Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2020-10-15

00:13:56speachyefqw: can you run 'amixer contents' on the console please?
00:27:35Stanley00yes, you may need to check if the sound driver is loaded too
00:28:06Stanley00on my m5, the module is loaded by the "player" application, so I have to do it manually
00:38:29***Saving seen data "./dancer.seen"
00:51:06 Join Soap_ [0] (~Soap@rockbox/staff/soap)
00:52:55 Quit Soap (Ping timeout: 246 seconds)
00:54:05 Join Soap [0] (~Soap@rockbox/staff/soap)
00:56:23 Quit Soap_ (Ping timeout: 260 seconds)
02:10:25 Quit kugel (Ping timeout: 264 seconds)
02:11:25 Join kugel [0] (
02:11:25 Quit kugel (Changing host)
02:11:25 Join kugel [0] (~kugel@rockbox/developer/kugel)
02:38:33***Saving seen data "./dancer.seen"
03:23:20 Join petur [0] (~petur@rockbox/developer/petur)
03:25:02 Join Rower [0] (
04:12:33Stanley00phew... I playing around with the button code for m5 and it's working now (sort of)... yay
04:13:08Stanley00it was my misunderstood how the button_read_device should return that messed up
04:38:36***No seen item changed, no save performed.
04:50:10 Quit braewoods (Quit: WeeChat 2.8)
05:03:10 Quit prof_wolfff (Ping timeout: 246 seconds)
05:10:49 Join braewoods [0] (~braewoods@learnprogramming/staff/braewoods)
05:11:42 Quit speachy (Quit: WeeChat 2.8)
05:20:01 Join prof_wolfff [0] (
05:49:50 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
06:38:38***Saving seen data "./dancer.seen"
06:57:00 Quit kugel (Ping timeout: 256 seconds)
06:58:25 Join kugel [0] (~kugel@rockbox/developer/kugel)
07:04:22 Quit pamaury (Ping timeout: 272 seconds)
07:18:54 Join speachy [0] (~speachy@
07:20:02 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
07:26:34 Quit Stanley00 (Remote host closed the connection)
07:28:30 Join Huntereb [0] (~Huntereb@
07:32:05speachyefqw: ok, I see the problem, guess that's why the code was commented out in the vortex tree.
07:32:17speachyand, ugh, poking through fiio's sources, they're bitbanging the i2c comms
07:44:17fs-bluebotBuild Server message: New build round started. Revision 4873a1a, 291 builds, 9 clients.
07:52:16speachyefqw: when this one's done it will at least get past that panic. But I'm glad the launcher worked!
07:59:30fs-bluebotBuild Server message: Build round completed after 912 seconds.
07:59:33fs-bluebotBuild Server message: Revision 4873a1a result: All green
08:06:59_bilgus_#g2845 framrbuffer rewrite appreciate any tesing anyone can do
08:08:11mendel_munkis_bilgus_: g#2845 or g#2811?
08:08:14fs-bluebotGerrit review #2845 at : Fix compile warnings (set-but-not-used) on big endian targets by Solomon Peachy
08:08:15fs-bluebotGerrit review #2811 at : LCD core move buf ptr and address look up function viewport struct WIP by William Wilgus
08:08:24_bilgus_lol g#2811
08:08:25fs-bluebotGerrit review #2811 at : LCD core move buf ptr and address look up function viewport struct WIP by William Wilgus
08:09:39mendel_munkisI a running a few other builds right now hopefully I will try it sometime this week.
08:10:28_bilgus_I haev the testing down for quite a few players sitting here but I really need some 2 bit screens and some remotes
08:10:50mendel_munkisah. my clip+ has a broknn screen so not much help there.
08:10:54speachythey expose something like 19 mixer controls, only three actually do anything on this hardware
08:11:32speachy_bilgus_: I think the ipodmini2g qualifies as 2bit?
08:11:37_bilgus_On the road w/ me I now have a Clip+ zip fuzev2 x3 , Rocker
08:11:58_bilgus_yeah one of those ipods has a reverse 2 bit screen
08:12:43speachy2811 builds on 2415?
08:13:35_bilgus_yeah mendel asked me to include his fb excise
08:14:19_bilgus_I figured the easiest way was just to make it parent
08:14:20mendel_munkisfor the record I don't mind being referred to as he
08:18:09_bilgus_the framebuffer code is (was?) a freaking mess hopefully this makes it better
08:20:19speachyanything in particular I should try?
08:20:24speachy(building now)
08:29:13speachymusic playback with a spinning cube is good.
08:35:46speachy_bilgus_: seems to work with everything I've thrown at it so far
08:37:11_bilgus_trouble areas are the core stuff that statusbar code is scary!
08:37:35_bilgus_for the most part plugins should JUST WORK (tm)
08:38:37speachyyeah, I didn't see anything out of whack with the main statusbar. granted I was just using the stock cabbiev2 theme
08:38:40***Saving seen data "./dancer.seen"
08:39:37_bilgus_backdrops store the offset from the current veiwport buffer it was really hard to track across the code
08:41:24_bilgus_I think even cabbie actually has quite a few viewports
08:43:42speachybut I'm not exactly a "power user" :D
08:47:19_bilgus_I already found one bug viewport_set_fullscreen I moved the sizing to set_default which works because they most of the code calls both but still...
08:47:35speachydo you think this would have any effect on the occasional display artifact that happens when eg a splash notification leaves behind someting?
08:51:37_bilgus_possibly but it should still be using the same underlying buffer so I don't think having too big a viewprt would do that
08:52:38speachyit's basically that the splash code expects the underlyng framebuffer to get completely repainted and (eg) the wps screen only repaints updates.
