| 05:12.35 | *** join/#openjtag purl (ibot@apt.bot.TimRiker.active.supporter.pdpc) |
| 05:12.35 | *** topic/#openjtag is This is the place to discuss development of Open JTAG tools and Bootloaders |
| 05:35.02 | *** join/#openjtag rwhitby-away (~rwhitby@rwhitby.nslu2-linux) |
| 06:24.32 | *** join/#openjtag qbot (qbot@netblock-66-159-209-60.dslextreme.com) |
| 06:26.01 | *** join/#openjtag nslu2-log (dyoung@netblock-66-159-209-60.dslextreme.com) |
| 07:23.01 | *** join/#openjtag qbot (qbot@netblock-66-159-209-60.dslextreme.com) |
| 07:24.03 | *** join/#openjtag nslu2-log (dyoung@netblock-66-159-209-60.dslextreme.com) |
| 07:39.20 | *** join/#openjtag qbot (qbot@netblock-66-159-209-60.dslextreme.com) |
| 07:40.03 | *** join/#openjtag nslu2-log (dyoung@netblock-66-159-209-60.dslextreme.com) |
| 12:53.33 | *** join/#openjtag [g2] (~g2@g2.nslu2-linux) |
| 14:32.30 | *** join/#openjtag rwhitby-away (~rwhitby@rwhitby.nslu2-linux) |
| 14:47.11 | *** join/#openjtag prpplague (~dave@mailhost.amltd.com) |
| 17:07.01 | *** join/#openjtag kas11 (~kas11@193-63.73-24.tampabay.res.rr.com) |
| 17:16.39 | *** join/#openjtag beewoolie-away (~beewoolie@florence.buici.com) |
| 17:23.52 | ka6sox-away | morning kas11 beewoolie-away |
| 17:24.01 | beewoolie-away | morning. |
| 17:24.01 | ka6sox-away | cya from the mine in a few |
| 17:24.27 | kas11 | Hi ka6sox-away ;) |
| 17:24.35 | kas11 | k |
| 17:25.39 | ka6sox-away | headed into the office. |
| 17:25.52 | beewoolie-away | go get 'em |
| 17:27.07 | [g2] | morning |
| 17:35.12 | beewoolie-away | [g2]: have any insights into the ehci driver? |
| 17:39.42 | [g2] | beewoolie-away, not yet |
| 17:39.48 | [g2] | I got a good sleep |
| 18:21.21 | [g2] | beewoolie-away, I did scratch down some notes at http://www.nslu2-linux.org/wiki/OpenSlug/EhciIssue |
| 18:26.04 | beewoolie-away | [g2]: I added a couple of comments. |
| 18:35.54 | *** join/#openjtag prpplague (~dave@mailhost.amltd.com) |
| 18:47.35 | kas11 | did anyone look into nslu2_pci_map_irq yet?...dropping INTD and changing to mod 3 looks wrong..and I was the moron responsible. |
| 18:50.56 | kas11 | sorta depends on bus and slot...stuff a printk in before return to see what the inputs are and what gets mapped |
| 18:57.35 | kas11 | make than slot and pin |
| 19:03.07 | [g2] | beewoolie-away, thx for the note in the wiki |
| 19:03.15 | prpplague | beewoolie-away: lo |
| 19:03.28 | beewoolie-away | prpplague: bi |
| 19:03.36 | [g2] | kas11, I'd like to check that in a bit |
| 19:04.10 | prpplague | beewoolie-away: whats cookin |
| 19:05.02 | beewoolie-away | prpplague: little things. Some small bug fixes in apex. The ext2 driver is almost done. |
| 19:05.12 | prpplague | lovely |
| 19:05.22 | prpplague | beewoolie-away: still tweaking my mmc driver |
| 19:06.12 | beewoolie-away | prpplague: I'm gonna have to implement some kind of driver chaining soon in order to make these fs drivers viable. |
| 19:10.32 | prpplague | beewoolie-away: yea, i see that there is considerable fat stuff in the drv-cf.c |
| 19:11.21 | beewoolie-away | prpplague: I don't think any of it is used anymore. |
| 19:11.35 | beewoolie-away | prpplague: the first implementation put all of the code in the cf driver, but I've pulled it all out. |
| 19:11.36 | prpplague | beewoolie-away: yea, i've yanking most of it |
| 19:11.49 | prpplague | beewoolie-away: i think alot of it was just commented out |
| 19:45.30 | kas11 | [g2-away] nobody cared comes from do_level_IRQ iirc...at certain places in PCI code, 4 slots and ints is assumed...somewhere this needs to be filtered for the NSLU2 since we only have 3 ints...and no physically defined slots |
| 19:49.55 | beewoolie-away | kas11: I don't see how that can cause the nobody-cared message. |
| 19:50.36 | beewoolie-away | kas11: My understanding is that the error comes from the fact that there is an interrupt triggered and detected by the kernel, but the irq handler has nothing to do. |
| 19:50.41 | kas11 | 0,1 and 2 are valid indicies into the table...a bogus index is readily generated by slot or pin out of our bounds |
| 19:51.00 | beewoolie-away | kas11: When I've seen these before, it is caused by a race in the driver. |
| 19:51.16 | beewoolie-away | kas11: I still don't see how that matters. |
| 19:51.46 | beewoolie-away | kas11: how is a bogus index generated? |
| 19:51.55 | kas11 | iic...I just saw something in the code that looked problematic...but may have nothing to do with that problem |
| 19:53.16 | kas11 | slot and pin seem to be constants of 4...that won't work since only 3 ints are available for the PCI bus on the nslu2 |
