| 07:44 | sanjeevs | Hello All |
| 07:44 | sanjeevs | For hummingboard family of board do we have production grade Linux/BSP avilable |
| 07:47 | sanjeevs | I believe that we can use FSL BSP too |
| 07:48 | sanjeevs | jnettlet:ping |
| 08:44 | rabeeh | sanjeevs: https://github.com/SolidRun/linux-fslc |
| 08:52 | sanjeevs | Thanks @rabeeh: I have couple of question with respect to kernel |
| 08:53 | sanjeevs | this kernel looks like similar copy of linux-fsl kernel except 1)defconfig 2)dts file for the humming board correct ? |
| 08:55 | sanjeevs | myself and jnettlet had an long discussion some time back when he was upgrading the linux kernel to 3.14 |
| 08:55 | sanjeevs | is he not avilable ? |
| 09:11 | sanjeevs | ping rabeeh |
| 09:12 | mk01 | sanjeevs: jnettlet is on business trip somewhere |
| 09:13 | sanjeevs | ok |
| 09:13 | sanjeevs | Since jnettlet is not avilable,can someone clarify my questions ? |
| 09:14 | mk01 | just ask |
| 09:43 | sanjeevs | https://github.com/SolidRun/linux-fslc kernel looks like similar copy of linux-fsl kernel except 1)defconfig 2)dts file for the humming board correct ? |
| 09:46 | sanjeevs | I know jnettlet started this exercise because fsl 3.14 kernel was not avilable but FSL had already released kernel 3.14 stabvle series |
| 09:46 | sanjeevs | Question :why there is need of separate kernel now ? |
| 09:48 | mk01 | bugfixes, backports, other platform(s)/special models support |
| 09:48 | mk01 | (or even features) |
| 09:49 | sanjeevs | features for solidrun devices ? |
| 09:49 | mk01 | no, not only. imx6 |
| 09:51 | sanjeevs | AFAIK this is forked from the linux-fslc which is opensource kernel and do you update/fetch regulary festures/bug fixes from opensource and freescale BSP |
| 09:51 | mk01 | sanjeevs: what is your concern, specifically? |
| 09:51 | mk01 | ok, up-to-date kernel |
| 09:52 | mk01 | I can't speak for jnettlet |
| 09:52 | sanjeevs | I don't have any concern just asking, is this stable enough to use for humming Board or fsl bsp is more stable |
| 09:52 | mk01 | .... |
| 09:52 | mk01 | but |
| 09:52 | mk01 | the fact it is forked from FSL is |
| 09:53 | sanjeevs | no |
| 09:53 | mk01 | wait |
| 09:53 | mk01 | I didn't finish the sentence |
| 09:53 | sanjeevs | it is forked from Freescale/linux-fslc and jnettlet ported FSL bsp on it |
| 09:54 | sanjeevs | Please comlete your sentence ,sorry |
| 09:56 | mk01 | jnettlet is integrating (trying) the best from imx6 code laying everywhere |
| 09:56 | sanjeevs | ok |
| 09:57 | mk01 | so yes, it is fork of fslc - to have the core code from creators |
| 09:57 | mk01 | and to provide sustainable code base |
| 09:57 | sanjeevs | For your board like humming board or micro som you would suggest to use this kernel instead of fsl bsp |
| 09:57 | sanjeevs | Thanks,make sense |
| 09:58 | mk01 | sanjeevs: yes |
| 09:58 | mk01 | mainly because of two things |
| 09:59 | sanjeevs | BTW,your introduction |
| 09:59 | mk01 | 1. bugfixes on top of fslc (either community, or upstream -> backport) |
| 09:59 | mk01 | 2. features - which FSL will never integrate or has no interest |
| 10:00 | mk01 | also look at the releases cycle |
| 10:01 | mk01 | there is a release - which stops at some point - with no further maintanance |
| 10:01 | sanjeevs | Are you talking about additonal features for imx6 |
| 10:01 | mk01 | yes |
| 10:01 | sanjeevs | also you pull bugfixes from FSL BSP |
| 10:02 | mk01 | sure, it is not only fork |
| 10:02 | mk01 | FSL upstream is still actively used parent tree |
| 10:02 | mk01 | the fork is not (was not) one shot deal |
| 10:03 | sanjeevs | I didn't get you |
| 10:03 | mk01 | ok - rephrase |
| 10:04 | sanjeevs | This kernel is based on fslc which is opensource kernel exculding fsl bsp code. So far solid run kernel=opensource+fsl bsp+bugfixes |
| 10:05 | mk01 | FSLC is of course still merged actively and regularly |
| 10:06 | mk01 | look, I'm not here a jnettlet's tree propaganda |
| 10:06 | sanjeevs | simple question: is this tree contain fsl bsp drivers/featurs |
| 10:06 | mk01 | SURE |
