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 2021-07-25

00:09:19 Quit dconrad ()
00:13:15 Quit JanC (Remote host closed the connection)
00:15:14 Join JanC [0] (~janc@user/janc)
00:20:38 Quit akaWolf (Ping timeout: 265 seconds)
01:05:38braewoodsspeachy: something else that's been available since C99... array declarators. it's an alternate way of managing pointers via array syntax. but it only works in function parameter lists afaik.
01:05:58braewoodsit might be useful to use
01:06:16braewoodsone example...
01:06:32braewoodsint test(char arr[static 45]);
01:06:56braewoodsyou can also apply storage modifiers like const, volatile, restrict, etc to it
01:07:23braewoodsthe 45 allows the compiler to check a static sized array contains at least the stated number of elements
01:07:31braewoodsgcc supports this in 11 but no earlier
01:07:56braewoodsthis could help with managing complex pointer arrangements being passed to functions
01:08:11braewoodsbut otherwise limited use case that i can tell
01:08:53braewoodsit lacks the flexibility of being able to work with both dynamic and static arrays
01:09:01braewoodsso i'm honestly not sure the point of it
01:09:35braewoodsjust pointer salad is a giant mess to deal with
01:21:16***Saving seen data "./dancer.seen"
01:25:03 Quit massiveH (Quit: Leaving)
02:53:32 Join i [0] (
02:53:56 Nick i is now known as Guest541 (
03:12:14 Join lebellium [0] (~lebellium@2a01:cb10:2e:2000:39a0:d3f4:c82f:13b7)
03:20:14 Join ZincAlloy [0] (
03:21:18***Saving seen data "./dancer.seen"
03:52:34 Quit ZincAlloy (Quit: Leaving.)
03:52:51 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:9fc:edf4:35fc:55d5)
03:53:28Guest541hi, i am trying to unbrick my partially bricked sansa clip+ from fs corruption, player can boot but can not read or write anything from internal storage (can't save settings, can't shutdown without home+power), it also cant be mounted as storage device on win or linux (mount as ?media remote? instead). Player sometimes can resume last track and can play anything from SD card but only one file per boot. I followed this
03:53:28Guest541iki/SansaAMSUnbrick guide and now i can see internal storage in fdisk but what to do next is unclear. Is there any way to solve this? rockbox version some 3.15+ daily.
03:54:31 Quit ZincAlloy (Client Quit)
03:55:24 Join akaWolf [0] (
05:21:20***Saving seen data "./dancer.seen"
06:17:00 Join ZincAlloy [0] (
06:19:42 Join ZincAlloy1 [0] (
06:19:43 Quit ZincAlloy (Read error: Connection reset by peer)
06:33:19 Join johnb2 [0] (
06:34:37johnb2Guest541 This sounds like your internal storage is dying. I'd advise to not touch it anymore, but boot from and only use a sdcard from now on.
06:36:55johnb2Assuming you have a somewhat recent (last two years) bootloader installed.
06:58:41 Quit johnb2 (Ping timeout: 255 seconds)
07:04:56 Quit Guest541 (Quit: Leaving)
07:21:23***Saving seen data "./dancer.seen"
07:58:48 Quit ZincAlloy1 (Quit: Leaving.)
08:14:05 Join ZincAlloy [0] (
09:15:09 Join speachy [0] (~speachy@rockbox/developer/speachy)
09:15:09Mode"#rockbox +v speachy" by ChanServ (
09:21:26***Saving seen data "./dancer.seen"
09:33:31 Join dconrad [0] (~dconrad@
09:53:29 Join amachronic [0] (~amachroni@user/amachronic)
09:54:16rb-bluebotBuild Server message: New build round started. Revision 77ec752248, 303 builds, 9 clients.
10:05:08rb-bluebotBuild Server message: Build round completed after 651 seconds.
10:05:10rb-bluebotBuild Server message: Revision 77ec752248 result: All green
10:07:43rb-bluebotBuild Server message: New build round started. Revision e532714d1f, 303 builds, 9 clients.
10:17:14speachyamachronic: I could have sworn something was using the get_peak_buffer() crap. I remember checking and being quite grumpy at having to implement it.
10:17:34speachy(and ended up nuking a few other pcm_() API functions around the same time
10:17:41amachronici did a quick grep and nobody seemed to be using it
10:17:52speachybut maybe the final user was in the hwcodec code and that's why it's all gone now
10:18:16rb-bluebotBuild Server message: Build round completed after 633 seconds.
10:18:19rb-bluebotBuild Server message: Revision e532714d1f result: 0 errors 13 warnings
10:18:33amachronicwell that wasn't totally unexpected
10:18:42speachyall MIPS
10:18:54speachyno wait, not all
10:19:09speachyit's nice seeing green in the delta table
10:21:56speachyoh, I bet this was obsoleted by the pcm->dsp migration way back in the olden times
10:22:54rb-bluebotBuild Server message: New build round started. Revision 148fac6f34, 303 builds, 9 clients.
