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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2017-01-22

00:02:09pamauryno, as I said, I haven't even try to investigate the problem yet
00:02:29pamaurythe only thing I've done so far is: select an audio file, play it, notice sound is garbage, stop here
00:02:37pamaurylitterally :)
00:02:45Mihailok :)
00:03:09pamauryI need to try aplay and investigate the codec interface that Sony is using
00:03:24pamauryit uses some nonstandard stuff
00:04:08Mihailyes, some linux SoC do that
00:05:51 Quit ender` (Quit: "I'm the head wizard now. I've only got to give an order and a thousand wizards will ... uh ... disobey, come to think of it, or say 'What?', or start to argue. But they have to take notice."— Archchancellor (Terry Pratchett: Lords and Ladies))
00:06:08Mihailbut maybe you should just adjust alsa swparms/hwparms
00:08:34pamauryI haven't had a look at what rockbox alsa driver does, I don't know if it's only a pcm driver or if it can set sampling rate and all, the problem might just be there
00:09:33furrywolfMihail: why does it work with the old code? it's just so much more efficient than it handles the events without a problem? also, it does it even if the battery isn't charged.
00:11:02Mihailfurrywolf: IMHO old code just hide this problem
00:11:42furrywolf-8 does not work either.
00:11:47 Quit skapazzo (Quit: leaving)
00:12:13Mihaillogf.txt?
00:12:14furrywolfsame thing, freezes when cable is plugged, flashes usb image after cable unplugged, doesn't get enumerated.
00:12:57 Quit atsampson (Ping timeout: 255 seconds)
00:13:22furrywolfhttp://fw.bushytails.net/tmp/logf.06.txt
00:13:57furrywolfby "hides the problem", do you mean "handles interrupts efficiently"? :P
00:15:12Mihailit stop INT_AUDIO interrupts for some time, so others parts can work
00:17:46Mihailtry http://knk.square7.ch/rb/rockbox-fuzev1-usb-9.zip - try do as datasheet say :)
00:20:00 Join atsampson [0] (~ats@cartman.offog.org)
00:20:15furrywolfare you turning the charger off when you get the end-of-charge interrupt?
00:20:48furrywolfalso, if you %x your debug, it makes translating it to bits much easier. :)
00:21:08Mihailrockbox already do that
00:21:27furrywolfare you sure?
00:21:44Mihail99% :)
00:21:45furrywolfbecause it shouldn't generate an end-of-charge interrupt if the charger is off!
00:22:48Mihailyes, so I not sure is it software issue
00:25:24furrywolf-9 seems to work
00:25:36furrywolfshows usb image, enumerates
00:25:52furrywolfbah, I think I uploaded the wrong log file last time. sorry.
00:26:16Mihailgreat, datasheet was right :)
00:27:01*pamaury spots an unlikely claim
00:27:09pamaurydatasheets are never right ;)
00:28:01furrywolfhttp://fw.bushytails.net/tmp/logf.07.txt there's the logfile from -9. do you want me to reinstall -8 and upload the proper logf for it?
00:28:34furrywolfI forgot to dump the log, so you got -7's log again. heh.
00:28:47Mihailno, as -9 work
00:29:17MihailI'll do clean patch for gerrit
00:29:45furrywolfis it a chipset feature, that it generates the endofch interrupt continually until the charger is turned off, but the new code is too busy handling the interrupt to turn the charger off?
00:29:59furrywolfwhat does -9 do differently?
00:30:44furrywolfsince the device seems to go out to lunch on plugging it in, it's quite possible it's never turning the charger off.
00:31:57Mihaildatasheet say about EoC: "The interrupt must not be enabled if the charger block is disabled"
00:32:06furrywolflol
00:32:28furrywolfso if the charger is off, the pin floats, and generates random interrupts?
00:33:06Mihailprobably, on AMSv2 datasheet don't say this
00:33:16furrywolfalthough I'm much more fond of the theory that it's wasting so much time busy-looping to handle the interrupt, it never makes it to the charger off step.
00:33:28furrywolfalso, it was failing before with the battery only half charged.
