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 2020-09-02

00:21:11 Quit TheEaterOfSouls (Remote host closed the connection)
00:39:52***Saving seen data "./dancer.seen"
00:40:41 Join TheEaterOfSouls [0] (~souls@unaffiliated/theeaterofsouls)
00:53:12 Quit Tsesarevich (Read error: Connection reset by peer)
00:53:20 Join Tsesarevich [0] (Tsesarevic@fluxbuntu/founder/joejaxx)
01:20:59 Join ZincAlloy [0] (
01:25:25 Quit ZincAlloy (Ping timeout: 264 seconds)
01:32:44 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
01:54:10 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:b121:52e3:8bc5:2942)
01:59:15 Quit ZincAlloy (Ping timeout: 272 seconds)
02:02:44 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
02:39:54***Saving seen data "./dancer.seen"
02:56:53 Quit pamaury (Ping timeout: 260 seconds)
02:58:13 Join petur [0] (~petur@
02:58:13 Quit petur (Changing host)
02:58:13 Join petur [0] (~petur@rockbox/developer/petur)
03:03:59 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
03:22:23 Quit sakax (Quit: Leaving)
03:46:31 Nick mendel_munkis_ is now known as mendel_munkis (
04:06:44 Quit pamaury (Ping timeout: 240 seconds)
04:36:10 Join lebellium [0] (
04:39:57***Saving seen data "./dancer.seen"
05:41:47 Quit scorche (Disconnected by services)
05:41:50 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
05:41:53 Join lebellium_ [0] (
05:45:59 Quit lebellium (Ping timeout: 272 seconds)
06:03:27 Join Rower [0] (
06:03:31 Quit Rower (Client Quit)
06:17:32 Quit Oksana (Ping timeout: 244 seconds)
06:40:00***Saving seen data "./dancer.seen"
07:05:06 Join vmx [0] (
07:07:01 Join ac_laptop [0] (~ac_laptop@
07:11:37 Quit prof_wolfff (Ping timeout: 264 seconds)
07:19:14 Join Oksana [0] (~Wikiwide@Maemo/community/ex-council/Wikiwide)
07:19:27 Quit ac_laptop (Ping timeout: 265 seconds)
07:35:40 Join Moarc_ [0] (
07:35:44 Quit Moarc (Ping timeout: 240 seconds)
08:33:41fs-bluebotBuild Server message: New build round started. Revision bb6fc21, 280 builds, 8 clients.
08:36:59 Quit Oksana (Ping timeout: 260 seconds)
08:40:02***Saving seen data "./dancer.seen"
08:55:25fs-bluebotBuild Server message: Build round completed after 1304 seconds.
08:55:27fs-bluebotBuild Server message: Revision bb6fc21 result: All green
09:44:41 Join __Bilgus [0] (41ba23be@
10:13:42 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:14:52 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
10:22:56 Quit __Bilgus (Remote host closed the connection)
10:26:47 Join Huntereb [0] (~Huntereb@
10:35:23 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
10:40:05***Saving seen data "./dancer.seen"
11:25:28 Join __Bilgus [0] (41ba23be@
11:50:06 Quit petur (Quit: Connection reset by beer)
12:13:25 Quit __Bilgus (Ping timeout: 245 seconds)
12:38:45fs-bluebotBuild Server message: New build round started. Revision 35371df, 280 builds, 8 clients.
12:40:07***Saving seen data "./dancer.seen"
12:40:11 Join __Bilgus [0] (41ba23be@
12:44:13__Bilgusspeachy 2726 did you find registers getting clobbered?
12:44:50__Bilgusdebug menu hw info is finished it should be pretty easy to add new items
12:56:19speachyI saw, I like!
12:56:26fs-bluebotBuild Server message: Build round completed after 1062 seconds.
12:56:28fs-bluebotBuild Server message: Revision 35371df result: All green
12:57:07fs-bluebotBuild Server message: New build round started. Revision 1a8939c, 280 builds, 8 clients.