10:23:16amachronicit was just unused function + variable warnings, that should fix it
10:24:59amachronicwhat's it even for? peak meters? (must be a hwcodec thing)
10:26:39amachronicspeachy, could you update the download site? i uploaded a new version of jztool which supports the Q1 and Eros Q.
10:26:54amachronicand the q1 bootloader too
10:28:52speachyuniversal macos or still intel-only?
10:29:32amachronicdconrad built it, pretty sure it's intel only
10:30:00dconradtested on brand new m1 macbook air, works there as well
10:30:14dconradcan't vouch for very old osx, however
10:31:36dconradlike I said before (and it turned surprisingly contentious), apple did a good job with making the transition seamless from intel to arm
10:32:12speachyI only asked because the current directory is called 'macos_intel'
10:32:35speachyso is it a fat binary, or does it require rosetta installed?
10:33:12dconradoh, yeah probably
10:33:24dconradprobably rosetta
10:33:29rb-bluebotBuild Server message: Build round completed after 635 seconds.
10:33:31rb-bluebotBuild Server message: Revision 148fac6f34 result: 0 errors 1 warnings
10:33:36dconradI'll have to ask what's on that laptop
10:33:48speachyok, all binaries updated.
10:35:20amachronicwhoops, I missed the jz4760
10:35:53 Join skipwich [0] (~skipwich@user/skipwich)
10:36:48rb-bluebotBuild Server message: New build round started. Revision 05d4d6a4f2, 303 builds, 9 clients.
10:39:13speachysweet, the q1 and m3k manuals were generated overnight.
10:40:19amachronicdconrad, may I ask what was wrong with g#3603 approach? why move it to pcm-x1000?
10:40:21rb-bluebotGerrit review #3603 at : Start of 32-bit software volume for native players by Dana Conrad
10:41:36dconradtbh, I think either could work
10:41:43dconrador rather, I think that could work
10:42:04sporkthanks amachronic and speachy, with the bootloader/jztool available the q1 manual is even more correct
10:42:16dconradbut I don't know if that could be reasonably done without transitioning the rest of the generic pcm stuff to 32 bit
10:42:34dconradso I guess the thought was doing it in the driver is "safer"
10:42:59dconradspeachy might have a better grasp on why?
10:43:31amachronichmm I see what you mean, it's the callback hell that makes it tricky.
10:43:39speachybasically there's a ton of hardcoded assumptions in the core around 16-bit sample sizes, especially with respect to buffer sizing.
10:43:54speachyall of the internal pcm/dsp routines also assume 16bpp
10:44:20amachronicmy only concern is that the HW PCM code runs totally in interrupt context so scaling there might be unworkable
10:44:53speachyamachronic: yeah, that's most likely going to be a major issue. but going to double-buffering _might_ help there.
10:45:55amachronicbuffering is done in the mixer afaict, so probably not
10:46:16dconradpardon my ignorance, but why would it work fine in the hosted port but not the native? is that code in the hosted port not run in interrupt context?
10:46:18amachronicwith scaling in the interrupt we actually want a lot of small buffers
10:46:39speachyall hosted code is running in userspace
10:46:45rb-bluebotBuild Server message: Build round completed after 596 seconds.
10:46:47rb-bluebotBuild Server message: Revision 05d4d6a4f2 result: 4 errors 0 warnings
10:46:58dconradI see
10:47:14speachydouble-buffering in the x1000 pcm driver; ie triple-buffering from the core's perspective
10:47:45amachronicand I suppose that's why the linux driver uses double buffering −− to cover the huge latency of calling back to userspace
10:48:03speachythat way we will always have something to hand to the hardware while we're processing the data from the callback
10:48:17speachyok, another random build glitch.
10:48:29speachyon my builder too
10:48:57 Quit ZincAlloy (Quit: Leaving.)
10:48:57speachywe _might_ have some sort of dependency issue in the build that doesn't manifest itself unless you have a ton of cores..
10:49:20speachybut I've not been able to get this to break when firing it up manually
10:49:37speachy(16c/32t on that box)
10:49:49amachronicmaybe CODECS_LIBS vs CODEC_LIBS
10:50:42amachronicyeah CODECLIB=libcodec.a is in CODEC_LIBS, it's just a typo in the makefile
10:50:46speachyyeah, you're right
10:50:53dconradjust as a curiosity, it would appear the 32-bit volume scaling method works better than the current 16-bit scaling method, even when truncated back down to 16-bit
10:51:01dconrad g#3623
10:51:02rb-bluebotGerrit review #3623 at : HACK: Improved 16-bit volume scaling by Dana Conrad
10:52:01speachydconrad: so.. perhaps the solution is to scrap all of this and make the core pcm_volume stuff work using 32-bit math, and re-truncate..
10:52:16dconradwell, I'm unsure
10:52:18amachronicI see. I don't understand the math well enough to understand why one is better than the other.
10:52:31dconrad(me neither, amachronic haha)
10:52:33amachronicthe thing is, both are doing 32bit math
10:53:12speachy16-bit intermediates
10:53:14dconradI swear it still distorts audibly at the low end of the volume, but it doesn't pop/click
10:53:32speachyamachronic: you want to fix that dependency typo or should I?