08:53:36_bilgus_I imagine that would happen at head then
08:53:39speachyhmm. I wonder if there's any way to force a complet screen repaint
08:53:47_bilgus_there is
08:54:31_bilgus_the action system has redraw and there are force full update flags in the them engine too
09:00:40speachythe softlock/unlock action already fires off ACTION_REDRAW apparently
09:01:11_bilgus_IIRC I had to do that to remove the messages
09:01:49 Join p3tur [0] (~petur@
09:01:49 Quit p3tur (Changing host)
09:01:49 Join p3tur [0] (~petur@rockbox/developer/petur)
09:02:54speachyI wonder if the problem is that the areas not getting repainted are the _background_ elements.
09:05:20 Quit petur (Ping timeout: 272 seconds)
09:08:54_bilgus_hmm and now the fuze+ sim is broken :/
09:09:13 Join massiveH [0] (
09:18:29speachybackdrop is not repainted under a full refresh.
09:19:32_bilgus_does taht happen at head too or only after the patch?
09:22:18speachyI didn't check under the patch; this was definitely on head
09:22:59speachyI think I have a fix
09:24:39speachyyikes, on the x3ii display is quite busted
09:25:17_bilgus_ I know what destroyted the sim I had to move to a void pointer to allow the lcd/lcd_remote probably the smae thing
09:25:35speachyonly the top 1/3 of the screen is dispalying anything
09:25:48_bilgus_luckily I can test that!
09:26:08_bilgus_hehe yeah moving to void* is throwing off the elem count
09:26:23speachychecking to see if the sim is busted the same way
09:27:12 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
09:27:35speachyquite broken, albeit different visually
09:29:35_bilgus_and this is why we test lol
09:32:03 Quit Stanley00 (Ping timeout: 260 seconds)
09:38:49_bilgus_ok fixed
09:41:37_bilgus_not sure how I didn't catch that I even checked it lol must have been 'lucky'
09:52:11speachyoh, there's also a race condition going on between the rocker/etc hotplug stuff I hooked up rockbox's own remount hooks.
09:52:42speachyif hotplug remounts things prior to rockbox getting to it, then rockbox will panic.
09:53:08_bilgus_oh I bet thats the issue on unplug 'invalid control file'
09:53:43speachypossibly. haven't seen that in a while though.
09:53:53_bilgus_that bug is so old every time I find something fkd I hope it fixes it lol
09:54:18_bilgus_I found its dirty cousin a while back
09:54:33speachyI've fixed a ton of stuff that could have led to it
09:57:21 Nick p3tur is now known as petur (~petur@rockbox/developer/petur)
10:00:12_bilgus_ok latest patch is up x3 works now along with the sims
10:02:49braewoodsit's amusing. the h1x0 series... the most common variant is the h120
10:03:11braewoodsi can't practically test the h100 bootloader simply because i can't find any units
10:03:46braewoodsbut they're sister units so... if the h120 works, they should both work... main differences are the RAM and storage.
10:03:53_bilgus_we run into that issue a lot braewoods
10:04:08braewoodsi ee.
10:04:34_bilgus_also totally appreciate if you could test g#2811
10:04:36fs-bluebotGerrit review #2811 at : LCD core move buf ptr and address look up function viewport struct WIP by William Wilgus
10:04:40_bilgus_on that remote screen
10:04:47braewoodssometime this week
10:05:25_bilgus_no hurry I still have to clean up & try to optimize
10:05:56braewoodsi'll eventually try to build an official V7 bootloader for the H120/H140
10:06:08braewoodsi'm hestitant to do the same for H100 since i can't test it
10:06:58braewoodsi'll see if the community has anyone that can help
10:07:05_bilgus_yeah especially if they are difficult to recover
10:07:18braewoodsthe H120, i can just buy a junker to test with
10:07:24braewoodsif it bricks it, so what? plenty more on ebay
10:07:28_bilgus_geaverts has a tower of rockbox unrivaled
10:08:01braewoodsgevaerts: you own any iriver h100 units? the h110, h115, etc?
10:08:06_bilgus_saying for the end user we don't want to break anyones baby
10:08:51braewoodsthe iriver h1x0 and h3x0 are riskier but i like their approach more
10:09:13braewoodsi think i will try to publish new bootloaders for the h3x0 series at some point
10:09:27braewoodsbut later
10:10:02braewoodsi don't want to push something out unless i'm fairly certain it actually works
10:10:10_bilgus_now the fun part lets see how much my address translation slows down the lcd
10:16:47_bilgus_nice on the x3 its within measurement error < 5%
10:16:58speachybut that's a pretty overpowered target
10:17:03_bilgus_some went up some went down
10:17:15_bilgus_yeah I'll be testing the zip next
10:17:16speachyeven with the impact IMO this is worth it
10:17:37_bilgus_agreed but i'll try to minimize it if at all possible
10:21:11_bilgus_I really don't like passing that void* around and having to cast every where but I think the alternative is going to make it more complicated
10:21:39braewoods_bilgus_: why is it cast?
10:21:41braewoodsor void*
10:21:51braewoodsi usually see that used because
10:22:00braewoodsyou need to take variable types
10:22:11braewoodsdue to C lacking union types
10:22:16braewoodsor derived types even
10:22:17_bilgus_remote lcd has a (usually) different data size for the frame buffer
10:22:55_bilgus_the address function has to handle both so it needs to be void* and then cast
10:23:03braewoods_bilgus_: meaning the underlying data is a different type or just different sized array?
10:23:23braewoodsI see.
10:23:34braewoodsif it helps you can try using type punning
10:23:38braewoodsabout the only alternative
10:23:52braewoodsi've seen people use that instead of void*
10:24:00braewoodsand somehow know which type is in use in any given usage
10:24:59braewoodsif our toolchains were C11 capable...
