| 09:50 | jnettlet_ | I think we should start trying to organize information about what cards have had problems so we can start attacking some of these problems in a more organized manner. |
| 09:51 | jnettlet_ | unfortunately with having multiple SPL images there is margin for user error as well. |
| 10:13 | MikeSeth | I will start a known issues section in the wiki, and that's where it will all go |
| 10:18 | hste | MikeSeth: have u tried dd the image from the working sdcard to the one with problems and see if it boots then? |
| 10:44 | MikeSeth | hste: no, but I will try to figure out what's up in the evening |
| 10:45 | MikeSeth | I currently don't have a native linux box and the card is written via virtualbox usb driver so it's all a bit.. fishy |
| 11:17 | sickness | morning all :) |
| 11:26 | MikeSeth | 'lo |
| 12:53 | MikeSeth | hste: tried a bunch of tests, nothing helps, uboot can't read the card |
| 12:54 | MikeSeth | this isn't a big issue, but I am curious why it does this |
| 13:38 | _Adik_ | MikeSeth: did you try jnettlet_'s build of u-boot that seems to fix this? |
| 13:39 | MikeSeth | _Adik_: no, I use rabeeh's one, but isn't jnettlet_'s fix for voltage issues in UHS cards? |
| 13:40 | _Adik_ | maybe, but in my case it works much better with this u-boot build |
| 13:40 | MikeSeth | do you have the git repo URL? |
| 13:41 | _Adik_ | no, but if jnettlet_ have not yet removed these builds they should still be available from his dropbox |
| 13:41 | _Adik_ | want a link for it? |
| 13:42 | MikeSeth | If you don't mind |
| 13:42 | jnettlet_ | They are still there |
| 13:42 | MikeSet | 13:42 * MikeSeth pets jnettlet_ |
| 13:42 | MikeSeth | you really love u-boot don't you ;) |
| 13:42 | _Adik_ | jnettlet_: I was wondering if you have time to sleep sometimes ;) |
| 13:42 | jnettlet_ | _Adik_, yeah like last night after one too many beers at the dinner party I went out to |
| 13:43 | jnettlet_ | :-) |
| 13:43 | _Adik_ | lol ;) |
| 13:44 | jnettlet_ | MikeSeth, the u-boot fixes that I had _Adik_ try are actually for an errata that some of the iMX6 chips have, where the PFD's don't reset and lock on Power On Reset |
| 13:44 | jnettlet_ | There was a fix in u-boot already but I found that the existing fix didn't always fix the problems, and in some cases could actually trigger it. |
| 13:44 | MikeSeth | so clocks get skewed? |
| 13:45 | _Adik_ | MikeSeth: https://dl.dropboxusercontent.com/u/736509/i4pro/u-boot/SPL https://dl.dropboxusercontent.com/u/736509/i4pro/u-boot/u-boot.img |
| 13:45 | jnettlet_ | they don't get set properly at all, because the PFD's need to be locked for clock changes to take effect. |
| 13:47 | _rmk_ | jnettlet: I have wifi botched to work with ojn's stuff... botched in the sense that I work around the oscillator power issue by just declaring an always-on regulator for it |
| 13:47 | _Adik_ | jnettlet_: what I have seen so far is that even if the new u-boot build works better, when it boots kernel and the kernel freezes, then there is still a chance that if I unplug the power and power it again u-boot can lock and cannot boot then |
| 13:47 | jnettlet_ | the FSL errata fix is to just ungate and the re-gate each pfd. I found that if the SOC was in this state and one pfd was locked and the other wasn't that doing this just flip-flopped one being lock and the other not. I am just selectively doing the clock gate cycle based on the lock status debug registers |
