sailus | freemangordon: Pong. | 00:01 |
---|---|---|
freemangordon | sailus: hi! I am trying to make cameras working on n900, linux 4.6-rc1. No matter what I do, all I can get is a green image, the same as with 3.x kernels. So, is CSIb supported in ISP driver and what needs to be done to implement that support if not? | 00:48 |
freemangordon | I am using Pali's linux-n900 tree, there are a couple of fixes from you (and sre) to make N9/50 cameras working | 00:50 |
freemangordon | In dmesg log everything seems ok, there are no errors, etc | 00:50 |
freemangordon | sailus: One thing I saw missing, is enabling the CSI1 receiver in SCM CONTROL_CSIRXFE register, but even after I added the needed code, still no success | 00:52 |
DocScrutinizer05 | silly question: you got proper initialization for the camera(s)? | 01:30 |
DocScrutinizer05 | SMIA95_AF_camera module_APL_090511.doc ET8EK8 | 01:33 |
DocScrutinizer05 | http://wstaw.org/m/2016/04/11/plasma-desktopUk2219.png | 01:34 |
DocScrutinizer05 | http://wstaw.org/m/2016/04/11/plasma-desktopqJ2219.png http://wstaw.org/m/2016/04/11/plasma-desktophw2219.png http://wstaw.org/m/2016/04/11/plasma-desktopIL2219.png | 01:45 |
DocScrutinizer05 | https://www.google.de/search?q=SMIA95_AF_camera+module_APL_090511.doc | 01:46 |
DocScrutinizer05 | http://mg.pov.lt/maemo-irclog/%23maemo.2013-03-02.log.html#t2013-03-02T08:21:15 | 01:49 |
*** Pali has quit IRC | 02:30 | |
sailus | freemangordon: I think the CCP2 receiver should be ok. No-one has verified the changes though. And AFAIR there's no DT support (in the mainline kernel). | 02:58 |
sailus | I remember there were issues with clocks on 3430, I have to say I don't know whether the issue was resolved since. :-I | 02:59 |
sailus | Do you have a working setup with any kernel at the moment? | 03:00 |
DocScrutinizer05 | no DT support sounds like fun | 03:03 |
DocScrutinizer05 | sailus: well Nokia kernel works just fine ;-) | 03:04 |
DocScrutinizer05 | a tad long in the teeth though | 03:04 |
DocScrutinizer05 | I think freemangordon got it semi-working with a 3.x or 4.x kernel, at times | 03:05 |
DocScrutinizer05 | I dunno if the relevant bit in Nokia kernel (and userland) are FOSS though | 03:06 |
DocScrutinizer05 | bits even | 03:07 |
sailus | They're not. | 03:07 |
sailus | The omap3camd is closed, and it implements the automatic exposure, white balance and focus algorithms. | 03:08 |
DocScrutinizer05 | there's /lib/firmware/et8ek8-0002.bin | 03:08 |
sailus | The APIs it expects are not upstreamed, but one could implement a wrapper to convert the IOCTLs to what's in the mainline kernel. | 03:08 |
DocScrutinizer05 | yeah, thought as much | 03:08 |
sailus | I think it'd be good to get it working on a recent kernel first. | 03:09 |
sailus | Either of the cameras. | 03:09 |
DocScrutinizer05 | yes, for sure | 03:09 |
DocScrutinizer05 | you know if there's any init of the sensor done in mainline? | 03:09 |
DocScrutinizer05 | see my URL wall above | 03:10 |
sailus | There's no DT support for CCP2 in upstream. | 03:10 |
DocScrutinizer05 | then anything beyond is moot to discuss | 03:10 |
sailus | Other than that, the front camera should be supported in upstream with the smiapp driver. | 03:10 |
DocScrutinizer05 | may I query you? | 03:11 |
sailus | What about the screenshots? | 03:12 |