10:25:12braewoodscould use type punning with _Generic to set it automatically on the caller's side
10:25:23braewoodsvia macros
10:25:25braewoodsthank you CPP
10:25:48_bilgus_I'll keep my C thanks :p
10:26:17braewoods_bilgus_: well it's one thing i thought of but i believe gcc 4.9.x doesn't have _Generic yet
10:26:27*braewoods shrugs
10:26:53_bilgus_so If I put in a union for the data field that would get rid of some of it
10:27:19braewoodsit's useful but requires extra work
10:27:30braewoodssome of it can be automated with modern C11 stuff
10:27:35braewoods_Generic is rather useful
10:27:43braewoodsyou can automatically choose a type enum with it
10:27:54braewoodsso it has no runtime overhead, the setup anyway
10:28:01braewoodsyou still have runtime overhead for checking it if you choose to do so
10:28:15fs-bluebotBuild Server message: New build round started. Revision d544ce4, 291 builds, 9 clients.
10:28:16braewoodsbut void* has no such way to check it
10:28:20braewoodsyou just assume
10:28:33braewoodsthis has some manual type checking
10:28:38braewoodsmade possible at runtime
10:29:14speachyok, that should solve the mount-already-happened panic.
10:29:51speachyand the m3k is completely dependent on hotplug to do the remounting
10:30:53_bilgus_yes definitely a fan of builtin error checking
10:32:00_bilgus_I need to fire up my pc next time i'm home and leave it up to do builds for me clean builds take sooo long
10:36:18_bilgus_speachy hotplug as in a service running on the base OS?
10:38:43***Saving seen data "./dancer.seen"
10:39:25speachy_bilgus_: good news, x3ii display works now. until I segfaulted the oscilloscope plugin (also a bug with HEAD)
10:39:56_bilgus_that damn o-scope
10:40:14speachymore like damn hosted target issues
10:40:45_bilgus_I figured it was the last bug I fixed in it re appearing
10:41:27speachyI get some visual corruption in the FFT demo too when switching modes
10:42:01 Quit massiveH (Quit: Leaving)
10:42:32speachylooks almost like a picture-in-picture going on; a default viewport is being initiaalized that doesn't respect the screen properly. it does self-correct.
10:43:12speachywait, it's like it tries to render stuff in the statusbar.
10:45:48fs-bluebotBuild Server message: Build round completed after 1053 seconds.
10:45:49fs-bluebotBuild Server message: Revision d544ce4 result: All green
10:51:12gevaertsbraewoods: no. I have a h120 somewhere I think, but I don't know what state it's in, really. No 100 or 110
10:51:23braewoodsi thought as much.
10:51:48braewoodsi'll see if i can find anyone to help test a new bootloader though that unit is very rare from what i've seen
10:51:56braewoodsthe h120 is the most common to see pop up
10:51:59braewoodssame with h320
10:53:00_bilgus_speachy do you still get the statusbar while running fft?
10:53:30speachyno, it's .. maybe 20 pixels high and 100 pixels wide in the upper-left corner
10:53:35speachythe area of interest
10:54:05_bilgus_weird not seeing it on the x3
10:54:12_bilgus_i'll try the clipzip
10:54:42speachyrocker is probably a better one to try
10:55:28_bilgus_plugins get their viewprot based on whatever is default when they start
10:55:39speachyI don't recall seeing it before but I don't reember if I ever launched this before on the x3ii
10:56:28_bilgus_so maybe the better idea would be to have it explicitly set it up and use that as the default for people that want to access directly
10:56:49_bilgus_I already added explicit cleanup on plugin_exit
10:57:45 Quit pamaury (Ping timeout: 265 seconds)
10:58:05speachywhen you switch modes, you'll see the full fft going on-screen, but in that corner thre's a sorta-mini-fft being rendered
10:58:13speachyit goes away after a second or two
10:58:18_bilgus_my guess* is its probably getting remnants of the sb viewport
10:59:41_bilgus_speachy if you change the sb position to bottom does it move to the lower corner?
11:01:15speachyno change
11:01:47speachybut! there's also a central corruption too, and it's due to the splash text
11:02:13_bilgus_hmm, I also have something weird showing up in the fps test plugin too sounds like I still have a bit of work yet
11:03:22speachyyeah, it's related to the splash text.
11:03:48speachyit's the same size/dimensions as the splash text, same garbage in both
11:05:06_bilgus_FFS i'm testing on head!
11:05:39_bilgus_so the testfps plugin is broken on the clipzip at head
11:06:09_bilgus_makes me wonder when that happened
11:06:55speachyoh, new toolchain or old toolchain?
11:06:58speachymight matter
11:07:21speachynevermind then
11:07:40_bilgus_I figured i'll upgrade once my tree is empty again
11:07:41speachyhavne't managed to figure out where in the theme/skin code one forces the background to get repainted
11:15:46_bilgus_it uses a switch in lcd_clearviewport
11:20:01_bilgus_what I found off was that it switches to the default fb to do screen clear and then switches back
11:20:03 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:20:08_bilgus_odd not off
11:25:15 Quit petur (Quit: Connection reset by beer)
11:26:46_bilgus_ok so splash sets its own vp and then restores the default instead of the one inflight
11:26:46_bilgus_<_bilgus_> maybe set_vp should return the current one
11:28:25 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
11:33:31 Quit Stanley00 (Ping timeout: 265 seconds)
11:46:47speachywell.. it's leaving behind part of the splash
11:47:27speachythe "full repaint" that the wps is triggering isn't restoring the background over that area
11:47:42speachyeven when I force the wps to repaint all lines
11:50:54_bilgus_thats because it should be restoring the theme buffer but its restoring default
11:51:40_bilgus_worked before because the fb doesn't get changed when you set the default
11:51:48_bilgus_now it does
11:53:48_bilgus_now If I can only figure out who has it defined as returning void probably an extern somewhere
11:54:45_bilgus_great now I need to hunt for all users of this in core and change accordingly
11:55:02speachyjust remember, you're doing this for fun!
