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 2017-01-11

00:05:27chrisjjSo assuming the current fail really is a power-off, does that mean the trigger must be a freeze that prevents your timer IRQ from pacifying the watchdog? Nothing else, to your knowledge?
00:06:55pamauryNo that I can think about but how knows. What kills me is that 45697a0bf should be the same as 8e82839f_lcdfix. If you can observe a difference, this is very puzzling to say the least
00:11:45pamaurythe only thing that I can think about would be different configure options, but I think I built both as (N)ormal builds, no debug options
00:15:45chrisjjPuzzling, agreed. I'll repeat the tests with sterilised test tubes :-)
00:18:18chrisjjThat you built the two with same Normal/Build looks true from the fact the binary sizes are identical.
00:18:36chrisjjThe larger version number is presumably absorbed by word alignment.
00:20:20chrisjjI know you don't have binary stability, but any guess as to the binary difference we see?
00:23:11 Quit Ruhan (Ping timeout: 240 seconds)
00:24:31chrisjjAll of the first few single-byte differences are differences of one e.g. EB -> EA . Addresses of shifted data?
00:25:49BilgusIt is hard to say in a compiled binary Disasm then compare
00:26:43 Join Ruhan [0] (uid76353@gateway/web/
00:27:10 Quit ender` (Quit: All great truths begin as blasphemies. –– George Bernard Shaw)
00:27:13pamaurychrisjj: the two contains different strings (ie the revision number), this can change the data section size alignment, make the entire binary to shift by 4 byte, change all addresses
00:29:58[Saint]FWIW, it's not so much that the panic screen 'disables the watchdog', and more that the fact that we panicked means something very fundamental fell over.
00:29:58chrisjjUnderstood, but ISTM the large version number is absorbed by word alignment in this case. There's no data section displacement.
00:30:29chrisjjs/large version number/largeR version number/
00:30:55 Join alexweissman [0] (
00:31:32[Saint]I just wanted to be clear because your verbiage seems to suggest that despite repeated clarification that you think that the watchdog service is being disabled deliberately.
00:31:58[Saint]when it's more akin to 'shit totally hit the fan'.
00:32:56[Saint]if a panic was a recoverable state, it wouldn't happen to begin with.
00:33:02pamaurychrisjj: most of the changes I see between the two bins are jumps with slightly different addresses. Comparing diff is inevitably going to be tricky, litterally anything can trigger massive changes in the compile process
00:34:25chrisjjUnderstood. And I see the gcc version changed.
00:38:37pamaurythat's the host gcc, it should make no different, what matters is the arm compiler and it's the same
00:38:58chrisjjGood to hear the compiler is the same.
00:39:03***Saving seen data "./dancer.seen"
00:39:32chrisjjBTW, re strings, UNIX strings command finds no string changed, other that the version number we know of.
00:41:57chrisjjSo, of the 'literally anything can trigger massive changes', it might be useful if we knew even one that might cause today's small changes :-)
00:41:58[Saint]...which is entirely expected.
00:42:54 Quit skapazzo (Quit: leaving)
00:42:59[Saint]even order of operations can change the outcome of non-reproducible builds.
00:43:58pamaurychrisjj: the version number ( 8e82839feM) is included in the binary, so yes there is at least one string that is different. I'm not going to discuss binary changes anymore unless you have something *interesting* to say on the topic.
00:44:21chrisjjBilgus: pamaury has said source is unchanged so I'm confident disasm would be likewise unchanged.
00:44:35[Saint]well, evidently his definition of interesting is incredibly broad.
00:44:35pamauryrockbox does not have reproducible builds
00:45:40[Saint]you can compile the same build on the same host, with the same host configuration, and still get different yet non-substantive results.
00:45:48[Saint]it's really neither important nor interesting.
00:46:00pamauryand in this case, the source IS DIFFERENT, because the version number is different and it is included in the binary, as I already said 2 times already
00:46:15[Saint]Oh, yes, I know.
00:46:19chrisjjOK. The most interesting this I can say (not saying it is interesting) is that: 1) code behaviour difference, cause unknown 2) code binary differs, cause unknown 3) In the absence of any other explanation for 1, it might be worth considering that there is a causative realtionship.
00:47:13[Saint]Yes, I'm aware this is a different build. Just clarifying that even the same source revision can still produce a different binary and it's non-interesting.
00:47:38chrisjjGood point. Corrting myself: Version number aside, pamaury has said source is unchanged so I'm confident disasm would be likewise unchanged.
00:47:46[Saint]at this stage the only interesting metric is pass/fail.
00:47:49[Saint]that's really it.
00:49:15[Saint]chrisjj: your confidence is unwarranted and no one is really interested in correcting that misunderstanding.
00:49:38[Saint]at this stage it's very clear you can be both confident, and wrong.
00:49:46[Saint]they're not mutually exclusive things.
00:50:11chrisjjAnyone sufficiently unconfident to fancy spending their time comparing the two disasms, please go ahead.
00:50:55pamauryWell let's just hope that it doesn't come to diffing the two bin to understand the difference, because that's going to be a lot of work.
00:51:13pamauryThere is another option that we have no considered
00:51:27pamaurychrisjj: did you use RoLo to load the new binary when you changed the firmware ?
