**** BEGIN LOGGING AT Fri Jan 09 02:59:57 2009 Jan 09 03:06:39 nbd * r13946 /trunk/target/linux/ (3 files in 3 dirs): b44: fix misalignment and space-saving in rx packet handling (significant throughput and cpu load improvement) Jan 09 03:08:34 nbd * r13947 /branches/8.09/target/linux/ (2 files in 2 dirs): merge r13946 to 8.09 Jan 09 08:04:26 what command can I use to compile alsa-utils and alsa-lib ? I tried "make package/alsa-lib-clean V=99" and it tells me there is no rule to make target 'package/alsa-lib-clean' Jan 09 08:19:31 anyone know how to compile lib's? Jan 09 08:19:38 eg alsa-lib ? Jan 09 08:23:45 dbro: you might try make package/alsa-{clean,compile} V=99 Jan 09 08:24:17 or if there is an separate alsa-libs... Jan 09 08:24:32 dbro: you might try make package/alsa-libs-{clean,compile} V=99 Jan 09 08:25:26 I am in the middle of a build... so I can't test it ti see if it works. Jan 09 08:28:00 ok, I'll try that Jan 09 08:30:14 same error as before Jan 09 08:30:24 (tried both of those) Jan 09 08:30:43 i'm using kernel 2.4, by the way Jan 09 12:40:59 juhosg * r13948 /trunk/package/mac80211/patches/ (15 files): [package] mac80211: update ath9k patches Jan 09 13:01:11 nbd * r13949 /branches/8.09/scripts/metadata.pl: merge 13877 to 8.09 Jan 09 13:02:19 nbd * r13950 /packages/lang/python/Makefile: fix python libffi dependency for targets which do not support libffi Jan 09 13:02:29 nbd * r13951 /packages/libs/libdnet/Makefile: fix dnet python dependency Jan 09 13:50:51 Ahem... Jan 09 13:51:04 Anybody listening? Jan 09 13:54:17 no Jan 09 13:54:27 lol Jan 09 13:57:31 i know too many johns on freenode :/ Jan 09 13:57:59 always mix them up Jan 09 13:58:22 Ok... well I'd like to ask how come MateIn4 is ignored...? Jan 09 13:58:53 I mean the guy is ready and willing to help Openwrt become even greater... Jan 09 14:00:57 but really... his port for the Actiontec mi424wr... should stand on it's own merit... They are dreadfully cheap...(I bought one for 4.99 on ebay) and have a brawny Intel 533/Ixp425 cpu... Jan 09 14:02:38 I would say that is one helluva Opening Act... Jan 09 14:03:31 Actually... I wound up buying 3 of them for a couple dollars... none over 12.95... all on ebay. Jan 09 14:03:59 booooooooooooooooo Jan 09 14:04:03 10/1000 Jan 09 14:04:06 err Jan 09 14:04:08 10/100 Jan 09 14:04:23 boo what...? Jan 09 14:04:41 no 1000Mbit...? Jan 09 14:05:10 oh Jan 09 14:05:15 minipci Jan 09 14:05:16 nice Jan 09 14:06:13 I am just going to say... if it takes Openwrt 6 months to recognize talent like this... I just have to add... talented people don't hang around... waiting for discovery. Jan 09 14:07:23 I have enjoyed working with MateIn4... Jan 09 14:08:09 but if you guys don't grab him quick... he's got better things to do than throw patches into space. Jan 09 14:08:32 Weedy: You see what I mean? Jan 09 14:08:42 Take a look at ebay... Jan 09 14:09:01 place is crawling with them... mi424wr's for pennies... Jan 09 14:11:31 My wife... she uses www.speakeasy.net/speedtest to check my toys out... if they don't pass that... she's demanding her Linksys back. Jan 09 14:12:21 I clocked 33,000 kb/sec on one of those actiontecs... better than I ever got on her Linksys :) Jan 09 14:12:36 at speakeasy that is... Jan 09 14:15:04 http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=180313928864 Jan 09 14:15:18 That is the one I got for 4.99 Jan 09 14:17:49 Yeah... I popped out that minipci and threw in an Atheros... in a heartbeat... Jan 09 14:24:38 you got it for $15, some ebay