| 08:34.27 | p2-mate | hi |
| 08:34.40 | ka6sox | lo |
| 08:45.13 | AchiestDragon | hi |
| 08:45.43 | ka6sox | yo...how goes it? |
| 08:45.59 | AchiestDragon | ok ty :) |
| 08:46.19 | ka6sox | cool |
| 08:46.29 | ka6sox | I printed out the latest schematics. |
| 08:46.50 | ka6sox | they look good. |
| 08:47.05 | ka6sox | but I was wondering about the 2 jtag chains. |
| 08:47.31 | AchiestDragon | 3 |
| 08:47.41 | AchiestDragon | arm , cpld , fpga |
| 08:48.06 | ka6sox | are we going to reprogram the FPGA on the EP board? |
| 08:48.15 | ka6sox | on the fly I mean. |
| 08:48.39 | AchiestDragon | had not planned on it |
| 08:49.02 | ka6sox | okay we could put the cpld and fpga on one chain then...save some space. |
| 08:49.23 | ka6sox | the 20 pin ARM is important for debugging and bringing it up. |
| 08:49.29 | AchiestDragon | the problem will be getting a 2.5v i/o for it on the epboard |
| 08:50.03 | AchiestDragon | as the arm is a seperate jtag the cpld jtag is at 3v3 while te fpga is 2.5 |
| 08:50.06 | ka6sox | I got a ARM debugger/programmer going to be at the office when I get in tommorrow to check out. |
| 08:50.13 | ka6sox | got it. |
| 08:50.17 | AchiestDragon | koo |
| 08:50.19 | AchiestDragon | l |
| 08:50.31 | ka6sox | ya...so I'll be able to check things out really well. |
| 08:50.37 | p2-mate | ka6sox: what are you trying to build If I may ask ? |
| 08:51.28 | AchiestDragon | p2-mate: http://www.whipy.demon.co.uk/geep.pdf |
| 08:51.40 | ka6sox | a FAST jtag programmer/debugger with 32-128Channels of Logic Analyzer. |
| 08:51.44 | p2-mate | aha |
| 08:51.50 | AchiestDragon | p2-mate: and http://www.whipy.demon.co.uk/geep-sch.pdf |
| 08:52.28 | AchiestDragon | this is the sbc control board for it , |
| 08:54.55 | ka6sox | time for me to get some sleep here. |
| 08:55.02 | ka6sox | I'll poke around tommorrow. |
| 08:55.15 | AchiestDragon | k |
| 08:55.18 | ka6sox | got names for the parts yet? |
| 08:55.29 | AchiestDragon | not as yet |
| 08:55.34 | ka6sox | (or did I miss that) (/me is really sleepy( |
| 08:56.03 | ka6sox | okay I'll look forward to those |
| 08:56.21 | ka6sox | we are looking at 208's for the s3's in 400 sizes? |
| 08:56.52 | ka6sox | XC3S400? |
| 08:57.02 | ka6sox | pqfp? |
| 08:57.08 | AchiestDragon | yes |
| 08:57.22 | ka6sox | okay I'll get a UCF percolating |
| 08:57.41 | ka6sox | same for the GE boards or just the EP? |
| 08:57.50 | AchiestDragon | same |
| 08:57.53 | AchiestDragon | :) |
| 08:57.59 | ka6sox | makes it easire |
| 08:58.00 | ka6sox | er |
| 08:59.04 | ka6sox | we can treat the expansion boards like SRAM and that will make the software easy. |
| 09:00.00 | AchiestDragon | paged sram , theres more ram on the GE board than address space from the arm |
| 09:00.32 | p2-mate | AchiestDragon: is the geep already produced or is it still in development as well ? |
| 09:00.36 | ka6sox | yes...since we are going to have to multiplex A/D we can page it as well. |
| 09:00.57 | AchiestDragon | ka6sox: my girlfriendwants to know if you have six feet for your 6 sox |
| 09:01.19 | AchiestDragon | p2-mate: in development |
| 09:01.31 | ka6sox | AchiestDragon, yes...mine, wife's and Son's |
| 09:03.51 | AchiestDragon | although since the onboard fpga is beeing used to do the addressing for the flash and the expantion bus , the ram on the expantion board's could be paged by boards accross 4 banks of 64mb |
| 09:04.08 | AchiestDragon | with the flash also |
| 09:04.15 | ka6sox | ah.. |
| 09:04.45 | ka6sox | I was hoping we could use a DMA channel to send raw frames out the ethernet |
| 09:05.01 | ka6sox | (from the expansion boards) |
| 09:05.12 | AchiestDragon | that way each expantion card can have a 256mb address space |
| 09:05.27 | AchiestDragon | yes the fpga also has one of the dma chans |
| 09:05.42 | AchiestDragon | the eide and cf cpld has the other |
| 09:05.42 | ka6sox | how many bits of addressing 24 or 26? |
| 09:06.04 | ka6sox | the CF only runs in PIO. |