00:34:34*furrywolf leaves it running unplugged for a while to draw the battery down a bit. random track pick was cult of luna - mariner - wreck of the ss needle. about ten minutes long, should run the battery down a little. :)
00:35:31furrywolfcompletely random feature request: an option not to use the very uppermost row of pixels. at least on my unit, it's under the bezel.
00:36:34pamaury__builtin: are you *sure* you want to port SDL to rockbox ?
00:37:07furrywolfalso, a hardware question: is the line out dock better audio quality? because the headphones jack sounds like it has some annoying equalization applied to it.
00:38:46furrywolfso your -9 version disables the charger interrupt on turning the charger off. does it get re-enabled when the charger is turned back on?
00:39:16Mihailnot sure, I have plans for future - check and improve audio quality and battery life time on AMSv1
00:39:30Mihailyes, it back on
00:39:41furrywolfpresumably if I let it fully charge, unplug it, listen to music for a while, then plug it back in, it had better turn the charger back on, and any interrupts it needs. :)
00:43:22furrywolfdid you reverse-engineer the proper charge voltage and current limit from the stock firmware?
00:43:32__builtinpamaury: too late now :)
00:43:49__builtinthough it's surely not impossible, it's been done for ipodlinux in the past
00:44:30***Saving seen data "./dancer.seen"
00:44:39 Quit xorly (Ping timeout: 248 seconds)
00:45:18Mihailpamaury: BTW if you can, just play sound in original FW and "cat /proc/asound/card0/pcm0p/sub0/hw_params" than do same in rockbox and compare
00:47:33 Quit pamaury (Ping timeout: 252 seconds)
00:53:24__builtinhmm, this is a tough bug to fix
00:53:38__builtinwhen I try to build SDL under the simulator it thinks it's being built for linux
00:57:32 Quit Milardo (Quit: Page closed)
00:59:04 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
00:59:48Mihailfurrywolf: was wrong, try: http://knk.square7.ch/rb/rockbox-fuzev1-usb-10.zip
01:00
01:00:32Mihail-9 was wrong
01:21:48furrywolfok
01:25:08 Join Jinx [0] (Dojo@unaffiliated/jinx)
01:25:33furrywolfam I missing something? no dump log file option?
01:25:44furrywolfin any case, it did not work. back to freezing, then showing usb image after unplug, no enumeration.
01:26:27Mihailit was clean build without debug
01:27:32Mihailit looks like your player have broken end of charging detection
01:27:42furrywolfalso, it's been playing music since I installed -9, unplugged, so you can rule out anything to do with the fully charged detection...
01:28:59furrywolfgiven as the stock firmware works fine, I highly doubt that...
01:29:06furrywolfand the old code works fine.
01:29:06furrywolfheh
01:29:39furrywolfit could well be "doesn't behave the way rockbox thinks it does", but I doubt it's behaving differently than sandisk thinks it does.
01:30:24dyspamaury: to add insult to injury: According to a review on some german site, these things also have a ADSP-BF606 on board (dual-core blackfin DSP)
01:30:46furrywolfLOL, checking the news. there was a women's rights (i.e. anti-trump) march in town. it was supposedly going to be peaceful. now the city police are calling for backup from both the county sheriff and the highway patrol... lol
01:33:14furrywolfperhaps you want the endofch to be edge, not level?
01:33:40furrywolfit sounds like it stays high, according to the datasheet.
01:33:59furrywolfwith the interrupt set to trigger on high level, not low-high edge, it might keep firing them.
01:34:05furrywolfbut, again, the battery is no longer fully charged.
01:34:53furrywolfaccording to the view battery debug, the battery is at 81%
01:36:09furrywolfwhen I plug in the usb cable on the show battery screen, it shows charger: present immediately before freezing, but still shows discharging. when I unplug the usb cable, it flashes charging.
01:40:49furrywolfI assume installing the build environment is non-trivial?
01:43:03Mihailsure, I build it manually :)
01:43:11Mihailnot sure
01:43:53furrywolfI mean I assume I need to install various cross-compilers and such, and it's not easy to make it all work.
01:44:23Mihailtry http://knk.square7.ch/rb/rockbox-fuzev1-usb-11.zip it have bigger logf
01:45:04Mihailare you have linux?