sellers jack shipping Jan 09 14:26:06 well any ebayer knows that 10$ is about the minimum... you can get a router shipped for... Jan 09 14:26:46 yes... some charge more... I paid 15 to get one shipped from brooklyn nyc to jersey city... Jan 09 14:26:55 but that was fed ex Jan 09 14:27:16 and that one had a bad power supply... Jan 09 14:27:18 roundsparrow: be forwarned that there's a high probability that the power supply doesn't work Jan 09 14:27:57 but MateIn4 has found a good source in china for about 4 dollars each... and 4 shipping... Jan 09 14:28:23 some day we dream that router companies and such standardize PSU's at least to a common voltage/etc ;) So many are crap. Jan 09 14:28:23 quantity prices are combined shipping... Jan 09 14:29:00 buy 100 and get a bulk shipping rate... Jan 09 14:29:22 he got it in about 7 days... Jan 09 14:30:24 Out of the 3 units I bought... 1 psu was bad... Jan 09 14:31:43 I ordered a mega mutha from pacificgeek.com... and they have to be smoking crack over there... first I got a unit with no power cord. Then they sent me a power cord for a PC... Jan 09 14:32:19 That one was 20$ +$10 shipping... for 1/2 a power supply... lol Jan 09 14:32:43 But it is a 4 amp job... Jan 09 14:34:02 MateIn4... I think I am going to send this one back to pacificgeeks.com ; I'll take one of yours off your hands... if you have any left. Jan 09 14:39:13 Johnrw: I think the whole issue of performance is one of lots more software optimization needed in Linux. Look at this reviw in detail, http://en.network01.net/modules/newbb/viewtopic.php?topic_id=15&forum=2 - you will find that ZynOS consistantly kicks Linux ass on performance. this is some clue of the need to optimize. openWRT now runs on this platform (Atheros AR9132 / AR9102 or AR9103) ... and it isn't he hardwa Jan 09 14:40:51 I agree on the optimization... part... but with all the cloak and dagger... with these chip companies... it is just amazing that linux runs at all Jan 09 14:45:30 I am in need to discuss a slow usb when the data stored on it... goes over the network through samba... Jan 09 14:46:20 I did a hdparm -tT /dev/sda and it could read at 8MB/sec i think it was... Jan 09 14:46:51 but when I get files from it... over samba... I am lucky to get 1MB second Jan 09 14:52:51 I think less to do with the chips and more to do with optimization of things like NAT for better memory usage and system design. ZynOS was heavily optimized for small systems since like 1995. Company called Stagecoach in 2000: http://download.modem-help.co.uk/mfcs-Z/ZyXel/p1600/firmware/p1600-V2.50-X.06--Standard.zip.php?showFile=p1600_ZyNOS_v2.50(X.06).rtf - but it goes back earlier. I say this as in general it is i Jan 09 15:09:26 johnrw: pastebin your smb.conf Jan 09 15:09:35 i can max 100mbit Jan 09 15:10:06 I don't have it anymore... new image... Jan 09 15:10:12 ah Jan 09 15:10:14 built it last night... Jan 09 15:10:14 well Jan 09 15:10:21 if it still sucks Jan 09 15:10:23 poke me Jan 09 15:10:45 but that was one of the 20 things I need to do... Jan 09 15:14:48 fine Jan 09 15:15:18 socket options = TCP_NODELAY IPTOS_LOWDELAY Jan 09 15:15:25 if you have anything else Jan 09 15:15:30 replace that line Jan 09 15:24:38 Weedy I actually threw in the socket options = IPTOS_LOWDELAY TCP_NODELAY SO_SNDBUF=4096 SO_RCVBUF=4096 Jan 09 15:25:15 and smbd was taking all the cpu time... i mean it was at 90% Jan 09 15:25:58 my distro config was 8196 Jan 09 15:26:11 topped out at 4MB/s Jan 09 15:26:19 now i can do 12 Jan 09 15:28:10 