00:51:37pamauryor did you do a proper power-down and power up ?
00:51:47chrisjjGood question. I tried both.
00:52:06chrisjj... thinking that perhaps a ROLO start is unclean.
00:52:14pamauryand both failed ?
00:52:48chrisjjWhoops, my [23:51] crossed with yours.
00:53:52chrisjjHang on... checking notes.
00:57:03pamauryThere is actually another possible explanation, very logical but also very bad: it's a cache+dma problem
00:58:16*chrisjj reaches for garlic
01:01:30chrisjjA: Not recorded. Sorry. I can't say for sure that ROLO-started sessions failed or Power-on-started sessions failed, but I can say for sure Reset-started sessions failed.
01:02:04chrisjjI can rerun any tests you need, in the three variants.
01:02:53pamaurychrisjj: please re-run after a clean boot
01:03:05chrisjjDefine clean boot, wise one.
01:03:13pamaurypower up, power down, power up
01:03:37chrisjjOK. And do you think battery v. external power might affect?
01:04:38pamauryin doubt, just battery. It shouldn't make a difference though
01:04:50*pamaury goes to bed
01:05:46chrisjjOK, though I have to say it makes more of difference that I expect elsewhere. E.g. with external power from charger (not PC), noot attempts to connect to USB. But no matter. Will do.
01:05:56 Quit paulk-collins (Quit: Leaving)
01:06:07chrisjjs/noot/boot / (should go to bed too)
01:06:25chrisjjpamaury: And which Version should I test first?
01:09:11 Quit The_Prospector (Read error: Connection reset by peer)
01:10:16 Quit pamaury (Ping timeout: 260 seconds)
01:12:08 Quit cttttt (Read error: Connection reset by peer)
01:12:40 Join cttttt [0] (sid135570@gateway/web/
01:12:41 Quit anonus (Ping timeout: 255 seconds)
01:13:10 Quit Galois (Ping timeout: 240 seconds)
01:13:30 Quit megal0maniac (Ping timeout: 240 seconds)
01:13:32[Saint]chrisjj: as with every other target holding any key during plug prevents it from launching USB screen
01:14:26 Join megal0maniac [0] (~megal0man@unaffiliated/megal0maniac)
01:17:00chrisjj[Saint], it is not the USB screen, and holding plug doesn't prevent it. It is boot's attempt to connect to USB, before the .rockbox app launches.
01:17:21chrisjjThanks anyway for the thought.
01:18:47[Saint]Sorry, I thought you were referring to code that Rockbox actually has control over.
01:22:26 Nick __builtin is now known as _builtin (~xray@rockbox/developer/builtin)
01:22:34 Nick _builtin is now known as __builtin (~xray@rockbox/developer/builtin)
01:23:06*chrisjj finds builtin's nick changes... fascinating
01:24:04chrisjj[Saint] I was. This code is the Rockbox bootloader.
01:24:33[Saint]chrisjj: at that stage, it isn't. It's a fairly complex process.
01:25:59chrisjjSo if this is not Rockbox code, how come the lines above show the Rockbox version number?
01:26:11 Join anonus [0] (
01:26:35*__builtin can type "Linux" into windows notepad
01:27:27[Saint]Perhaps I'm misinterpreting what you're seeing.
01:31:28chrisjjOr I didn't describe it well enough.
01:31:44[Saint]My understanding is these things have a slightly unusual boot process, not in and of themselves, but in comparison to other targets. I thought you were referring to a host USB mode before the jump to the bootloader.
01:32:13chrisjjOn RBed ZEN boot, after CREATIVE and some flashes, I get an image with black background and white text.
01:32:29 Quit __builtin (Quit: - Chat comfortably. Anywhere.)
01:32:39chrisjjFirst line says 'Boot version:' and Rockbox version number.
01:33:31chrisjjWhen on charger, a line below says "USB connecting".
01:33:44[Saint]And it just stalls there if USB is plugged? Cute.
01:34:00 Join __builtin [0] (~xray@rockbox/developer/builtin)
01:34:29[Saint]Apparently it can't tell the difference between charge-only and a USB host.
01:34:58[Saint]It's probably waiting for an imaginary host in the charger to enumerate.
01:35:29chrisjjA: No. It enters "Bootloader USB mode".
01:35:45chrisjjOn a charger, it soon times out. No harm done.
01:36:14chrisjjYes, I agree it can't tell the difference.. until the charger doesn't respond :)
01:37:55chrisjjBut that timeout is very impatient. 1-2s.
01:39:22chrisjjNot a problem... unless that same impatience is the cause of the poor engagement I get on the USB mode of the .rockbox app. The device can bounce on and off Windows like a rubber ball.
02:30:31 Quit elensil (Ping timeout: 240 seconds)
02:39:04***Saving seen data "./dancer.seen"
02:45:46 Quit ZincAlloy (Quit: Leaving.)