| 09:06.13 | ka6sox | but Eide can run in DMA |
| 09:06.17 | AchiestDragon | yes |
| 09:07.23 | AchiestDragon | the spi is going to have 2 onboard devices on it |
| 09:07.54 | ka6sox | DDS and RTC? |
| 09:07.58 | AchiestDragon | yes |
| 09:08.18 | ka6sox | which RTC? |
| 09:08.29 | AchiestDragon | not decided yet |
| 09:09.08 | AchiestDragon | the ac97 will probablay be an analog devices device |
| 09:09.31 | ka6sox | you looking for an SO8 devic on the RTC? |
| 09:10.00 | AchiestDragon | so8 to so16 should fit np |
| 09:11.11 | ka6sox | Maxim has some that are pretty nice. |
| 09:11.38 | ka6sox | do we have access to the configuration eeprom onboard of the SoC? |
| 09:12.14 | ka6sox | I'm thinking about where to store the Ethernet MAC and IP Address. |
| 09:12.14 | AchiestDragon | would prefer a generic part that is made by a few manufactures |
| 09:12.23 | ka6sox | okay... |
| 09:13.22 | AchiestDragon | the config eeproms need a way to program them so got to look at that |
| 09:13.49 | ka6sox | k |
| 09:14.15 | ka6sox | I thougth they were done via the ARM jtag..but I wasnt' sure. |
| 09:14.26 | AchiestDragon | i want to be able to ensure that the system can be initaly programmed up using only say a simple parallel port cable /programmer |
| 09:15.20 | AchiestDragon | that will also have a pcb done for it |
| 09:15.23 | ka6sox | if we can get a bootloader that is network savvy then we can load that up and serial into it. |
| 09:15.49 | ka6sox | then load up the kernel and then a rootfs. |
| 09:16.03 | ka6sox | (via tftp or some other faster means) |
| 09:16.04 | AchiestDragon | but theres still the cpld's to program |
| 09:16.20 | AchiestDragon | and the config eeprom for the fpga |
| 09:16.23 | ka6sox | the CPLD and the FPGA can be programmed via the same parallel port. |
| 09:16.37 | ka6sox | the config eeprom can also be done via the wiggler. |
| 09:16.42 | ka6sox | (very small device) |
| 09:17.00 | ka6sox | I'm sure that beewoolie/lennert can get that sorted. |
| 09:18.55 | ka6sox | I'm comfortable that I can get the FPGA and CPLD done |
| 09:19.29 | AchiestDragon | theres an open i/p core for the vga |
| 09:20.08 | ka6sox | lennert did one too for his pong device (via the S3) |
| 09:20.09 | AchiestDragon | i found a print out i had of it yesterday , although not looked though it yet |
| 09:20.48 | lennert | vga is easy :) |
| 09:21.21 | AchiestDragon | the hard bt is going to be deciding what colour scheam is best |
| 09:22.59 | AchiestDragon | the colour dac is = to 8 leveles per colour so 3+3+3 bits and we only have 8 bits if muxing the vram data |
| 09:23.37 | lennert | so take one bit away from green? |
| 09:23.39 | AchiestDragon | but theres mor than enough ram so could use a single 16 bit access per pixel |
| 09:24.09 | ka6sox | I leave that to you guys. |
| 09:24.31 | AchiestDragon | i mean 64mb video ram , and only a requirement for 1.2mb at 1240*1024 in 256 colours |
| 09:25.22 | AchiestDragon | so a 16 bit access would give us 512 colours with the other 7 bits unused , and take 2.4mb ram |
| 09:26.13 | p2-mate | AchiestDragon: is the idea not equip the production boards intended for the jtag/logic analyzer application ? |
| 09:26.29 | lennert | hey p2-mate, didn't know you were here too |
| 09:26.31 | p2-mate | not equip VGA i mean :) |
| 09:27.05 | AchiestDragon | theres alot of space in the fpga so (not saying we should but theres the posibilaty for ) adding windowing and sprites in hardware |
| 09:27.51 | AchiestDragon | the idea is that the ep board should also work as well as we can as a sbc for other uses |
| 09:28.21 | p2-mate | ic |
| 09:28.21 | AchiestDragon | while providing us with the fast bus and dma for the ge board |
| 09:28.54 | AchiestDragon | the ge board has a fpga , ram and ports for the anaizer and jtag tester |
| 09:29.11 | p2-mate | hi lennert |
| 09:30.10 | AchiestDragon | btw we can only have 2 usb ports , theres no spare i/o on the fpga for the extra 2 |
