#milkymist IRC log for Monday, 2011-02-14

kristianpaulFallenou: MAXIMUM_FRAME_SIZE is not implemented on your driver, why? how do you handle the  ethernet frames then?00:00
Fallenouin the driver I use MLEN as the max len00:16
kristianpaulin wich part fo the netwroking documentation is MLEN mentioned?00:18
FallenouMLEN = MSIZE - sizeof(struct m_hdr)00:18
kristianpaulhmm may be mine is outdate.. Edition 4.10.99.0, for RTEMS 4.10.99.000:18
Action: Fallenou is grepping through the source code, 1 min00:19
Fallenouhum hum cannot find MSIZE00:24
Fallenoubut I am pretty sure MLEN is something like 150000:24
Fallenoudamn I cannot find any #define MSIZE anywhere00:28
Fallenoucan you ?00:28
Fallenoumaybe it's in the compiler header :o00:28
Action: Fallenou is having a look00:29
Fallenouok yes it is00:29
FallenouMSIZE = 12800:29
Fallenouhum hum so it's the starting size of an mbuf00:29
FallenouI guess it grows afterward00:30
Action: Fallenou looking at his code00:30
Fallenouhum hum strange I cannot find where I allocate memory inside the mbuf :o00:32
Fallenoucould be our bug :)00:32
Fallenoumaybe we just have the basic 128 bytes00:32
Fallenouand bigger frames will just overwrite memory00:32
FallenouI will ask Joel about that :)00:33
Fallenouwhy the hell am I using MLEN in the packet structure instead of ETHERNET_FRAME_LENGTH00:35
Fallenoulet  fix that00:37
CIA-40rtems-milkymist: Yann Sionneau master * r73572ad / c/src/lib/libbsp/lm32/shared/milkymist_networking/network.c : fixed length of data member of mm_packet struct to be ETHERNET_FRAME_LENGTH instead of MLEN - http://bit.ly/gTn78500:38
Fallenoukristianpaul: I pushed a fix for the ethernet driver00:39
FallenouI don't know if it will fix the FIFO problem00:39
Fallenoubut it surely was a big issue00:39
Fallenouthat would corrupt memory00:39
Fallenoukristianpaul: please compile the bsp and try again to make it crash :)00:40
FallenouI emailed the rtems ML about that00:53
Fallenouthanks for the feed back kristianpaul :)00:53
kristianpaulFallenou: he, i hope i works, i'll try it tomorrow or later midnight01:00
kristianpaulI need sump to work on the avnet, but thanks to jevin that will be quick i think :-)01:00
kristianpaulergg i mean jackgassett:01:00
Fallenouok great, keep me posted on your tests :)01:03
FallenouI think this fix fixes something, but it may not be the problem you are talking about :)01:04
Fallenouwe'll see !01:04
kristianpaulwhat is something?01:05
Fallenouwell I think it fixes a big memory corruption issue01:06
Fallenouinside the ethernet driver01:07
Fallenouwhen packets were > 128 bytes memory was overwritten somewhere01:07
Fallenouwhich isn't really cool :)01:07
kristianpaulheh01:09
Fallenoulook at the patch, it's a one-liner-patch01:10
adamw_larsc, last time I used: modprobe ftdi_sio vendor=0x20b7 product=0x071303:21
adamw_ to activate the JTAG/Serial pod03:21
larscadamw_?03:23
adamw_if I erased VID/PID info, then I use UrJtag tool & tried to 'cable milkymist', but I got 'Couldn't connect to suitable USB device.03:24
adamw_Error: usbconn/libftdi.c:372 usbconn_ftdi_common_open() ftdi/ftd2xx error: ftdi_usb_open_desc() failed: device not found03:24
adamw_'03:24
adamw_so this is reasonable. right?03:25
kristianpaul20591:09:33 < lekernel> rmmod ftdi_sio03:25
kristianpaul20592:09:34 < lekernel> modprobe ftdi_sio  vendor=0x... product=0x...03:25
kristianpaul20597:09:46 < wolfspraul> 1. "rmmod ftdi_sio"03:25
kristianpaul20598:09:47 < wolfspraul> 2. "modprobe ftdi_sio vendor=0x20b7 product=0x0713"03:25
kristianpaulthats better03:25
larscadamw_: if the vid/pid info is erased it makes sense that the driver does not handle the device03:26
larscwhat does `lsusb` say?03:26
adamw_yeah...Yanjun helped us to find the JTAG/Serial high speed bug on h/w03:26
adamw_Bus 002 Device 038: ID 0403:6010 Future Technology Devices International, Ltd FT2232C Dual USB-UART/FIFO IC03:27
adamw_this info( be noticed that I erased VID/PID already)03:27
