| 01:34.49 | CIA-91 | rhombus-tech: lkcl master * r80451ae2c1ba / (Pippo_Kennedy.mdwn allwinner_a10/orders/Pippo_Kennedy.mdwn): order processing cleanup |
| 02:12.58 | *** join/#arm-netbook QingPei (~qingpei@inet-hqmc07-o.oracle.com) |
| 03:39.06 | hno | mnemoc, that asm part have barely changed at all. And is highly unlikely to be related to any early hangs. It's about reserving memory for gps and something more. |
| 03:39.16 | hno | in an hackish way. |
| 05:50.51 | *** join/#arm-netbook lkcl (~lkcl@host217-43-148-3.range217-43.btcentralplus.com) |
| 08:47.45 | mnemoc | hno: what would be the right way of doing the same? |
| 08:53.27 | *** part/#arm-netbook QingPei (~qingpei@inet-hqmc07-o.oracle.com) |
| 08:59.59 | mnemoc | hno: 9aae53f907709010497ac6db7338786cb0c1fbe6 (as a whole) made the difference |
| 09:00.22 | mnemoc | including the chunk that hardcodes their blesses machine type |
| 09:00.27 | mnemoc | blessed |
| 09:00.57 | mnemoc | amery/linux-2.6.36-sun4i |
| 09:38.30 | mnemoc | hno: I didn't move along this commit because I only understood the part that hardcodes r1 (machine type), thought all the rest was related (wrong) and harmless (more wrong), and had no clue how to port it forward. As I had no way of testing anything at that time I found better to skip and file at ticket (https://github.com/amery/linux-allwinner/issues/1) |
| 09:41.16 | mnemoc | all the "referenced commits" are unrelated as github tries to be smart about "#1" written on commits |
| 09:57.46 | CIA-91 | rhombus-tech: Larry master * r934d1974ce7d /allwinner_a10/orders/10052101.mdwn: |
| 10:01.18 | CIA-91 | rhombus-tech: Larry master * rea2798f4399d /allwinner_a10/orders/10052101.mdwn: |
| 10:03.13 | CIA-91 | rhombus-tech: Larry master * r2d8c39d52890 /allwinner_a10/orders/10052101.mdwn: |
| 10:04.48 | arokux1 | hi guys. what is the status of our connectors? |
| 10:05.26 | mnemoc | A0Sheds: ---^ |
| 10:05.44 | CIA-91 | rhombus-tech: Larry master * rac3760ddafae /allwinner_a10/orders/10052101.mdwn: |
| 10:54.11 | CIA-91 | rhombus-tech: Giancarlo master * rd429cc8df52a /allwinner_a10/orders/sentiretina__64__yahoo.it.mdwn: |
| 10:55.37 | CIA-91 | rhombus-tech: Giancarlo master * r1b1d3fa1b2ad /allwinner_a10/orders/sentiretina__64__yahoo.it.mdwn: |
| 10:57.13 | CIA-91 | rhombus-tech: Giancarlo master * r63137ca71873 /allwinner_a10/orders/sentiretina__64__yahoo.it.mdwn: |
| 12:02.53 | CIA-91 | rhombus-tech: oldwanderer master * r7c15a33e8603 /allwinner_a10/orders/oldwanderer.html: |
| 12:05.40 | CIA-91 | rhombus-tech: oldwanderer master * r2cd17148ae87 /allwinner_a10/orders/oldwanderer.html: |
| 13:14.46 | hno | should not need hardcoding machine type. But uboot and kernel machine type need to match. |
| 13:15.38 | hno | mnemoc ^ |
| 13:15.53 | mnemoc | hno: tried using the same blessed machine number used in that commit and doesn't solve the problem |
| 13:16.12 | hno | odd. |
| 13:17.20 | hno | your tree is working now? |
| 13:17.26 | mnemoc | yes |
| 13:17.56 | mnemoc | amery/linux-2.6.36-sun4i I mean |
| 13:18.12 | hno | ok. was just about to ask which one of them.. |
| 13:18.17 | mnemoc | :) |
| 13:19.41 | hno | I forward ported the memory reservation hacks to my 3.3 tree. But did not seem to help. |
| 13:22.35 | mnemoc | after lots of headdesk'ing I decided to reduce the greed and work on a minimal patchset upon 2.6.36 before continue messing with newer kernels |
| 13:22.53 | hno | That's ok. |
| 13:23.36 | hno | A minimal patchset which can drop us into an initrd shell on 2.6.36 is a great start. |
| 13:33.53 | hno | mnemoc, you hand-merged the __s8 bugfix? |
| 13:34.15 | mnemoc | cherry-pick'ed your commit |
| 13:35.12 | hno | Hm... git history management is odd.. |
| 13:35.28 | mnemoc | commits have both, author and commiter |
| 13:35.44 | hno | that commit is listed twice now in my tree. |
| 13:35.52 | hno | with exact same changes. |
| 13:36.09 | mnemoc | git merge does odd things |
| 13:36.45 | mnemoc | I prefer to keep topic branches as stacks of commits (patches) instead of merged blobs |
| 13:36.53 | hno | did a git pull which resulted in an automatic 'recursive' merge. |
| 13:37.05 | mnemoc | git pull = git fetch + git merge |