| 10:07 | sanjeevs | Are you also maintainer of this tree ? |
| 10:07 | sanjeevs | when jnettlet's will come back from conference ? |
| 10:07 | mk01 | look |
| 10:08 | mk01 | (yes, I push commits now and then) |
| 10:08 | sanjeevs | Ok and you work for solidrun |
| 10:08 | sanjeevs | BTW Thanks for your support ! |
| 10:09 | sanjeevs | can you also let me know the official solidrun bootloader code ? |
| 10:09 | mk01 | what is important - I'm maintaining mine 4.1.y tree - and I simply changed from fslc |
| 10:09 | mk01 | to jnettlets tree |
| 10:10 | mk01 | because it's much easier (all the various trees of fslc consolidated) - and much safer than fslc |
| 10:10 | sanjeevs | https://github.com/SolidRun/linux-fslc is official tree from solidrun for cubebox and humming board |
| 10:10 | mk01 | (safer -> i don't have to retest what comes from jnettlet's tree) |
| 10:11 | sanjeevs | ok |
| 10:11 | mk01 | sanjeevs: yes, uri is right |
| 10:11 | sanjeevs | for bootloader |
| 10:12 | sanjeevs | I came across two version 1) https://github.com/SolidRun/u-boot-imx6/tree/408544d61f230060f18ffe2e06565deadbcf3451/board/solidrun 2) https://github.com/SolidRun/u-boot-imx6 |
| 10:12 | sanjeevs | which one is official version |
| 10:13 | mk01 | https://github.com/SolidRun/u-boot-imx6 |
| 10:14 | mk01 | AFAIK |
| 10:14 | sanjeevs | does it support humming baord also https://github.com/SolidRun/u-boot-imx6/tree/imx6/board/solidrun |
| 10:14 | mk01 | yes, it support cubox + humm + humm2 |
| 10:14 | sanjeevs | who is the right guy to autheticate this info |
| 10:14 | mk01 | and the new models as well |
| 10:14 | sanjeevs | Thanks again |
| 10:15 | sanjeevs | going for lunch will pinig you later |
| 10:15 | mk01 | of course official maintainer is right person to ask, but I have almost all boards running |
| 10:15 | mk01 | (and testing) |
| 11:21 | sanjeevs | Thanks mk01,who is the official maintainer |
| 12:30 | mk01 | sanjeevs: jnettlet |
| 13:34 | sanjeevs | ok thanks |
| 13:50 | mk01 | np |
| 14:04 | jnettlet | Hey, sorry I am around. My irc bounce got knocked off of freenode. |
| 14:05 | mk01 | jnettlet: so you are alive after all :D |
| 14:05 | jnettlet | mk01, yeah sorry about that. |
| 14:05 | jnettlet | I was wondering how I was being so productive :) |
| 14:05 | mk01 | :D |
| 15:02 | mk01 | jnettlet: the last uboot commit - is it part of the "allow memory accesses to wrap" topic ? |
| 15:03 | jnettlet | mk01, no it fixes a problem with passing buffered memory from the GPU to userspace. |
| 15:04 | mk01 | so it can or can't have effect on the v5 topic (I) have outstanding ? |
| 15:04 | jnettlet | mk01, the memory wrapping should be fixed. |
| 15:05 | jnettlet | you are still seeing problems? |
| 15:05 | mk01 | still running on v4 |
| 15:05 | jnettlet | yeah it should be fixed in u-boot though |
| 15:05 | mk01 | but quick test compiled the last package you pushed , ... so will start (re)testing perhaps later today |
| 15:06 | mk01 | ... |
| 15:06 | mk01 | do you remember the 'flicker' or how to call it |
| 15:06 | mk01 | I was complaining |
| 15:06 | mk01 | with doublerate DEINT for some media ? |
| 15:06 | mk01 | by accident solved it |
| 15:06 | mk01 | "accident" |
| 15:07 | mk01 | if you remember it was specific to interlaced content with Bot/BT frame first |
| 15:07 | jnettlet | oh, I didn't realize it was working but not properly fixed. |
| 15:07 | jnettlet | yes |
| 15:07 | mk01 | I found the stream from IPTV is simply switching |
| 15:08 | mk01 | with or without FLAG |
| 15:08 | mk01 | Kodi if DEINT is forced and has no FLAG, flags with TOP |
| 15:08 | mk01 | changed to BOT |
| 15:09 | mk01 | and it deinterlaces like my plasma natively (meaning LITTLE flicker is there, but HW samsung is doing on the stream the same) |
| 15:10 | mk01 | also retested all INT content I had if this default change (for frame not having flag) has some adverse effect - has not |
| 15:10 | mk01 | ...... |
| 15:11 | mk01 | you have all the code in treeJnettlet |
| 15:11 | jnettlet | yes, I have been playing with it, but not the interlaced content so much. |