02:50:26 Quit uwe_ (Ping timeout: 258 seconds)
02:52:47 Join elensil [0] (~edhelas@2001:1c02:1903:d800:7c38:ddef:a0bd:d3a7)
02:55:23 Join uwe_ [0] (
03:10:55 Quit soap (Read error: Connection reset by peer)
03:20:39 Join Galois [0] (
03:22:18 Join The_Prospector [0] (~The_Prosp@unaffiliated/cornman)
03:26:25 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
04:12:25__builtinhmm, looks like the plugin buffer on the c200v2 is too small to run any puzzles without crashing
04:13:59[Saint]can't you just ask for the playback buffer temporarily?
04:14:09__builtinI'll probably use gevaerts's recommended approach and dynamically grab the playback buffer, so yes
04:22:00__builtinproblem is, it's impossible for me to test
04:22:17__builtinI don't think the sim accurately simulates how memory is laid out, or does it?
04:23:18*chrisjj is reassured to hear impossible to test is considered a problem for a RB update :)
04:23:30*chrisjj goes to bed.
04:35:56 Join Strife1989 [0] (
04:38:50 Quit Strife89 (Ping timeout: 240 seconds)
04:39:05***Saving seen data "./dancer.seen"
04:39:21 Quit jhMikeS (Ping timeout: 245 seconds)
04:42:02 Quit WakiMiko (Max SendQ exceeded)
04:42:40 Join WakiMiko [0] (~WakiMiko@unaffiliated/wakimiko)
05:46:52 Join jhMikeS [0] (
05:48:57BilgusI'm having a hard time deciding if a user altered the buffer in a UDF formatting function with embedded nulls short of walking the damn thing
06:16:18Bilgusso what I ended up doing was getting rid of the nulls till after I checked user defined string, ORing the characters in the buffer and then walking the string again ORing the chars again if the user didn't just return a string const from elsewhere
06:16:35Bilgusanyone have a better idea>
06:29:19 Quit TheSeven (Ping timeout: 258 seconds)
06:29:41 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:34:30Bilgusoh I mean XOR and amazingly enough it didn't really slow it down by much
06:39:09***Saving seen data "./dancer.seen"
06:52:18 Join Senji [0] (~Senji@
07:41:11 Quit girafe (Read error: Connection reset by peer)
08:24:44 Join ender` [0] (
08:39:13***Saving seen data "./dancer.seen"
08:51:35 Join parchd [0] (~parchd@unaffiliated/parchd)
09:23:35 Join petur [0] (
09:23:35 Quit petur (Changing host)
09:23:35 Join petur [0] (~petur@rockbox/developer/petur)
09:52:40 Part elensil
09:52:58 Join elensil [0] (~edhelas@2001:1c02:1903:d800:7c38:ddef:a0bd:d3a7)
10:02:32 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:19:34 Quit krnlyng (Ping timeout: 252 seconds)
10:25:30 Quit pamaury (Ping timeout: 240 seconds)
10:30:45 Quit [Saint] (Read error: Connection reset by peer)
10:32:20 Quit mc2739 (Ping timeout: 258 seconds)
10:32:33 Join [Saint] [0] (~sinner@rockbox/staff/saint)
10:32:35 Join krnlyng [0] (
10:36:09 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
10:39:14***Saving seen data "./dancer.seen"
10:42:58 Join fujisan [0] (~fujisan@unaffiliated/fujisan)
10:52:11 Join robertd1 [0] (~root@
11:12:33 Join wodz [0] (
11:25:21 Join skapazzo [0] (~skapazzo@
11:32:12wodzpamaury: (log) have you seen ?
11:46:55 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:47:19pamaurywodz: no but it seems very similar to other C/C++ interprets (like cint), why ?
11:49:13wodzpamaury: 1) based on llvm 2) still in active development/support 3) I still miss C prototyping environment in hwstub :-)
11:49:56wodzpamaury: btw. did you try to disasm SpiderApp?
11:50:07wodzpamaury: I know it is huge but still
11:51:13pamauryI remember having a quick look but it's very huge
11:51:25 Join xorly [0] (
11:51:33pamaurywodz: I see, well we could add cling support to hwstub I guess
11:52:22pamauryIf we do so, ideally I'd like to be able to call lua from cling and cling from lua
11:52:36pamaury(or at the very least, lua from cling)
11:53:14 Join pamaury_ [0] (
11:55:05pamaurywodz: I'm busy right now but I'll have a look at how cling works and how one interacts with it
11:55:56pamauryideally I'd like to have hwstub_shell be able to be able to handle lua and cling transparently, so it just works
11:55:56wodzpamaury: I'd be nice to have BUT it is really low priority.
12:01:39pamaurywodz: at least with something based on clang with jit, I am more confident that this interpreter is not going to die so soon
12:03:17wodzpamaury: Being used in CERN is advantage too
12:20:03 Join elchapolin [0] (
12:21:02 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:3c2a:da40:8fef:17cf)
12:23:12 Quit elchapolin (Client Quit)
12:29:20 Quit petur (Read error: Connection reset by peer)
12:33:35 Quit xorly (Ping timeout: 248 seconds)
12:39:16***Saving seen data "./dancer.seen"
12:39:29 Quit fujisan (Quit: Textual IRC Client:
12:41:11chrisjjZEN on eBay:
12:47:30 Quit JdGordon_ (Quit: Lost terminal)
12:53:22 Join soap [0] (~soap@rockbox/staff/soap)
13:13:51chrisjjpamaury, I'm installing on a further ZEN for testing, and notice the latest .nk is surprisingly small - 113Kb. Is that as it should be? The previous was 24,621Kb.