sailus | For what it's worth, I have preliminary patches for CCP2 support: | 03:22 |
sailus | http://git.retiisi.org.uk/?p=~sailus/linux.git;a=shortlog;h=refs/heads/rm696-056-ccp2 | 03:23 |
sailus | (I'm pushing right now so it won't work until after a while.) | 03:23 |
sailus | I didn't upstream those as some of the patches are hacks and they haven't been verified with CCP2 working. | 03:24 |
*** RedW has quit IRC | 05:09 | |
*** RedW has joined #maemo-ssu | 05:11 | |
*** RedM has joined #maemo-ssu | 05:16 | |
*** RedW has quit IRC | 05:16 | |
*** Sicelo009N has quit IRC | 05:49 | |
*** RedM has quit IRC | 06:10 | |
*** RedW has joined #maemo-ssu | 06:11 | |
*** DocScrutinizer05 has quit IRC | 06:23 | |
*** DocScrutinizer05 has joined #maemo-ssu | 06:23 | |
*** corvinux has quit IRC | 08:40 | |
*** amiconn has quit IRC | 08:55 | |
*** arcean has joined #maemo-ssu | 08:56 | |
*** amiconn has joined #maemo-ssu | 08:57 | |
freemangordon | sailus: There is *some* DT support/setup in Pali's tree, no idea if it is correct, but ooks like. see https://github.com/pali/linux-n900/commit/ec98148da1b1665facb13f49757742952d8eb47d | 09:12 |
freemangordon | that video-bus-switch is a driver from sre, that controls camera switch gpio based on which pad is linked. I verified the clocks and power supplies, all seems fine, smiapp is ok too, according to the logs (it has debug logs enabled) | 09:14 |
freemangordon | but as I told you, CSI1 never gets enabled, so I had to patch that, see https://github.com/pali/linux-n900/blob/v4.6-rc1-n900/drivers/media/platform/omap3isp/ispccp2.c#L840 | 09:16 |
freemangordon | ccp2->phy is always NULL, as there is no PHY for CSI1. No idea if this is the way it should be, however I added a code enabling CSI1 (CONTROL_CSIRXFE) in ccp2_if_configure() | 09:18 |
freemangordon | I also got all the values from ISP_CSI1B register array, while front camera is active in stock kernel and put them back while trying to get some video with mainline (with devmem2), no sucess | 09:21 |
freemangordon | sailus: this is the pipeline I use http://pastebin.com/AtbriatN , I see on obvious mistakes in it | 09:24 |
freemangordon | also, in DTS there vdds_csib-supply was missing, but in the DTS here it is added | 09:26 |
freemangordon | the only suspicious things in the logs is "omap3isp 480bc000.isp: Entity type for entity OMAP3 ISP XXXXX was not initialized!", where XXXXX is CCP2, CSI2a, CCDC... | 09:28 |
freemangordon | sailus: yavta, from the pastebin ^^^ just hangs waiting for a frame, mplayer times out. When I try to stop mplayer or yavta, there are messages like "OMAP3 ISP resizer: module stop timeout." and "CCDC stop timeout!" | 09:33 |
freemangordon | sailus: http://pastebin.com/EZKs6Gst , ISP registers from one of my attempts, CCP2 enable bit is 0, but that's normal as registers are printed just before CCP2 is enabled | 09:35 |
*** futpib has joined #maemo-ssu | 09:40 | |
freemangordon | sailus: http://pastebin.com/Ne3zE0tz full log with printks on various places, ignore those bus errors, it is me playing with devmem2 :) | 09:41 |
*** futpib has quit IRC | 09:48 | |
freemangordon | sailus: BTW I am unable to open http://git.retiisi.org.uk/?p=~sailus/linux.git;a=shortlog;h=refs/heads/rm696-056-ccp2 | 09:52 |
