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 2020-11-02

00:05:00_bilgusif it was the plugin itd show when the screen wasn't fliiped I'd think
00:05:51_bilgusI feel into the same idea that its the plugin but the plugin is right
00:27:35mendel_munkisI cant figure out the oscilloscope but I did catch the other off by one I knew was there.
01:32:01 Join lebellium [0] (
01:38:47 Nick Oksana_ is now known as Oksana (~Wikiwide@Maemo/community/ex-council/Wikiwide)
01:42:11***Saving seen data "./dancer.seen"
02:08:00 Quit lebellium (Quit: Leaving)
02:16:11 Join petur [0] (~petur@
02:16:11 Quit petur (Changing host)
02:16:11 Join petur [0] (~petur@rockbox/developer/petur)
02:18:19 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
02:18:20 Quit pixelma (Quit: .)
02:20:56 Join amiconn [0] (jens@rockbox/developer/amiconn)
02:20:56 Join pixelma [0] (marianne@rockbox/staff/pixelma)
02:36:19 Quit t0mato (Quit: Ping timeout (120 seconds))
02:52:11 Join t0mato [0] (~t0mato@
03:32:44 Quit Stanley00 ()
03:42:12***Saving seen data "./dancer.seen"
03:51:42 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
04:46:45 Join vmx [0] (
04:53:30 Join skx_ [0] (~r0b0t@unaffiliated/r0b0t)
04:57:02 Quit sakax (Ping timeout: 272 seconds)
05:26:08 Join ufdm_ [0] (
05:27:54 Quit ufdm (Read error: Connection reset by peer)
05:42:14***Saving seen data "./dancer.seen"
06:08:04 Quit prof_wolfff (Ping timeout: 240 seconds)
06:53:37 Quit ufdm_ (Quit: Leaving)
06:53:52 Join ufdm [0] (
07:10:32 Join l0g [0] (~unix@
07:11:49***Invited to ##chatterz by l0g!~unix@
07:20:27 Quit Stanley00 ()
07:22:00 Quit ubervison (Quit: Leaving)
07:27:33 Part l0g
07:42:16***Saving seen data "./dancer.seen"
08:03:23 Quit Acou_Bass (Ping timeout: 268 seconds)
08:05:11 Join Acou_Bass [0] (
08:16:48speachy_bilgus: I'd like to pick back up g#2568 −− it's a feature long requested. Since your vp rework the experiment works much better too
08:16:51fs-bluebotGerrit review #2568 at : usb: prompt user about what to do upon usb insertion (WIP) by Solomon Peachy
08:17:28speachybut now while it works great on standard contexts, in wps it breaks badly.
08:37:01edhelasspeachy "proof of concept for discussion and flagellation" :D
08:38:05 Nick Slasherii is now known as Slasheri (
08:39:18 Quit Slasheri (Changing host)
08:39:18 Join Slasheri [0] (~miipekk@rockbox/developer/Slasheri)
08:45:49 Quit Acou_Bass (Ping timeout: 264 seconds)
08:47:18 Join Acou_Bass [0] (
09:17:07_bilgusspeachy, does it just have issues staying on the screen? it probably needs to hold the screen or it needs to mark the vp as clean so the wps doesn't try to take it back
09:17:54speachyit's just using the standard gui_syncyesno_run() stuff.
09:18:12_bilgusoh that should already be aware
09:18:21_bilgusok i'll try it
09:18:44speachyunder the wps, it paints, and then goes a little nuts; the screen is partially repainted over, and the wps gets the keypresses.
09:19:10speachybut it doesn't recover unless I leave and then go back to the wps
09:23:26 Quit vmx (Ping timeout: 264 seconds)
09:42:00 Join massiveH [0] (
09:42:20***Saving seen data "./dancer.seen"
09:43:24_bilgusI don't see that in the usb code
09:43:45_bilguswps isn't aware of your screen perhaps
09:44:18_bilgusi'll still make guisync work but ultimately you need to push and pop your activity
09:44:28 Join johnb3 [0] (
09:45:45_bilgusmisc.h push_current_activity(enum current_activity screen); pop_current_activity(void); get_current_activity(void)
10:01:11speachyokeydokey, will try
10:07:56_bilgusyep that did the trick
10:08:22_bilgusI think you'll find the wps very agressive about taking its viewport back
10:12:59_bilguslooks like maybe you need to stop playback too though I guess it might be the wanted action
10:13:50speachywell, either you care about using your device in USB mode (in which case playback would stop), or you'd tell it to only charge, in which case you'd want playback to continue. either way you don't want to interrupt playback until the choice is made.