13:57:53pamaurychrisjj: it does not contain the OF
13:58:14pamaurybecause we don't dualboot at the moment, so I figured it doesn't make sense to embed 20MB for nothing
14:00:11 Quit elensil (Ping timeout: 240 seconds)
14:01:04pamaurychrisjj: did you test 8e82839f_lcdfix ?
14:13:11BilgusSpeaking of lua @pamaury I made a script that chrisjj ran on the zen that seemed to be missing rb.pcm_is_playing() is that function tied to specific hardware? or a specific version?
14:17:41pamauryBilgus: based on the name I would say no but I no nothing about lua for plugins
14:19:02pamauryis it documented somewhere ?
14:19:51 Quit Rower (Ping timeout: 272 seconds)
14:20:53pamauryI guess it tries to export all the functions of the plugin API but maybe some of them were forgotten
14:22:21Bilgushell IDK is anything regarding lua documented? I pulled it from the source header
14:23:26BilgusWeird I mean at least it throws up rather than silently failing
14:25:18pamauryBilgus: it's quite possible tat some functions were added to plugin.h but no one wrote the glue code to export them to lua
14:25:23pamauryLet me have a look
14:26:10Bilgusthey work fine on a bunch of sansas no clue about others besides the zen
14:26:53pamauryare you saying that rb.pcm_is_playing() works on some players but not all
14:28:31Bilgusfailed specifically for chrisjj and his zen replaced it with rb.audio_status() and it worked
14:28:37pamauryBilgus: afaict, pcm_is_playing is not exported at all by lua
14:29:03Bilgusextra weird then
14:29:43pamaurybut I don't see audio_status() either
14:30:26BilgusI think lua actually just parses the header file and makes its own functions from them
14:31:14pamauryah, I get it
14:31:18Bilgussome black magik stuff
14:31:34pamauryit does some magic to auto export
14:32:09Bilgusyep thats the file
14:32:49pamaurywell maybe the parser is broken, let me try to do a zen build, see how it works and what it generates
14:32:55Bilgusjust struck me as odd that the zen didn't have it
14:37:09 Quit jhMikeS (Ping timeout: 240 seconds)
14:38:24pamauryBilgus: pcm_is_playing has a wrapper generated apparently
14:38:46pamaury(on the ZEN)
14:39:19***Saving seen data "./dancer.seen"
14:39:48pamauryBilgus: maybe you could export the rb table to a file to see if it is indeed exported
14:39:53pamaurythere is code for that on
14:39:59 Quit The_Prospector (Quit: when in doubt, kernel panic)
14:40:55BilgusWonder why it failed then chrisjj sent this SS
14:42:49Bilgusmaybe chrisjj was using an old release or something IDK
14:43:13pamaurywell you can always try it in a ZEN sim
14:43:18 Quit Moarc (Quit: i znowu NADMUCHAŁ BALONA)
14:44:22BilgusWorked for me
14:44:52 Join paulk-blaze [0] (
14:45:41Bilgusheaded to work thanks
14:49:46 Join Moarc [0] (
14:52:09 Part robertd1
14:59:41 Quit shmibs (Quit: leaving =o)
15:01:05 Join shmibs [0] (
15:09:51 Quit wodz (Ping timeout: 258 seconds)
15:14:26 Quit paulk-blaze (Quit: Leaving)
15:23:44 Join robertd1 [0] (~root@
15:28:04 Quit michaelni (Ping timeout: 240 seconds)
15:32:52 Join michaelni [0] (
15:38:34 Quit michaelni (Ping timeout: 252 seconds)
15:39:59 Join michaelni [0] (
15:56:36parchdouch... any setting in rockbox for resetting volume on startup? Last night I had my player hooked up to speakers - just put my headphones on and had a blast of max-volume drums :(
16:03:41pamauryparchd: last volume is restored on boot normally
16:04:13pamaurylook in the manual
16:08:02 Join Bilgus_ph [0] (~Bilgus_ph@
16:08:13 Quit Bilgus_ph (Remote host closed the connection)
16:10:43 Join Bilgus_ph [0] (~Bilgus_ph@
16:12:03Bilgus_phParchd yes, google ' custom config file'
16:12:20 Quit Bilgus_ph (Remote host closed the connection)
16:12:37parchdBilgus: will do - thanks
16:13:13parchdpamaury: yeah, that is what I expect as the default so it was my own fault, but it was a nasty shock so I wanted to prevent it happening again
16:15:35pixelmalook for fixed.cfg (I think it is called)
16:18:35parchdpixelma: thanks - searching for that gave it me immediately
16:26:46 Join xorly [0] (
16:30:23 Quit Moarc (Ping timeout: 248 seconds)
16:30:55 Join Moarc [0] (
16:36:31 Nick Strife1989 is now known as Strife89 (
16:39:23***Saving seen data "./dancer.seen"
17:20:27 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
17:30:20__builtindammit these PMs on the forum are annoying as heck
17:41:10 Quit pamaury_ (Remote host closed the connection)
17:44:02 Part __builtin (" - Chat comfortably. Anywhere.")