| 13:48 | jnettlet_ | _Adik_, I think that is also partially because of the soc |
| 13:48 | _rmk_ | jnettlet: argh, that's annoying... but sounds very similar to the hdmi workaround |
| 13:48 | _rmk_ | jnettlet: sometimes doing the workaround fixes it, and sometimes it causes it |
| 13:52 | _rmk_ | jnettlet: on the wifi, the brcmfmac driver looks like it falls apart - it spews errors and can't keep a list of nearby APs... I've sent broadcom a kernel log and request for help |
| 13:53 | _rmk_ | that's why it didn't apparantly see the networks in range when I first tested it - it /can/ see the RF signals, it's just that it falls over very quickly. |
| 13:54 | _rmk_ | _dab_: which patches of mine were you after? |
| 13:54 | jnettlet_ | _rmk_, really? |
| 13:55 | jnettlet_ | I think there is another driver that they use for Android. |
| 13:55 | _rmk_ | yep, it will occasionally get a boot-time glimse of the other networks around here, and then its game over: you can never get another AP list out of the driver because it just times out |
| 13:55 | _rmk_ | the initialisation of the card is littered with errors as well. |
| 13:56 | _rmk_ | errors returned by the firmware on the card |
| 13:56 | jnettlet_ | sounds very similar to the mess with went through at OLPC with Marvell's libertas wifi cards and driver |
| 13:57 | _rmk_ | I've gone through and confirmed the firmware blob is what Rabeeh has in his repo, which is what ubuntu has in their linux-firmware package, and cross-checked that with another source too |
| 13:57 | jnettlet_ | we just got that straightened out before Marvell retired the card and then it all started again with the 8787 |
| 13:57 | jnettlet_ | well at least they are distributing a consistent firmware |
| 13:57 | jnettlet_ | instead of one vendor gets one and another gets another |
| 13:57 | _rmk_ | yep |
| 13:58 | _rmk_ | it looks like this: |
| 13:58 | _rmk_ | brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0: Sep 2 2011 14:48:19 version 4.220.48 |
| 13:58 | _rmk_ | brcmfmac: brcmf_fil_cmd_data: Failed err=-23 |
| 13:58 | _rmk_ | brcmfmac: brcmf_fil_cmd_data: Failed err=-23 |
| 13:58 | jnettlet_ | so bcmdhd is the driver that broadcom released |
| 13:58 | _rmk_ | brcmfmac: brcmf_fws_init: failed to set bdcv2 tlv signaling |
| 13:58 | _rmk_ | brcmfmac: brcmf_fil_cmd_data: Failed err=-23 |
| 13:58 | _rmk_ | |
| 13:58 | _rmk_ | brcmfmac: brcmf_sdio_bus_rxctl: resumed on timeout |
| 13:58 | _rmk_ | |
| 13:58 | _rmk_ | brcmfmac: _brcmf_set_multicast_list: Setting BRCMF_C_SET_PROMISC failed, -23 |
| 13:58 | _rmk_ | and eventually: |
| 13:58 | _rmk_ | brcmfmac: brcmf_escan_timeout: timer expired |
| 13:59 | _dab_ | _rmk_: I will happily test any patches you consider worthwhile. Something that gets HDMI going would be good. I like the idea of latest stable + patchset = operable cubox |
| 13:59 | _rmk_ | and at that point, the list of APs will be permanently empty :( |
| 13:59 | _rmk_ | _dab_: how about 3.13.1 source plus my 3.13 patches at http://www.home.arm.linux.org.uk/~rmk/cubox/ ? |
| 14:00 | _rmk_ | I'm assuming you're talking about cubox-i not cubox? |
| 14:00 | _dab_ | Thank you. and yes cubox-i |
| 14:00 | _dab_ | I have an i4pro |
| 14:00 | _rmk_ | Note that I work on both, so being clear about which you mean is important :) |
| 14:02 | _rmk_ | jnettlet: I've been through the driver trying to work out what the -23 means, and it's an error code returned by the firmware on the wifi itself |
