| --- Log opened Wed Mar 09 00:00:26 2016 | ||
| stekern | shorne: yes, that seems like a completely redundant thing | 01:04 |
|---|---|---|
| stekern | ...but looking at that, I see that I introduced a bug when I added SMP support | 01:04 |
| shorne | stekern: thanks for confirming about r4, I traced it back to existing in the inital commit so not sure | 04:24 |
| shorne | In terms of SMP, i havent been looking at it since I am pretty sure I have it disabled... double checking | 04:24 |
| shorne | hmm, maybe its on | 04:28 |
| shorne | no, I have swa lwa disabled, so its not on | 04:32 |
| shorne | ok been working on debugging my boot failure, not sure I used to get further | 06:21 |
| shorne | but definitely now its failing on head.S line 1067 | 06:22 |
| shorne | https://github.com/openrisc/linux/blob/master/arch/openrisc/kernel/head.S#L1067 | 06:22 |
| shorne | the flag is set in SR, but the bf is not jumping | 06:22 |
| shorne | I get a bad tlb | 06:23 |
| shorne | itlbw0mr2 0x384001 , itlbw0tr2 0x40384050 | 06:23 |
| shorne | when we return from the boot_itlb_miss_handler, it goes to 0x200 (bus excp) | 06:25 |
| shorne | because of that bad tlb entry | 06:25 |
| shorne | Ill see if I can write some test code to test the bf instruction | 06:26 |
| shorne | nevermind, it seems like it might be a debug issue, if I single step it doesnt work, but if I conitnue thorugh the miss_handler I get | 07:35 |
| shorne | itlbw0mr2 0x384001, itlbw0tr2 0x384050 | 07:35 |
| shorne | ok, but back the the last issue, on the jump to or32_early_setup | 08:02 |
| shorne | the code jumps to just past 300mb (I dont have the match ram) then I get a bus error | 08:03 |
| shorne | 0x384174: l.jalr r24 | 08:04 |
| shorne | 0x384178: l.nop 0x0 | 08:04 |
| shorne | 0xc038b190 in or32_early_setup () | 08:04 |
| shorne | 0x00000200 in ?? () | 08:04 |
| shorne | Its not going into the tlb_miss_hanlder for that 0xc038.. jump | 08:05 |
| shorne | maybe this debug interface is buggy | 08:05 |
| stekern | shorne: I wouldn't bet that stepping through the mmu enable is bullet proof | 12:27 |
| stekern | shorne: let me dust off my de0-nano and see if I can reproduce your issue | 12:38 |
| stekern | de0-nano free of dust, let's see if I can remember how to build and run stuff still | 14:49 |
| olofk | p1oooop: The plusargs warning is nothing to worry about. I just haven't migrated all the cores in orpsoc-cores yet, since I wanted to give people some time to update their FuseSoC versions | 16:06 |
| olofk | mithro: I synthesized mor1kx for an iCE40 a while ago. That was with Lattice's iCEcube2 tools (using synplify for synthesis), but it seemed to work. Didn't have any board to test it with though | 16:07 |
| stekern | why am I not getting any uart output... | 16:09 |
| shorne | stekern: the uart pins were changed some time ago | 16:40 |
| stekern | yeah, I have them at a completely different place, but I have changed them | 16:41 |
| shorne | do you have gdb up? | 16:41 |
| stekern | yes | 16:42 |
| shorne | does the program counter get passwed where I was having issues? | 16:43 |
| stekern | well, I haven't got to try that, since I don't get any uart output | 16:43 |
| shorne | oh, how are you testing your uart output? you have a separate program for that? | 16:44 |
| stekern | it's just a printf("hello world") | 16:45 |
| stekern | ...+ some led blinking | 16:45 |
| stekern | so the program is running... | 16:45 |
| shorne | I see you have the uart as the uart16550 from orpsoc-cores/de0_nano git? | 16:48 |
