| 00:16.11 | hno | odd.. semi-blinded still. Something stomps on the kernel log buffer on reset |
| 00:38.16 | hno | worked around, but nothing interesting logged. |
| 01:11.53 | hno | Heh, 3.3 fails hard very early if CONFIG_DEBUG_LL is not set. Crashes in DRAMC_get_dram_size() as the DRAM controller is not yet mapped in. |
| 01:12.26 | hno | wonder how that worked in older kernels.. |
| 01:14.03 | hno | It didn't... CONFIG_DEBUG_LL is set in all configs. |
| 02:17.00 | *** join/#arm-netbook hp__ (~hp_@p5B38B9B4.dip.t-dialin.net) |
| 03:35.32 | *** join/#arm-netbook QingPei (~qingpei@123.118.96.98) |
| 03:52.16 | CIA-91 | rhombus-tech: lkcl master * r2cfa3a35a4fc /allwinner_a10/orders/ (.mdwn johang.qf__64__gmail.com.mdwn): order processing cleanup |
| 04:33.06 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 04:33.26 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 06:11.11 | hno | Still at loss on why things crashes. |
| 06:13.22 | hno | or.... f1c00000 + 512MB is out of bounds. Maybe that is related |
| 06:13.46 | hno | seems it do not really manage to clean up the initial temporary mapping. |
| 06:31.08 | hno | Ah, it's the physical location that is limited to 512MB. 01c00000 + 512MB is ok. |
| 07:02.59 | *** join/#arm-netbook zenitraM (~zenitraM@2607:f358:1:fed5:11:0:1018:14) |
| 09:29.39 | hno | crosses fingers |
| 09:31.00 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 09:35.12 | mnemoc | hno: success? :) |
| 09:49.27 | hno | :( |
| 09:49.39 | hno | But it feels close. |
| 09:50.27 | hno | It's crashing when trying to set up the first I/O map in the page table, disliking the status of the page table. |
| 09:52.13 | hno | at least I am not blinded any more |
| 09:54.10 | mnemoc | added a few hundred printascii calls? :) |
| 09:56.46 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 10:11.47 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 10:37.06 | hno | mnemoc, printk works fine. |
| 10:41.28 | mnemoc | uhm... do you have unpushed commits? 3.3 halts on me right after initializing earlycon0 |
| 11:06.35 | hno | only a couple of ugly hacks to unblind me, and lots of printk. |
| 11:08.32 | hno | mnemoc, it has always worked up to Memory policy: ECC disabled, Data cache writeback for me. Remeber http://fpaste.org/MfU6/ |
| 11:09.31 | hno | unless CONFIG_DEBUG_LL is disabled, then it crashes in [ 0.000000] Lichee System fixup |
| 11:09.54 | mnemoc | paste expired... what was on it? |
| 11:10.25 | hno | the first successful earlyprintk log. It's still viewable if you enter the chapta. |
| 11:11.29 | mnemoc | for some reason I never reach the Memory Reserved: |
| 11:12.34 | hno | something different in uboot perhaps? |
| 11:12.35 | mnemoc | may be related to not having merged the gps allocation thingy |
| 11:12.51 | mnemoc | I'm using your uboot |
| 11:13.05 | hno | I don't think that's related. But I can try. |
| 11:13.07 | mnemoc | but with my old env |
| 11:13.22 | hno | have not tried the latest default env at all. |
| 11:13.36 | hno | but should work.. |
| 11:13.59 | mnemoc | I'll play tonight... and start adding printk()s all around :) |
| 11:15.16 | hno | I have a little hack to dump printk messages in two memory regions, that way I can read them from u-boot after reset. |
| 11:15.40 | hno | for some strange reason the normal printk ring buffer gets trashed on reset. |
| 11:15.55 | mnemoc | "security" :) |
| 11:16.29 | hno | it's not wiped. Only partially overwritten by zeroes, like 8 bytes every 60 bytes or so. |
| 11:16.48 | hno | sufficient to make messages almost unreadable. |
| 11:17.31 | hno | and it does not seem to be address dependent which is the really odd thing. |
| 11:18.15 | hno | I mean, i can understand if u-boot were to overwrite some memory, but then it should work to move the ring buffer to another address, but then that address gets trashed instead. |
| 11:21.38 | mnemoc | odd |
| 11:22.22 | hno | http://fpaste.org/MQO2/ is the extrta printk buffer patch. grep __log_buf2 System.map to find the address, only need chaning from 0xcXXX to 0x4XXX to access from u-boot. |
| 11:22.35 | hno | print with md command, or similar. |
| 11:29.20 | hno | mnemoc, without those 3 lines it crashes at bootconsole [earlycon0] enabled |