10:53:41amachronici'll do it
10:54:19speachycrazy how it's managed to mostly work for so long..
10:55:10*speachy feeds his good little builder a byte of bits.
10:55:36rb-bluebotBuild Server message: New build round started. Revision 2e9443104f, 303 builds, 9 clients.
10:56:22amachronicit's weird how it managed to work at all :D
10:58:18speachycrap. just broke the wiki.
11:06:00rb-bluebotBuild Server message: Build round completed after 624 seconds.
11:06:03rb-bluebotBuild Server message: Revision 2e9443104f result: All green
11:07:22yangIs it possible to flash fiio m3k rockbox firmware without SD card inserted? or does the rockbox firmware sits on SD card?
11:07:44amachronicyes, you need the SD card.
11:07:59speachyok, wiki should be back up now.
11:08:47speachy(for future reference, next time, double-check 'rm -Rf fos<tab>' *before* pressing <enter>
11:09:59 Quit Maxdamantus (Ping timeout: 258 seconds)
11:12:05 Join johnb2 [0] (
11:18:45yangamachronic: is your first entry in the forum the accurate flashing description ?,53858.msg248529.html#msg248529
11:18:54yangor were any later changes to it
11:19:11amachronicthat's still correct
11:20:08dconradspeachy: I just asked, the macos jztool does require rosetta for arm macs (though I doubt many don't already have rosetta anyway)
11:20:27speachyokeydokey. so I'll leave he directory names as-is
11:21:29***Saving seen data "./dancer.seen"
11:24:02braewoodsdconrad: i assume that's just until we compile a native version?
11:25:43 Quit johnb2 (Quit: Nettalk6 -
11:36:10yangamachronic: so, the rockbox firmware is then residing on SD card? asking because I would likely replace SD card with a bigger one in a few days.
11:36:34sporkyou can copy the .rockbox folder to your new card
11:44:53dconradbraewoods: I imagine so, but it seems to work pretty seamlessly anyway
11:45:57 Join Maxdamantus [0] (~Maxdamant@user/maxdamantus)
11:50:42 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:39f0:1b64:1df7:1a3c)
12:19:42speachyrosetta isn't installed/active out of the box on the new macs
12:19:51 Quit pixelma (Ping timeout: 252 seconds)
12:19:52speachyone has to take steps to install/enable it, including a click-through agreement of sorts
12:20:26 Quit amiconn (Ping timeout: 255 seconds)
12:22:00 Join amiconn [0] (
12:22:00 Join pixelma [0] (
12:26:13braewoodsspeachy: what's stopping us from cross-compiling for mac?
12:26:17braewoodsproper toolchain?
12:30:42speachypretty much ahve to use xcode in order to do anything these days.
12:42:18dconradmore in the saga of software volume: the "improved 16-bit" hack is distorted and noisy at minimum volume, whereas the hosted port is (mostly) clean and noise-free at minimum volume, so there's definitely an advantage to having the better resolution going to the hardware
12:42:28dconrad(from actual testing)
12:43:30dconradthe (mostly) is that podcasts in particular for some reason, tend to pop at the end of an edited-in silence, for instance between segments
12:45:08dconradas well as between/starting/stopping songs, but we already knew that (and the native does that too, another thing to iron out)
12:57:47 Quit Maxdamantus (Ping timeout: 255 seconds)
13:00:25 Quit ZincAlloy (Quit: Leaving.)
13:06:56 Quit amachronic (Ping timeout: 256 seconds)
13:09:47 Join Maxdamantus [0] (~Maxdamant@user/maxdamantus)
13:21:33***Saving seen data "./dancer.seen"
14:24:00 Join ZincAlloy [0] (
14:32:56 Quit dconrad (Remote host closed the connection)
14:41:33 Part dbohdan
14:59:57 Join dconrad [0] (~dconrad@
15:21:35***Saving seen data "./dancer.seen"
15:59:55 Quit dconrad (Remote host closed the connection)
17:00:19 Quit lebellium (Quit: Leaving)
17:00:20 Join dconrad [0] (~dconrad@
17:04:57 Quit dconrad (Ping timeout: 258 seconds)
17:10:04 Join dconrad [0] (~dconrad@
17:15:33 Quit dconrad (Remote host closed the connection)
17:21:39***Saving seen data "./dancer.seen"
17:36:31 Join dconrad [0] (~dconrad@
17:53:03 Join cockroach [0] (~blattodea@user/cockroach)
17:57:11 Quit dconrad ()
18:43:28 Quit ZincAlloy (Quit: Leaving.)
19:21:40***Saving seen data "./dancer.seen"
20:59:20 Join dconrad [0] (~dconrad@
21:21:44***No seen item changed, no save performed.
22:01:01 Quit cockroach (Quit: leaving)
23:21:45***Saving seen data "./dancer.seen"
23:32:58 Quit dconrad ()

Previous day | Next day