11:58:47_bilgus_it is fun and itll be even more fun when I get to revisit luas image code
11:58:59_bilgus_or thats what I'll tell myself
12:38:47***Saving seen data "./dancer.seen"
12:47:04_bilgus_nice I think that should resolve the rest of the issues there were a couple spots that I wasn't totally sure what the intent was so I left them be as restoring default_vp one was the eq screen but it appears to work properly
12:47:59_bilgus_and agter removing the giant block of color in test_fps its within 1% of the readings I could see at HEAD
12:48:09 Join MrZeus [0] (~MrZeus@2a02:c7f:70d0:6a00:95e7:97d6:1fb5:7a52)
12:56:32 Quit MrZeus (Read error: No route to host)
12:58:05 Join MrZeus [0] (~MrZeus@
13:01:40_bilgus_plugins still do the same probably a similar situation
13:17:53_bilgus_the plugins taht have an issue all seem to use the OSD plugin lib
13:18:46speachyugh. looks like the de-emphasis filter on the X3 should be disabled by default.
13:19:28speachywould be nice to make it an option.
13:20:08 Quit pamaury (Ping timeout: 260 seconds)
13:20:55 Quit Huntereb (Read error: Connection reset by peer)
13:25:58 Quit pixelma (Quit: .)
13:25:58 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
13:28:32 Join amiconn [0] (jens@rockbox/developer/amiconn)
13:28:33 Join pixelma [0] (marianne@rockbox/staff/pixelma)
13:29:07efqwspeachy: tried this one and I can see the launcher now
13:29:20efqwStill getting the same panic if I launch rb though
13:29:34 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
13:29:38speachyyou'll need the latest dev build
13:30:05speachyie newer than the last nightly
13:31:12efqwwhere are those builds located?
13:32:27 Join S|h|a|w|n [0] (~shawn156@unaffiliated/shawn156)
13:32:29speachynot on the builds page as there's no way to actually look at them yet.
13:33:58fs-bluebotBuild Server message: New build round started. Revision 0cde20f, 291 builds, 9 clients.
13:34:30 Quit Stanley00 (Ping timeout: 272 seconds)
13:35:58efqwgetting a lot of spam in the console :P
13:37:02efqwOh, and [ 109.744900] graphics fb0: Unknown ioctl 0x0 when the screen turns on
13:37:04speachyefqw: that's from the of, not rockbox
13:37:25efqw[ 131.483810] graphics fb0: Unknown ioctl 0x4 when the screen turns off
13:37:37speachythe ioctl is probably our doing.
13:39:15_bilgus_OK I got it narrowed down to the code in OSC that attempts to restore the original framebuffer data with bitmap_part
13:39:46_bilgus_same with the fft plugin
13:40:07_bilgus_and probably anyone else using OSD
13:41:08efqwI'll have to grep from init.d to find what exactly is doing this rmmod thing
13:41:13speachythat probably explains the segfaults that crop up on plugins on hosted targets
13:43:10_bilgus_quite possible< there is a bounds issue or just failure to set the data
13:43:37efqwPressing the power button doesn't seem to turn off the screen.
13:44:00efqwIt can turn on the screen after the backlight times out, but not the other way
13:44:30speachyefqw: as in the only mention of that module is in a commented-out init script, and the OF player binary
13:44:48speachyso you're inside rockbox?
13:44:50_bilgus_I don't remember us having that feature s that something extended from the OF efqw?
13:44:53braewoodsi posted about the new bootloaders.
13:45:00braewoodshopefully i'll get some community testing.
13:45:12efqwbut the player binary isn't actually running
13:45:27efqwso technically we shouldn't get the rmmod spam
13:45:49efqwI killed everything that is not a kernel process (other than rb) and it's still doing it
13:47:16_bilgus_braewoods, Pinned your topic
13:47:23braewoods_bilgus_: thanks
13:47:25speachyefqw: ok, found the offending bit.
13:47:37speachy(a serious wtf in the vortex code)
13:48:25fs-bluebotBuild Server message: Build round completed after 868 seconds.
13:48:27fs-bluebotBuild Server message: Revision 0cde20f result: All green
13:48:28fs-bluebotBuild Server message: New build round started. Revision 455a23b, 291 builds, 9 clients.
13:49:07efqwI'll stay on for a couple of hours from now so we could hopefully get some testing done.
13:49:43efqwShould we remove the adb stuff from the launcher?
13:49:46speachyefqw: building a fix.
13:49:49efqwIt's not like it's doing anything
13:50:05speachyefqw: eventually, yes, but due to how the menu is implemented it's a PITA
13:50:19speachyefqw: I'm still holding out hope of being able to actually use adb though
13:50:21*efqw nods
13:50:56speachyI figure that until we can come up with a non-flastool method of getting the launcher onto the m3k, there's no rush to polish it.
13:51:34efqwOh, and shutdown doesn't work
13:52:00efqwthis happens if I hold the power button to tell rb to shut down the machine
13:52:11efqwsame happens if I do `shutdown` in the console
13:52:38_bilgus_lol and yet you still went there!
13:52:50efqwlook at the timestamps of the messages, the UART Rx indicator lights up _solid_
13:55:04speachythis port is such a hacky mess.
13:55:43efqwwell, at least we have the source now, so it's possible to clean it up gradually at a bare minimum
13:55:45speachyefqw: does a short press on the power button activate the soft lock, at least?