| 09:32.40 | ka6sox | we can always use a hub external |
| 09:32.46 | AchiestDragon | yes |
| 09:33.11 | AchiestDragon | and not as important now we got eide onboard |
| 09:33.23 | ka6sox | yes |
| 09:34.58 | AchiestDragon | :) spaire connector space ,,, ps2 to rear or add audio jacks ? |
| 09:36.24 | ka6sox | audio jacks |
| 09:36.28 | ka6sox | its Geep! |
| 09:36.33 | AchiestDragon | yes |
| 09:40.54 | ka6sox-zzzz | cya in aobut 6hrs. |
| 09:41.04 | AchiestDragon | k , nn |
| 09:41.11 | ka6sox-zzzz | nn |
| 10:08.18 | *** join/#openjtag z3ro___ (n=z3ro@60-234-138-172.bitstream.orcon.net.nz) |
| 10:23.34 | *** join/#openjtag bullet (n=bullet@230.218.62.81.cust.bluewin.ch) |
| 12:50.15 | *** join/#openjtag prpplague (n=billybob@69.73.203.9) |
| 13:00.33 | prpplague | ~seen beewoolie-afk |
| 13:00.39 | purl | beewoolie-afk <n=beewooli@206.124.142.26> was last seen on IRC in channel #openjtag, 5d 7h 27m 11s ago, saying: 'On one device I've looked at, the largest packet is 128 bytes.'. |
| 14:07.19 | *** join/#openjtag bullet_ (n=bullet@104.50.203.62.cust.bluewin.ch) |
| 14:29.20 | *** join/#openjtag bullet (n=bullet@78.53.62.81.cust.bluewin.ch) |
| 14:54.38 | *** join/#openjtag vmaster_ (i=vmaster@p549B5077.dip.t-dialin.net) |
| 15:05.05 | *** join/#openjtag prpplague (n=billybob@72.22.143.178) |
| 17:17.56 | *** join/#openjtag beewoolie-afk (n=beewooli@206.124.142.26) |
| 17:18.09 | beewoolie-afk | lennert: hello |
| 17:18.46 | vmaster | hey beewoolie-afk |
| 17:18.55 | beewoolie-afk | Hey |
| 17:19.26 | beewoolie-afk | I'm really close to having something releasable. |
| 17:19.48 | beewoolie-afk | Another sad discovery is that some of the chips I'm using don't have boundary scan chains. |
| 17:20.02 | beewoolie-afk | Only Armv4 stuff such as the embedded ice chain. |
| 17:20.02 | vmaster | mostly -s cores, i guess |
| 17:20.21 | beewoolie-afk | I don't think so, but I'm not sure. |
| 17:20.28 | beewoolie-afk | Isn't S the synthesizable core? |
| 17:20.31 | vmaster | yes |
| 17:20.45 | beewoolie-afk | We talked about this before and I thought you said that these weren't -s cores. |
| 17:20.49 | beewoolie-afk | The Sharp LH parts. |
| 17:20.52 | vmaster | ah, ok |
| 17:20.54 | vmaster | no, they aren't -s |
| 17:21.22 | beewoolie-afk | I think that one of the problems is that they goofed on their first implementation and didn't setup the external chain properly. |
| 17:21.37 | beewoolie-afk | It latches outputs on the rising edge instead of the falling clock edge. |
| 17:21.44 | vmaster | ah, yeah |
| 17:21.59 | beewoolie-afk | Perhaps they dropped it altogether in the second implementation of the ARM7. |
| 17:22.08 | beewoolie-afk | The ARM9's have external scan chains that work. |
| 17:23.48 | vmaster | ok, seems boundary scan is left out randomly |
| 17:24.06 | vmaster | the sam9261 (9-ejs) has a boundary scan interface |
| 17:24.06 | prpplague | beewoolie-afk: hey hey |
| 17:24.11 | prpplague | beewoolie-afk: whats cookin |
| 17:24.25 | beewoolie-afk | prpplague: Hey. |
| 17:24.36 | beewoolie-afk | prpplague: I send mail about those drivers. |
| 17:24.47 | beewoolie-afk | vmaster: I'm not following you. |
| 17:25.25 | beewoolie-afk | Are you saying that other -S parts don't have boundary scan chains, but the samsung part does? |
| 17:26.23 | vmaster | beewoolie-afk: no. i originally thought that boundary scan was missing at -s parts, because all i've looked at missed it - but as the at91sam9261 (-s core) has one, and the samsungs (not -s) don't, my assumptions proved to be wrong |
| 17:27.14 | ka6sox | morning |
| 17:27.56 | beewoolie-afk | ka6sox: |
| 17:28.02 | beewoolie-afk | hi |
| 17:28.14 | beewoolie-afk | vmaster: I've gotten most of the pieces working. |
| 17:28.34 | beewoolie-afk | The trouble is that there are so many holes in the hardware, it's been hard to make sure that everything is good. |