| 14:02 | jnettlet_ | _rmk_, wifi is connected through sdhc3? |
| 14:02 | _rmk_ | sdhc1 |
| 14:03 | _rmk_ | it's not a sdio communication error |
| 14:03 | _rmk_ | it sends a command packet to the wifi firmware, and the wifi firmware responds with a properly formatted response packet which has an 'error' bit set and an error code of -23 |
| 14:04 | MikeSeth | _rmk_: fyi brcmfmac is horrible in 3.0 series |
| 14:04 | MikeSeth | all br* drivers are |
| 14:04 | jnettlet_ | sounds like a lot of fun to be had there. |
| 14:04 | _rmk_ | MikeSeth: agreed, it took me a long while to unwind what the driver was doing |
| 14:04 | jnettlet | 14:04 * jnettlet_ has to head out for another hour to finish Sunday errands. |
| 14:05 | _rmk_ | you look in one file. it calls a function, that function is not in that .c file but somewhere else |
| 14:05 | jnettlet_ | luckily until I get my replacement CBi4pro this won't be much of a distraction for me |
| 14:05 | _rmk_ | it's very rare for a function to call another function in the same file! |
| 14:06 | _rmk_ | and the built-in debugging that they have in the driver is woefully inadequate |
| 14:06 | MikeSeth | I am still subscribed to linux-wireless because of all the interesting things I had to deal with in brcmsmac/mrcmfmac |
| 14:06 | _dab_ | I just spent the day battling with brcmsmac on a Dell laptop. The wifi driver craps itself when CPU utilzation is high. The downloadable BC drivers from http://www.broadcom.com/docs/linux_sta are less featured but seem to be more stable |
| 14:06 | _dab_ | an option for the cubox? |
| 14:06 | MikeSeth | _dab_: which kernel version? |
| 14:07 | MikeSeth | _dab_: no, no arm port for wl.ko |
| 14:07 | _dab_ | 3.11 |
| 14:07 | _dab_ | bummer |
| 14:08 | MikeSeth | brcmsmac worked for me on 3.11 pre-releases, on that bt and wifi combo board Lenovo likes to use in cheap laptops |
| 14:09 | _dab_ | It works for me until CPU hits ~100% for 15 minutes. Then it always dies with - brcmsmac bcma0:0: wl0: brcms_c_watchdog: dead chip |
| 14:09 | _rmk_ | I also did a load of digging around on brcm wifi/bt combos. it seems that it's only possible to talk to the BT via the UART and not via SDIO, despite the SDIO declaring that there is an interface function for BT. |
| 14:10 | _rmk_ | that's quite annoying because it causes btsdio to get loaded - that needs blacklisting to prevent that behaviour |
| 14:13 | _rmk_ | mikeseth: was there much help forthcoming from broadcom on that stuff? |
| 14:14 | MikeSeth | _rmk_: broadcom itself? None, but a couple of broadcom devs are active on the linux-wireless mailing list |
| 14:14 | _dab_ | gotta be better than mvwifi |
| 14:14 | _rmk_ | I was thinking of [email protected] |
| 14:14 | _rmk_ | 'cause that's where I sent my report yesterday... |
| 14:16 | MikeSeth | _rmk_: brcmXmac developers are in [email protected] |
| 14:16 | MikeSeth | oh wait |
| 14:17 | MikeSeth | they got themselves a separate list indeed |
| 14:17 | MikeSeth | _rmk_: Arend is a nice fellow |
| 14:18 | _rmk_ | yea, there's four people, that list and the linux-wireless list in MAINTAINERS for the brcmXmac stuff |
| 14:19 | _rmk_ | we'll see next week what happens :) |
| 14:19 | _dab_ | The linuxmint forum has lots of "not working BCM" wifi posts |
| 14:20 | MikeSeth | I'm seeing SDIO patches from two weeks ago for brcmfmac, dunno if that'd help you _rmk_ |
| 14:23 | _rmk_ | http://comments.gmane.org/gmane.linux.kernel.wireless.general/118014 ? |