| 15:12 | mk01 | np, more or less I was telling this that - sometimes one tries hard for weeks and nothing. and then simply the brain splits solution - with no obvious external motivation. |
| 15:14 | mk01 | ((change of pinctrl)) |
| 15:23 | mk01 | have you had time to give a look to the EDID firmware load option ? and the EDID itself unifying / tidy up a bit ? |
| 15:24 | jnettlet | mk01, I already merged that thanks :) |
| 15:24 | jnettlet | it all looked reasonable to me |
| 15:26 | mk01 | ... you really was productive. |
| 15:26 | mk01 | heh. |
| 15:26 | mk01 | I will send the fix for the doublebuffer / screen destroy on EDID change (AVR case for instance) |
| 15:29 | jnettlet | sure thing. thanks |
| 15:29 | jnettlet | my openelec builds should be out soon. I think most the bugs are sorted at this point. |
| 15:32 | mk01 | and as always - consulting. https://github.com/SolidRun/linux-fslc/blob/3.14-1.0.x-mx6-sr/drivers/mfd/mxc-hdmi-core.c#L95 <<< have you ever followed this ? it should re-start audio stream after hdmi is operational again , but shouldn't this be more https://github.com/SolidRun/linux-fslc/blob/3.14-1.0.x-mx6-sr/sound/soc/fsl/imx-hdmi-dma.c#L1024 ? |
| 15:33 | mk01 | the 'problem' is that plain SNDRV_PCM_TRIGGER_RESUME/START triggers alsa reinit what drains buffer and ends up as buffer underrun in KODI |
| 15:35 | mk01 | ((i'm running the SNDRV_PCM_TRIGGER_PAUSE_RELEASE for few days and no alsa errors whatsoever - does it make sense too ? )) |
| 15:35 | jnettlet | so you just enable the HDMI for ANDROID option in the kernel? |
| 15:36 | jnettlet | I haven't messed with that yet. I am just getting Android to the point I am testing multimedia |
| 15:36 | mk01 | nono it is #ifndef |
| 15:37 | mk01 | ((so the code path is noadroid)) |
| 15:38 | jnettlet | sorry, yep just took a quick glance at it. |
| 15:41 | mk01 | np |
| 15:45 | mk01 | last two distractions for today (promise). |
| 15:45 | jnettlet | go ahead. |
| 15:46 | jnettlet | I am just working on some enhancements for the the HB Edge |
| 15:46 | mk01 | 1. there was commit changing hextexted edid output to binary -> this directly affects EDID readout in EGLIMX - maybe you already changed fixed. only 'issue' is the code is mainstream already |
| 15:46 | mk01 | so no only treeJnettlet |
| 15:46 | mk01 | .... |
| 15:47 | jnettlet | yeah, I already have a patch that is exporting our edid info as binary in sysfs |
| 15:47 | mk01 | malte promised new boards ... and month and nothing |
| 15:47 | mk01 | :-/ |
| 15:47 | jnettlet | there is a backlog |
| 15:47 | jnettlet | you are on the list |
| 15:47 | mk01 | ok |
| 15:48 | jnettlet | rmk is waiting as well |
| 15:48 | mk01 | so he (malte) is alive as well :D |
| 15:48 | mk01 | 2 . |
| 15:48 | mk01 | 2. |
| 15:48 | jnettlet | yeah, he has been busy with university |
| 15:48 | mk01 | oh, forgot that |
| 15:49 | jnettlet | so what is 2? |
| 15:49 | mk01 | 2. are you closer to the way how to push 1000/1001 modes to userspace ? |
| 15:50 | jnettlet | not at all. I have been buried in other things. |
| 15:51 | jnettlet | I think for now we go with what is easiest and then work from there |
| 15:52 | jnettlet | mk01, not sure if you are interested but CAAM is working and stable now |
| 15:53 | mk01 | I'm just asking because I updated the patch for this and beside it is almost no intrusion to mxc_hdmi, the audio (hdmi) works for ANY pixclk + I added own khz->ps calculation (it just considers remainder after /) what made the clocks to match desired clk with 0,001% |
| 15:54 | mk01 | ad CAAM |
| 15:55 | mk01 | I will return: I already merged that thanks :) |
| 15:55 | jnettlet | okay |
| 15:55 | mk01 | (and provided libssl to APT as replacement) |
| 15:56 | mk01 | just Q |
| 15:56 | mk01 | is this enough for tasks like openswan ? |
| 15:57 | mk01 | I have the feeling the engine crypto still needs to be specified as default , .. ? |
| 16:01 | mk01 | btw: I know you want it right, but the idea with lowercasing character in mode spec (eg S: -> s:) costs only one line in FBDEV layer and is transparent to userspace (xserver is no case sensitive etc) |
