| jonibo | nollan: you are probably using the default config which requires that you telnet into the console | 11:41 |
|---|---|---|
| jonibo | see you or1ksim.cfg file and check if it is setup to use telnet or xterm | 11:41 |
| nollan | jonibo: i checked and channel is xterm | 11:42 |
| jonibo | do you get an xterm window? | 11:42 |
| nollan | yep | 11:42 |
| nollan | same message in that "ok. booting the kernel" | 11:42 |
| jonibo | ok | 11:42 |
| nollan | but i've made some progress on this, i used or32-elf-sim from openrisc toolchain | 11:42 |
| jonibo | what linux version? | 11:42 |
| nollan | now i've downloaded and compiled or1ksim and it now boots, with stack dumps, but still | 11:43 |
| nollan | 3.4 | 11:43 |
| jonibo | from kernel.org? | 11:43 |
| nollan | it's the version that comes from openrisc repo, i think his name was jonas | 11:43 |
| jonibo | good | 11:43 |
| jonibo | so Linux boots for you, but you get a stack dump? | 11:45 |
| jonibo | care to put it into a pastebin? | 11:45 |
| nollan | yes. some folks in #opencores helped me with it and they got the same results | 11:45 |
| nollan | sure, np just wait a sec | 11:45 |
| nollan | jonibo: http://pastebin.com/Dd2D43T3 | 11:51 |
| jonibo | ok, thanks... | 11:52 |
| jonibo | interesting, I'm not seeing that here | 11:52 |
| jonibo | your xterm should still be responsive though... | 11:53 |
| jonibo | is it? | 11:53 |
| nollan | it is | 11:53 |
| jonibo | ok, good | 11:53 |
| jonibo | that's with a defconfig? | 11:53 |
| jonibo | i'll look into it | 11:54 |
| nollan | it dumps loads and then works as normal, tho i haven't exactly stress-tested it or something | 11:54 |
| nollan | yep default config from arch/openrisc/or1ksim.cfg | 11:54 |
| nollan | ok | 11:54 |
| jonibo | thanks | 11:54 |
| nollan | np, thank you! | 11:55 |
| jonibo | stekern, juliusb_: do we consider the DSX bug to be a "acceptable"? | 12:06 |
| jonibo | i don't see it documented anywhere, in the or1200 spec, for example... | 12:07 |
| jonibo | it really is a terrible handicap | 12:07 |
| stekern | DSX bug? | 12:07 |
| jonibo | the DSX bit doesn't get set when it's supposed | 12:07 |
| jonibo | supposed to | 12:07 |
| jonibo | it's supposed to get set when you get an exception in a delay slot | 12:08 |
| jonibo | is it really so difficult to fix? | 12:08 |
| stekern | I don't think it would be, but that's me guessing ;) | 12:09 |
| stekern | I though that was marked as optional in the arch spec, but I can't find it being that now when looking | 12:10 |
| jonibo | funny... that was exactly what I remembered too... why do we remember that as being optional? I can't find it now either | 12:10 |
| stekern | since it's not optional and it's not working, I'd say that qualifies as not-acceptable | 12:18 |
| jonibo | stekern: ok... I think I'll proceed accordingly then | 12:19 |
| jonibo | I hope somebody else chimes in if they have a different opinion... | 12:20 |
| stekern | jonibo: trying to reproduce nollan's bug with a newly built kernel from the head of your tree I get this: http://pastebin.com/mgptaFD6 | 13:37 |
| stekern | with trace on I get this: http://pastebin.com/kXH6e43d | 13:38 |
| stekern | c00047e0 <arch_local_save_flags>: | 13:39 |
| stekern | c00047e0: b5 60 00 11 l.mfspr r11,r0,0x11 | 13:39 |
| stekern | c00047e4: 44 00 48 00 l.jr r9 | 13:39 |
| stekern | c00047e8: a5 6b 00 06 l.andi r11,r11,0x6 | 13:39 |
| stekern | c00047ec <arch_local_irq_restore>: | 13:39 |
| stekern | c00047ec: d7 e1 17 fc l.sw 0xfffffffc(r1),r2 | 13:39 |
| stekern | c00047f0: 9c 21 ff fc l.addi r1,r1,0xfffffffc | 13:39 |
| stekern | c00047f4: b4 80 00 11 l.mfspr r4,r0,0x11 | 13:39 |
| stekern | c00047f8: 9c 40 ff f9 l.addi r2,r0,0xfffffff9 | 13:39 |
| stekern | c00047fc: e0 84 10 03 l.and r4,r4,r2 | 13:39 |
| stekern | c0004800: e0 64 18 04 l.or r3,r4,r3 | 13:39 |
| stekern | c0004804: c0 00 18 11 l.mtspr r0,r3,0x11 | 13:39 |
| stekern | c0004808: 9c 21 00 04 l.addi r1,r1,0x4 | 13:39 |
| stekern | c000480c: 44 00 48 00 l.jr r9 | 13:39 |
| stekern | c0004810: 84 41 ff fc l.lwz r2,0xfffffffc(r1) | 13:39 |
| stekern | c0004814 <or1k_pic_mask>: | 13:39 |
| stekern | gaaah | 13:39 |
| stekern | sorry | 13:39 |
| stekern | that was suppose to go to a pastebin :/ | 13:40 |
| stekern | c0004810 is in arch_local_irq_restore as you can see | 13:41 |
| nollan | :) | 13:44 |
| nollan | stekern: that looks like my output when I used or32-elf-sim | 15:36 |
| jonibo | stekern: thanks... I can't reproduce that warning, for some reason | 15:59 |
| jonibo | it's realated to some core change in Linux RCU (obviously) | 15:59 |
| jonibo | it may go away by itself | 15:59 |
| jonibo | or it may require a tweak to the architecture bits in order to match some expectation in the core code | 15:59 |
| jonibo | i'll look into it this evening | 16:00 |
| -!- o01eg_ is now known as o01eg | 21:45 | |
| jonibo_ | juliusb_: you around? | 22:47 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!