| 14:23 | _dab_ | _rmk_: Dumbo question - where is your 3.13.1 source? |
| 14:24 | _rmk_ | I don't have 3.13.1, but... if you grab the 3.13.1 kernel source from kernel.org, and then apply my 3.13 patches, they should apply on top of 3.13.1. |
| 14:25 | _dab_ | ok |
| 14:25 | _rmk_ | from my site, you want: |
| 14:25 | _rmk_ | hummingboard-cubox-i-v3.13-20140120.diff |
| 14:25 | _rmk_ | hummingboard-cubox-i-v3.13-20140120-errata.diff |
| 14:26 | _dab_ | ok |
| 14:27 | _dab_ | Which BCM firmware (if any) are you guys using? |
| 14:28 | _rmk_ | _dab_: it's the one in the linux-firmware package: you may need to rename the firmware file though |
| 14:29 | _rmk_ | but... wifi isn't supported in the above patches, that's something I've been tinkering with only recently |
| 14:31 | _dab_ | ok |
| 14:35 | _dab_ | _rmk_: I will report back after sleep, (12:30am here) |
| 14:38 | MikeSeth | _rmk_: http://www.spinics.net/lists/linux-wireless/msg117942.html but looking at the patches I doubt it's useful |
| 14:38 | MikeSeth | gmane won't let me find those for some reason |
| 14:41 | _rmk_ | I'm not a fan of gmane anyway :) |
| 14:42 | _dab_ | b4 I go, I think rabeehs cubox-i uboot is now pretty good. Works for what I have tried (nfs boot, dhcp and sd boots) |
| 14:42 | _rmk_ | brcmfmac: enable firmware console logging functionality |
| 14:42 | _rmk_ | that might be useful |
| 14:43 | MikeSeth | hmmmm |
| 14:43 | MikeSeth | apparently I can get myself a BusPirate |
| 14:43 | MikeSeth | that could be useful |
| 14:44 | _rmk | 14:44 * _rmk_ imagines Johnny Depp on a London bus. |
| 14:59 | dv_ | _rmk_: what is your opinion on armv8? good step forward? |
| 18:20 | MikeSeth | jnettlet_: your uboot images have no effect, still can't read this particular card |
| 18:25 | crypt0s_ | do I need a boot.scr to boot the Cubox-i |
| 18:25 | crypt0s_ | and if I do, how do I get one? |
| 18:26 | crypt0s_ | And does anyone have a vanilla uENV.txt they could pastebin me? |
| 18:32 | MikeSeth | crypt0s_: not for the new uboot you don't |
| 18:33 | MikeSeth | bootfile=/boot/uImage |
| 18:33 | MikeSeth | mmcargs=setenv bootargs root=/dev/mmcblk0p1 rootwait video=mxcfb0:dev=hdmi consoleblank=0 |
| 18:33 | MikeSeth | ^^ uEnv.txt |
| 18:33 | crypt0s_ | MikeSeth: thank you :) |
| 18:33 | crypt0s_ | i am guessing i still need the uEnv.txt |
| 18:39 | crypt0s_ | MikeSeth: hrm, i'm still getting invalid partition errors |
| 18:40 | crypt0s_ | i have a ext4 FS on /dev/sdb2 |
| 18:40 | crypt0s_ | so i set the root partition as mmcblk0p2 |
| 18:40 | crypt0s_ | MikeSeth: any ideas? |
| 18:40 | MikeSeth | crypt0s_: you get invalid partition from u-boot? |
| 18:40 | crypt0s_ | yea |
| 18:40 | crypt0s_ | MikeSeth: Correct |
| 18:41 | MikeSeth | and your partition table is valid and can be seen by fdisk -l on the host machine? |
| 18:41 | MikeSeth | do this in uboot console: mmc read 0 1 1, do you get an error? |
| 18:41 | crypt0s_ | MikeSeth: one sec while i get a keyboard. |
| 18:41 | crypt0s_ | MikeSeth: yes i see it in fdisk |
| 18:43 | MikeSeth | crypt0s_: run the command I gave you in uboot console, if you get an error then it's a broken card, I have one that does the same |
| 18:43 | crypt0s_ | MikeSeth: do i put the serial connection line in uEnv.txt to enable the serial mode? |
| 18:43 | MikeSeth | tell me the type of the card |
| 18:43 | MikeSeth | crypt0s_: for the kernel, yes |
| 18:44 | crypt0s_ | Sandisk 16g |
| 18:44 | crypt0s_ | card reads good AFIAK |