10:14:48_bilgusthe only other thing might be a timeout
10:15:25speachyok, the activity fixed the wps going nuts, but the WPS doesn't repaint properly afterwards. I guess we need to flag it as dirty?
10:15:36_bilguswell I guess in a build doesn't work situation you could do it through USB mode in the bootloader
10:16:04_bilgusit should be dirty automagically
10:16:22_bilgusyou popped the activity when you were done?
10:16:33speachywhile the message is on-screen, the wps progress bar continues to be displayed on the bottomw
10:17:00_bilgusyeah thats what I was saying about stopping pb
10:17:48speachyand after answering no, it doesn't repaint the background properly
10:18:06_bilgusok i'll mess with it for a few
10:18:16speachy(active elements on the wps are updated)
10:18:49speachymight be a bug in yesno.c instead
10:19:27speachyhmm, or wps not handling the activity switch properly.
10:20:39speachythere's a much more elaborate usb-insertion prompt screen in pamaury's old MTP patch
10:21:08_bilgusmy guess would be something in the wps not being aware
10:21:44_bilgusits refreshing through another mechanism most likely
10:22:19_bilgusand it marks stuff as clean when its done
10:27:28speachybut carringt this thing forward, my idea is to add a default preference for the mode −− usb, charge, prompt, and we'd default to prompt.
10:32:25 Quit beencubed (Ping timeout: 240 seconds)
10:33:03_bilgusthat sounds good
10:33:12_bilgusbecause it quickly gets annoying lol
10:35:26 Join beencubed [0] (~beencubed@
10:38:27speachyit's been a constant user complaint for quite a while now
10:44:03_bilguswps has its own stuff for this
10:44:17_bilgusgwps_enter and leave
10:46:42_bilguslooks like currently its all handled in the event loop for the wps
10:47:25_bilguswe don't want it to be exclusively its domain though so maybe there is another way
10:50:40 Join vmx [0] (
10:51:30_bilgusspeachy I assume we kick out the action on usb plug
10:52:11_bilgusI think what we will have to do is tread usb insertion as an action
10:52:27_bilguslike ACTION_USB_INSERT_PROMPT
10:52:44 Quit Acou_Bass (Ping timeout: 260 seconds)
10:52:51_bilgusthe wps will have to have its own handler to fit within the provided constraints
10:53:05_bilguswps.c has the loop
10:53:50_bilgus case ACTION_USB_INSERT_PROMPT
10:53:50_bilgus {
10:53:50_bilgus gwps_leave_wps();
10:53:50DBUGEnqueued KICK _bilgus
10:53:50_bilgus if (1 == gui_syncpitchscreen_run())
10:53:50_bilgus return GO_TO_ROOT;
10:53:51***Alert Mode level 1
10:53:51_bilgus restore = true;
10:53:53_bilgus }
10:53:55_bilgus break;
10:55:52 Join Acou_Bass [0] (
10:56:40_bilguswe can check at your current spot and pass the action if needed
10:58:06_bilguslike if (current activity == wps) shit we do we have a way to send actions from core I hope
10:59:03 Quit massiveH (Quit: Leaving)
10:59:07speachyI suppose anther appropach is to _leave_ the wps (ie not unlike the existing usb screen) and re-enter it upon exiting the prompt?
11:00:33_bilgussounds like a good idea
11:00:45_bilgusit probably should just return to root
11:00:58_bilgusthen we don't have to fight to keep ontop
11:01:34_bilgusthe menus will probably fight you as well since they are constantly refreshing
11:02:13_bilgusor instead of guisync if you make it a menu then nothing else will get it
11:02:27speachyI'd rather return to WPS if the choice is to ignore the insertion.
11:02:32_bilgusthe menu keep their own event loop
11:02:35speachyreturn to <last activity>
11:03:23_bilgusif you use a menu I think it will happen automagically when it release the vp let me try real quick
11:03:52***Alert Mode OFF
11:05:10 Quit Acou_Bass (Ping timeout: 256 seconds)
11:16:06 Join Acou_Bass [0] (
11:17:31_bilgusdamnit the menus look to auto kill when usb is in
11:23:40CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
11:23:40*speachy grumbles at the cat bumping mouse focus
11:28:09johnb3speachy The Hifi Walker I bought from *zon has a firmware version 1.3 on it. It refuses both to update to your patched image and to the official v1.2 I downloaded from their website :-(
11:28:55speachyjohnb3: huh. that's disappointing. any specific error, or just "Failed" ?