| 19:53.21 | beewoolie-away | kas11: It seems that an interrupt is being asserted for irq26, the ehci vector, that doesn't match a valid event. |
| 19:53.43 | beewoolie-away | kas11: Sure, but if nothing generates the fourth interrupt, it ought not matter. |
| 19:54.29 | beewoolie-away | kas11: Unless, for some reason, that interrupt is mapping to the ehci vector *and* it is generating spurrious interrupts. |
| 19:54.50 | kas11 | but what if something is assigned to what would normally be INTD?...higher levels assume that would be valid...*shrug* |
| 19:55.55 | beewoolie-away | If the intd pin has been reassigned, sure that could be an issue. That may account for the spurrious interrupt. I'm skeptical that it is an issue because we only see the nobody-cared messages when the USB driver is under load. |
| 19:56.43 | kas11 | makes me wonder if someone is trying for an extra irq for something under load |
| 19:57.00 | beewoolie-away | What is your hypothesis? |
| 19:58.13 | kas11 | the code is wrong anyhow...may have nothing to do with anything but what would normally be valid a pin and/or slot will generate a bad mapping |
| 19:59.00 | kas11 | that code being specific to the NSLU2... |
| 20:00.00 | kas11 | someone needs to put a printk there and make sure that it isn't called later when the load occurs... |
| 20:01.02 | kas11 | simple enough to do if you are setup to build an openslug kernel...unfortunately, I'm not |
| 21:03.24 | *** join/#openjtag ka6sox-office (~tking@netblock-66-159-209-52.dslextreme.com) |
| 21:10.08 | *** join/#openjtag prpplague (~dave@mailhost.amltd.com) |
| 21:21.11 | *** join/#openjtag prpplague (~dave@mailhost2.amltd.com) |
| 21:57.35 | [g2]-away | ka6sox-away, ping do you know where FatSlug is ? |
| 21:58.02 | ka6sox-office | I'm packing up your now. |
| 21:58.11 | ka6sox-office | (the other one) |
| 22:00.02 | [g2]-away | cool thx |
| 22:18.35 | dyoung | IIRC, INTD is reassigned to power. |
| 23:05.39 | beewoolie-away | dyoung: I'm still not clear how this difference in the number of interrupt pins could have any effect on the kernel. |
| 23:06.29 | beewoolie-away | dyoung: If the feature isn't present, and the PCI controller doesn't generate bogus interrupts for the unused pin, then there should be no issue. |
| 23:07.16 | beewoolie-away | dyoung: BTW, I released a version of APEX with an ext2 filesystem driver. We're not quite all of the way there, but that is the last hard part. |
| 23:09.30 | [g2] | beewoolie-away, I think error just happend on flash during a "rm" command |
| 23:10.01 | beewoolie-away | [g2]: have you had a chance to try bonnie? |
| 23:10.08 | [g2] | no |
| 23:10.18 | [g2] | :( |
| 23:10.33 | dyoung | beewoolie-away, I dont understand either. A lot of this PCI stuff is too much for my feeble brain. |
| 23:11.00 | [g2] | actually thing relating to the PCI is the suspend/resume stuff |
| 23:11.20 | [g2] | I tried running without CONFIG_PM but that seems busted |
| 23:12.21 | [g2] | beewoolie-away, how big is your flash stick ? |
| 23:12.37 | beewoolie-away | The reason that I think bonnie is an interesting test is that we can determine if the throughput is reasonable. |
| 23:12.56 | [g2] | Ok it does occur on flash :) |
| 23:12.57 | beewoolie-away | [g2]: if it is, we can assume, for the time being, that the lost interrupt is innocuous. |
| 23:13.12 | beewoolie-away | [g2]: I see that you are getting errors on flash. |
| 23:13.21 | [g2] | yes |
| 23:13.28 | beewoolie-away | [g2]: The largest flash I've got is 256MiB. |
| 23:13.29 | [g2] | not as often but yes |
| 23:13.40 | [g2] | is it empty ? |
| 23:13.50 | beewoolie-away | [g2]: It probably could be. |
| 23:14.17 | [g2] | Ok it happens on flash and hd with ext2 and ext3 |
| 23:14.19 | beewoolie-away | I'll tell you what, I'll see about putting openslug on one of my slugs. |
| 23:14.49 | beewoolie-away | Let me pull the latest from oe and from your bk repo and see what happens. |
| 23:15.01 | [g2] | Ok that's what I'm running |
| 23:15.11 | [g2] | I pulled about an hour ago |
| 23:16.46 | beewoolie-away | something seems to be odd with the slug repo. |
| 23:17.03 | dyoung | hehe is it a little sluggish? |
| 23:17.28 | beewoolie | I'm a little scared of trying to ^c. |
| 23:17.39 | beewoolie | Ah, there it goes. |
| 23:18.58 | beewoolie | ...it's in super-slug-slo-mo |
| 23:21.06 | beewoolie | ...building |
| 23:21.52 | [g2] | dd if=/dev/zero of=zero bs=1M count=240 |
| 23:21.57 | [g2] | and rm zero |
| 23:22.00 | [g2] | both do it |
| 23:22.09 | [g2] | not always but often |
| 23:22.40 | [g2] | mount -t ext2 /dev/sdb1 /media/cf |
| 23:22.47 | [g2] | and cd /media/cf |
| 23:23.17 | [g2] | kids need food, Mr. Mom's gotta run for a bit |