01:45:08furrywolfyes
01:46:06Mihailyou need binutils-arm and gcc-arm cross-compiler
01:46:33Mihailwe have some script to download and build it
01:48:13Mihailhttps://www.rockbox.org/wiki/HowToCompile
01:49:04furrywolfI'll check that after -11 finishes downloading... my connection can't do two things at once and have either one be useful.
01:52:46furrywolfhttp://fw.bushytails.net/tmp/logf.08.txt
01:56:49furrywolfis the charger enabled at boot, or disabled until enabled by some event?
01:58:40Mihailenabled by event
01:59:16Mihailbut I enable EoC in same event
01:59:24furrywolfdoes the endofch bit read high whenever the charger is disabled?
01:59:47furrywolfI assume the 68 is the first irq register, in decimal?
01:59:53Mihailin -10 -11 - no
02:00
02:00:00Mihailhex
02:00:41furrywolfso you changed it from decimal to hex. that's why it doesn't make sense now. lol
02:01:00Mihail:)
02:06:47furrywolfdo we have a datasheet for the I2C control and interrupt registers?
02:07:04Mihailno
02:07:09 Quit ZincAlloy (Quit: Leaving.)
02:07:47furrywolfhrmm
02:07:55furrywolfI really don't like doing actual reads/writes inside an isr.
02:08:04furrywolfover the bus the interrupt is for
02:08:28Mihailhttp://knk.square7.ch/rb/rockbox-fuzev1-usb-12.zip with more info
02:09:20 Quit dys (Ping timeout: 240 seconds)
02:10:29 Quit jhMikeS (Ping timeout: 258 seconds)
02:11:39furrywolfthat build works
02:12:58 Join dys [0] (~dys@tmo-099-104.customers.d1-online.com)
02:13:12Mihaillogf.txt?
02:13:13furrywolflet me guess, you didn't change anything, so it works because of some timing change with more debug? :)
02:13:22furrywolfhttp://fw.bushytails.net/tmp/logf.09.txt
02:13:26furrywolfit takes a while to upload things. it's 500k!
02:13:26Mihailno :)
02:25:46Mihaillast try for today: http://knk.square7.ch/rb/rockbox-fuzev1-usb-13.zip
02:26:20furrywolfI tried doing a git clone... didn't realize how large it is.
02:27:16Mihail~200 MB in packed state
02:27:57furrywolfyeah, that'll take forever. heh.
02:31:38furrywolfpanic
02:31:46furrywolfsd read error. hrmm.
02:32:18__builtinohhhh
02:32:35furrywolfwell, I did use a sd card that I'd written "BAD" on with a sharpie...
02:32:49__builtinI figured out why SDL won't work... it's including the system SDL header...
02:33:08Mihail__builtin: :)
02:33:35furrywolfyour last build is corrupting ram
02:33:58furrywolfsd panics, garbage written onto the lcd, etc. heh.
02:34:19furrywolfit fails hard quite quickly after connecting the usb cable.
02:34:32furrywolfnot going to get a logf out of it.
02:35:41Mihailvery strange
02:35:49__builtinthat explains everything...
02:36:29furrywolfthe last crash overwrote the fonts. :P
02:37:36__builtinwow... now it works :D
02:38:38furrywolfmy last attempt ate the firmware.
02:41:02furrywolf-12 still works fine. heh.
02:41:36furrywolfhttp://fw.bushytails.net/tmp/rockbox02.jpg that's the closest you're going to get to a logf from -13. :)
02:41:56Mihail-13 just bad number :)
02:42:24furrywolfit works until the usb cable is connected, then self-destructs. I'm assuming you're clobbering ram.
02:43:11furrywolfor the stack is growing into other things
02:43:33Mihailhttp://knk.square7.ch/rb/rockbox-fuzev1-usb-14.zip
02:44:32***Saving seen data "./dancer.seen"
02:50:07furrywolfthat one worked, however my internet connection is being so crappy I don't know if I'll get the logf uploaded
02:50:33furrywolfuploading
02:50:39 Join MrZeus1 [0] (~MrZeus@2a02:c7f:7025:ed00:9835:fcfc:58a8:daa8)
02:51:46furrywolfit worked, but a little oddly. it acted like it was plugged, unplugged, then plugged, and then it worked. but everything flashed by pretty quickly, so I'm not sure.