| 18:44 | MikeSeth | if you see uboot messages then you don't need a serial, the uboot console can be summoned by pressing any key on the keyboard |
| 18:44 | MikeSeth | so in the first 3 seconds of boot |
| 18:45 | MikeSeth | which class is your card? |
| 18:45 | crypt0s_ | MikeSeth: having issue finding a keyboard that isn't on a laptop :P |
| 18:45 | MikeSeth | no USB one around? |
| 18:46 | MikeSeth | in any case try another card, I have the exact same issue with a Transcend Micro while 3 other Sandisk cards work |
| 18:46 | MikeSeth | the card reads and writes fine on the host machine, but uboot can't read from it even though it can load itself from it |
| 18:48 | crypt0s_ | MikeSeth: i did the mmcread |
| 18:48 | crypt0s_ | MikeSeth: outputs blocks read: ERROR |
| 18:48 | crypt0s_ | MikeSeth: remember though i have the data partition for the OS on the second logical part |
| 18:51 | crypt0s_ | MikeSeth: and when i rescan and do mmc list i do get something |
| 18:52 | MikeSeth | if it says error and invalid partition table then it is unable to access the card at all |
| 18:53 | MikeSeth | I have spent almost a day trying to figure out a way around this, no dice, the card is doing something that U-boot can't handle, use another card |
| 18:54 | crypt0s_ | MikeSeth: damn...are SanDisk SDHC cards just not supported? |
| 18:56 | MikeSeth | Mine works just fine |
| 18:56 | MikeSeth | what class is your card? |
| 18:59 | MikeSeth | crypt0s_: if you followed the instructions from the wiki do note I've just updated them |
| 18:59 | MikeSeth | http://imx.solid-run.com/wiki/index.php?title=Building_the_kernel_and_u-boot_for_the_CuBox-i_and_the_HummingBoard#Flashing_U-boot_and_the_kernel |
| 19:00 | MikeSeth | the old procedure was incomplete and intended for older uboot |
| 19:00 | crypt0s_ | MikeSeth: class 10 i think. |
| 19:00 | crypt0s_ | MikeSeth: OK I'll take a look that is what I was using though i did screw up. |
| 19:05 | MikeSeth | crypt0s_: please try another card and report if it solves your problem; I maintain a list of incompatible cards |
| 19:06 | crypt0s_ | MikeSeth: OK. |
| 19:07 | crypt0s_ | MikeSeth: stupid question though so i can follow the wiki exactly: how do i create an offset partition with fdisk |
| 19:15 | MikeSeth | it asks you where to begin the partition |
| 19:15 | MikeSeth | the default value on my system is 2048 which is sufficient |
| 19:17 | MikeSeth | also if fdisk feels irky to you, use cfdisk |
| 19:17 | crypt0s_ | that's usually what I use |
| 19:17 | crypt0s_ | question though: the jessi disk image i deployed to test the SD card worked |
| 19:18 | crypt0s_ | MikeSeth: also: that Jessi image only had one partition with 1MB of free space at the beginning. |
| 19:18 | MikeSeth | that is correct |
| 19:19 | MikeSeth | the jessi image has an older bootloader which might very well work for this card |
| 19:19 | MikeSeth | in which case this is a regression |
| 19:19 | crypt0s_ | Ah. |
| 19:19 | MikeSeth | I am quote sure I tried the jessi bootloader and it didnt work |
| 19:19 | MikeSeth | are you 100% certain? |
| 19:20 | MikeSeth | quite* |
| 19:20 | crypt0s_ | Yea i was using it yesterday to make sure the device worked before I started making my custom gentoo for it |
| 19:20 | MikeSeth | oh |
| 19:20 | MikeSeth | okay lemme verify that real quick |
| 19:20 | crypt0s_ | kk |
| 19:23 | MikeSeth | I will be making several Debian images for different card sizes soon |
| 19:23 | MikeSeth | need to finish my vagrant environment first |