*** freemangordon_ has joined #maemo-ssu | 10:37 | |
sailus | freemangordon: video-bus-switch sounds like the right thing, I haven't looked at it though. | 10:51 |
bencoh | regarding automatic exposure/wb/focus we have fcam drivers/userland as well | 10:53 |
bencoh | I dunno if it depends on omap3camd in some way | 10:54 |
freemangordon_ | sailus: yeah. https://github.com/pali/linux-n900/blob/v4.6-rc1-n900/drivers/media/platform/video-bus-switch.c | 10:54 |
freemangordon_ | I had to do a couple of fixes though | 10:55 |
freemangordon_ | sailus: however, to my understanding, vbs doesn't look related to the problems I face | 10:56 |
sailus | I think I must have suspended my laptop before the push finished. | 10:58 |
sailus | I'll redo that noe. | 10:58 |
freemangordon_ | :) | 10:58 |
sailus | Hmm. I pushed it to wrong repository. :-P | 11:04 |
sailus | Pushing again now... | 11:04 |
sailus | There it is. | 11:05 |
sailus | freemangordon_: Do you happen to get interrupts from the isp? | 11:06 |
freemangordon_ | sailus: I suspect no (IRQ debug is not enabled in the config) | 11:13 |
freemangordon_ | sailus: "Firefox can't find the server at git.retiisi.org.uk." | 11:13 |
freemangordon_ | are you sure this server is connected to internet? | 11:14 |
*** freemangordon_ has quit IRC | 12:09 | |
*** freemangordon_ has joined #maemo-ssu | 12:25 | |
sailus | I'm pretty sure it is. I'm running irssi on the server and I'm at the office right now. | 13:10 |
sailus | What's the problem you have? | 13:11 |
sailus | I.e. can you connect to it using telnet on that port, or not, and ipv4 or ipv6? | 13:11 |
sailus | Both should work. | 13:11 |
freemangordon_ | sailus: trying to open "http://git.retiisi.org.uk/?p=~sailus/linux.git;a=shortlog;h=refs/heads/rm696-056-ccp2" in ff results in "Firefox can't find the server at git.retiisi.org.uk.". trying to open "http://retiisi.org.uk/?p=~sailus/linux.git;a=shortlog;h=refs/heads/rm696-056-ccp2" brings me to "http://retiisi.org.uk/~sailus/", unfortunately the page is not in English, so I am not sure what is written there | 13:16 |
*** freemangordon_ has quit IRC | 13:32 | |
sailus | Ah, I added git recently, and there's probably a negative match or stale data in a secondary DNS server. | 13:49 |
sailus | http://salottisipuli.retiisi.org.uk/cgi-bin/gitweb.cgi?p=~sailus/linux.git;a=shortlog;h=refs/heads/rm696-056-ccp2 | 13:50 |
*** arcean has quit IRC | 14:06 | |
*** arcean has joined #maemo-ssu | 14:16 | |
*** RedW has quit IRC | 14:40 | |
*** RedW has joined #maemo-ssu | 14:43 | |
*** RedW has quit IRC | 14:51 | |
*** RedW has joined #maemo-ssu | 14:51 | |
*** RedW has quit IRC | 15:14 | |
*** RedW has joined #maemo-ssu | 15:14 | |
*** RedW has quit IRC | 15:39 | |
*** RedM has joined #maemo-ssu | 15:39 | |
*** RedM has quit IRC | 15:44 | |
*** RedW has joined #maemo-ssu | 15:44 | |
*** RedW has quit IRC | 15:48 | |
*** RedW has joined #maemo-ssu | 15:49 | |
*** NishanthMenon has joined #maemo-ssu | 15:50 | |
*** RedW has quit IRC | 15:59 | |
*** RedM has joined #maemo-ssu | 15:59 | |
*** corvinux has joined #maemo-ssu | 16:22 | |
*** RedW has joined #maemo-ssu | 16:28 | |
*** RedM has quit IRC | 16:28 | |
*** corvinux has quit IRC | 16:30 | |
*** RedW has quit IRC | 16:40 | |
*** RedW has joined #maemo-ssu | 16:40 | |
*** RedW has quit IRC | 16:46 | |
*** RedW has joined #maemo-ssu | 16:46 | |
freemangordon | sailus: those patches are already in the tree I use | 16:48 |