12:57:21speachyI didn't see 'em clobbered, but we were inconsistent with our option usage so I cleaned it up.
12:58:27__BilgusI did notice ROLO doesn't re-enable the ADC on reboot I'm guessing it has something to do with disabling all the interrupts
12:58:58speachyour startup/init code should explcitly enable everything we care about
12:59:49__Bilgushmm wonder whats up then better check at head again before I even start looking
13:01:20__BilgusI bet one of those gpios get setup by the bootloader but not by us
13:01:39speachyeither the xvortex bootloader build or stage1
13:02:06speachyI committed a change to disable the MSC0 and LCD controller
13:02:21__Bilgusi'll investigate it while I do the button patch its probably pertinent to my goal
13:02:37speachyI think we can de-clock the AIC too, as we only use the I2S peripheral.
13:04:22 Join johnb2 [0] (
13:04:30__BilgusMSC0 isn't the first sd card?
13:04:52speachywe're using MSC1 + MSC2.
13:05:17__Bilgusthe more we shutdown the more power available to the rest so if its a PS issue that might help a lot
13:05:21speachy2 is sd0 and 1 is sd1
13:05:38__Bilguswhy, because thats why
13:06:13speachyeven if we don't have a PS issue it'll save a bit of juice. I suspect the analog portions of this player dwarf the SoC though.
13:06:33__BilgusI figured thats whay the relay click for lol
13:08:18__Bilgusnope rolo seems to be fine now glad I checked @Head
13:12:52fs-bluebotBuild Server message: Build round completed after 945 seconds.
13:12:53fs-bluebotBuild Server message: Revision 1a8939c result: All green
13:12:54fs-bluebotBuild Server message: New build round started. Revision f913829, 280 builds, 7 clients.
13:17:07speachyhmm, I need to dial down PIXCLK too; we're not using it on the X3.
13:18:51__Bilgusstill get crashes while playing + cube / fire
13:20:01speachyyeah, and the cache rework is further destabilizing
13:26:21__BilgusAs we keep going through code we might find that smoking gun
13:26:46fs-bluebotBuild Server message: Build round completed after 834 seconds.
13:26:48fs-bluebotBuild Server message: Revision f913829 result: All green
13:39:26fs-bluebotBuild Server message: New build round started. Revision 546212a, 280 builds, 8 clients.
13:43:02speachyPIXCLK dialed way back
13:48:04 Quit sakax (Quit: Leaving)
13:49:07__Bilgusdid you note the note in the manual too?
13:49:09__BilgusNOTES: 1 It is strongly recommended that software set the “LCD Module Stop Bit” in PMC to shut down LCDC clock supply to save power consumption after disable LCDC. Please refer to PMC for detailed information.
13:51:53speachyyeah, I did that earlier, via __cpm_stop_lcd()
13:54:12__Bilgusoh you meant disable the clock I was thinking the LCD controller was enabled. duh
13:54:13speachythere are a lot more bits in the clock gating registers than we're reporting
13:58:58 Quit vmx (Ping timeout: 260 seconds)
13:59:25fs-bluebotBuild Server message: Build round completed after 1199 seconds.
13:59:26fs-bluebotBuild Server message: Revision 546212a result: All green
14:16:18__BilgusSpeachy you mentions 480GB of traffic to any ideas on quelling that?
14:21:49speachydownload was 282GB, themes was 61GB, www was 30GB, build* 36GB, gerrit 13GB, forums 2.9GB, and translate is a rounding error.
14:22:16speachyeverythign hitting download is ignoring robots.txt, so short of banning user agents and IP ranges there's nothign that can be done.
14:26:07__Bilgushide the files and break them out to a script to serve em?
14:27:24 Quit johnb2 (Quit: Nettalk6 -
14:30:17speachythe analytics I'm using is a bit flaky, unfortunately.