13:56:39efqwConsole spam is gone.
13:57:01efqwAnd no, it does not activate soft lock. The touch panel is fully responsive
13:58:43efqwWell, not *all* of the rmmod stuff is gone, but at least it's not spamming the console anymore.
13:58:48speachybut it wakes the screen up, huh. ok, probably a broken keymapping.
13:59:23efqwThere is still a little bit of it during the startup process
14:00:12speachyok, that's probably in the launcher code.
14:00:25speachy(I'll have to rebuilt it)
14:00:36efqwThe hamburger menu button on the touch panel seems to be mapped to play/pause, when I touch it the player says "nothing to resume"
14:01:05efqwThe hardware play/pause button doesn't do this however, it's equivalent to "confirm" in the menu.
14:01:46speachyI didn't muck with this stuff at all
14:02:21efqwWhenever I press the volume buttons, the vol indicator on top of the screen disappears for 1s and then comes back.
14:04:07 Quit Rower (Ping timeout: 258 seconds)
14:05:22speachyok, the ioctl errors are from the screen blank/unblank code. it seems FB driver doesn't actually respond to standard commands. WTF.
14:06:38speachyefqw: do you have /sys/class/lcd on there?
14:07:10efqwYup, I have /sys/class/lcd/truly_tft240240_slcd
14:07:42speachyecho "4" > /sys/class/lcd/truly_tft240240_slcd/lcd_power
14:07:49speachyshould turn it off
14:08:02speachyecho "0" > /sys/class/lcd/truly_tft240240_slcd/lcd_power (back on)
14:09:05efqwNothing happens unfortunately.
14:09:38speachy(one more hacky lie..)
14:09:47*efqw shakes head
14:09:58speachydoes audio playback work?
14:10:15efqwI haven't tried, I'll get another MicroSD and dig up some music
14:10:24speachyno worries.
14:10:36speachyok, one thing I'd like you to do.
14:11:09efqwas you can tell the card I'm using is an ancient crap that can barely sustain 1MB/s sequential writes
14:11:30efqwcopying music to this would be a huge pain
14:12:50braewoodsefqw: ouch. my CF modded RBs can normally do around 10
14:13:01braewoodsmy H10 being an odd exception
14:13:53speachycat /dev/input/event* | hexdump -C
14:14:02speachyhit each key one by one
14:14:25fs-bluebotBuild Server message: Build round completed after 1557 seconds.
14:14:26fs-bluebotBuild Server message: Revision 455a23b result: All green
14:14:31speachy(please keep track of the order so we know what each one corresponds to)
14:16:28efqwHardware buttons do not show up at all.
14:17:40speachyok, try that again with /dev/input/event1 (vs event0)
14:20:19efqwWorks. I'll go grab something to eat and compose a table soon.
14:33:28fs-bluebotBuild Server message: New build round started. Revision 2e07223, 291 builds, 9 clients.
14:38:49***Saving seen data "./dancer.seen"
14:48:17 Join johnb7 [0] (
14:49:15fs-bluebotBuild Server message: Build round completed after 946 seconds.
14:49:18fs-bluebotBuild Server message: Revision 2e07223 result: All green
14:51:51_bilgus_nope nm still doesn't have the UI one
14:52:09speachy_bilgus_: good find, but I think we already have all of that stuff
14:52:15speachy(except perhaps the 4770 stuff)
14:52:46 Join lebellium [0] (
14:53:27_bilgus_came through the forum
14:53:50speachycloning it all just in case
14:55:15_bilgus_same lol
14:55:41speachywhat's another 800M of disk space
14:56:50efqwaudio playback works great
14:57:01speachyhmm. it's probably safe to nuke the pre-migration rockbox snapshots.
14:57:22_bilgus_I wonder how closly the 70 series ui manual matches the 60's
14:57:25efqwonly tried flac but I assume if flac works everything else should work too?
14:57:43_bilgus_those 4770 manulas are something I haven't seen before
14:57:58_bilgus_efqw a faulty assumption..
14:58:00speachyvolume control?
14:58:30efqwvolume control works just fine
15:01:11efqwSoftlock works in
15:01:28speachyefqw: well, that's not from anything I changed!
15:05:31lebelliumbraewoods: "only" €99 :)
15:05:42braewoodstalk about rare
15:06:02speachyit has the rockbox tax pre-applied
15:09:03braewoodsthat didn't show up in my searches
15:09:06braewoodsnot surprising i guess
15:10:13efqwhw buttons done
15:11:01lebelliumBTW, aside the store capacity, what's the difference between the H100/H115 and H120/H140? I never understood why those are 2 different Rockbox targets with separated builds
15:11:21braewoodslebellium: RAM.