02:52:01furrywolfhttp://fw.bushytails.net/tmp/logf.10.txt
02:52:46furrywolfwow, lots of values for ENRD0 in that one.
02:56:12Mihailgood, now I go to sleep
02:56:37furrywolfso what did you figure out?
02:57:49Mihailnot sure, what is with usb, is it work normal or reconnect after few seconds?
02:58:01furrywolfit reconnected at least once
02:58:36furrywolftried again, now it's looping continually.
02:58:50furrywolfshows usb plug, shows loading, shows main menu, shows usb plug, etc etc.
02:59:39Mihailbad, try upload new logf after few loops
03:00
03:02:51furrywolfthis is going to take a while to upload
03:02:55furrywolf1.6MB!
03:03:47furrywolffortunately it compresses well
03:05:15furrywolfhttp://fw.bushytails.net/tmp/logf.11.txt
03:07:44Mihailok, I'll check it tomorrow
03:07:58furrywolfok, cyas
03:08:12 Quit Mihail (Quit: http://www.kiwiirc.com/ - A hand crafted IRC client)
03:08:42furrywolfbah, I spent all afternoon on this again, and didn't get done the things I needed to get done.
03:18:23__builtincrap, the ported SDL is screwing with the sim a lot
03:39:20 Quit dys (Ping timeout: 240 seconds)
03:48:04 Join dys [0] (~dys@tmo-113-173.customers.d1-online.com)
04:00
04:17:06 Quit athidhep (Ping timeout: 256 seconds)
04:44:33***Saving seen data "./dancer.seen"
04:55:37 Join Senji_ [0] (~Senji@85.187.103.250)
04:58:29 Quit Senji (Ping timeout: 240 seconds)
05:00
05:53:39 Join naleo [0] (~naleo@unaffiliated/naleo)
05:59:51 Nick JdGordon_ is now known as JdGordon (~jonno@rockbox/developer/JdGordon)
06:00
06:15:05 Join athidhep [0] (~afoakf@unaffiliated/athidhep)
06:17:23 Quit TheSeven (Disconnected by services)
06:17:35 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
06:44:37***Saving seen data "./dancer.seen"
07:00
07:09:35 Quit dys (Ping timeout: 240 seconds)
07:12:09 Quit Strife89 (Ping timeout: 260 seconds)
07:21:23 Quit furrywolf (Ping timeout: 252 seconds)
07:42:29 Quit alexweissman (Remote host closed the connection)
07:45:39 Quit amiconn (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
07:45:39 Quit pixelma (Quit: .)
07:45:50 Join pixelma [0] (~pixelma@rockbox/staff/pixelma)
07:45:51 Join amiconn [0] (~amiconn@rockbox/developer/amiconn)
07:50:35 Quit amiconn (Client Quit)
07:50:35 Quit pixelma (Client Quit)
07:55:03 Join pixelma [0] (~pixelma@rockbox/staff/pixelma)
07:55:04 Join amiconn [0] (~amiconn@rockbox/developer/amiconn)
08:00
08:00:32 Join naleo_ [0] (~naleo@unaffiliated/naleo)
08:00:44 Quit naleo (Read error: Connection reset by peer)
08:05:48 Join ender` [0] (krneki@foo.eternallybored.org)
08:16:56 Quit naleo_ (Ping timeout: 248 seconds)
08:20:29 Quit shmibs (Quit: leaving =o)
08:28:31 Join shmibs [0] (~shmibs@shmibbles.me)
08:30:18 Join Barbadus [0] (62e117a1@gateway/web/freenode/ip.98.225.23.161)
08:33:26BarbadusAre themes and forums currently down?
08:44:41***Saving seen data "./dancer.seen"
09:00
09:01:24 Quit Barbadus (Ping timeout: 260 seconds)
09:20:30 Quit MrZeus1 (Ping timeout: 255 seconds)
09:25:33 Quit kugel (Read error: Connection reset by peer)
09:44:59 Join fishbulb [0] (~fishbulb@unaffiliated/fishbulb)
10:00
10:08:01 Join dys [0] (~dys@tmo-109-78.customers.d1-online.com)
10:09:20 Join Echo [0] (c25a5869@gateway/web/freenode/ip.194.90.88.105)
10:09:43 Nick Echo is now known as Guest44936 (c25a5869@gateway/web/freenode/ip.194.90.88.105)
10:10:43Guest44936Hi
10:11:08Guest44936I cant brows to the theme page on rockbox website
10:11:39Guest44936Any idea why and how to fix it?