11:29:07johnb3Just failed
11:29:20speachygrab the eros q image instead and see what happens
11:29:48speachy(I wonder if that means that hifiwalker finally switched to a different target/platform ID)
11:30:46_bilgusok so same deal with menus wps fights it for focus
11:31:29_bilgusso I'd say store last activity and kill everything dump to root
11:36:09 Quit petur (Quit: Connection reset by beer)
11:40:21 Quit vmx (Quit: Leaving)
11:42:23***Saving seen data "./dancer.seen"
11:42:47johnb3speachy "Insert TF ... File error ... Failed". I tried both erosq and k.
11:43:07johnb3I haven't found the 1.3 on the web though.
11:47:16 Quit Acou_Bass (Quit: ZNC 1.7.5 -
11:51:14 Join Acou_Bass [0] (
11:53:47speachyjohnb3: it's most likely the platform id string. not much that I can do without a ROM dump though.
11:54:13speachy(well, unless you're willing to take yours apart and get a serial console)
11:55:25johnb3I was about to take apart the one with the broken usb socket, but not this new one.
11:55:55speachyyou could use the ingenic burn tool to extract the full ROM
11:56:02johnb3Any idea where to search for the v1.3 on the web?
11:56:45johnb3where can I find/download that?
11:57:55speachy(Alternatively it could be a datecode match that prevents downgrades?
11:58:31johnb3I was also thinking of prevention of "downgrade".
11:58:44johnb3by version string or date
11:59:07speachyIIRC there's no version string, just the date.
12:01:36 Join MrZeus [0] (~MrZeus@
12:02:50speachyeg the agptek h3 uses "eros_agptek" instead the "erosq" every other fw target uses
12:03:11speachyhifiwalker probably did something similar
12:03:14speachyfor v1.3 I mean
12:07:05johnb3for the ROM dump: is this the "cloner tool"? on
12:08:47speachywhoops, you barely beat me to it
12:09:45johnb3Windows complains the drivers lack a signature. I need to look it up how to do that without.
12:10:14 Join wodz [0] (
12:11:27 Quit johnb3 (Quit: Nettalk6 -
12:11:41wodzspeachy: How did you run my discovery util? I do it like this 1) in bootloader select tools -> adb 2) select tools second time but do not select anything 3) connect usb cable 4) adb push.... 5) adb shell 6) run discovery
12:11:52speachywodz: that's it
12:12:12speachyI'm running it on the erosq and x3ii though, my rocker is with bilgus
12:12:38wodzdamn. I tried hard on agptek and in wheezy chroot and couldn't crash it
12:13:39speachygranted my toolchain build is slightly different, but that shouldn't matter since it's dynamic linking anyway
12:13:42_bilgusif you need me to do something with it I can run stuff tonight
12:14:51speachyadb push that to /mnt/sd_0, adb shell, then ./discovery
12:15:29speachyif it goes boom, the problem's (presumably) with the toolchain/etc, if it works, then we have some subtle platform differences between teh rocker and the rest
12:15:33speachy(either is possible)
12:15:38_bilgusok i can try that here shortly
12:16:00wodzpreferably with some a2dp sink advertising device in discoverable state
12:16:55speachyjohnb3: btw, I emailed hifiwalker's support folks asking about the v1.3 firmware
12:17:00 Quit wodz (Quit: Leaving)
12:19:24 Join johnb3 [0] (
12:21:08_bilgusDefaultAdapter: /org/bluez/682/hci0
12:21:08_bilgusDiscovering : 1
12:21:12_bilgusDiscovering : 0
12:21:21speachyok, so it works.
12:21:30 Quit johnb3 (Client Quit)
12:21:32speachythe problem is some target difference
12:28:57speachyDefaultAdapter: /org/bluez/284/hci0
12:29:01speachy Discovering : 1
12:29:08speachySegmentation fault
12:29:12speachy(ignore the 'y')
12:29:16speachyidentical binary
12:29:58 Join johnb3 [0] (
13:12:17 Join lebellium [0] (
13:20:17johnb3speachy I installed the driver, but I am overwhelmed by the Cloner GUI and don't know what to do.
13:22:38johnb3So unless someone can give me intructions or have a Teamviewer session, I won't be of much help.
