| 00:02.29 | vmaster | beewoolie: if you're looking for something cheap that's known to work, have a look at the olimex arm-jtag (sparkfun distributes to the u.s., iirc) - it's a simple parport wiggler clone, with working nSRST |
| 00:02.45 | beewoolie | vmaster: Cool. |
| 00:10.39 | beewoolie | vmaster: $20. Not a lot of corn. |
| 00:11.50 | vmaster | i doubt any commercial offering could be a lot less |
| 00:12.08 | beewoolie | Indeed. |
| 00:12.39 | vmaster | i have one on my desk, and i know that i'm able to control both nTRST and nSRST separately |
| 00:13.23 | vmaster | which is what you need for xscale debugging - you hold the core in reset while loading the mini i-cache |
| 00:13.33 | beewoolie | That's exactly the issue. |
| 00:16.09 | beewoolie | Oddly, they don't have a schematic for the arm-jtag. There is one for the MSP hip |
| 00:17.28 | vmaster | there's a posting in the sparkfun forum from olimex documenting the arm-jtag connections |
| 00:18.26 | beewoolie | do you have a link? |
| 00:18.49 | vmaster | http://www.sparkfun.com/cgi-bin/phpbb/viewtopic.php?t=2068&highlight=armjtag |
| 00:19.31 | vmaster | there are typos, DB is always DB25, not DB26, but the signals are correct |
| 00:19.59 | vmaster | and NTRST is really nSRST, TRST is nTRST |
| 00:20.37 | beewoolie | That can be deduced from the JTAG pin numbers. Looks good. |
| 00:20.59 | beewoolie | Someone just pointed me to a device called jlink. |
| 00:21.04 | beewoolie | I'm going to check that out, too. |
| 00:21.04 | vmaster | segger |
| 00:21.13 | vmaster | it's proprietary withouting documentation |
| 00:21.20 | beewoolie | Not good. |
| 00:22.03 | beewoolie | Is it the ftdi2322? |
| 00:22.37 | vmaster | I don't think so, as it's faster than the FT2232 ever could be |
| 00:23.04 | beewoolie | Looks like it's windows centric? |
| 00:23.12 | vmaster | yes |
| 00:23.19 | beewoolie | Non starter |
| 00:26.10 | *** join/#openjtag purl (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
| 00:26.10 | *** topic/#openjtag is This is the place to discuss development of Open JTAG tools and Bootloaders - http://www.openjtag.net/ |
| 00:28.01 | vmaster | i'm looking to get an xscale development board, as it's always been on my wish-list for the openocd |
| 00:28.50 | vmaster | but xscale debugging is a whole new world - most of what i've done one the host for arm7/9 targets has to happen on the xscale |
| 00:38.22 | vmaster | I'm off - night everyone |
| 00:38.53 | lennert | vmaster: nite |
| 00:47.42 | [g2] | vmaster nite |
| 00:47.52 | [g2] | I'll catch you tomorrow |
| 00:48.31 | [g2] | lennert beewoolie can I ask you a quick question ? |
| 00:48.44 | beewoolie | as long as I can ask you one. :-) |
| 00:49.54 | [g2] | sure I've got to run in a couple minutes but can ask now |
| 00:51.36 | *** join/#openjtag [g2-lap] (n=g2@cpe-066-057-008-035.nc.res.rr.com) |
| 00:51.41 | [g2-lap] | http://pastebin.ca/38858 |
| 00:52.03 | [g2-lap] | The madwifi driver is oopsing on an address that seems valid |
| 00:52.25 | [g2-lap] | The PCI space is mapped as a 64K region at 0x48000000 |
| 00:52.41 | [g2-lap] | and the address being referenced is 0x48004004 |
| 00:54.44 | [g2] | beewoolie what's your Q ? |
| 00:54.55 | beewoolie | hang on, I'm reading your Q |
| 00:56.12 | beewoolie | It could be that those addresses are PCI mapped and that it only goes for 0x4000 bytes |
| 00:56.25 | beewoolie | An lspci would tell you. |
| 00:57.02 | beewoolie | That would be my guess. |
| 00:57.06 | beewoolie | The error is bonafide |
| 00:57.16 | beewoolie | It is saying that the descriptor at that address is zero. |