17:48:33 Join __builtin [0] (~xray@rockbox/developer/builtin)
17:54:25 Quit pamaury (Ping timeout: 256 seconds)
18:05:50 Quit xorly (Ping timeout: 240 seconds)
18:32:33 Join The_Prospector [0] (~The_Prosp@unaffiliated/cornman)
18:37:46 Quit robertd1 (Quit: Leaving.)
18:39:26***Saving seen data "./dancer.seen"
18:49:33 Join johnb2 [0] (
18:56:34 Quit parchd (Ping timeout: 240 seconds)
19:14:50__builtinooh, good news
19:15:00__builtinmy antialiased line draw is now 51% efficient :D
19:15:14__builtin(in comparison to rb's lcd_drawline())
19:16:24 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
19:18:21__builtinoh crap, no
19:18:24__builtinthat was a fluke :(
19:19:31pamaurychrisjj: I have a test build with code for the LED, could you test it on the ZEN ? (I don't have a ZEN anymore, and the ZEN has a different LED setup then ZEN X-Fi)
19:20:47__builtinstill at 32%
19:30:23 Join parchd [0] (~parchd@unaffiliated/parchd)
19:54:49 Quit parchd (Quit: real life)
19:59:10 Join petur [0] (~petur@rockbox/developer/petur)
20:16:00 Join girafe [0] (
20:33:37 Join xorly [0] (
20:33:45 Quit johnb2 (Quit: Nettalk6 -
20:39:29***Saving seen data "./dancer.seen"
20:55:41 Join Finfi [0] (5f29cd2f@gateway/web/freenode/ip.
20:56:13Finfi...So I've cross-wired pins on my Clip+ and plugged it while in Ubuntu
20:56:39FinfiAnd this is what sudo fdisk -l gives me:
20:56:55FinfiDisk /dev/sdc: 30,6 MiB, 32096256 bytes, 62688 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes
20:57:09FinfiIs flash drive in this Clip+ dead?
20:58:15chrisjjpamaury, re your 'did you test 8e82839f_lcdfix' - not since we last IRCs last night. I picked be68b6a7b.
20:59:58pamaurychrisjj: and ?
21:00:07chrisjjI retested Verions be68b6a7b using clean install (no added/retained files except audio) and clean boot (power-on).
21:00:59chrisjjOn Unit Q (the unit that showed fail on this version before, in possibly ROLO or Reset boot), success on two runs (>=6hrs).
21:01:42chrisjjTest had power source = battery.
21:02:59chrisjjWould you like the next test to be 8e82839f_lcdfix? Or the new build with LED blink on panic?
21:05:06pamauryok so may the fails you saw were false alarm. Could you run ?
21:05:23pamauryit doesn't blink on panic, at the moment it just gives an extra entry in the debug menu to play with the LED
21:05:31pamauryIn System > Debug > Show HW Info > led
21:06:33pamaurythe code highligh the led color channel (in your case, only one: blue), press enter to start editing, then press up/down (or left/right can't remember) to dim it.
21:08:58chrisjjI think the many fails I saw were a true alarm. They weren't 100% then, so no big surprise they aren't 100% now. I'd bet they depend on a variable ingredient. My list of candidates includes: Boot type (power-on/reset/ROLO); user settings e.g. theme; power source (battery/external). When I'm sure we've got a stable recipe, i'll retest to find the culprit.
21:11:32FinfiAm I even readable?
21:12:33 Quit Moarc (Ping timeout: 240 seconds)
21:13:53 Join Moarc [0] (
21:15:08pamauryFinfi: yes
21:15:22pamauryFinfi: wait until someone knowledgeable answers you
21:20:27 Quit Moarc (Ping timeout: 255 seconds)
21:20:52 Join Moarc [0] (
21:24:36chrisjjpamaury: that is 3cda546d5-170111 and succeeds in an unclean install with clean boot on my ZEN Unit P.
21:25:31chrisjjBy succeeds I mean e.g. System > Debug > View HW Info > led , press Select, press Down -> LED lights.
21:26:08FinfiThere's another thing: the output of sudo fdisk -l doesn't change no matter if pins are cross-wired or not
21:31:48chrisjjpamaury, The UI's step values are unstable. The scale has the poor brightness linearity that we'd expect from uncorrected pulse width. The menu offers (in addition to 'blue', where Up/Down alter %age and brightness) 'unknown' on which Up/Down alter a displayed frequency temporarily
21:32:19chrisjjLooks to me ample for PANIC (and other) debugging!
21:32:55pamaurychrisjj: the unknown is a bug, I'll fix. Yeah it's not very linear but for LEDs, I'm not going to calibrate it
21:33:22chrisjjYou planning to use intermediate brightness for something?
21:34:52pamauryProbably not but since they are on the PWM channel, you get for free
21:35:18pamaurythe debug UI only changes the duty cycle, not the frequency, it's just a test, the code can affect both
21:35:20 Join TheLemonMan [0] (~root@irssi/staff/TheLemonMan)
21:36:15chrisjjOK. If you change your mind, sqrt().