| 19:32 | crypt0s_ | MikeSeth: I wonder if I want to just write the Jessi image down, then wipe out the data in the EXT4 partition and replace with the stage3 tarball |
| 19:35 | MikeSeth | you can but then you'd be living off an unsupported bootloader |
| 19:36 | MikeSeth | and you are right the old uboot works which means this is a regression |
| 19:36 | MikeSeth | jnettlet_: ^ |
| 19:36 | MikeSeth | rabeeh: ^ |
| 19:36 | crypt0s_ | woo i found a bug |
| 19:36 | crypt0s_ | MikeSeth: I can always update the bootloader though, that's just a DD command away |
| 19:37 | MikeSeth | oh and you'll have to deal with uboot.scr and mkimage etc |
| 19:37 | MikeSeth | ....at least I think so, that is |
| 19:37 | crypt0s_ | well i already have a uImage of the kernel |
| 19:38 | MikeSeth | uboot mkimage |
| 19:38 | crypt0s_ | yea |
| 19:38 | crypt0s_ | wait -- what's the difference |
| 19:38 | crypt0s_ | I made the mkimage from the mxi6 uboot build |
| 19:38 | crypt0s_ | added it to my toolchain |
| 19:38 | crypt0s_ | then made the kernel with make uImage |
| 19:44 | MikeSeth | you dont need the mkimage step for SPL uboot |
| 19:44 | MikeSeth | the default configuration is provided from uboot source and embedded in second stage bootloader |
| 19:44 | MikeSeth | uEnv.txt amends it |
| 19:45 | MikeSeth | it was mentioned in the flashing instructions because of the old pre-SPL uboot procedure |
| 19:46 | MikeSeth | (again this is what I learned through exploration and I may be wrong) |
| 19:47 | MikeSeth | hste: would you happen to know from which uboot tree and commit point your jessi uboot was generated? |
| 19:47 | MikeSeth | hste: I want to try bisecting |
| 19:57 | hste | MikeSeth: it was from http://imx.solid-run.com/wiki/index.php?title=Building_the_kernel_and_u-boot_for_the_CuBox-i_and_the_HummingBoard&oldid=96 |
| 19:59 | MikeSeth | thanks, so it's the same tree |
| 19:59 | hste | MikeSeth: patches up tp dec 15 |
| 19:59 | MikeSeth | hste: ok, that will probably isolate it |
| 19:59 | MikeSeth | thanks |
| 20:02 | crypt0s_ | hrm ok i think the kernel uImage that I built with the new uboot tools build is not compatible with old bootloader |
| 20:02 | MikeSeth | crypt0s_: that shouldn't be the case |
| 20:02 | crypt0s_ | so i will try rebuilding the kernel with mkimage from the older version of the uboot tools |
| 20:02 | crypt0s_ | MikeSeth: ok -- maybe my build is bad then |
| 20:03 | crypt0s_ | I tried using the Jessi config.gz as a base though |
| 20:03 | crypt0s_ | so not sure what (if anything) i would have done wrong -- uboot starts loading the kernel and then resets |
| 20:03 | crypt0s_ | "data abort" |
| 20:04 | MikeSeth | uhhh |
| 20:04 | MikeSeth | no idea here sorry |
| 20:04 | MikeSeth | during uboot or during kernel? |
| 20:05 | crypt0s_ | MikeSeth: during uboot loading kernel image |
| 20:06 | crypt0s_ | I'm just trying to get a path to get gentoo on the thing, having quite the learning experience |
| 20:09 | MikeSeth | well, you'd save yourself a lot of trouble by using another card |
| 20:11 | crypt0s_ | MikeSeth: I have a 32g class 4 samsung SDHC |
| 20:11 | crypt0s_ | are those known-good? |
| 20:11 | MikeSeth | I have no report on those |
| 20:28 | hste | MikeSeth: I have the old u-boot files here: http://stende.no-ip.info/imx6/solidrun-hb/uboot/ |
| 20:29 | MikeSeth | hste: it's ok, I already found the culprit commits, now studying them |
| 20:43 | MikeSeth | commit a577b broke it |