| 00:57.56 | beewoolie | Oh. I see. |
| 00:57.58 | beewoolie | You have that. |
| 00:58.50 | beewoolie | Also remember that the memory reported by PCI is physical. |
| 00:59.04 | beewoolie | The read is a virtual address read. |
| 00:59.14 | beewoolie | There isn't a guarantee that they'll be identity mapped. |
| 00:59.24 | beewoolie | Usually, they aren't, but I'm not so familiar with PCI. |
| 01:00.27 | beewoolie | My question is easier. |
| 01:00.42 | beewoolie | I haven't been able to find Harvold's router distro. |
| 01:00.48 | beewoolie | Do you have a keyword or a link? |
| 01:03.15 | *** join/#openjtag ka6sox-office (n=tking@netblock-66-159-209-52.dslextreme.com) |
| 01:08.23 | [g2-lap] | https://svn.gnumonks.org/trunk/grouter/ |
| 01:08.31 | [g2] | beewoolie how's that ? :) |
| 01:08.41 | beewoolie | sweet |
| 01:09.08 | beewoolie | googling grouter got a lot of tile stuff |
| 01:09.24 | [g2-lap] | http://gnumonks.org/~laforge/weblog/2005/ |
| 01:09.33 | [g2-lap] | search that page for grouter / linwrap |
| 01:09.43 | [g2-lap] | bbl |
| 01:12.59 | beewoolie | thanks |
| 01:13.16 | [g2] | np |
| 01:13.26 | beewoolie | He says that it is x86 only. |
| 01:23.42 | *** join/#openjtag vmaster_ (i=vmaster@p549B4EAA.dip.t-dialin.net) |
| 02:34.53 | [g2-lap] | re |
| 02:36.01 | [g2-lap] | beewoolie I think we'd be able to take those apps and the configs and run them pretty easily |
| 02:44.24 | *** join/#openjtag ByronT (n=byron-po@nslu2-linux/ByronT) |
| 03:50.35 | *** join/#openjtag ka6sox (n=ka6sox@nslu2-linux/ka6sox) |
| 08:05.27 | beewoolie | vmaster_: I'm looking at the LH 79520 datasheet again to see how I missed the fact that it's an r3 part. |
| 08:05.35 | beewoolie | The file I'm looking at: http://document.sharpsma.com/files/tec_datasheet_lh79520.pdf |
| 08:05.48 | beewoolie | If it says r3, it's well hidden. Is there something else you were reading? |
| 08:07.09 | beewoolie | Ah, now I see. It's burried in the description. |
| 11:17.33 | lennert | [g2]: the driver is accessing physical addresses as virtual addresses |
| 11:17.37 | lennert | [g2]: this is ixp4xx, i guess? |
| 12:34.02 | *** join/#openjtag bullet (n=bullet@83.217.203.62.cust.bluewin.ch) |
| 12:53.50 | *** join/#openjtag eFfeM (n=eFfeM@ip545787f8.direct-adsl.nl) |
| 12:53.57 | *** part/#openjtag eFfeM (n=eFfeM@ip545787f8.direct-adsl.nl) |
| 14:02.28 | *** join/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 14:06.32 | *** join/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 15:34.16 | *** join/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 17:07.50 | *** join/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 18:14.59 | *** join/#openjtag bullet (n=bullet@83.217.203.62.cust.bluewin.ch) |
| 19:13.23 | *** join/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 19:47.08 | *** part/#openjtag velinp (n=velinp@c88-163.cable.netissat.bg) |
| 20:03.49 | ka6sox | see velinp bounce |
| 20:09.28 | lennert | hehe |
| 20:15.03 | ka6sox | lennert, the glue logic is mostly gone...now 3 cpld's |
| 20:18.48 | lennert | cool |
| 20:59.11 | *** join/#openjtag ByronT (n=byron-po@nslu2-linux/ByronT) |
| 21:38.17 | [g2] | lennert yeah its on the Loft and that's what I thought too about access |
| 21:40.32 | lennert | [g2]: ? |
| 21:40.55 | lennert | oh, i see |
| 21:41.03 | [g2] | lennert sorry :( I was responding to you comment from 10 hrs ago |
| 21:41.07 | lennert | yeah, it's trying to access iomap cookies directly, probably |
| 21:41.11 | lennert | or something like that |
| 21:41.13 | lennert | do you have a backtrace? |