21:37:06*__builtin gives chrisjj a friendly reminder that PANIC has no meaning
21:37:23pamaurywell it's doesn't make sense to do it in the interface, the API exposes frequency and duty cycle. The duty cycle does not necessarily corresponds to brightness
21:38:25*pamaury is trying to think if any other imx233 target has LEDs
21:38:40chrisjjThe duty cycle definitely does not correspond to brightness. Hence my suggestion of sqrt(). However yes, that can be left to the app rather than RB.
21:39:26 Quit Finfi (Quit: Page closed)
21:39:51pamauryI mean, if you set freq=1Hz and the duty cycle controls the blinking frequency, thus sqrt() makes no sense
21:39:52chrisjjDid the ZEN I sent die?
21:40:13pamaurychrisjj: it got stolen with my bag last year, along with other players
21:40:23chrisjjAw, bummer.
21:40:44chrisjjYou should have said. I can't replace the others, but I'll send you a ZEN.
21:40:45pamauryI still have the ZEN Mozaic, V, V Plus, MX, X-Fi, X-Fi2, X-Fi3
21:41:14chrisjjShame they didn't steal the crappy ones instead :-)
21:41:43pamauryWell that's was quite random, to be honest losing my computer was more annoying
21:41:56chrisjjSounds like you need a ZEN and a ZEN X-Fi Style.
21:42:14pamauryI have a ZEN X-Fi Style but the battery is dead :-(
21:42:28chrisjjAny others missing from your set?
21:43:05pamauryNot that I can think of. I also have the Touch 2 but not a good device. And I'm not interested in the pre-ZEN generation
21:43:29pamauryI also have the Style M300
21:44:02TorCchrisjj: pamaury: Was it just one ZEN that is having trouble? If so, have you verified that sleep timer isn't set to turn on on boot set for a couple hours?
21:44:32pamaurychrisjj: please wait a moment, I will dig into my player box to make sure which players I have exactly
21:44:34chrisjjIs your ZEN X-Fi Style on external power good enough for development? Or shall I send you a replacement?
21:44:40chrisjjOK, I'll wait.
21:45:33pamaurychrisjj: if you have a spare X-Fi Style that would be cool. The port should be functional but there are features I cannot test without the battery
21:45:41 Join MaxLeiter [0] (~MaxLeiter@unaffiliated/maxleiter)
21:45:45 Part MaxLeiter
21:46:13chrisjjOK, it's yours. Re ZEN, what colour do you want? :-)
21:47:17pamauryare there non-blak ZEN ?
21:47:23*pamaury didn't know
21:47:24 Join lebellium [0] (
21:47:42pamauryOk just checked, the only I miss are the ZEN and ZEN X-Fi Style
21:47:54pamaurychrisjj: any color is find, I don't really care
21:48:05TheLemonManthe Zen V came also in white
21:48:24chrisjjOK, black.
21:49:08 Join Finfi [0] (5f29cd2f@gateway/web/freenode/ip.
21:49:12FinfiI'm back
21:49:17Finfidid somebody answer my question?
21:49:27FinfiMy case is similar to this:
21:49:59*chrisjj hides pink ZEN from girlfriend.
21:50:06TorCFinfi: You can read the logs for that (see topic), but I had just read the scrollback, and it wasn't answered yet.
21:50:11pamauryFinfi: iirc, if you see a 30MB I think the flash is dead
21:51:01FinfiWhat is surprising me tho is that it doesn't matter if pins are short-wired or not
21:51:26TorCThat's what I recall too. You might want to replace it. There's a build of bootloader that will load RB from uSD if int flash is dead, but IIUC you have to install it before the player dies.
21:52:01TorCFinfi: The thread with links is,51304.msg237134.html#msg237134
21:52:10 Join johnb2 [0] (
21:53:07TorC*by replace it, I mean the player. dongs was working on replacing the flash nand, but I haven't seen that he succeeded.
21:53:35johnb2Finfi: I had a similar clip+ recently, i.e. showing the same behaviour. I was able to dd the bin file, but that did not change it.
21:54:27johnb2I gave up on it.
21:55:46chrisjjpamaury, re sending replacement ZENs, you have mail.
21:56:05chrisjjMeanwhile, which build do you want me to test? I could return to be68b6a7b+memDFS which will hopefully now give us the battery benchmark.
21:56:13pamaurychrisjj: thanks, answered
21:56:26 Quit johnb2 (Client Quit)
21:56:39pamaurychrisjj: yeah, we still need to compare memdfs vs no memdfs
21:56:44FinfiWell, thank you all
21:56:46pamaurywhich logs do you have so far ?
21:57:08FinfiSadly I don't see any way to bring my Clip+ to life
21:59:00chrisjjpamaury, none complete, and none I would now trust anyway! :) I'll rerun, with clean install & boot.
22:00:02TorCFinfi: Not easily, if at all, most likely. It was right around 0000 1/1/1 UTC-8 that the IRC logs would be found if you want to look up on replacing the nand. Some work a little later. Nothing conclusive has showed up on IRC that I've seen.
22:00:03pamauryok, so in this case, re-run be68b6a7b+memDFS and be68b6a7b
22:00:14chrisjjOK, will do.