| 13:37.31 | hno | but I think the problem is cherry-picking changes. |
| 13:38.45 | mnemoc | the merge shows the list of all remote changes, and makes one real commit (often empty) |
| 13:39.28 | mnemoc | that's awesome for mainline repos, but imo very noisy for topic branches |
| 13:40.13 | mnemoc | as the purpose of the topic branch is to become part of an upstream it's natural to want your changes always based on top of that upstream |
| 13:40.35 | hno | I prefer ignoring history while hacking and then prepare patches in a patch submission branch using rebase. |
| 13:42.13 | mnemoc | doing `git pull --rebase` when updating your local topic branch or working tree keeps it clean during development too ;-) without the confusions of seeming duplicated commits, etc |
| 13:42.45 | mnemoc | if that remote gets rebased hell happens |
| 14:04.16 | hno | git pull --rebase works fine for a private topic branch. But not on anything shared. |
| 14:05.25 | hno | as hell happens for anyone who have branched from it. |
| 14:08.30 | mnemoc | fully agree, but I thought those branches you have weren't intended for been the upstream of someone else... |
| 14:10.09 | mnemoc | but i suppose it's safe to asume than any non-local tree is potentially branched out even when not desired |
| 14:34.11 | mnemoc | anyhow, found the gremlin? :) |
| 14:35.38 | mnemoc | hno: anyhow, found the gremlin? :) |
| 15:18.19 | hno | Not yet. Birthday party today, remember? |
| 15:19.56 | CIA-91 | rhombus-tech: Frank master * rf67b3fca4fdb /allwinner_a10/orders/ftokc.mdwn: |
| 15:23.07 | mnemoc | hno: oh :D yes, your wife's ... forgot that |
| 15:46.52 | *** join/#arm-netbook QingPei (~qingpei@114.243.170.245) |
| 16:12.39 | hno | mnemoc, I wonder.. the R2 constant in the patch. 0x81000000. What is that about? |
| 16:15.20 | hno | code says r2 = atags or dtb pointer. Does not make sense to me to hardcode like this as it's passed by u-boot. |
| 16:16.33 | hno | confusing. |
| 16:20.07 | mnemoc | but somehow the passing from uboot is not working at all |
| 16:20.54 | mnemoc | or it does for you now? |
| 16:25.31 | mnemoc | the whole idea that the kernel *needs* such constants to be passed by the bootloader when it's built for only one machine anyway or it will die in painful silence puzzles me |
| 16:26.46 | mnemoc | hno: I always gave for granted "good defaults" are used and those hints are optional... sadly reality seems to disagree |
| 16:31.08 | mnemoc | it feels stupid that the same kernel can't boot on the stock bootloader and from ours because of a constant that's only used to decide if you have suicide if it's not the same used on compile time... |
| 16:37.37 | *** part/#arm-netbook QingPei (~qingpei@114.243.170.245) |
| 17:34.31 | CIA-91 | rhombus-tech: lkcl master * r128273789683 /allwinner_a10/orders/ (ftokc.mdwn sentiretina__64__yahoo.it.mdwn): order cleanup |
| 17:58.10 | MaDMaLKaV | got my a1000 a week ago and still have not time even to power it on :/ |
| 18:07.38 | hno | r2 is the boot parameters. I.e. kernel command line etc. |
| 18:08.54 | hno | mnemoc, I don't know the full story why arm uses a machid, but it's sure stupid for kernels only supporting a single board. |
| 18:09.43 | hno | or cpu |
| 18:13.53 | hno | hm.. the r2 mystery is intruiging. The Allwinner kernel hardcodes the location of boot paramaters elsewhere to 0x40000100, so it's most likely using the r2 parameter for something else entirely. |
| 18:14.18 | hno | two wrongs do not make it right. |
| 18:14.39 | mnemoc | narf |
| 18:30.02 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 18:30.17 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
| 18:46.19 | hno | Hm.. what is at 0x01000000 DRAM address? (0x41000000, 0x81000000, 0x01000000 all the same depending on context) |
| 18:47.21 | hno | nothing ifaict. |
| 19:04.21 | mnemoc | maybe as 1<<20 ? |
| 19:10.37 | mnemoc | err, 1<<28 |
| 19:10.51 | mnemoc | missed a zero :) |
| 19:38.55 | hno | Hmm.. still no go. |
| 19:43.17 | mnemoc | hipboi! :-/ |
| 19:44.00 | hno | will dig a little more. |
| 20:27.12 | CIA-91 | rhombus-tech: FrankiR master * r9afbbdb650eb /allwinner_a10/orders/frankiR.mdwn: |
| 20:33.20 | *** join/#arm-netbook Gumboot (~gumboot@rev.bovine.muck.net.nz) |