13:42:26***Saving seen data "./dancer.seen"
13:45:48 Quit beencubed (Quit: Leaving)
14:04:47mendel_munkisany advice on compiling a 2012 rockbox dev build? make tells me a bunch of things don't match the target pattern.
14:06:31speachyew. you're presumably going to need 2012-era toolchains too.
14:07:01speachyjohnb3: I haven't figured out how to get this thing to _read_ yet.
14:31:58 Quit johnb3 (Quit: Nettalk6 -
14:37:17mendel_munkisdo we have old snapshots available somewhere?
14:40:39speachywe have old releases, there's probably one from that era
14:44:38 Join petur [0] (~petur@rockbox/developer/petur)
14:44:47mendel_munkisbah didn't make it into stable until 3.14 wwhich I can build myself.
15:29:18_bilgusmendel I may have a backup at home of the OOOOLLLD toolchain on a vm
15:29:57_bilgusmay its so old its hard telling if I tossed it but probably still have it on a backup if not
15:31:37_bilgusoh and I think I made a vm at some point or at least had a backup of the one rb provided
15:32:04mendel_munkisthe oscilloscope bug exists in 3.14 (the earliest fuze+ version I can get my hands on) so it's definitely not because of any recent changes
15:32:32mendel_munkisI wanted to do a bisect but I suspect there is no good version to bisect against
15:33:08_bilgusthe logic of oscope isn't changing though
15:33:30_bilgusit has no idea the framebuffer is copied backwards when flipped
15:33:36mendel_munkisalso the commit message on the originaloscope commit sounds like the horizontal update may be hacky.
15:34:56_bilgusthe whole thing is hacky (backend) but it is sane
15:35:12_bilgustelling you I vetted the backend for a week straight
15:35:35mendel_munkisI'm not saying it's not
15:36:08_bilgusevery time I thought it was wrong, it was wrong but at driver level
15:36:49mendel_munkisI meant to say I think it's a bug in the driver since paumery wrote it
15:37:22_bilgusif its at driver level that gets broad real fast
15:37:46_bilgusthen you have to look at all trhe places people hacked around it
15:38:14*_bilgus did not find it fun
15:40:33mendel_munkisI want to check commit d9177c and seeif it shows up there.
15:41:40_bilgusif its always reproducible if you don't find any smoking guns then make the screen backend aware of the flip and move the vp 1 px to right or left accordingly
15:42:30***Saving seen data "./dancer.seen"
15:42:51_bilgusthe way the screen backend works is that it snapshots a portion of the bitmap and it copies it back scrolled
15:43:19_bilgusso that update line is drawn as a single pixel at screen edge
15:44:18_bilguswhats happening I suspect is that instead of copying Vert col 0 and pasting at v col 1 its overlapping a bit
15:44:34_bilgusgrabs col 0 too and propagates into a wall
15:45:28_bilgusbut the screen flip shouldn't have that power
15:46:01_bilgushow would it alter the coords above it we aren't snapshotting the actual LCD buffer
15:46:31_bilgusits more likely a driver bug as in it isn't set up right
15:47:13_bilgusbut we then shift everything by 1 px further and keep it out of that region in the LCD
15:47:57_bilgusis there some weird backporch scrolling function coming into effect here?
15:49:10_bilgusIDK just wandering atm I guess I have hardware in hand I'll start screwing around
15:49:58mendel_munkisI have a few theories but I have a class soon. I'll try them out in a few hours.
16:19:06braewoods_bilgus: how do i find out how much stack space is available in RB?
16:20:25 Quit JanC (Remote host closed the connection)
16:20:48 Join JanC [0] (~janc@lugwv/member/JanC)
16:28:36speachybraewoods: under the debug menu there's thread/stack dump that includes stack usage
16:29:05braewoodsspeachy: ok... i was wondering if a single 4K allocation would overload the stack.
16:40:05speachybraewoods: almost certianly
16:41:06speachyI think default stack size is in the range of 512B, though individual stacks can be larger, especially codecs.
16:48:47 Quit MrZeus (Ping timeout: 268 seconds)
17:12:20braewoodsspeachy: ok... good to know
17:17:38speachyplugins get a decent stack I think
17:18:32speachydefault stack size is 1K.
17:19:50speachybut individual plugins can and do go (much) larger.
17:30:20 Quit lebellium (Quit: Leaving)
17:42:34***Saving seen data "./dancer.seen"
18:08:28braewoodsspeachy: what's the difference between plugin_get_buffer and plugin_get_audio_buffer?