| 17:28.51 | beewoolie-afk | I've been looking for a part that has both boundary scan and embeddedice. |
| 17:29.04 | vmaster | i have one, the at91rm9200 |
| 17:29.11 | beewoolie-afk | *and* that works with the ftdi2232. |
| 17:29.17 | vmaster | the at91rm9200 |
| 17:29.29 | beewoolie-afk | vmaster: :-) That will help, eventually. |
| 17:29.37 | prpplague | beewoolie-afk: did you get your sd/mmc driver working in apex? |
| 17:29.45 | beewoolie-afk | prpplague: Nope. |
| 17:29.50 | vmaster | i can run the ft2232 at the full 6mhz with the rm9200 |
| 17:29.52 | beewoolie-afk | prpplague: I haven't looked at it in a while. |
| 17:30.02 | beewoolie-afk | vmaster: I haven't had speed problems. |
| 17:30.12 | prpplague | beewoolie-afk: ok, i wasn't sure if you still wanted my patch, but i'll get that ready to submit as well |
| 17:30.36 | beewoolie-afk | vmaster: ATM, I have a part that isn't responding to a IDCODE scan with the ftdi2232. It works with a parallel port dongle and I can use the same code on a different board. |
| 17:31.12 | vmaster | beewoolie-afk: what chip? |
| 17:31.18 | beewoolie-afk | prpplague: I do. I've been working on the JTAG code for some time now. I'm eager to get it to a working state before I start a new job. |
| 17:31.43 | prpplague | beewoolie-afk: understood |
| 17:31.49 | prpplague | beewoolie-afk: i hate leaving a project |
| 17:31.59 | prpplague | beewoolie-afk: takes twice as long to get back into the game |
| 17:31.59 | beewoolie-afk | vmaster: I've got two, the LH7A400 and LH7A40X. I thought that these used to work, but I suspect that I broke something when I was working on the IXP42x code. |
| 17:32.06 | beewoolie-afk | prpplague: Yeah. |
| 17:32.31 | beewoolie-afk | Once this JTAG code is really useful, I'll release it and see if it works for other folks. |
| 17:33.09 | prpplague | beewoolie-afk: cool beans |
| 17:33.40 | beewoolie-afk | vmaster: I know that there is room for wrong code producing OK results on some targets. |
| 17:34.32 | vmaster | beewoolie-afk: heh, yeah, had enough of this |
| 17:34.34 | beewoolie-afk | vmaster: in my parallel port driver, I forgot to assert the TDI data on the low clock pulse. It would assert on the rising pulse with the TMS signal and TCK. One parallel cable worked fine with this. Another did not. |
| 17:34.58 | beewoolie-afk | vmaster: Huh? |
| 17:35.18 | vmaster | beewoolie-afk: my code worked with one device, but not with the other - mostly due to bugs in the code |
| 17:35.38 | beewoolie-afk | That is our job, after all. |
| 17:35.55 | lennert | beewoolie-afk: hello |
| 17:36.00 | beewoolie-afk | lennert: Hey! |
| 17:36.13 | vmaster | hey lennert |
| 17:36.40 | lennert | vmaster: hi :) |
| 17:40.54 | prpplague | hehe, this channels just comes alive when the bee shows up |
| 17:41.03 | lennert | hehe |
| 17:41.06 | lennert | the Big Bee |
| 17:41.13 | lennert | vmaster: you had a nice talk, btw |
| 17:41.26 | lennert | vmaster: you're quite knowledgable about arm cores, for one |
| 17:42.24 | vmaster | lennert: heh, thx. well, i hope it was interesting for some |
| 17:42.36 | lennert | definitely interesting |
| 17:42.47 | prpplague | vmaster: the openocd speech? |
| 17:42.53 | vmaster | prpplague: yeah |
| 17:43.04 | prpplague | vmaster: cool, sounds like it went over well then |
| 17:43.47 | beewoolie-afk | vmaster: did you give a presentation? |
| 17:44.15 | vmaster | beewoolie-afk: yeah, i presented the openocd and arm-debugging in general at the fosdem embedded softwares room |
| 17:44.54 | lennert | the embedded room was disappointing except for one or two talks |
| 17:45.07 | vmaster | heh, yeah, it was |
| 17:45.09 | lennert | there were people from montavista teaching the audience how to best circumvent the gpl, etc. |
| 17:45.20 | vmaster | i left after he said that |
| 17:45.22 | beewoolie-afk | OUCH! |
| 17:45.22 | lennert | btw, in case you didn't know yet, montavista wrote the linux kernel |
| 17:45.30 | beewoolie-afk | OUCH! |