15:11:27braewoodsH100 has 16MB
15:11:32braewoodsH120 has 32MB
15:11:36braewoodsso they have to be different
15:12:06braewoodsthe H3x0 devices apparently all have the same RAM
15:12:29braewoodsthough i must say i love the H120 series
15:12:38braewoodsthey're very featureful
15:12:39lebelliumah interesting
15:12:49braewoodsone of the better rockbox ports
15:12:54_bilgus_ok so what is going on with OSD is that it is copying from the same buffer for the clear so it needs to tak into account the direction
15:13:14braewoodsi can actually load rockbox into ROM with the iriver h1x0
15:13:21braewoodssomething i didn't know was possible lol
15:13:40lebelliumYes I like my H140 too although I never use it like pretty much all my DAPs :D
15:15:41lebelliumI tried the optical out with my AV receiver optical in but it seems that the red light is not powerful enough, my receiver doesn't get the signal. I don't know if the optical out gets weaker over time or if it was an iriver design fault
15:17:01_bilgus_lebellium, might try a different interconnect I know the yamaha ones I got were glass but Ive seen some lower quality ones that look to be fishing line
15:20:51 Join pixelma_ [0] (marianne@rockbox/staff/pixelma)
15:20:52 Quit pixelma (Killed ( (Nickname regained by services)))
15:20:52 Nick pixelma_ is now known as pixelma (marianne@rockbox/staff/pixelma)
15:20:55 Quit amiconn (Ping timeout: 240 seconds)
15:21:03 Join amiconn_ [0] (jens@rockbox/developer/amiconn)
15:21:04 Nick amiconn_ is now known as amiconn (jens@rockbox/developer/amiconn)
15:22:16efqwspeachy: this is the soft buttons sans the scrolling stuff
15:22:37efqwI'll put the scrolling stuff in a separate paste
15:30:37 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
15:32:26efqwok, this should cover all of the cases
15:35:37 Quit Stanley00 (Ping timeout: 264 seconds)
15:41:28 Quit johnb7 (Ping timeout: 265 seconds)
15:45:06braewoodsi think next month i will try to acquire an iriver h320 to test an updated bootloader for as well
15:45:38braewoodswould be nice to do all 3 but...
15:45:50braewoodsthe hardest to find unit is the h100
15:53:02braewoodsspeachy: is rbutil capable of using HTTPS?
15:53:40speachythere are some issues with (I think) v1.4.0
15:54:10braewoodsi'm asking because i would like to update it in git to use https to download files
15:54:16braewoodsfor greater confidence in their integrity
15:54:38braewoodsthe website appears to have no issue with it
15:54:38speachythat's bluebrother's purview
15:55:07braewoodsso only obvious problem is whether SSL is supported in rbutil
15:57:24braewoodsbluebrother: what're your thoughts? i was hoping to update rbutil to use HTTPS by default in newer releases.
15:58:01braewoodsit appears to be a simple set of changes is all that are needed
16:00:16bluebrotherbraewoods: the problem with https isn't Rockbox Utility, it's Qt. Qt itself can do ssl, but for older buikds it was disabled. That includes 1.4.0.
16:00:47braewoodsbluebrother: so what's the harm in enabling it in newer ones?
16:00:51braewoodsthe older ones continue to work, no?
16:00:59bluebrotherI actually wanted to enable ssl, after bringing everything up to date and before making the next release.
16:01:11braewoodsOk. i'll leave it be then.
16:02:15bluebrotherwell, if you change the urls in rbutil.ini there is no problem for older binaries. But the webserver forcing ssl breaks them.
16:03:26braewoodsbluebrother: ah. i wasn't going for that.
16:03:26bluebrotherwe had a discussion about ssl some years ago The main thing is that Qt uses Openssl, which uses a different license. Since we wanted to have an all-in-one binary on Windows this caused some concerns w/ licensing.
16:04:04braewoodsopenssl's license changed awhile ago.
16:04:20braewoodsnot sure if it matters for that
16:04:30bluebrothersome dev binaries I made already have Qt w/ openssl linked. That was mainly because I used mxe to build them and didn't realize that :)
16:04:32braewoodsor is in the process
16:04:39braewoodsapache 2 license? apparently
16:04:54braewoodsis that GPL incompatible?
16:04:57braewoodsi'm not sure honestly
16:05:42bluebrotherwith the changes I want to make wrt bootloader installation we won't have a single binary anymore, so we can as well dynamically link, leaving those issues aside.
16:05:55speachywell, RB is GPLv2+, and GPLv3 is ASL2 compatible, so it is technically kosher. :D
16:06:18bluebrotherand I'm not sure if those concerns would actually turn out to be a oroblem. We're oss anyway.
16:06:31braewoodsin 2018 openssl was completely switched to
16:06:53bluebrotherback when we had that discussion openssl wasn't ASL.
16:06:59braewoodsi see that
16:07:18bluebrotherso I guess that isn't an issue anymore.
16:08:27braewoodsspeachy: eventually i intend to make some bootloaders that we'll try to finalize as version 7 for the iriver h100, h120, h320 series
16:08:38braewoodspretty sure h320 also has issues with CF cards that is fixed in git
16:08:48speachyyep, it's the same basic platform
16:09:10bluebrotherThe missing ssl support basically stems from the goal to have a static binary on Windows, and for that I ended up building Qt myself.. And that disabled ssl support, and since ssl waasn't really used back then (like 10 years ago) nobody noticed / bothered.
16:11:16bluebrotherbtw, the ipod 5.5g came as 32M and 64M variant. At some point a way was found to determine the RAM size during runtime and those were merged. Maybe something similar could be done for h100 / h120. Though afaiu they'd still need different bootloaders.
16:14:13efqwLong pressing the power button opens the pandora's box
16:14:53efqwrb panicked with "Cannot get control 'DACL Playback Volume' info"
16:17:44speachysounds like I need to bite the bullet and set up the cloner tool (and disassemble it so I can get a serial console)
16:18:52efqwI mean the cloner should be sufficient
16:19:08efqwYou can load the serial gadget module and get a console over USB
16:19:23efqwThat should be good enough for most use cases anyway
16:32:10speachybluebrother, I'm wanting to add the Eros Q/K (and clones) to rbutil, but in a way that preserves the appropriate OF for each OEM. Unfortunately so far it looks like they share the same USB VID+PID, possibly differing only in the descriptor strings.
16:32:28speachycan rbutil match on those, or is it vid/pid only?
16:35:09braewoodsspeachy: should be viable... i know the linux kernel has all sorts of ways to match hardware specific workarounds or quirks to specific platforms
16:37:18_bilgus_but does windows?
16:37:24braewoodsno idea.
