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 2021-08-01

00:15:25 Quit saratoga (Quit: Connection closed)
01:01:52***Saving seen data "./dancer.seen"
02:22:04 Quit massiveH (Quit: Leaving)
02:53:34 Join ZincAlloy [0] (
03:01:54***Saving seen data "./dancer.seen"
05:01:58***No seen item changed, no save performed.
05:09:30 Join lebellium [0] (
05:21:59 Quit S|h|a|w|n (Read error: Connection reset by peer)
07:02:02***Saving seen data "./dancer.seen"
08:24:29 Join dconrad [0] (~dconrad@
08:41:25dconradspeachy: 3633 should be updated now
09:02:03***No seen item changed, no save performed.
09:05:57 Quit f1refly (Read error: Connection reset by peer)
09:07:42 Join f1refly [0] (
10:41:28 Quit rando25892 (Ping timeout: 272 seconds)
10:42:59 Join rando25892 [0] (
11:02:06***Saving seen data "./dancer.seen"
11:29:40 Quit Moriar (Ping timeout: 272 seconds)
12:04:51 Quit dconrad (Remote host closed the connection)
12:11:08 Join dconrad [0] (~dconrad@
13:02:09***Saving seen data "./dancer.seen"
13:07:17 Join saratoga [0] (
13:11:47saratogaspeachy: (just catching up to the logs) I think PCM_SW_VOLUME_FRACBITS is intended to be 15, so full scale PCM is +/-0x7FFF0000, which leaves you room for rounding
13:12:38saratogagrepping through the code I see the other definition in config.h which sets it to 16, which seems questionable to me
13:13:19 Quit dconrad (Remote host closed the connection)
13:21:32saratogaLooking at the logs, JhMikeS set that to 16 after the SW volume control patch (which was set to 15), I think that might be a mistake
13:22:48saratogai'm surprised no one noticed though, does anything actually use the SW volume control? Should be obvious if ever remotely loud sample overflows ...
13:26:31speachyI was wondering that
13:27:24speachythe erosq obviously does, and several of the hosted targets too (most Sonys, for example)
13:28:01saratogaone possibility is that whatever device that was programmed on feeds in the data shifted over by one bit
13:28:39saratogathis device has working file system right? can you record the samples being written to the DAC to disk and see what they are?
13:28:45speachyok, ibassodx50/dx90, hifiman 60x/80x, fiio m3k, ihigi 760/960, onda vx747/777
13:29:03saratogasome of those are pretty popular, seems surprising no one complained about corrupted audio
13:29:04 Join dconrad [0] (~dconrad@
13:29:42speachyof that list, only the ibassos ahd any real use
13:30:52speachythe artifacts are only really there at either MAX or MIN volume.
13:32:01speachythe sonys are popular but implement sw volume control inside the alsa driver, much like the erosq/m3k (hosted) targets.
13:32:36saratogahow does this port work? You use our codecs and buffering normally, but feed it to the embedded linux driver ?
13:32:46saratogaor do you have your own driver
13:33:12speachyfeed teh PCM data to ALSA
13:33:30speachythose swvol hosted ones send the data to ALSA in 32-bit format
13:33:42 Quit dconrad (Ping timeout: 265 seconds)
13:34:29saratogaartifacts and also that suspicious -1 offset make me think that the fixed format is off by 1 bit
13:34:57speachythat -1 thing is to work around a hw quirk.
13:35:08speachyby preventing the cocec from powering down
13:35:17saratogaoh it powers down if it gets all zeros?
13:35:26saratogathat is strange
13:35:36saratogathis is the ESS9018?
13:35:47speachyactually fairly typical... but unuusally, it cant' be disabled
13:35:54speachyTI PCM5102
13:37:41speachyand doing an analog mute (on the codec or other parts of the audio path) apparently adds louder artifacts.
13:38:48saratogaoh wrong codec
13:41:23saratogaaside from the analog mute function, and the overflow from rounding on 16.16 samples, did it otherwise work or were there other problems?
13:42:22speachythe rounding wasn't correct either, assumed positive samples.
13:42:41saratogayeah that is interesting
13:43:09saratogarounding is a good idea if you actually have a > 16 bit dac since otherwise you're losing half a bit
13:44:17speachysure, but that doesn't do any good when we're unconditionally truncating to 16bit. :)
13:45:19speachyas I opined yesterday, we could probably enable >16bit output into the HW PCM API relatively easily.
13:46:07speachyI think all of the current-ish-gen targets we have now all support 24 or 32-bit audio.