14:32:45 Join lebellium__ [0] (
14:34:20speachy$dayjob's overall upstream bandwidth usage is way up due to everyone working from home and pushing around multi-GB files, but I do need to figure out how to cut this back if I'm to continue keeping it under the radar
14:36:43 Quit lebellium_ (Ping timeout: 272 seconds)
14:40:11***Saving seen data "./dancer.seen"
14:43:34__Bilguswell blocking by useragent might be a decent stopgap
15:02:41speachythere's already a 20-odd-entry user agent blocklist going on
15:18:21__BilgusSo update on the ADC Play Option Home can be used as the first of two keys [vol up dn; prev; next]
15:19:04__Bilgusturns out I can pull that other gpio high and block the BOP buttons so I read the secondaries
15:20:17 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
15:20:50__BilgusI still have a bit of figuring to do as I'm getting double key presses on the secondary keys
15:23:04 Quit yosafbridge (Quit: Leaving)
15:24:54 Join Self-Perfection [0] (~self@2a02:2168:874f:1a00:d63d:7eff:fede:7f54)
15:27:17 Join yosafbridge [0] (
16:06:54__Bilgusok got that taken care of g#2721 now just need to clean it up a bit and figure out how the action system likes a button being released during the action scan
16:09:33speachyso each keypress can potentially take three IRQs before it gets picked up −− the keypress, the SADC complete, and then the OS tick
16:40:15***Saving seen data "./dancer.seen"
16:40:47__Bilgusyes and in the case of button press hold 1 more adc interrupt per cycle too
16:42:32__BilgusSpeachy do you think that is excessive I could actually poll the adc and mask its interrupt
16:44:24speachygiven that we can't preempt interrupts right now, adding more doesn't seem wise. at least until we can figure out what's causing our instability under load, anyway.
16:45:19speachypolling is worse though, because we'd be polling it in (another) interrupt context.
16:59:43 Quit akaWolf (Quit: leaving)
17:00:07speachyour MO in interrupt handlers needs to be get out as quickly as possible
17:01:23speachy... I'm working on a way to make the interrupts preemptible and thus allowing us to have meaningful prioritization
17:02:56speachysince we dont' have a lot a leeway with our audio DMA.
17:09:50 Quit Self-Perfection (Quit: Konversation terminated!)
17:14:12 Join akaWolf [0] (
17:14:46 Quit livvy (Remote host closed the connection)
17:15:02 Join livvy [0] (~livvy@gateway/tor-sasl/livvy)
17:15:44speachyif my math is correct we have to handle the audio dma buffer interrupt within 363us.
17:29:31speachyyeah, we're definitely underrunning our audio buffer.
17:29:42speachyI threw in a watermark check
17:30:04 Quit __Bilgus (Ping timeout: 245 seconds)
17:33:51 Join cockroach [0] (~blattodea@pdpc/supporter/active/cockroach)
17:56:04 Quit ecs (Remote host closed the connection)
17:56:32 Join ecs [0] (
17:59:54 Quit kugel (*.net *.split)
17:59:55 Quit Skyrider (*.net *.split)
17:59:55 Quit galaxy_knuckles (*.net *.split)
17:59:59 Join kugel [0] (
18:00:00 Quit kugel (Changing host)
18:00:00 Join kugel [0] (~kugel@rockbox/developer/kugel)
18:00:18 Join Skyrider [0] (~Skyrider@
18:00:42speachygreat, our internal pcm api doesn't support double-buffering. notify when the single buffer is complete, and you'll get a new one to dma.
18:03:11 Join galaxy_knuckles [0] (
18:03:12 Quit galaxy_knuckles (Changing host)
18:03:12 Join galaxy_knuckles [0] (~gknux@unaffiliated/galaxy-knuckles/x-1756549)
18:03:27braewoodsspeachy: how would another buffer help? i've only used basic IO buffers myself.