10:13:45TorCGuest44936: Because the server's been having problems. It's been up and down for at least a month.
10:14:40TorCscorche` is rebuilding it anew, but I'm not sure what the ETC is on that project, or what it means for the server in the interem.
10:15:27Guest44936Ow well, hope they manage to get it working
10:16:33TorCKeep checking randomly. It may be up and down through the process. You can also hang here for updates.
10:16:46Guest44936In the meanwhile, any one have an offline copy of "Information Beauty" theme for sansa clip plus?
10:17:07TorCThat, I can help with. Give me a couple minutes.
10:17:25Guest44936Greate, thanks
10:22:04 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:15fe:7dc0:97ad:53c)
10:22:40TorCGuest44936: Hmmm... Maybe I don't have it hanging around in it's original form. I've got an edit of it I did for the zip that just adds a couple lines.
10:24:18TorCAh. Looks like I do have an editor backup that should be it. Just a moment while I upload it.
10:28:01TorCGuest44936: http://pastebin.com/fqN2m8zq
10:28:42TorCI don't certify that it is an unmodified version, but looking at the diff between it and the clip zip version I made and my memory, I believe it is.
10:29:07TorCDrop it in .rockbox/wps
10:29:21TorCThat's all the theme is, AFAICT.
10:44:44***Saving seen data "./dancer.seen"
10:46:03 Join lebellium [0] (~chatzilla@89-93-177-91.hfc.dyn.abo.bbox.fr)
10:49:14 Join xorly [0] (~xorly@ip-89-176-102-19.net.upcbroadband.cz)
10:49:16Guest44936Thanks, but it doesn't work
10:50:33Guest44936I think there are more files needed (.sbs, .fms) or some dirrectory
10:56:56TorCEither that, or there was an extra save and it's bad.
10:57:07TorCGuest44936: Lemme check.
10:59:54TorCGuest44936: Ah. There was, and I just somehow never noticed. You can find it at http://pastebin.com/T9JZf5jp
11:00
11:02:15TorCHope that works. find tells me that should be the lot of the files from a backup of working system. Headed to bed now, but I'll read the scrollback in the morning.
11:03:00Guest44936It works now, thanks
11:03:41Guest44936Still a minor difference but it's good for now, until the theme site starts working again
11:03:59Guest44936Thanks a lot
11:04:44Guest44936And for all of you listening, the .cfg file should be in .rockbox/themes, not like the other files
11:05:23lebelliumI have informationbeauty 1.0 on my Clip+
11:05:28lebelliumshould I extract it?
11:05:46 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:05:49lebelliumInformative beauty*
11:18:13Guest44936If you can it would be grate
11:18:15Guest44936thanks
11:18:53 Quit athidhep (Ping timeout: 260 seconds)
11:19:17 Join n3m9 [0] (~n3m9@ANantes-652-1-4-13.w90-25.abo.wanadoo.fr)
11:22:38lebelliumGuest44936: https://drive.google.com/file/d/0B7z2DypmySvnMTZTNldId0dFYmc/view?usp=sharing
11:27:53Guest44936Thanks, that helps a lot
11:28:29lebelliumI don't know if that's the latest version on the theme site. I haven't updated the themes on my Clip+ for ages
11:28:46Guest44936TorC: There was only one difference- the first line in your version was not in the original file
11:36:42Guest44936That is good enough until the site gets back online, thanks
11:37:20 Join MrZeus1 [0] (~MrZeus@2a02:c7f:7025:ed00:a494:9314:2854:cf60)
11:41:07 Quit ZincAlloy (Quit: Leaving.)
11:44:26 Join robertd1 [0] (~root@186-90-12-124.genericrev.cantv.net)
11:49:39 Join parchd [0] (~parchd@unaffiliated/parchd)
11:56:30 Join jhMikeS [0] (~jethead71@d192-24-173-177.try.wideopenwest.com)
12:00
12:00:28lebelliumpamaury: NWZ-A820. Too old?