| 11:36.03 | *** join/#arm-netbook penguin42 (~dg@tu006.demon.co.uk) |
| 11:40.10 | hno | Looking at those three lines it looks like it's actually mapping the script.bin area, not reserving some memory. |
| 11:47.11 | hno | but it's 0x8000 off... odd. |
| 11:49.57 | hno | No, that's masked away. |
| 12:33.28 | hno | so it's really mapping script.bin into the kernel address space. |
| 13:12.24 | mnemoc | hno: very GPSish :p |
| 13:34.07 | A1Sheds | no student has stepped up to the RTEMS STM32 project |
| 13:34.23 | A1Sheds | nobody wants to be a part of history |
| 13:34.27 | A1Sheds | oh well |
| 13:43.56 | mnemoc | hno: can that memory be mapped in the fixup callback instead of head.S? |
| 13:59.42 | CIA-91 | rhombus-tech: Wojciech master * r4dd08558d8db /allwinner_a10/orders/.mdwn: |
| 14:01.36 | mnemoc | A1Sheds: I received my stm32f103 devkit already, but won't have time for it yet... at least until the linux mainlining of sun4i is near-done |
| 14:11.41 | A1Sheds | mnemoc: you don't count anyway |
| 14:11.55 | mnemoc | :) |
| 14:12.03 | A1Sheds | we'll see if any students turn up |
| 14:58.38 | hp__ | sun4i? |
| 14:59.03 | hp__ | "allwinner's A10" codenamed "sun4i crane" |
| 14:59.04 | hp__ | ah |
| 15:05.59 | mnemoc | iirc sun4i also covers other a1x socs |
| 15:06.23 | mnemoc | no clue which |
| 15:27.36 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 16:23.23 | hno | mnemoc, thining the same. |
| 16:39.33 | mnemoc | hno: it actually seems they are doing so within sw_reserve (CONFIG_SW_SYSMEM_RESERVED_BASE) ... but seems too late |
| 16:40.05 | traeak | good to see the lima driver is very actively being updated |
| 16:40.13 | traeak | maybe one of thee days we'll have good opengl es2 on it? |
| 16:42.29 | mnemoc | i haven't seen any comment about gallium3d support |
| 16:44.41 | traeak | it's probably too early for that yet |
| 16:44.59 | traeak | probably will make ARM upset since they want you to buy a commerical license to ge3t opengl on the mali |
| 16:45.22 | traeak | btw, i note that rhobust counted out the amlogic SOC |
| 16:45.36 | traeak | but the kde plasma tablet seems to have full working kernel with it |
| 17:19.42 | *** part/#arm-netbook QingPei (~qingpei@123.118.96.98) |
| 17:32.49 | hno | mnemoc, that just reserves address space. Requires the memory to be already mapped,. |
| 17:40.08 | mnemoc | hno: i was fearing so |
| 17:41.13 | hno | The mystery tightens. Learning more of the page mapping and a single page map entry can cover the whole I/O area nicely. |
| 17:42.49 | mnemoc | sounds good |
| 17:43.28 | hno | 0xf1cXXXXX |
| 18:47.26 | hno | Ok. with a single mapping for the I/O area and reduced early I/O mapping to minimum required then it gets a bit further. |
| 18:50.42 | mnemoc | how further? |
| 18:53.30 | hno | mnemoc, http://fpaste.org/nIJB/ |
| 18:54.51 | mnemoc | looks good :) |
| 18:55.35 | mnemoc | that null pointer probably comes from an incomplete API update on my part |
| 18:56.52 | hno | which API? |
| 18:56.53 | mnemoc | i did them blindly by looking at other similar commits |
| 18:57.06 | mnemoc | 1m |
| 18:57.26 | hno | cleans up his tree a bit for commit. |
| 18:58.00 | mnemoc | https://github.com/amery/linux-allwinner/commit/76765215b84ccd3242a3192a9182fb3628878d46 |
| 18:58.50 | mnemoc | the struct irq_chip may needs something else which wasn't required before |
| 18:58.54 | mnemoc | need* |
| 19:21.30 | hno | my changes pushed to github. |
| 19:22.21 | hno | mnemoc ^ |
| 19:24.44 | mnemoc | hno: thanks! |
| 19:27.05 | hno | I do not understand yet what really was failing, but it clearly did not like creating more than one I/O mapping in the same bigger region. |
| 19:28.48 | hno | both the kernel/head.S and mach-sun4i/core.c changes is needed to get past the pagetable setup. |
| 19:29.04 | hno | crashing in different ways without. |
| 19:30.03 | mnemoc | ic |
| 19:46.29 | hno | or actually, crashing in the same way but different mapping locations. |
| 19:47.05 | hno | both crashed in early_pte_alloc BUG_ON |
| 19:47.45 | lkcl | A1Sheds: that's ok, actually, thinking about it. i think things will have to move a bit quicker than august anyway. |
| 19:48.05 | A1Sheds | august? |