13:47:40saratogawould make sense, at least on devices that need SW volume control
13:47:58saratogais updating the PCM buffering code so easy though?
13:49:18speachyI think so, it's a small surface area.
13:50:11speachymost of the work would be in converting plugins to use the mixer api instead of directly interacting with the pcm layer.
13:50:25speachywhich IMO should be done anyway
13:53:08speachyThe mixer is already bitdepth aware
13:56:53saratogathis volume code is hard to follow, pcm_scale_sample takes signed arguments, but all calls to it pass unsigned for the first argument
13:57:31saratogais the intention that it would be converted to signed (so it is within the range of signed int) or is this a mistake and the whole function was meant to take unsigned (so the rounding makes sense)
13:57:59speachyI have no idea; this code well predates my active involvement
13:59:26speachybut given the use of signed values everywhere else..
14:00:34saratogaoriginally that function was a preprocessor macro, and so had whatever the type of the data was
14:28:00 Join dconrad [0] (~dconrad@
14:51:00 Join Moriar [0] (
15:02:13***Saving seen data "./dancer.seen"
15:26:49 Quit dconrad (Remote host closed the connection)
15:42:32 Join dconrad [0] (~dconrad@
15:46:42 Quit dconrad (Ping timeout: 240 seconds)
16:21:21 Quit ZincAlloy (Quit: Leaving.)
16:52:14 Join dconrad [0] (~dconrad@
17:02:16***Saving seen data "./dancer.seen"
17:10:36 Quit dconrad (Remote host closed the connection)
17:19:18 Join dconrad [0] (~dconrad@
17:34:24 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:25ac:6a7a:8e9:3c4d)
17:35:19 Quit dconrad (Remote host closed the connection)
17:39:24 Quit ZincAlloy (Quit: Leaving.)
17:39:32 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:25ac:6a7a:8e9:3c4d)
17:51:03 Join dconrad [0] (~dconrad@
17:55:56 Quit dconrad (Ping timeout: 272 seconds)
18:02:28 Quit lebellium (Ping timeout: 250 seconds)
18:14:50 Join dconrad [0] (~dconrad@
18:19:05 Quit dconrad (Remote host closed the connection)
18:34:49 Join dconrad [0] (~dconrad@
18:39:06 Quit dconrad (Ping timeout: 240 seconds)
19:02:18***Saving seen data "./dancer.seen"
19:35:11 Join dconrad [0] (~dconrad@
19:36:05 Join cockroach [0] (~blattodea@user/cockroach)
19:45:34 Quit dconrad (Remote host closed the connection)
20:01:17 Join dconrad [0] (~dconrad@
20:05:30 Quit dconrad (Ping timeout: 240 seconds)
20:37:22 Join dconrad [0] (~dconrad@
20:37:30 Quit dconrad (Remote host closed the connection)
20:37:42 Join dconrad [0] (~dconrad@
20:55:58rb-bluebotBuild Server message: New build round started. Revision 83653023d4, 303 builds, 9 clients.
21:02:22***Saving seen data "./dancer.seen"
21:08:15rb-bluebotBuild Server message: Build round completed after 737 seconds.
21:08:18rb-bluebotBuild Server message: Revision 83653023d4 result: All green
21:13:17_bilgusNice that knocked off most of the cost of sanitizing tagcache :)
21:43:00 Quit cockroach (Quit: leaving)
22:28:08rb-bluebotBuild Server message: New build round started. Revision 2346a7aa6d, 303 builds, 9 clients.
22:38:36rb-bluebotBuild Server message: Build round completed after 628 seconds.
22:38:38rb-bluebotBuild Server message: Revision 2346a7aa6d result: All green
22:38:41rb-bluebotBuild Server message: New build round started. Revision 247731fe8c, 303 builds, 9 clients.
22:48:58rb-bluebotBuild Server message: Build round completed after 618 seconds.
22:49:00rb-bluebotBuild Server message: Revision 247731fe8c result: All green
23:00:03rb-bluebotBuild Server message: New build round started. Revision 0501af8b06, 303 builds, 9 clients.
23:02:24***Saving seen data "./dancer.seen"
23:09:57rb-bluebotBuild Server message: Build round completed after 594 seconds.
23:09:59rb-bluebotBuild Server message: Revision 0501af8b06 result: All green
23:10:38 Quit Moriar (Quit: Leaving.)
23:37:55 Quit dconrad (Remote host closed the connection)
23:53:38 Join dconrad [0] (~dconrad@
23:58:16 Quit dconrad (Ping timeout: 272 seconds)

Previous day | Next day