| 17:45.36 | lennert | there's some guy in finland who contributed some bits, but montavista did most of the work |
| 17:45.38 | beewoolie-afk | They are not nice people. |
| 17:45.40 | lennert | oh, and by the way |
| 17:45.46 | beewoolie-afk | I met them at a show here in Seattle. |
| 17:45.59 | lennert | there are 25 million linux phones and 5 million linux PCs, so it's only fair that montavista get to decide what goes into the linux kernel |
| 17:46.19 | lennert | so the next time daniel walker gets flamed on lkml, remember taht |
| 17:46.49 | beewoolie-afk | Huh? |
| 17:47.08 | lennert | daniel walker is some montavista dude that wants to push all kinds of rt crap into the kernel |
| 17:47.39 | beewoolie-afk | OK. I'm assuming that your tongue has poked a hole in your cheek. |
| 17:47.47 | lennert | yeah |
| 17:48.03 | lennert | p2-mate: you didn't miss anything |
| 17:48.14 | p2-mate | otherwise I missed food :) |
| 17:48.17 | lennert | hehe |
| 17:48.20 | lennert | and nice beer |
| 17:48.27 | p2-mate | that as well |
| 17:48.30 | beewoolie-afk | vmaster: are there slides of your talk? |
| 17:48.34 | p2-mate | although I can always drink nice beer :) |
| 17:48.41 | lennert | i ate about six tuna sandwiches on sunday and got really stuffed |
| 17:48.59 | *** join/#openjtag bullet (n=bullet@253.19.79.83.cust.bluewin.ch) |
| 17:48.59 | vmaster | beewoolie-afk: yeah, they're going to be at the fosdem site, but i can put them on a server, too |
| 17:49.25 | lennert | the montavista dude's arguments were totally flawed |
| 17:49.29 | beewoolie-afk | No hurry. |
| 17:49.41 | lennert | like, linus has been wrong in the past, ergo, we can better leave the decisions to montavista |
| 17:50.04 | p2-mate | lennert: this was on the rt talk ? not on the boot time thing ? |
| 17:50.08 | beewoolie-afk | Fortunately, his confidence doesn't affect our ability to use Linux. |
| 17:50.22 | lennert | he described some really crazy setups where the cpu had to service some peripheral ten gazillion times per second, where i figured that you might as well add a $5 cpld and avoid all the complexity |
| 17:50.36 | lennert | p2-mate: i missed the boot time thing (by vitaly wool, right?) |
| 17:51.31 | p2-mate | lennert: yes |
| 17:51.33 | lennert | oh |
| 17:51.58 | lennert | and he also made a snide remark that if you mess up the licensing, harald welte is gonna get you |
| 17:52.08 | p2-mate | lennert: he isn't a very good speaker, but it was a reasonably nice talk |
| 17:52.29 | lennert | i dropped in at the end of vitaly's talk, he was mostly staring at his laptop it seemed |
| 17:53.24 | p2-mate | I liked the lock free data exchange talk |
| 17:53.30 | lennert | yeah |
| 17:53.40 | lennert | although it was kind of disappointing that it applied to UP only |
| 17:54.03 | lennert | still, he mentioned some pretty nice properties (on UP) that i wasn't aware of |
| 17:54.32 | p2-mate | yes. but we still have some time before embedded SMP becomes really mainstream to fix that :) |
| 17:54.40 | lennert | hehe |
| 17:55.03 | lennert | actually, at least one VM kernel patch didn't go into the arm tree because it didn't work on smp |
| 17:55.12 | prpplague | yea MV is struggling to find a place for themselves these days |
| 17:55.26 | p2-mate | arm11 SMP does seem to exist :) |
| 17:55.36 | lennert | rmk has a bunch of arm smp boxes, i think |
| 17:55.52 | lennert | whether those are fpga-based or not i don't know |
| 17:55.57 | p2-mate | is there real arm11 SMP silicon ? |
| 17:55.58 | vmaster | heh, yeah, and if you offer them your soul in return you might actually get an arm11 with docs to play with |
| 17:56.06 | p2-mate | I thought it was still all FPGAs |
| 17:56.14 | lennert | probably fpgas |
| 17:56.21 | lennert | vmaster: yeah :) |
| 17:56.54 | p2-mate | there has been MIPS SMP SoC silicon for a few years now :) |
| 17:56.59 | lennert | we're having lots of fun with the ixp2350 since intel decided to use the armv5 isa and the armv6 mmu |