larscso you should be able to get it working with `rmmod ftdi_sio; odprobe ftdi_sio vendor=0x0403 product=0x6010`03:27
kristianpaulrmmod ftdi_sio; modprobe ftdi_sio vendor=0x0403 product=0x601003:28
adamw_hmm...with 'dmesg', I can see two device attached to ttyUSB0 and ttyUSB103:29
kristianpaulhmm03:29
adamw_so I found that strange things that if I used "modprobe ftdi_sio vendor=0x20b7 product=0x0713" & programmed ftdi to be correct '0x20b7' & '0x0713'03:30
adamw_then I just got only ttyUSB0 only, ttyUSB1 disappeared...what could cause this?03:30
kristianpaulttyUSB1 is actually jtag so should be okay03:31
kristianpauldoes it said too "Ignoring serial port reserved for JTAG" ?03:32
adamw_no..03:32
adamw_ttyUSB0 is jtag, ttyUSB1 is serial03:32
adamw_kristianpaul, http://en.qi-hardware.com/wiki/File:M1_jtagserial_run1_C27_ground_rework3.png03:33
adamw_this fixed the high speed bug03:33
adamw_so after I fixed this, then I tried to follow all steps on http://en.qi-hardware.com/wiki/JTAG_Serial_Cable_run_1_for_Milkymist_One#Tests_and_Results03:35
adamw_then I got only ttyUSB0 only.03:35
adamw_so I erased VID/PID then 'dmesg'...i got two device attached!03:36
larscshould be fine. there is code in the driver which tells to don't create a ttyUSB for the jtag port, when VID/PID matches those of the milkymist jtag adapter03:36
adamw_but i got failed while running 'jtag''s 'cable milkymist'. :-) hope you understand my descriptions.03:36
kristianpaulwhy you erased VID/PID? you mean in the eeprom?03:38
larscadamw_: did it fail for the MM VID/PID or the default VID/PID or both?03:39
adamw_i didn't see the replied msg of 'dmsg' about attaching two device, so after I erased VID/PID then my laptop can detect both.03:41
kristianpaulyou dont need reply about attaching two devices03:41
adamw_with MM VID/PID, I got ttyUSB0 only; with default VID/PDI, i got both.03:42
larscadamw_: thats the correct result03:42
adamw_so what action I need to do now?03:42
kristianpauljtag03:42
kristianpaulcable milkymist03:43
kristianpauldetect03:43
adamw_erase back to MM VID/PID...then jtag..cable milkymist ..detect ?03:43
larscyes03:43
kristianpaulyes03:43
larscthat should work03:43
adamw_good...second.03:43
kristianpaulnice bug fix btw :-)03:44
rohat openmoko we always had 2 serials as far i can remember... and one vanished as soon as openocd was started03:44
larscurjtag checks the VID/PID and if it doesn't match you'll get the error you've seen03:44
kristianpaullarsc: did you finally try openocd with lm32?03:45
kristianpaulor was mwalle ..03:45
larsckristianpaul:  i couldn't getit to work03:46
kristianpauladamw_: does it works?03:52
adamw_good, jtag port/coonection surely work well.03:53
adamw_but my ttyUSB1 can't work well since I need to use it to test serial port connected to M1.03:53
kristianpaulhmm?03:54
adamw_flterm --port /dev/ttyUSB1 --kernel boot.bin03:54
adamw_i used this to test 'serial' port03:55
kristianpaulwhy /dev/ttyUSB1 and not /dev/ttyUSB0 ??03:55
adamw_with 'dmesg', I still didn't the ttyUSB103:55
kristianpauli still confused about your setup03:55
kristianpaulyou should not get ttyUSB1 jut ttyUSB003:56
kristianpaultry03:56
kristianpaulflterm --port /dev/ttyUSB0 --kernel boot.bin03:56
adamw_hm...oah...MAN! I totally made mistakes on last 97pcs productions...No one found.03:58
adamw_yes, now just used ONLY ttyUSB0 to both 'jtag' & 'serial' ports.03:58
adamw_god! my steps of productions were totally wrong!03:59
kristianpauldid detect command in urjtag work?03:59
adamw_yes, it works well now.;-)04:00
kristianpaulYou got ginally fpga id?04:00
kristianpaulgreat !04:00
kristianpaulgn8 then !04:00
adamw_kristianpaul, larsc  Much thanks for helps04:00
adamw_I'll send rework steps to list for high speed bug04:01
adamw_kristianpaul, I think you can directly rework on your side.;-)04:01
adamw_the ground of C3/R14/C27/C28 doesn't connect with system ground.04:02
kristianpauladamw_: yeah !04:02
kristianpauladamw_: i'll wait impatietly for the steps04:04
adamw_you could just follow:04:06