| 19:48.06 | lkcl | A1Sheds: we have to do a laptop sample like... well before august. |
| 19:48.12 | lkcl | or ... whenever gsoc is. |
| 19:48.18 | lkcl | whenever it starts/ends. |
| 19:48.19 | A1Sheds | oh yeah |
| 19:48.27 | A1Sheds | we could keep somebody busy |
| 19:48.47 | lkcl | yeahh... but for a laptop i need like most of the functionality anyway. |
| 19:48.53 | A1Sheds | the stm32 stuff will be working long before that |
| 19:49.02 | A1Sheds | sure |
| 19:49.04 | lkcl | yes, precisely. |
| 19:49.21 | lkcl | which sort-of defeats the object of giving an already-done job to a gsoc student. err :) |
| 19:49.35 | A1Sheds | send connectors directly to me btw |
| 19:49.41 | lkcl | ack. |
| 19:49.46 | A1Sheds | they can port more utils |
| 19:49.51 | A1Sheds | always something to do |
| 19:49.54 | lkcl | yep that's what i asked bret |
| 19:49.56 | A1Sheds | adjust whitespace |
| 19:49.58 | lkcl | good point. |
| 19:50.02 | lkcl | snorts |
| 19:50.05 | A1Sheds | check code for holes |
| 20:05.01 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:05.20 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:27.56 | *** join/#arm-netbook CaCtus492 (~Kent@123-243-197-152.static.tpgi.com.au) |
| 20:31.05 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:31.17 | *** join/#arm-netbook cjoe (~customerj@fibhost-66-7-177.fibernet.hu) |
| 20:31.18 | *** join/#arm-netbook mike__ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:39.47 | hno | lkcl, there seems license concerns in the Allwinner kernel & uboot sources. There is no license indication in the Allwinner sources. There was even binary blobs but we have cleaned those. |
| 20:41.01 | hno | s/seems/is some/ |
| 20:42.13 | hno | the exception is toms uboot-mmc, but it's missing a lot with no nand or usb support. |
| 20:45.10 | mnemoc | hno: tom said everything there was GPL (except the gps.ko file) |
| 20:46.56 | mnemoc | IANAL but when the time of cleaning comments and whitespacing issues we can all inject the missing headers pointing (C) to allwinner |
| 20:47.56 | mnemoc | after all we used the official 2.6.36 tree published by allwinner itself on github, not a leak |
| 20:51.44 | hno | Ok. Good. |
| 20:51.58 | mnemoc | but IANAL :) |
| 20:52.00 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:52.30 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 20:52.56 | hno | Is Toms statement recorded somewhere? |
| 20:54.26 | mnemoc | not sure, but we can ask him (in a clean mail without distractions) on the ML and take the response as "statement" |
| 20:56.08 | mnemoc | but in a way that doesn't scare him or his manager :p |
| 20:58.11 | hno | Have already asked him gently for clarification on why he removed all those files from his u-boot tree. |
| 20:58.48 | hno | no answer yet, but have not seen him around for quite some time now. |
| 20:59.32 | mnemoc | iirc he is on vacations |
| 21:15.06 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 21:15.24 | *** join/#arm-netbook mike__ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 21:34.53 | A1Sheds | and I thought ubuntu on 386 was slow http://dmitry.co/index.php?p=./04.Thoughts/07.%20Linux%20on%208bit |
| 21:36.39 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 21:36.57 | *** join/#arm-netbook mike__ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 21:46.25 | mnemoc | A1Sheds: did you see when Fabrice Bellard run linux on a browser using a virtual machine written in javascript? |
| 21:49.24 | mnemoc | A1Sheds: http://bellard.org/jslinux/ |
| 21:59.07 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 21:59.21 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 22:23.17 | CIA-91 | rhombus-tech: Micha master * rce1b1c63cb01 /allwinner_a10/orders/.mdwn: |
| 22:23.40 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 22:24.06 | *** join/#arm-netbook mike__ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 22:45.18 | traeak | mnemoc: what use is this VM ? |
| 22:46.18 | traeak | youch x86 machine |
| 22:53.32 | *** join/#arm-netbook Gumboot (~gumboot@rev.bovine.muck.net.nz) |
| 22:59.45 | CIA-91 | rhombus-tech: Jared master * r29b2344a2a55 /allwinner_a10/orders/.mdwn: |
| 23:02.50 | CIA-91 | rhombus-tech: Jared master * r6d62c2dac827 /allwinner_a10/orders/jwright.mdwn: |