| 17:57.18 | vmaster | i talked to someone from arm at the e-world in nuremberg, and there might actually be an off the shelf cortex available sometime this year |
| 17:57.19 | lennert | and the armv6 mmu only allows supersection (16M granularity) mappings of physical addresses > 4 GB |
| 17:57.31 | lennert | (the ixp2350 has 33 bit physical address space) |
| 17:57.50 | p2-mate | right. they should use MIPS :) |
| 17:57.57 | lennert | p2-mate: i know of the bcm1250, where you generally just run linux on one core and packet processing on the other one |
| 17:58.07 | p2-mate | lennert: it can do SMP linux too |
| 17:58.08 | lennert | vmaster: a $10000 development kit? :) |
| 17:58.30 | vmaster | mhh, no, like a real core in quantities |
| 17:58.59 | lennert | p2-mate: i have some dual ixp2000 boards but the external bus isn't cache coherent so you have to run two kernels... :) |
| 18:00.52 | p2-mate | right :) |
| 18:01.04 | p2-mate | the 1250 is cache coherent |
| 18:01.16 | p2-mate | and 64bit :) |
| 18:07.07 | p2-mate | would be cool to have openocd support for. no open docs though :( |
| 18:09.02 | lennert | i asked like 5 times what a bcm1250 board would cost.. never even got a reply |
| 18:09.13 | lennert | so i guess the pricing structure isn't open either ;) |
| 18:09.25 | p2-mate | no :( |
| 18:09.34 | p2-mate | debian got some |
| 18:09.41 | lennert | yeah |
| 18:10.04 | lennert | tbm was looking for a fast arm board.. he thought 600mhz was too slow since he has a dual 800mhz mips and he's used to that :) |
| 18:10.14 | p2-mate | :) |
| 18:11.27 | p2-mate | I suggested IOP boards to several debian people already, but they somehow do not seem to get it |
| 18:11.45 | p2-mate | (for fast arm boards) |
| 18:11.52 | lennert | you mean like the allnet nas thingy? i recommended that one to tbm, but he thought it was too slow. |
| 18:11.57 | lennert | i don't know what other iop boards are out there. |
| 18:12.03 | p2-mate | lennert: intel devboards |
| 18:12.11 | lennert | hmmmmm, what do those cost? |
| 18:12.15 | vmaster | hehehhe |
| 18:12.21 | lennert | we have a dual ixdp2850 dev board which cost like $25000 or so |
| 18:12.25 | p2-mate | right :) |
| 18:12.41 | p2-mate | there is one for 500USD iirc |
| 18:12.46 | lennert | last time vince asked intel for a donation they were willing to donate a 200 mhz pxa board |
| 18:12.47 | p2-mate | or was it 1000 USD ? |
| 18:13.03 | lennert | there's a dual ixp2350 pci-e board for $3000 |
| 18:13.05 | p2-mate | 200Mhz PXA I hope :) |
| 18:13.09 | lennert | which is also too pricey |
| 18:13.15 | p2-mate | but still :) |
| 18:13.21 | lennert | yeah, not 200 millihertz :-D |
| 18:13.47 | lennert | (200 millihertz.. that means a pipeline stall takes 1 sec :) |
| 18:13.48 | p2-mate | but intel did pay for all of the wireless at dc5 including someone staying there for 1 week or so |
| 18:14.14 | p2-mate | costs much more then a retail priced 1000USD board :) |
| 18:14.22 | lennert | i think it depends on whether you find the right people within intel |
| 18:14.27 | p2-mate | must be |
| 18:14.33 | lennert | intel is a 100-headed monster |
| 18:15.46 | vmaster | definitely |
| 18:15.59 | p2-mate | devboard prices are also a bit arbitrary |
| 18:16.37 | lennert | arbitrary, yes, very much so |
| 18:18.40 | vmaster | beewoolie-afk: http://openocd.berlios.de/OpenOCD_Fosdem2006_talk.sxi |
| 18:19.27 | p2-mate | vmaster: you are dominic ? |
| 18:19.32 | vmaster | p2-mate: yeah |
| 18:19.37 | p2-mate | ah ok :) |
| 18:20.32 | beewoolie-afk | vmaster: thx |
| 18:22.29 | beewoolie-afk | vmaster: What is sxi? |
| 18:22.34 | lennert | staroffice ? |
| 18:22.50 | beewoolie-afk | I have openoffice, but it doesn't use that extension by default... |
| 18:23.00 | p2-mate | ooo yes |
| 18:23.25 | vmaster | it's openoffice 1.1 |
| 18:23.30 | [g2] | HEY beewoolie-afk's in da house !! |