adamw_http://en.qi-hardware.com/wiki/File:ECO_Yanjun_jtagserial_run1.png04:06
adamw_http://en.qi-hardware.com/wiki/File:M1_jtagserial_run1_C27_ground_rework1.png04:07
adamw_http://en.qi-hardware.com/wiki/File:M1_jtagserial_run1_C27_ground_rework2.png04:07
adamw_http://en.qi-hardware.com/wiki/File:M1_jtagserial_run1_C27_ground_rework3.png04:07
adamw_well...starts to edit wiki now...We do apology.04:08
kristianpaulah, just two points?04:08
adamw_right!04:08
kristianpauli'll do it NOW04:08
adamw_you can handle i think ...but for s/w and whole documents I must edit more...04:09
adamw_ok..good to have you guys supports.04:09
adamw_one question again: why my last time of production steps can work with ttyUSB1 (for flterm...) and ttyUSB0( for Urjtag..)?04:12
adamw_i confuse myself now..:-(04:13
kristianpaulno04:20
adamw_kristianpaul, ?04:21
kristianpaulactually you dont need to tell urjtag about ttyUSB04:21
adamw_no, i meant that my last time production i used 'flterm --port /dev/ttyUSB1 ...' then it can work?04:22
kristianpauli wonder how/why it worked for you04:23
kristianpaulIt should not.. as serial port is attached as /dev/ttyUSB004:23
adamw_and now with fixed high speed problem...i can't use ttyUSB1 anymore...?04:23
kristianpaulright04:23
kristianpaulwell04:23
kristianpaulat least, you plug another usb2ttl cable or somthing that already take /dev&ttyUSB first so the jtagserial port will be ttyUSB104:24
adamw_so last time must have some things I still made mistake somewhere? o04:25
adamw_you are right!04:26
adamw_it indeed also shows 'Ignoring serial port reserved for JTAG' on ttyUSB0.04:27
kristianpaulyes04:28
kristianpaulhttp://paste.debian.net/107552/04:28
kristianpaul!!!04:28
adamw_great! yours now is high speed!04:29
kristianpaulyes, was easy rework04:29
kristianpauland i'm bit sleepy now04:29
adamw_what's command you typed?04:29
adamw_ha...ok04:29
adamw_sleep first04:30
kristianpaulwell i just ram some already made scripts for jtag to both flash bitstream and the app04:31
kristianpaulit works as usual, now a bit faster it seems04:31
kristianpaulok, bed time now04:31
kristianpaulchao !04:31
adamw_cu!04:32
wpwrakhmm, regarding that jtag fix, did anyone actually measure what the difference between full-speed and high-speed mm1 jtag is ?14:34
lekernelrejon: playing with Ron a bit? ;)17:22
wpwrakheh, ron isn't here. that's kinda convenient ;-)17:26
rejonhahahah17:31
rejonits good to expose what we need to be working on17:31
rejon:)17:31
rejonbbiab gotta eat17:31
wpwrakrejon: i wouldn't mind the _exposing_ part. but if you had a limp, would you truly appreciate it if everyone you meet would mention it in every other sentence ?17:34
wpwrakrejon: or, for that matter, if most people would drop that topic after a while, but there's one guy who keeps on with it.17:36
FallenouIs the minimac ethernet driver working well on Linux ?20:40
Fallenoucan someone give me the link of the repository ?20:40
larschttps://github.com/tmatsuya/linux-2.620:41
Fallenouthanks20:41
Fallenoulekernel: the linux driver is using it's own crc32 function too btw :)20:42
larscit should use the generic one though20:45
Fallenouyes that's what i've been told :p20:45
Fallenouwas just joking around20:46
lekernelFallenou: linux drivers aren't necessarily a good example20:50
lekernelif I didn't go after the Linux driver, it's only because I don't depend on Linux20:50
Fallenousure sure20:50
Fallenouanyway the linux driver is not doing anything in the RX top half irq handler20:51
lekernellast time I tested, the ethernet linux driver didn't work at all20:51
Fallenouis there any RX fifo problem with linux ?20:51
Fallenouoooh20:51
Fallenouok ='20:51
Fallenouso my looking at the code is irrelevant20:51
lekernellol: http://designcon.techinsightsevents.com/video_contest21:00
Fallenoulol :')21:07
Fallenouawesome21:07
mwallegood evening22:28
Fallenouhi mwalle22:29
lekernelhi!22:31
--- Tue Feb 15 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!