18:04:52 Quit pamaury (Ping timeout: 256 seconds)
18:05:55 Join Oksana [0] (~Wikiwide@Maemo/community/ex-council/Wikiwide)
18:06:39speachyit's to manage latency. instead of (eg) transferring 10 units every second, refilling all 10 after the last one is emptied, we transfer 5 units every half second, refilling that half while we're emptying the other half.
18:07:28speachyit's slightly more work overall, but we have far, far more time to get it done before it becomes critical.
18:12:39speachysevakis did a WIP significant PCM rewrite that at least partially implemeted double buffering, but I don't want to destabilize things _that_ much.
18:12:44speachy g#1720
18:30:23 Join ac_laptop [0] (~ac_laptop@
18:40:17***Saving seen data "./dancer.seen"
18:40:51 Join SiliconExarch [0] (sewiredci@gateway/shell/
18:56:50 Quit lebellium__ (Quit: Leaving)
19:18:03 Quit Oksana (Ping timeout: 260 seconds)
19:23:03 Quit bluebrother (Disconnected by services)
19:23:08 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
19:24:55 Join fs-bluebot_ [0] (
19:27:06 Quit fs-bluebot (Ping timeout: 256 seconds)
19:33:07 Quit sakax (Quit: Leaving)
19:33:34 Join __Bilgus [0] (41ba23be@
19:33:55__BilgusSpeachy I'll hold off on 2721 then till things are sorted
19:40:28__Bilgusactually I'm going to still move the handler to kick off the ADC in the ISR
19:40:58speachyhave a look at g#2729, give it a try if you feel inclined.
19:41:15__BilgusI think overall that makes the most sense
19:41:34speachyit's a possible minor powersaving I guess
19:42:13speachyoh! before you do that, keep in mind that the ADC is also checking the battery level
19:42:29speachyso you don't want that to only kick off in response to a keypress
19:43:30__Bilgusyeah I plan to do that from elsewhere with far less frequency
19:43:35speachyif you kick that off in _battery_voltatge() instead that would be fine
19:43:49braewoodsspeachy: sounds like like a circular log where you cannibalize older entries to make room for newer ones.
19:43:55speachyevery N invocations of that or something
19:44:12__Bilgussounds good
19:44:29speachyI forget how often battery_voltage() is called but 1HZ comes to mind
19:45:39braewoodsspeachy: does it cache the last result if it's too early to check again?
19:47:15braewoodsthen again i'm used to mainly optimizing for system call reduction since that's what is usually expensive in my code
19:49:39 Quit __Bilgus (Ping timeout: 245 seconds)
19:51:09speachyno, it currently uses a 0 value upon underflow. I have WIP code that changes that, and several other things
19:56:50 Join __Bilgus [0] (41ba23be@
20:08:56 Quit __Bilgus (Remote host closed the connection)
20:15:10 Join __Bilgus_ [0] (41ba23be@
20:15:19__Bilgus_Damn net
20:22:06 Join massiveH [0] (
20:29:49 Quit akaWolf (Ping timeout: 260 seconds)
20:33:50 Join prof_wolfff [0] (
20:40:20***Saving seen data "./dancer.seen"
21:24:20 Join Oksana [0] (~Wikiwide@Maemo/community/ex-council/Wikiwide)
21:41:28 Quit cockroach (Quit: leaving)
21:41:39 Join sakax [0] (~r0b0t@unaffiliated/r0b0t)
22:24:26 Quit __Bilgus_ (Remote host closed the connection)
22:27:49 Quit Huntereb (Read error: Connection reset by peer)
22:34:38 Quit ac_laptop (Ping timeout: 260 seconds)
22:40:22***Saving seen data "./dancer.seen"
22:58:23 Quit livvy (Ping timeout: 240 seconds)
23:01:14 Join akaWolf [0] (
23:12:47 Quit TheSeven (Ping timeout: 244 seconds)
23:13:07 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
23:20:43 Quit sakax (Quit: Leaving)

Previous day | Next day