| 18:23.32 | vmaster | but openoffice 2 opens that just as well |
| 18:23.35 | beewoolie-afk | Must have gotten dumped when I updated. |
| 18:23.41 | beewoolie-afk | [g2]: Hey |
| 18:23.47 | [g2] | howdy :) |
| 18:24.03 | vmaster | i had to use sxi as the laptop my university gave me didn't have ooo2 |
| 18:28.53 | vmaster | http://openocd.berlios.de/OpenOCD_Fosdem2006_talk.odp converted to open document presentation (OOo2) |
| 18:31.36 | beewoolie-afk | vmaster: yeah, it was that OO had been uninstalled in the upgrade. |
| 18:55.09 | prpplague | beewoolie-afk: hey, i sent a response email about the spi and such |
| 18:55.15 | beewoolie-afk | OK. |
| 18:55.18 | prpplague | beewoolie-afk: but i figured i'd tell you here anyway |
| 18:55.31 | beewoolie-afk | :-) |
| 18:55.33 | prpplague | beewoolie-afk: just curious about the spi drivers for apex |
| 18:55.50 | prpplague | beewoolie-afk: till now, i've been putting them in the hardware specific type drivers |
| 18:56.17 | prpplague | beewoolie-afk: do we want to do some generic spi framework? |
| 18:56.50 | prpplague | beewoolie-afk: because i'm gonna have atleast 4 platforms, all with different spi drivers, all gonna be used for different things |
| 18:57.03 | prpplague | beewoolie-afk: sd/mmc, enc28j60, and such |
| 18:57.18 | beewoolie-afk | OK. |
| 18:57.43 | beewoolie-afk | Since I've never been successful in using the MMC, I never worked on the spi driver. |
| 18:59.16 | prpplague | beewoolie-afk: basically my question is, should i just have spi drivre for each platform, which is what i'm doing now, or start work on an api? |
| 18:59.36 | prpplague | beewoolie-afk: i mean its basically like adding the other hardware specific drivers, like serial and timer |
| 19:00.44 | prpplague | beewoolie-afk: the enc28j60 and sd/mmc drivers will be generic, but they will depend on the spi driver for the specific platform |
| 19:01.11 | prpplague | beewoolie-afk: i.e. if the platform has no spi driver, then the enc28j60 and sd/mmc won't be available |
| 19:01.30 | beewoolie-afk | That can be OK. |
| 19:01.39 | beewoolie-afk | What I've been doing is defining a standard name. |
| 19:01.54 | beewoolie-afk | We can have a standard SPI device. |
| 19:02.07 | beewoolie-afk | The region-spec will point to the correct device. |
| 19:02.11 | beewoolie-afk | or chip select. |
| 19:02.57 | prpplague | beewoolie-afk: ok |
| 19:03.19 | prpplague | beewoolie-afk: i'll go with that then |
| 19:17.21 | [g2] | vmaster nice talk! |
| 19:17.25 | [g2] | congrats! |
| 19:17.37 | prpplague | vmaster: yea congrats |
| 19:17.48 | prpplague | its always nice to see someone who works hard get some credit |
| 20:47.00 | prpplague | beewoolie-afk: hey, thoughts on adding a lock/unlock command for flash? |
| 20:47.14 | beewoolie-afk | Right now, the driver does it automatically. |
| 20:47.27 | beewoolie-afk | Or are you *not* talking about the block lock? |
| 20:47.42 | prpplague | beewoolie-afk: more or less a write protect |
| 20:47.53 | beewoolie-afk | There are two levels. |
| 20:48.08 | beewoolie-afk | There is a hardware VPEN line that some boards control elswhere via GPIO. |
| 20:48.16 | beewoolie-afk | The block unlock/lock I do automatically. |
| 20:48.32 | prpplague | beewoolie-afk: right |
| 20:48.40 | beewoolie-afk | I suppose I could add a requirement that users commit to a flash write. |
| 20:49.05 | beewoolie-afk | I am not sure what more we need. |
| 20:49.15 | beewoolie-afk | The user has to enter a nor: prefix to write to flash. |
| 20:49.18 | prpplague | beewoolie-afk: just that i've fucked my kernel twice today by typing in the wrong copy command |
| 20:49.41 | prpplague | beewoolie-afk: you sure of that? |
| 20:49.49 | beewoolie-afk | Duude. Don't fuk the kernel. It's not worth the hassle. |
| 20:50.00 | prpplague | beewoolie-afk: hehe |
| 20:50.34 | beewoolie-afk | What do you imagine the command would be? |
| 20:50.43 | beewoolie-afk | The drivers don't presently have an unlock/lock method |