with no SO_SNDBUFF stuff? Jan 09 15:28:24 i guess i have good tcp settings and the samba settings were just screwing with stuff Jan 09 15:28:43 ok... i'll try that. Jan 09 15:28:46 brb Jan 09 15:28:58 # 5. Browser Control and Networking Options: Jan 09 15:28:59 # Most people will find that this option gives better performance. Jan 09 15:28:59 # See speed.txt and the manual pages for details socket options = TCP_NODELAY IPTOS_LOWDELAY Jan 09 15:29:12 no buffer stuff Jan 09 15:35:46 root@OpenWrt:~# mount /dev/sda1 Jan 09 15:35:46 mount: mounting /dev/sda1 on /mnt/sda1 failed: Invalid argument Jan 09 15:35:46 root@OpenWrt:~# cat /etc/fstab Jan 09 15:35:46 # WARNING: this is an auto generated file, please use uci to set static filesystems Jan 09 15:35:46 /dev/sda1 /mnt/sda1 ext3 rw,async,users 0 0 Jan 09 15:36:35 no ext3 loaded? Jan 09 15:36:42 .... Jan 09 15:36:51 does the folder exsist? Jan 09 15:37:10 dmesg? Jan 09 15:37:18 exsist-s Jan 09 15:37:20 -s Jan 09 15:37:22 ffffffffffffffffff Jan 09 15:37:26 need to sleep Jan 09 15:37:48 also Jan 09 15:38:31 /dev/mapper/system-home /home ext3 noatime 0 2 Jan 09 15:46:41 okay well that got me up to 1.5 MB/se Jan 09 15:47:08 wait Jan 09 15:47:17 is samba runing on a router? Jan 09 15:47:25 root@OpenWrt:/mnt/sda1# lsmod | grep ext3 Jan 09 15:47:25 ext3 85796 1 Jan 09 15:47:25 jbd 29172 1 ext3 Jan 09 15:47:38 yes Jan 09 15:47:43 yeah Jan 09 15:47:46 your fucked Jan 09 15:52:04 dmesg http://pastebin.ca/1304313 Jan 09 15:52:22 Weedy... can ya be a little more verbose? Jan 09 15:52:25 :) Jan 09 16:01:31 Weedy: what you mean is clear... but the why part... is a little thin... Jan 09 16:02:08 samba needs horse power Jan 09 16:02:24 200-300mhz mips wont cut it Jan 09 16:02:52 Well running Jungo... it got 2.8 on USB and 4MB on Firewire Jan 09 16:03:03 reference: http://www.nslu2-linux.org/wiki/Info/Performance Jan 09 16:03:07 It's an Intel... Jan 09 16:04:14 hmm Jan 09 16:04:15 3 Jan 09 16:05:55 I just tweak the following stuff on my NSLU2: Jan 09 16:05:55 - Use of writeback journal (available in ext3fs) and noatime mount option - In smb.conf, use of : socket options = TCP_NODELAY SO_KEEPALIVE IPTOS_LOWDELAY SO_RCVBUF=16384 SO_SNDBUF=16384 Jan 09 16:06:04 i guess try that Jan 09 16:08:30 well for some reason the cpu is pegged at 90% Jan 09 16:09:26 could it be that the cpu has a cache that is off or something? You know, like the old days... when L2 cache was off on the pc... Jan 09 16:10:03 11:02:07 < Weedy> samba needs horse power Jan 09 16:10:09 johnrw: DMA? Jan 09 16:10:16 might be dma Jan 09 16:10:22 now there could be my problem Jan 09 16:10:32 hdparm -tT Jan 09 16:10:36 I don't know how to set that up... Jan 09 16:10:54 hdparm -tT /dev/sda Jan 09 16:11:00 11:02:07 < Weedy> samba needs horse power Jan 09 16:11:00 11:02:07 < Weedy> samba needs horse power Jan 09 16:11:06 shit Jan 09 16:11:08 my bad Jan 09 16:11:19 ubuntu doesnt like my touchpad Jan 09 16:11:46 /dev/sda: Jan 09 16:11:46 Timing cached reads: 148 MB in 2.02 seconds = 73.27 MB/sec Jan 09 16:11:46 Timing buffered disk reads: 56 MB in 3.04 seconds = 18.44 MB/sec Jan 09 16:12:04 ok Jan 09 16:12:07 no dma Jan 09 16:12:09 not Jan 09 16:12:26 dma is not the problem...? Jan 09 16:12:41 no dma would be like 2MB/sec Jan 09 16:12:57 anything over 5 = DMA is on Jan 09 16:13:04 ok... so I can relax about dma then... cool Jan 09 16:13:14 so Jan 09 16:13:20 MateIn4 did that right then... Jan 09 16:13:21 samba still hates you Jan 09 16:13:41 I have already thrown one of these against