22:00:22 Quit michaelni (Read error: Connection reset by peer)
22:00:23 Quit Finfi (Quit: Page closed)
22:00:44chrisjjTorC: no, it was both ZENs I tried.
22:00:44pamauryI also uploadded one here:
22:00:52pamauryDid I upload be68b6a7b+memDFS ?
22:02:35chrisjjI already received be68b6a7b and be68b6a7b+memDFS from you.
22:03:13*pamaury doesn't remember uploading be68b6a7b+memDFS and can't find it on his dropbox
22:03:16chrisjjYour [21:00] be68b6a7b is binary identical to the previous send.
22:04:58chrisjjRecd. here 08 ‎January ‎2017, ‏‎12:52:39
22:05:27pamauryI must have erased it, but if you have it that's fine
22:07:01chrisjjSo we're together:
22:12:21pamaurylebellium: here ?
22:16:51 Join michaelni [0] (
22:17:29lebelliumpamaury: yep
22:19:08pamaurylebellium: wanna try rockbox on the A15 ?
22:21:30lebelliumyes, good thing to do while eating and drinking a beer :P
22:24:39pamaurylebellium: ok so get g#1481 (I updadted it since last time)
22:24:40fs-bluebotGerrit review #1481 at : Initial commit for the Sony NWZ linux port and NWZ-E460 (WIP) by Amaury Pouly
22:28:13lebelliumthen do the same thing as 2 days ago?
22:28:43pamauryexcept you select NWZ-A10 in configure
22:28:58pamauryI couldn't test it obviously (it compiles though) so I could have made a mistake
22:29:54lebelliumnote that YP-R0 and YP-R1 are considered as Application but Linux NWZ as Sony :D
22:32:23pamauryis it ?
22:32:46lebelliumIsn't it a RaaA port?
22:32:53pamauryyes it's a RaaA port
22:33:06pamauryalthough RaaA is kind of vague
22:33:16pamauryit basically means "there is an OS"
22:35:48lebelliumI guess I muse have forgot to update something
22:36:01lebelliumubuntu@ubuntu-VirtualBox:~/rockbox/A10/bootloader110117$ ../../rbutil/mknwzboot/mknwzboot -b -o NW_WM_FW.UPG
22:36:02lebellium[ERR] This player is not supported: a10
22:36:04lebellium[ERR] Invalid boot file
22:36:06lebelliumResult: 2
22:36:19 Join chrisb [0] (
22:38:12pamaurylebellium: ah shit, I forgot mknwzboot
22:38:17pamauryI knew I had forgotten something
22:38:45pamaurysorry, swearing is bad
22:39:16lebelliumno problem; it still will be much faster than 2 days ago if we only have one issue tonight :D
22:39:30***Saving seen data "./dancer.seen"
22:40:20pamaurylebellium: gerrit updated
22:43:16lebelliumalready done
22:43:37pamauryWell you kinda of encountered all the problems last time, now it's much better
22:49:43lebelliumpamaury: on device when upgrading: "mount fail"
22:50:49pamauryah yeah I forgot about that, I think the A10 is using ext4, please wait 5 min
22:53:21pamauryah yes
22:57:15lebelliumI'm building rockbox meanwhile
22:59:45pamauryI have a patch, I'm trying it on my E460 first
22:59:59lebelliumIf there is any French here interested in NWZ and Rockbox, the E580 is on flash sale
23:01:27pamaurythat's expensive
23:01:32pamaurylebellium: I updated gerrut
23:01:42pamauryyou will need to recompile mknwzboot:
23:01:50pamaurymake -C rbutils/mknwzboot
23:02:02lebelliumpamaury: your price is definitely not the right one. You stole it :P
23:03:02pamaurystill 100€ for that is expensive
23:03:18*pamaury wouldn't spend more than 10 ;)
23:08:51pamaurylebellium: is it working now ?
23:09:09lebelliumI'm not flash_pamaury!
23:10:05*pamaury works on more bootloader features in the mean time
23:10:20lebelliumif you saw me using Ubuntu, you would probably laugh
23:12:27pamaurySo let's agree on the bootloader, my plan is:
23:12:28pamauryFirst boot: show the menu
23:12:28pamaurySubsequent boots: don't show the menu (unless swtch is on) and proceed with the previously chosen option
23:12:28DBUGEnqueued KICK pamaury
23:12:28pamauryWhen the bootloader is invoked and switch is initially on, power off after X seconds if there are no actions / not switch unhold (which one is better ?)
23:12:28pamauryIf switch is off and there is not activity for Y seconds: power off/boot (boot is probably better)
23:14:08TorCI'm not sure what you mean by "not switch unhold", but otherwise I think your plan makes sense.
23:14:41TorCI'd go for boot on timeout and figure that [OF|RB] will handle power off on timeout if needed.
23:15:01pamauryTorC: ah yeah, sorry, the device can detect key stroke even when hold is on. Which means that there is the option to consider a key stroke when hold is on as "activity"
23:15:21TorCSince it is reasonable to figure on accidental boot menu show request.
23:15:34TorCAh. I don't have the device to know that.