| shorne | TX <-> GPIO_2[1] (Pin 6, bottom header) (RX at pin 5) | 16:48 |
| shorne | and uart offset=0x90000000 | 16:49 |
| stekern | yes, but I have changed the pins to what I have connected | 16:49 |
| shorne | I see | 16:50 |
| stekern | and it works with some old image | 16:50 |
| stekern | bah, I added a signaltap to uart tx and now it works... | 17:18 |
| shorne | ah... no chance debugging : | 17:22 |
| shorne | If I boot an old image (downloaded from the wiki) it gets a git further... | 17:24 |
| shorne | CPU: 0 PID: 0 Comm: swapper Tainted: G B 3.11.0-rc3-10831-gd706305-dirty #306 | 17:24 |
| shorne | Stack dump [0xc01ffef4]: , sp + 00: 0xc01ffef4 | 17:24 |
| shorne | hmm, let me see what is different in this image that it can get paste the immu part | 17:25 |
| shorne | maybe this doesnt have a tlb flush stuff and thats about it | 17:25 |
| stekern | try this: https://www.dropbox.com/s/bi5vx8kmqnjdldx/vmlinux-de0_nano | 17:25 |
| stekern | or was that the one you tried? | 17:25 |
| stekern | I'm able to boot that on my board now | 17:26 |
| shorne | yup exactly that one | 17:26 |
| shorne | with no errors? | 17:26 |
| stekern | yes | 17:26 |
| shorne | hmm, I get swapper errors, but at least it means the kernel is mostly booting | 17:26 |
| stekern | I'm loading directly with openocd though | 17:26 |
| shorne | just halt, load_image, reset? | 17:27 |
| stekern | reset | 17:27 |
| stekern | and then | 17:27 |
| stekern | halt; load_image /home/stefan/Downloads/vmlinux-de0_nano; reg r3 0; reg npc 0x100; resume | 17:27 |
| shorne | ok | 17:27 |
| shorne | let me try exactly that | 17:27 |
| shorne | hmm, with that last line on serial console is "Console: colour dummy device 80x25" | 17:30 |
| shorne | let me clear the console and double check | 17:30 |
| shorne | yeah, still getting swapper tained issue when running in openocd | 17:32 |
| shorne | I have a spanion 8mb flash on my board which seems to be (newer) | 17:33 |
| shorne | I wonder if that has anything to do with it? | 17:33 |
| stekern | Linux shouldn't be touching the flash in the default config | 17:34 |
| shorne | yeah, thats what I thought | 17:34 |
| shorne | it is defined in the latest dts | 17:34 |
| shorne | anyway, this is the dropbox image | 17:35 |
| stekern | I wonder if there are other differences on these new boards, maybe the memory settings needs to be tweaked differently | 17:36 |
| shorne | maybe, but at least I can reproduce it working partially on the old iamge, and not on my new image | 17:39 |
| shorne | I can compare the assembly | 17:39 |
| shorne | its probably less than 100 lines before it fails | 17:39 |
| stekern | mmm, but it's weird that the old image doesn't work fully for you | 17:39 |
| stekern | I can boot a 4.2 image too | 17:41 |
| shorne | yeah, I can look into that | 17:41 |
| shorne | let me try with my latest image 1 more time | 17:41 |
| stekern | if you pass me your image that doesn't work at all, I can try that as well | 17:41 |
| shorne | http://shorne.noip.me/uploads/vmlinux-shorne | 17:45 |
| shorne | I wouldnt' trust it much, now I am getting past the bus error in it | 17:46 |
| shorne | anyway, I need to get going thanks a lot for helping | 17:47 |
| stekern | me too, way past bedtime here | 17:47 |
| stekern | but your image boots here | 17:48 |
| shorne | ah... cool | 17:49 |
| shorne | thats 4.5 :) | 17:49 |
| shorne | good night, I have some hope | 17:50 |
| --- Log closed Thu Mar 10 00:00:27 2016 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!