| 16:02 | mk01 | (+1 line in EGLIMX at Kodi at we have fract modes) |
| 16:02 | mk01 | and it can be easily changed anytime later |
| 16:07 | mk01 | correction - xserver IS case sensitive so filters those modes out and just the usual stay (this I wanted to say) |
| 16:09 | jnettlet | mk01, send me patch and I will think about it. |
| 16:10 | mk01 | ok |
| 16:10 | jnettlet | recompile openssl and gnutls is good enough for most applications. libreswan/openswan requires OCF-linux to be added, which then uses cryptodev |
| 16:10 | jnettlet | I have it on my list but it is not a high priority right now |
| 16:14 | mk01 | jnettlet: the edid stuff I meant this: https://github.com/mk01/linux-fslc/commits/edid_get_fw |
| 16:14 | jnettlet | yes I didn't merge that? I am sure I did. |
| 16:15 | jnettlet | Or it got pulled into the wrong tree again :/ |
| 16:16 | mk01 | good to know I'm not only one :)) I tell you I would be reverting last day backup so many times (not having the export script from you) |
| 16:16 | mk01 | re IF not having this script ... |
| 16:16 | jnettlet | mk01, yep that is my bad. It is still in another branch |
| 16:17 | jnettlet | I will merge that it is fine |
| 16:18 | mk01 | no problem, just the other commits wasn't match. just ping me anytime you push |
| 16:20 | jnettle | 16:20 * jnettlet nods |
| 16:20 | jnettlet | thanks |
| 18:20 | mk01 | jnettlet: do you know what makes mxc_hdmi_set_mode_to_vga_dvi(hdmi); needed ? is it specs, or something different? I takes quite long and I found no issue with it removed (this is few months of our kernel releases). but don't want to touch it without you being OK. |
| 18:24 | mk01 | ((perhaps I never tested attaching different monitor on the fly - what is it intended for - so keep it)) |
| 22:44 | mk01 | jnettlet: with loglevel=7 I get this on your tree: |
| 22:44 | mk01 | [ 0.275981] hw perfevents: enabled with ARMv7 Cortex-A9 PMU driver, 7 counters available |
| 22:44 | mk01 | [ 0.276717] WARNING: CPU: 0 PID: 41 at crypto/testmgr.c:3334 alg_test+0x80/0x28c() |
| 22:44 | mk01 | [ 0.276736] Modules linked in: |
| 22:44 | mk01 | [ 0.276759] CPU: 0 PID: 41 Comm: cryptomgr_test Not tainted 3.14.53+ #9 |
| 22:45 | mk01 | [ 0.276801] [] (unwind_backtrace) from [] (show_stack+0x10/0x14) |
| 22:45 | mk01 | [ 0.276836] [] (show_stack) from [] (dump_stack+0x88/0x98) |
| 22:45 | mk01 | [ 0.276872] [] (dump_stack) from [] (warn_slowpath_common+0x6c/0x88) |
| 22:45 | mk01 | [ 0.276901] [] (warn_slowpath_common) from [] (warn_slowpath_null+0x1c/0x24) |
| 22:45 | mk01 | [ 0.276930] [] (warn_slowpath_null) from [] (alg_test+0x80/0x28c) |
| 22:45 | mk01 | [ 0.276958] [] (alg_test) from [] (cryptomgr_test+0x40/0x48) |
| 22:45 | mk01 | [ 0.276986] [] (cryptomgr_test) from [] (kthread+0xd0/0xe8) |
| 22:45 | mk01 | [ 0.277014] [] (kthread) from [] (ret_from_fork+0x14/0x34) |
| 22:45 | mk01 | [ 0.277063] testmgr: alg_test_descs entries in wrong order: 'authenc(hmac(sha512),cbc(aes))' before 'authenc(hmac(md5),cbc(aes))' |
| 22:56 | Artox | mk01: I have seen jnettlet complain about this |
| 22:56 | Artox | his solution seems to be "get rid of those warnings by disbaling those tests" |
| 22:57 | Artox | I very much hope that this does the job: https://github.com/mxOBS/deb-pkg_kernel-cubox-i/commit/0a117d069e395d054148f2684845a8234baba066 |
| 23:40 | kasvain | a question, any help with installing u-boot on a i2-ex, would like to use prebuilt binaries if there is any (didn't find by searching) |
| 23:41 | kasvain | also don't have a possibility to set up a compilation environment |
| 23:42 | kasvain | manage to drop a prebuilt uart image, but the rest including getting it to work |
| 23:45 | Artox | kasvain: a prebuild uart image? |
| 23:45 | Artox | you mean you are looking for a system that will have a console on serial? |
| 23:56 | mk01 | Artox: oh right, was wondering why it is shown when I also remembered a commit with this topic - but commit changes defconfigs, not .config :) |