23:15:48lebelliumpamaury: bootloader working, keys working, debug menu working, spiderapp working, rockbox can't load and crash even after a reset
23:16:04pamaurySo you would boot on timeout always (ie never power off) and let the OF/RB handle it ?
23:16:23TorCBut do leave a longish timeout on boot menu, because the assumption is it is wanted.
23:16:34pamaurylebellium: by crash you mean ?
23:17:38lebelliumpamaury: well I should rather say "stuck on the rockbox selection logo". The E580 crashed (rebooted) at first until I solved it somehow but this one doesn't reboot, it just gets stuck
23:17:50pamauryTorC: ok so what should be the values of X and Y ? For Y (switch on) it should be small (like 5/10 seconds). For X, not sure
23:17:52TorCWell, it appears the boot menu will only show on request, so there was an intention to turn the device on, IIUC. Unless it's a rattling-around-in-backpack boot, in which case all bets are off.
23:18:08pamaurylebellium: that sounds like a crash indeed
23:18:22pamaurylebellium: can you look at the root of your device, there should be a file called rockbox.txt
23:18:48pamaurydoes it contain any useful info ?
23:19:39pamauryTorC: yeah good point
23:19:54TorCIf hold switch on (and could be forgotten) will call boot menu, probably 5-10s is good. For other button pushed to more deliberately call bootloader, 30s or so.
23:20:14lebelliumpamaury: not sure
23:20:15pamaurysounds good
23:20:52TorCOf course, hold switch might be used to avoid backpack boot, in which case lock, and power off if hold not released within 5-10s.
23:21:06TorCAssuming you can do that.
23:21:08 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
23:23:47pamaurylebellium: it's going to be tricky
23:24:14lebelliumpamaury: I get more or less the same problem as the E580. I deleted .rockbox but it won't. Some files want to say
23:24:29pamaurysome files want to stay ?
23:25:15pamauryI still don't explain why it happened on the E580 though, that's weird
23:25:17lebelliumyes, when I delete it just comes back (funny) and I have "font" directory with 4 fonts left
23:26:03lebelliumI finally managed to deleted it on E580 but I must find how I did it :D
23:26:03pamaurylebellium: try to reset the player, then go straight to usb mode
23:28:27 Quit girafe (Read error: Connection reset by peer)
23:28:58lebelliumdidn't help. Windows just can't delete the files. Linux did it but with this stupid MacOS-like Trash directory. I hate that. Now the device says "cannot boot rockbox", which is expected. I try to copy .rockbox again
23:31:48lebelliumnow booting
23:31:58lebelliumexactly like with the E580
23:32:19pamaurythat sounds like your OS is doing crap
23:32:28pamaurylebellium: so it's working ?
23:32:44lebelliummaybe it's a problem with the VM when mounting and unmounting USB devices
23:32:50lebelliumworking, great :)
23:34:11lebelliumsame battery indicator not stable
23:34:28lebelliumit's moving like in charge status
23:35:25lebelliumand just powered off because of (untrue) empty battery
23:36:08lebelliumplayback not crashing but no sound
23:36:46pamauryit's strange, I don't have this funny battery reading problem
23:37:18lebelliumI don't know yet if it does it only after fresh install or at each reboot
23:37:27lebelliumbut it needs some time to stabilize
23:38:06lebelliumit said 25%; 1 second later 9% now 12%
23:38:13lebelliumnow 27%
23:40:48lebelliumas on E580 at first I got no sound and somehow now I get sound (crappy, max output)
23:41:15pamaurylebellium: can you go to System > Debug > HW info > power
23:41:44pamauryand tell me how "avg voltage" and "raw voltage" are changing
23:44:15lebelliumraw voltage is changing every millisecond or such, I don't even have time to read
23:44:31lebelliumavg voltage was stable
23:44:47lebelliumand now I got kicked off, powered off
23:46:06pamaurylebellium: is raw voltage changing a lot (like hugely different values) ?
23:46:43lebelliumlooks like it was switching always between 2 or 3 values (4060mV and down to 3xxxmV)
23:47:09lebelliumoh OK
23:47:13lebelliumI figured it out
23:47:23lebelliumit depends if you're playing a file or not
23:48:08pamaurylebellium: really ?? weird
23:48:11lebelliumwith no playback it seems stbale
23:48:23pamaurylebellium: I have updated gerrit
23:48:31 Quit xorly (Ping timeout: 240 seconds)
23:48:36pamaurycan you updated and recompile rockbox (not bootloader) and see if it's better
23:49:01lebelliumok, then I go to bed
23:49:40 Quit skapazzo (Quit: leaving)
23:49:59pamauryah I think I get it
23:50:36pamaurysometimes the kernel might return -1 because it cannot read the voltage (it needs to send a message to the PMU). And thus the battery reads 4V, 4V, -1V, 4V, etc
23:54:38lebelliumActually it doesn't seems to be related to playback. With no playback it switching between 4042 and 4060mV raw voltage every millisecond now
23:54:49pamaurysmall variations are normal
23:54:55pamauryhuge one not really
23:55:05pamaurybut try the new one, it should be much better
23:55:13lebelliumjust compiled
23:55:14lebelliumtrying now

Previous day | Next day