**** BEGIN LOGGING AT Sun Oct 25 02:59:57 2020 Oct 25 12:06:06 ouch, my N900 stopped detecting camera door open state Oct 25 12:06:40 isn't it just being slow? Oct 25 12:06:50 I mean, it can "ignore" it sometimes Oct 25 12:06:52 might be able to do some cleaning Oct 25 12:11:35 i'll look at it further a little later, but even reboot hasn't resulted in working camera detection Oct 25 12:12:39 :( Oct 25 12:14:28 while true; do cat /sys/devices/platform/gpio-switch/cam_shutter/state; done Oct 25 12:15:46 cam door is a magnet Oct 25 12:15:51 maybe it got weaker? Oct 25 12:15:56 do you have spare backpanel? Oct 25 15:24:37 KotCzarny: cam door is light sensor. magnet is the SD card present/cover sensor Oct 25 15:24:49 different back panel doesnt' change the situation Oct 25 15:25:17 i have always though cam door was also magnet based Oct 25 15:25:27 *thought Oct 25 15:25:59 it can't :) there would be too many magnetic fields to deal with, considering that all the magic is in the same area Oct 25 15:26:38 the other magnetic thing is the keyboard slide, since that's far from the camera mechanism Oct 25 15:31:29 i always believed the cam_shutter was some light sensor part of the reason the back of the shutter had blue film too on I recall Oct 25 15:31:41 yes Oct 25 15:31:52 did you try test above? Oct 25 15:32:03 is it hardware? Oct 25 15:32:09 yes, always "closed" Oct 25 15:32:14 hmm Oct 25 15:32:22 i believe it's h/w. guess my sensor has failed Oct 25 15:35:10 isn't the cam on a little daughterboard? does that have some flex or connections? I honestly can't remember Oct 25 15:49:57 the daughterboard thing is sd card reader, iirc Oct 25 21:18:46 sicelo: in case you need a workaround to use the camera, you can do something like `echo open >/tmp/open && mount -o bind /tmp/open /sys/devices/platform/gpio-switch/cam_shutter/state` Oct 25 21:19:06 then `killall camera-ui omap3camd` Oct 25 21:20:35 nice! will try that soonish Oct 25 21:21:16 i recall you have kernel patches for something in fremantle, or was it for the back cover removal? Oct 25 21:21:35 or maybe manually set file to "open" and make it 444 Oct 25 21:21:55 That's for the mmc driver, so it doesn't cut access to the SD card when the back cover is open. Oct 25 21:22:46 (doesn't affect what the rest of the system thinks about the gpio state, just disables that disabling behaviour in hsmmc) Oct 25 21:24:30 I don't think writing "open" to the existing "state" file will work, since it's in input mode, not output. Oct 25 21:24:46 Seems to just ignore it for me. Oct 25 21:25:22 and it doesn't seem to let me set it to be an output. Oct 25 21:26:04 yes. only the bind mount way seems to be working. thanks! Oct 25 21:30:10 I have done 444 trick for things like cam button. the back cover wouldn't be open tho, only the shutter. Oct 25 21:34:23 * Maxdamantus doesn't see how that would work. Oct 25 21:34:58 Normally, the kernel will just figure out the gpio state when you read from that file. Oct 25 21:35:28 You'd surely have to mount a regular file on top of it if you want it to behave like a regular file. Oct 25 21:37:22 and if you do the mount thing, you're probably not going to have other userspace processes writing into it, so 444 shouldn't be necessary. Oct 25 21:51:46 nice trick (mount -o bind) Oct 25 22:02:34 may need /disable to be ro too. as I say I have done this 444 trick to enable the cam_focus action while device is locked, which isn't normally possible. you can also do it to the proxy sensor but something in call-ui overrides/removes the 444 **** ENDING LOGGING AT Mon Oct 26 02:59:57 2020