>/zero bs=4096 count=$((1024*1024/4)); sync) 1GB to disk real 0m33.074s 0m34.596s 0m33.387s user 0m0.147s 0m0.163s 0m0.142s sys 0m7.872s 0m8.409s 0m8.395s 1GB to usb-flash real 3m21.170s 3m15.512s 3m23.889s user 0m0.135s 0m0.146s 0m0.127s sys 0m7.327s 0m7.328s 0m7.342s 2.6.20 device vs device 1GB disk vs disk real 1m30.736s 1m16.133s 1m42.068s user 0m0.204s 0m0.167s 0m0.222s sys 0m10.438s 0m7.958s 0m10.599s 1GB usb-flash vs background disk N/A 30m+ 1GB disk vs background usb-flash real 4m0.687s 2m20.145s 4m12.923s user 0m0.173s 0m0.185s 0m0.161s sys 0m8.227s 0m8.581s 0m8.345s 2.6.20-writeback 1GB disk vs disk real 0m36.696s 0m40.837s 0m38.679s user 0m0.161s 0m0.148s 0m0.160s sys 0m8.240s 0m8.068s 0m8.174s 1GB usb-flash vs background disk real 3m37.464s 3m49.720s 4m5.805s user 0m0.167s 0m0.166s 0m0.149s sys 0m7.195s 0m7.281s 0m7.199s 1GB disk vs background usb-flash real 0m41.585s 0m30.888s 0m34.493s user 0m0.161s 0m0.167s 0m0.162s sys 0m7.826s 0m7.807s 0m7.821s - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