16:37:36braewoodsi know you can probe information from /sys
16:37:38braewoodsfor Linux
16:37:44speachyI mean, libusb will do that for us just fine, the question is what rbutil currently supports
16:38:14braewoodsif it doesn't we can probably add it
16:38:33braewoodsthe question i start from is whether the underlying layers support it...
16:38:52***Saving seen data "./dancer.seen"
16:39:24speachyjust don't want to reinvent a wheel that's already there.
16:39:38speachy(and greatly preferring to not touch C++)
16:40:25braewoods(void) cpp;
16:58:39braewoods_bilgus_: remote lcd works. what did you want done?
16:59:29_bilgus_just fire it up, see if it displays properly on top of 2811
16:59:44_bilgus_assuming it works properly at head..
17:00:38 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
17:01:22_bilgus_ g#2811
17:01:24fs-bluebotGerrit review #2811 at : LCD core move buf ptr and address look up function viewport struct WIP by William Wilgus
17:02:59braewoods_bilgus_: are you sure this is for the h100 / h120 series?
17:03:06braewoods_bilgus_: i only see h300 modified
17:03:29fs-bluebotBuild Server message: New build round started. Revision df8b817, 291 builds, 9 clients.
17:03:49_bilgus_its linked in core those players were just ones using lcd_identifiers they had no business touching
17:04:13_bilgus_probably as an 'optimization'
17:04:20braewoodshow do i build from this?
17:04:27braewoodsor is there a build already?
17:05:10_bilgus_no builds
17:05:46_bilgus_see the guilde to compiling for hints to do it from patches or you can do it with git-review
17:06:46_bilgus_if you don't have git-review apt-install and thenm git-review -d I0d04c3834e464eca84a5a715743a297a0cefd0af
17:07:45mendel_munkishuh. android ndk no longer works without clang. this may be why I was having build issues.
17:08:11_bilgus_itll pull in as william/wilgus.... then create device folder cd into it do ../tools/configure pick device and press 'n' make && make zip
17:09:11braewoodshave to build a new toolchain first
17:09:46_bilgus_thatll be a bit go ahead with that but i'll put up a build for you
17:11:54 Join cockroach [0] (~blattodea@pdpc/supporter/active/cockroach)
17:11:57 Join ac_laptop [0] (~ac_laptop@
17:12:29speachy_all_ of the Eros-Q/K series and clones share the same USB VID/PID. Ditto for the xDuoo X3ii and X20.
17:12:43speachywhich is bad, because the latter definitely is _not_ compatible
17:14:35_bilgus_call the china USB id ?
17:15:31speachyalso the AGPTek H3 cannot share the same update image as the other Eros clones.
17:15:52speachy(Artificial restriction; they changed the model string that's checked against)
17:16:06lebelliumI should get mine in a few days btw
17:17:17fs-bluebotBuild Server message: Build round completed after 829 seconds.
17:17:23fs-bluebotBuild Server message: Revision df8b817 result: All green
17:31:39 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
17:31:55_bilgus_there is still a bug in the OSD code but it should work fine besides a bit of garbage on update
17:34:31fs-bluebotBuild Server message: New build round started. Revision 1a76bc4, 291 builds, 10 clients.
17:36:26 Quit Stanley00 (Ping timeout: 272 seconds)
17:39:07 Quit lebellium (Quit: Leaving)
17:47:45fs-bluebotBuild Server message: Build round completed after 794 seconds.
17:47:46fs-bluebotBuild Server message: Revision 1a76bc4 result: All green
17:55:25 Quit pamaury (Ping timeout: 264 seconds)
18:35:07 Quit _bilgus_ (Ping timeout: 260 seconds)
18:37:02 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
18:38:54***Saving seen data "./dancer.seen"
18:46:25 Quit _bilgus (Ping timeout: 240 seconds)
18:51:11 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
19:28:46 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
19:33:28 Quit Stanley00 (Ping timeout: 260 seconds)
19:47:46 Join fs-bluebot_ [0] (
19:48:02 Quit bluebrother (Disconnected by services)
19:48:07 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
19:50:10 Quit fs-bluebot (Ping timeout: 256 seconds)
20:10:01 Join Oksana [0] (~Wikiwide@Maemo/community/ex-council/Wikiwide)
20:10:12 Quit _bilgus (Ping timeout: 260 seconds)
20:10:22 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
20:14:55 Quit _bilgus (Ping timeout: 240 seconds)
20:15:02 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
20:20:07 Quit _bilgus (Ping timeout: 260 seconds)
20:20:33 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
20:31:57 Quit _bilgus (Read error: Connection reset by peer)
20:35:25 Quit MrZeus (Ping timeout: 240 seconds)
20:38:58***Saving seen data "./dancer.seen"
21:31:50efqwHow much effort would be required to implement a soft keypad?
21:32:33efqwThis would make those x1000[e] touchscreen-only devices viable targets.
21:33:00efqwWell, sans the tiny stuff like the FiiO M5
21:36:27 Quit ac_laptop (Ping timeout: 258 seconds)
21:55:25 Quit cockroach (Quit: leaving)
22:06:27 Join _bilgus [0] (~bilgus@2605:a000:1301:89f6:8880:3cbe:369:d070)
22:23:19 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
22:32:05 Join ac_laptop [0] (~ac_laptop@
22:38:59***Saving seen data "./dancer.seen"
23:18:58braewoodsefqw: it would probably require you to split the screen.
23:19:07braewoodspart would be for the regular UI
23:19:23braewoodsthe rest would be for translating inputs for the keypad
23:19:42braewoodsit would require some kind of touch translation system
23:20:04braewoodsyou can get the raw inputs from /dev stuff
23:20:38braewoodsso you're talking 1), changes to the UI so it only uses part of the screen.