| 20:51.12 | prpplague | beewoolie-afk: lock nor:256k+640k |
| 20:51.18 | beewoolie-afk | What you're asking for is a two-phase write. |
| 20:51.41 | beewoolie-afk | I believe that the locks are not saved to flash. They're in volatile storage. |
| 20:52.06 | prpplague | beewoolie-afk: yea, i would expect this to be stored in the environ space for apex |
| 20:52.16 | beewoolie-afk | That isn't what I mean. |
| 20:52.31 | beewoolie-afk | That's up to the user. |
| 20:52.44 | beewoolie-afk | Let me think about it. |
| 20:52.53 | prpplague | beewoolie-afk: okie dokie |
| 20:52.56 | beewoolie-afk | The downside is that I don't want to make the base image any larger. |
| 20:53.04 | prpplague | beewoolie-afk: yea |
| 20:53.05 | beewoolie-afk | Adding a command to unlock will make the smallest APEX build larger. |
| 20:53.13 | prpplague | agreed |
| 20:53.24 | beewoolie-afk | So, I may have to add an option to enable the command. Otherwise, it would remain automatic. |
| 20:53.26 | prpplague | it was just a thought |
| 20:53.31 | beewoolie-afk | Thots are good. |
| 20:56.16 | prpplague | i'm looking forward to having the s3c2410 part of apex |
| 20:56.28 | prpplague | hopefully more ppl will be using it |
| 20:56.43 | beewoolie-afk | cool. |
| 21:06.34 | prpplague | beewoolie-afk: i'll make a post about it as soon as the patches are accepted |
| 21:06.46 | prpplague | beewoolie-afk: i need to add support for the littlechips board next |
| 21:06.49 | beewoolie-afk | OK. |
| 21:17.06 | prpplague | beewoolie-afk: i really like apex it really picked up where blob left off |
| 21:17.29 | beewoolie-afk | thanks. |
| 21:18.05 | prpplague | beewoolie-afk: and the debug features are just lovely :) |
| 21:19.20 | prpplague | beewoolie-afk: i've got to find time to dump the rom contents of the pixter-mm and see what they are using for bootloader |
| 21:19.31 | beewoolie-afk | That would be interesting. |
| 21:19.45 | beewoolie-afk | I'd guess that they use the same thing they used in the first rev. |
| 21:31.59 | prpplague | beewoolie-afk: somehow if feel that they didn't as the the 3 rev was based on the LH75411 |
| 21:32.12 | prpplague | beewoolie-afk: the 4th (the pixter-mm) is on the lh79524 |
| 21:32.23 | prpplague | beewoolie-afk: just doesn't seem likely to me |
| 21:32.30 | beewoolie-afk | let me know |
| 21:32.54 | prpplague | beewoolie-afk: will do |
| 21:33.13 | prpplague | beewoolie-afk: i've got a real nasty feeling that they might be using linux and violating GPL |
| 21:33.34 | prpplague | beewoolie-afk: mattel and fisher-price have done so in the past when they use sub-contractors |
| 21:33.47 | prpplague | beewoolie-afk: that way they can disavow knowledge |
| 21:34.18 | beewoolie-afk | apparently, someone from montavista was giving advice on how to circumvent GPL |
| 21:34.31 | prpplague | beewoolie-afk: yea, thats what i heard |
| 21:50.49 | lennert | well, it was stuff like don't write kernel drivers cause then harald welte is going to get you |
| 22:40.12 | z3ro___ | hmm, one of the many reasons I don't like people form companies using the linux kernel (don't get me wrong, the ones who actually abide by the gpl are fine, but many do not) |
| 22:40.34 | z3ro___ | in fact, even the ones who do abide by the gpl usually have to be pushed to do so. |
| 22:41.13 | z3ro___ | which is a real shame, imho... seems that most of them don't realize that gpl is *NOT* public domain. |
| 22:41.25 | lennert | they are starting to realise now |
| 22:41.48 | z3ro___ | some are... but it's still not as good as it should be. |
| 22:41.57 | z3ro___ | there are many companies violating the gpl. |
| 22:44.46 | lennert | yeah |
| 22:45.08 | lennert | claiming that you wrote the linux kernel is one way of circumventing the gpl :P |
| 22:45.40 | z3ro___ | wouldn't supprize me if someone at montavista tried that... |
| 22:45.54 | z3ro___ | "I wrote one kernel module, my kernel! mine!" |