| 21:41.17 | lennert | can you find the offending function? |
| 21:41.17 | [g2] | sure |
| 21:41.32 | lennert | which kernel version? |
| 21:42.08 | [g2] | 2..6.1 |
| 21:42.10 | [g2] | 2..6.15 |
| 21:42.32 | [g2] | wow stellar typing from [g2] today |
| 21:43.05 | [g2] | this is the lastest Atheros madwifi-ng driver |
| 21:43.34 | lennert | s/lastest/latest/ :) |
| 21:43.47 | [g2] | I'm guessing either something isn't mapped right on the Loft or they are expecting something (like PCI support) to be handled differently like on x6 |
| 21:43.53 | [g2] | s/x6/x86/ |
| 21:45.15 | *** join/#openjtag [g2-lap] (n=g2@cpe-066-057-008-035.nc.res.rr.com) |
| 21:45.33 | [g2-lap] | http://pastebin.ca/38936 |
| 21:45.55 | [g2-lap] | ath_hal_reg_read is the offender |
| 21:46.04 | lennert | right |
| 21:46.25 | [g2-lap] | it's the instruction right after the "Before" printk |
| 21:46.39 | lennert | well |
| 21:46.55 | lennert | it might be doing ioremap() and then dereferencing the returned cookie |
| 21:47.05 | lennert | i'm not exactly sure |
| 21:47.22 | lennert | it's certainly doing something wrong |
| 21:47.26 | [g2] | nod |
| 21:49.10 | lennert | have you asked the atheros people? |
| 21:49.39 | [g2] | there's an open ticked on madwifi |
| 21:50.50 | lennert | ok |
| 21:52.36 | [g2-lap] | old_mmap(NULL, 223692, PROT_READ|PROT_WRITE, MAP_PRIVATE,Unable to handle kernel paging request at virtual address 48004004 |
| 21:52.37 | [g2-lap] | <PROTECTED> |
| 21:52.37 | [g2-lap] | init_module(".ELF...a", 0x369cc) = 0pgd = c2038000 |
| 21:52.37 | [g2-lap] | munmap(0x40142000, 223692) = 0 |
| 21:52.37 | [g2-lap] | close(3) [48004004] *pgd=00000000 = 0 |
| 21:53.24 | [g2-lap] | lennert does that mean the mmap is failing ? |
| 21:53.32 | [g2-lap] | old_mmap that is |
| 21:53.57 | [g2-lap] | that was from an strace of the modprobe |
| 21:54.12 | lennert | well, it's crashing in init_module() |
| 21:54.24 | lennert | but.. the oops is shown before strace has a chance to print old_mmap() |
| 21:54.32 | lennert | so that's why it looks like it's crashing in old_mmap() |
| 21:54.52 | lennert | init_module calls the module init function |
| 21:56.33 | [g2-lap] | I know the line of code that's failing I don't know why the memory map isn't properly setup |
| 21:57.51 | [g2-lap] | I'd guess that the ath_pci_probe or the pci_probe would be responsible for mapping the PCI address properly |
| 21:58.10 | [g2-lap] | the address is in the PCI config region of the card |
| 21:58.27 | [g2-lap] | lspci -v |
| 21:58.27 | [g2-lap] | 0000:00:02.0 Ethernet controller: Atheros Communications, Inc. AR5212 802.11abg NIC (rev 01) |
| 21:58.27 | [g2-lap] | <PROTECTED> |
| 21:58.27 | [g2-lap] | <PROTECTED> |
| 21:58.27 | [g2-lap] | <PROTECTED> |
| 21:58.28 | [g2-lap] | <PROTECTED> |
| 21:58.49 | [g2-lap] | we are getting quite far |
| 21:58.59 | [g2-lap] | I don't know if there's a race |
| 21:59.36 | [g2-lap] | as that area starts out disabled and the driver doesn't opps on the second modprobe, but things still aren't fully setup properly |
| 21:59.44 | [g2-lap] | things for the driver that is |
| 22:09.36 | lennert | can you check what that accessor function is doing? |
| 22:11.02 | [g2] | lennert sure |
| 22:13.41 | [g2] | lennert what do I need to do to turn on KERN_INFO message ? |
| 22:30.03 | [g2] | dinner bbiab |
| 22:37.09 | lennert | well |
| 22:37.13 | lennert | you should be able to see them in 'dmesg' |
| 22:37.16 | lennert | (i think) |
| 22:52.32 | [g2] | nod. the printk is right _after_ the ath_hal_probe (which we are oopsing in before the return) |