23:20:59braewoods2), also making it draw whatever keypad arrangement you devise
23:21:07braewoodsthen third
23:21:27braewoodsa system for translating the keypad zones to events internally
23:25:01efqwI have to admit that's much more than I could handle.
23:25:45braewoodsyea.. desktop linux has that all solved but these devices aren't nearly beefy enough for that.
23:26:03efqwThe m3k was easy because the kernel implemented the touchpad stuff at driver level, so we never had to deal with it in the first place.
23:26:38efqwAll we get are fixed keycodes that corresponds to the physical button regions on the hardware.
23:26:42braewoodswith Linux though touchscreen support is limited just to event reporting
23:26:52braewoodsit's up to userspace to do something useful with it
23:27:19braewoodsthere's a whole stack of support software in userspace for touchscreens
23:27:23braewoodsudev, etc
23:27:54braewoodswhich is ultimately exposed in xinput2
23:28:13efqwyea, that might be too much for the poor x1000[e] with only 32/64MB of ram
23:28:22braewoodsit would.
23:28:59braewoodsmy favorite TS to date were wacom tablets
23:29:05braewoodsvery nicely supported
23:30:12efqwWriting a simple daemon that processes the touchscreen events might be an option, and then we can just use swipe gestures for navigation.
23:31:23efqwup/down for scrolling, tap for enter
23:32:03efqwThis is the minimal amount of gestures that would make me consider the touchscreen to be "useful"
23:33:44braewoodsefqw: you basically have to create your own input stack to handle desktop-like devices.
23:33:53braewoodssince there's no HW buttons
23:33:59braewoodsthose are simple to handle
23:34:11braewoodstouch requires you to do a lot more
23:34:33braewoodsit shouldn't be too hard if you're used to writing GUI toolkits
23:34:51braewoodswhich is basically what you'd have to do to port to a touchscreen
23:34:55braewoodswith limited resources
23:35:13efqwThere are hw buttons. My target is the FiiO M3 Pro, which has the same amount of hw buttons as the M3K does.
23:35:14braewoodsmaybe there's some libraries that can help though
23:35:30braewoodsi recall directfb
23:36:36efqwso technically we're only missing ffwd/rewind/up/down/confirm/back
23:37:02braewoodsi don't know if directfb would work on ARM
23:37:15Stanley00I'm not sure, but when I tried with HAVE_TOUCHSCREEN for my M5, rockbox somehow can handle touch gesture itself, eg I can scroll the screen by swipe up/down
23:37:36efqwThat's good to know
23:38:24Stanley00it's just good part, bad part is that I cannot understand/trace how it worked, so I just skip HAVE_TOUCHSCREEN all together and use it just as simple touchpad, with 9 button region
23:38:27efqwI'm not all that interested in the M5 because the battery is way too small.
23:38:44Stanley00but you can trace HAVE_TOUCHSCREEN to see related code
23:39:36efqwAnd since it's quite a pricey device, I'm probably not gonna buy one just to test the touchscreen stuff. Bluetooth is nice but it's not all that useful if I can't get it to work with rb.
23:39:56Stanley00yup, I just let the battery die yesterday becasue I thought I already shutdown rockbox, but it still on, and draw battery overnight, doing nothing =]]
23:40:16efqwah, you might hit the same problem as I did
23:40:55efqwThe fiio kernel on my m3k does this if I just call `shutdown`
23:41:10efqwIt freaks out and doesn't actually power off the device.
23:41:33Stanley00ah, thanks for the info, let me check for that on my fiio later today
23:42:19efqwThere's no way to check it directly unless you take the player apart and connect the UART console
23:42:49efqwI pulled my m3k apart and hooked up the serial console.
23:43:59Stanley00hmm... too bad, I don't have tools for that
23:44:10Stanley00best I can do is just adb shell
23:44:50efqwoh, the m5 had adb shell?
23:45:27efqwthat's nice, the m3k had the android stuff disabled in the kernel
23:45:39Stanley00I *accidentally* discover that feature when looking in their firmware
23:46:02efqwwithout taking apart the device, you have to overwrite the system partition to get a root shell over USB gadget serial
23:46:38Stanley00lucky, adb shell is already in root, that's why I'm here :3
23:47:07efqwI wonder how's the m3 pro like. I can see the rootfs but I don't know about the kernel, I'd have to buy one of those to see if adb actually works or not.
23:47:34Stanley00anyway, M5 firmware looks more like in development state, many debuging info is still there
23:47:56efqwthe player binary on the m3k and m3 pro are unstripped
23:48:19Stanley00on m5, they have player.symbol and also gdb binary there
23:48:28Stanley00I heard that m3k also have gdb too
23:48:33efqwyeah, the m3 pro had gdb too
23:48:40efqwm3k doesn't
23:49:40Stanley00ah, I see
23:50:43efqwby the way, check your /data/userfs/app.log
23:51:05efqwyou might want to delete that and symlink it to /dev/null
23:51:43efqwon the m3k it writes some incredibly verbose log to the flash, with zero compression
23:51:49Stanley00thanks, I already did that, and also kernel.log
23:51:56Stanley00I don't need those for now
23:51:58efqwyeah, that's good :)
23:52:27 Quit [7] (Ping timeout: 260 seconds)
23:52:42 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
23:52:48efqwwhat annoys me is that the data partition is a separate ubi device, not an ubi *volume*
23:53:11efqwwear levelling doesn't work across different ubi devices
23:53:53efqwI understand why they did this (it's easier to update, since the recovery doesn't have to deal with ubi volumes) but it's incredibly bad for the flash
23:54:18efqwIt doesn't help that the database is written to that tiny partition too
23:55:20Stanley00I think the devs at FiiO just have so much pressure that they cannot think out carefully :P

Previous day | Next day