18:16:30gevaertsplugin_get_buffer() get free plugin RAM. plugin_get_audio_buffer() stops playback so you can have *all* RAM
18:19:54 Join MrZeus [0] (~MrZeus@
18:42:25mendel_munkistried taking an oscope screenshot and it hard paniced.
18:50:31_bilgusmendel_munkis, thats interesting
18:51:10_bilgusI assume its only on device?
19:05:22mendel_munkisthe panic or the bug?
19:08:48mendel_munkisyeah the bug is only on hardware
19:09:16mendel_munkisno testing until the battery drains. (from a full charge)
19:24:38 Quit MrZeus (Ping timeout: 256 seconds)
19:26:01 Quit petur (Quit: Leaving)
19:35:33 Quit bluebrother (Disconnected by services)
19:35:38 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
19:35:51 Join fs-bluebot_ [0] (
19:38:00 Quit fs-bluebot (Ping timeout: 256 seconds)
19:38:26_bilgusI had the same the second time it got wires
19:39:13_bilgushow do you screenshot from the device?
19:42:36***Saving seen data "./dancer.seen"
20:00:45 Quit ps-auxw (Ping timeout: 240 seconds)
20:06:49 Join ps-auxw [0] (
20:11:18 Quit ps-auxw (Disconnected by services)
20:11:26 Join ps-auxw [0] (
20:25:13 Quit speachy (Quit: WeeChat 2.8)
20:28:45mendel_munkisturn on screendump and plug in a usb
20:31:00_bilguswhat was the panic?
20:31:20mendel_munkisstkov USB
20:31:35mendel_munkisyeah I know it's not directly a lcd related panic
20:31:55_bilgusit quite possibly still is
20:57:19 Join kugel_ [0] (
20:57:20 Quit kugel_ (Changing host)
20:57:20 Join kugel_ [0] (~kugel@rockbox/developer/kugel)
21:00:54 Quit kugel (Ping timeout: 260 seconds)
21:06:23_bilgusmendel its just happens that the stack is too small it gets covered up by the extra afforded by the screendump but then its time to pay the piper
21:08:23mendel_munkisthat sounds like a bug to me
21:23:49_bilgusmendel_munkis, added 0x80 and screen dump says 94% used after screendump
21:24:01_bilgussounds like a good guess
21:24:37mendel_munkisadded 0x80 to reg_3_val?
21:24:57_bilgusno usb stack
21:25:00mendel_munkisoh to the stack size
21:39:43fs-bluebot_Build Server message: New build round started. Revision 611c187, 293 builds, 7 clients.
21:40:00_bilgusmendel there you go
21:42:37***Saving seen data "./dancer.seen"
21:47:11 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)
21:51:06braewoods_bilgus: random question. why do ROM chips come in different sizes for the smallest addressable unit?
21:51:15braewoodse.g., some are bytes some are words etc
21:51:42_bilgusmy guess would be speed
21:52:06braewoodsi see
21:52:12_bilgusalso arm likes word aligned but sheer speculation
21:52:24braewoodsi saw these on coldfire mainly
21:52:48_bilguscould just be whatever was available too
21:53:18_bilgusuh we have someone round here that is into ram stuff but hell if I could tell you whom
21:53:39_bilguslike in industry
21:53:47_bilgusmight ask in -community
21:54:16braewoodsmakes programming them a bit of a chore.
21:54:22mendel_munkismight it be time to remove experimental from there?
21:54:29braewoodsi suspect the byte order of these is dependent on the host CPU
21:54:54_bilguslol mendel IDK its still kinda an experiment
21:55:06mendel_munkisalso it's interesting that screendump usually doesn't panic. (although this isn't the first time)
22:03:37fs-bluebot_Build Server message: Build round completed after 1434 seconds.
22:03:41fs-bluebot_Build Server message: Revision 611c187 result: 4 errors 0 warnings
22:22:34_bilgusgotta figure it still might given the right / wrong circumstance..
22:24:15_bilgusI was incredulous when I saw 4 errors
22:25:02_bilgusI guess the builders failed
22:27:39 Quit efqw (Quit: Connection closed for inactivity)
22:28:53 Join beencubed [0] (~beencubed@
22:51:08 Quit tchan (Quit: WeeChat 2.8)
23:13:24 Quit [7] (Disconnected by services)
23:13:31 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
23:14:06 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
23:42:39***Saving seen data "./dancer.seen"

Previous day | Next day