*** arcean has quit IRC | 16:50 | |
freemangordon | sailus: https://github.com/pali/linux-n900/commits/v4.6-rc1-n900?page=4 | 16:50 |
* freemangordon feels stupid, interrupts are in /proc, not in /sys/kernel/debug :( | 17:01 | |
freemangordon | sailus: "480bd400.mmu, OMAP3 ISP" interrupt count stays at zero | 17:06 |
sailus | Yeah, I think it's the same problem that's been seen before. | 17:15 |
sailus | I can't give other suggestions than to bisect that, or check the hardware configuration at register level with a working setup. :-( | 17:15 |
freemangordon | sailus: which registers I should check? | 17:16 |
freemangordon | only CCP or CCDC etc as well? | 17:16 |
sailus | At least system clock tree configuration and the control block register for the receivers. | 17:20 |
sailus | I think there once was a suspicion that this could have been caused by changes in the clock control. | 17:21 |
sailus | pinchartl might remember. | 17:21 |
sailus | Pali as well. Googling should turn up results. | 17:21 |
freemangordon | sailus: well, I already checked clocks, see http://pastebin.com/dM8Zf60M | 17:22 |
freemangordon | looks ok, ain;t? | 17:22 |
freemangordon | sailus: there *is* a difference in what stock kernel sets in ISP_CSI1B compared with mainline, howecer writing the same values with devmem2 doesn't make a differenece | 17:25 |
sailus | What are the values? | 17:32 |
freemangordon | sailus: http://pastebin.com/fZaJ1uvy | 17:33 |
freemangordon | the major difference I see is in CSI1B_DAT_START/SIZE | 17:34 |
freemangordon | sailus: hmm, isn;t there a typo when getting clock-lane? http://salottisipuli.retiisi.org.uk/cgi-bin/gitweb.cgi?p=~sailus/linux.git;a=commitdiff;h=3710afacd0dfec5504ebfbf5d5850bbb07ee28c6 | 17:35 |
freemangordon | sailus: also, why bus type should be CCP2 but not CSI1? | 17:36 |
freemangordon | afaik cameras on n900 are connected to CSI1b, not CCP | 17:37 |
sailus | freemangordon: I'll have to check that later on, I have a meeting starting soon... | 17:44 |
freemangordon | sailus: ok, in the meanwile I think I am on something | 17:45 |
*** Pali has joined #maemo-ssu | 18:23 | |
*** RedW has quit IRC | 19:07 | |
*** RedW has joined #maemo-ssu | 19:11 | |
*** RedW has quit IRC | 19:16 | |
*** RedW has joined #maemo-ssu | 19:17 | |
*** RedM has joined #maemo-ssu | 19:23 | |
*** RedW has quit IRC | 19:23 | |
*** futpib has joined #maemo-ssu | 19:50 | |
*** M4rtinK has joined #maemo-ssu | 20:22 | |
*** kerio has quit IRC | 20:23 | |
*** kerio has joined #maemo-ssu | 20:28 | |
*** kerio has quit IRC | 20:32 | |
*** kerio has joined #maemo-ssu | 20:35 | |
* DocScrutinizer05 remarks that many N900 video/stillcam apps only worked decently after a genuine Nokia camera app been started before (and after last boot) | 20:47 | |
DocScrutinizer05 | sticky config in sensor? | 20:48 |
*** kerio has quit IRC | 21:21 | |
*** kerio has joined #maemo-ssu | 21:23 | |
*** kerio has quit IRC | 22:17 | |
*** kerio has joined #maemo-ssu | 22:18 | |
*** NishanthMenon has quit IRC | 23:06 | |
freemangordon | DocScrutinizer05: not in the sensor, but in v4l | 23:06 |
*** futpib has quit IRC | 23:43 |
Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!