the wall Jan 09 16:13:51 _awesome_ Jan 09 16:14:11 Rage... is a mighty thing... Jan 09 16:14:19 hard to overcome it... Jan 09 16:14:21 :) Jan 09 16:14:51 That was when I could not write to the thing using samba... Jan 09 16:14:56 dd if=/dev/zero bs=1k | ssh -c blowfish user@hostname "dd of=/dev/null bs=1k" Jan 09 16:14:58 still can't Jan 09 16:17:08 WARNING: Ignoring unknown argument '-c' Jan 09 16:17:08 ssh: exited: Error resolving: Name or service not known Jan 09 16:17:31 dd if=/dev/zero bs=1k | ssh -c blowfish root@Openwrt "dd of=/dev/null bs=1k" Jan 09 16:18:40 well Jan 09 16:18:54 drop the -c blowfish Jan 09 16:20:16 root@localhost worked... Jan 09 16:20:30 but it just sat there... Jan 09 16:24:40 just did some performance tests on a usr8200 (ixp422 running 233Mhz). the TCP test ran at 95.2 Mbits/sec! can't do better than that running NAT Jan 09 16:24:45 johnrw: kill it Jan 09 16:25:02 you streaming data to your self Jan 09 16:25:18 ok... Jan 09 16:25:23 killed it... Jan 09 16:25:30 that result was using iperf Jan 09 16:25:43 iperf is raw packets Jan 09 16:25:51 ofc its going to be fast Jan 09 16:26:38 first, need to establish that the basics are sound Jan 09 16:34:15 well if iperf can throw packets at the lan at 95MB/sec why can't samba do 1/10 of that? Jan 09 16:35:38 could it be some of the openwrt Samba patches? You know... to make it small enough to fit? Jan 09 16:36:18 well Jan 09 16:36:25 iperf throws zeros Jan 09 16:36:34 samba throws data Jan 09 16:36:38 huge diff Jan 09 16:37:24 could it be samba 3? Might samba 2 perform better? Jan 09 16:38:24 I would think the opposite... but at this point... Jan 09 16:51:52 Cpu(s): 1.3%us, 45.2%sy, 21.3%ni, 2.3%id, 2.0%wa, 2.3%hi, 25.6%si, 0.0%st Jan 09 16:52:12 25% on software interrupts... Jan 09 17:11:01 lars * r13952 /packages/Xorg/ (40 files in 40 dirs): Merge menuconfig Xorg libs and libraries section. Jan 09 17:24:38 nbd * r13953 /packages/Xorg/xorg/driver/xf86-video-intel/Makefile: fix xf86-video-intel compile Jan 09 17:49:16 xMff: ping Jan 09 17:49:32 johnrw: pong Jan 09 17:49:36 heya... Jan 09 17:49:48 hey :) Jan 09 17:49:54 where can I set the error=continue flag... Jan 09 17:50:09 I don't like that policy... Jan 09 17:50:27 I know it makes sense for flash sticks Jan 09 17:50:28 no idea right now Jan 09 17:50:45 not for a 200gb usb drive though... Jan 09 17:51:16 continue with errors seems kinda primitive... Jan 09 17:51:23 no idea? really? Jan 09 17:51:48 Did i finally stump the Answerman? I want my prize! Jan 09 17:52:15 Door number 3. I'll take door number 3. Jan 09 17:53:18 oh crap... sorry for the spam... guys... thought this was #openwrt Jan 09 17:53:28 johnrw: try setting option options "rw,error=continue" in /etc/config/fstab Jan 09 17:53:45 johnrw: oh usb disk... you have to edit /sbin/usb-storage then Jan 09 17:53:56 no... i want to get rid of it... Jan 09 17:55:06 noz * r13954 /trunk/target/linux/ixp4xx/ (4 files in 2 dirs): USR8200: Apply last change to 2.6.26 and 2.6.28 also Jan 10 00:55:40 nbd * r13955 /trunk/target/linux/brcm-2.4/patches/110-b44_alignment.patch: fix a memory corruption bug in the linux 2.4 version of b44 Jan 10 00:56:24 nbd * r13956 /branches/8.09/target/linux/brcm-2.4/patches/110-b44_alignment.patch: merge r13955 to 8.09 Jan 10 01:00:40 nbd * r13957 /branches/8.09/package/opkg/files/opkg.conf: set opkg feed to 8.09 RC2 url Jan 10 02:06:14 thepeople * r13958 /trunk/package/kernel/modules/netsupport.mk: fix kernel config **** ENDING LOGGING AT Sat Jan 10 02:59:56 2009