12:03:41pamauryyeah
12:05:27lebelliumI can't find a S7xx series :(
12:12:47duo8pamaury do you use any of the players daily?
12:20:52pamaurynot recently
12:26:12pamaurywhy ?
12:28:01lebellium*Panic* Incorrect CPU mode on Clip+ with fresh build
12:28:03lebelliumhum
12:30:00 Join Boltermor [0] (~Boltermor@subscr-46-148-172-69.dhcp-docsis.net.tomkow.pl)
12:30:14lebelliumso power the Clip+ on with a USB charger causes this panic screen everytime
12:31:05lebelliumwell, plugging the USB charger while already powered on too
12:32:39lebelliumoh, USB connection too
12:32:46lebelliumokay, so USB is broken
12:34:40pamauryjhMikeS identified some incorrect behaviours and added a panic to detect them so we can fix them
12:35:21lebelliumah
12:35:33fishbulbcould someone tell me why the h300 has a noticable hiss compared to (my laptop for example)
12:36:04lebelliumfishbulb: we already discussed it some days ago, no?
12:36:16fishbulbnot really, just that "it's not the dac"
12:36:52lebelliumbut if we couldn't answer the question for sure last time, why would we today?
12:37:35fishbulbok I can't quite remember.
12:38:00lebelliumjhMikeS: I have the panic screen on my Clip+ when connecting to USB. Tell me if I can help
12:38:32fishbulbI'm not sure where on the board to look, there have been people replacing caps but that only alters the "colour"
12:39:09 Part robertd1
12:41:05 Join robertd1 [0] (~root@186-90-12-124.genericrev.cantv.net)
12:42:04fishbulbTHD and all the rest of the amp specs helps knowing "this is good/bad" but I'm not sure what about the board is crappy
12:43:09lebelliumI'm not sure anyone can tell you.
12:43:49lebelliumas it was explained to you here https://www.rockbox.org/irc/log-20170117, the hiss probably comes from somewhere in analog domain after the DAC
12:44:23lebelliumI don't know how you can find out where exactly
12:44:45***Saving seen data "./dancer.seen"
12:44:50lebelliumthat looks like a job for a very knowledgeable person
12:45:16fishbulbah logs
12:47:00 Quit n3m9 (Read error: Connection reset by peer)
12:48:20fishbulbI should probably stick to pinterest or something right now
12:49:21 Join skapazzo [0] (~skapazzo@151.9.205.1)
13:00
13:00:51 Quit idonob (Read error: Connection reset by peer)
13:01:33pamaurylebellium: I will try on my clip+ this afternoon
13:01:55lebelliumok
13:02:06pamaurybasically Mihail was investigating a problem on amsv1 and in the process jhMikeS discovered that some code in apps was doing very dangerous operations
13:02:19pamauryso I suggested that we panic when that happen
13:02:25pamauryinstead of hoping for the best
13:02:29lebelliumno problem
13:02:40pamauryso it's possible we see a few panic in the future until we fix it :)
13:02:40lebelliumit's just that for now I have to use OF :D
13:03:09lebelliumno way to charge it, unless it charges on the panic screen
13:03:11pamaurylebellium: but I think on the clip+, you need to talk to Mihail
13:03:39pamaurybecause he is working on the usb charging issue, his code is what causes the panic but it's unclear how to solve it
13:07:05lebelliumI'm not sure I understood properly. Has the code done dangerous operations for a while?
13:11:38 Join idonob [0] (~Owner@S010610c37b922980.vs.shawcable.net)
13:19:16 Quit Boltermor (Quit: Leaving)
13:20:47pamaurylebellium: yes
13:22:44 Quit idonob (Ping timeout: 245 seconds)
13:24:02duo8would it be easy to port the jz4760 port to other devices, once one of them got support?
13:25:39pamauryyes and no
13:25:55duo8?
13:26:06pamauryfirst no because no port is easy, there is always some rever engineering necessary that might or might not be hard
13:26:06duo8what would need to be changed, other than UI?
13:26:27pamaurybutton, lcd drivers for starters
13:26:56duo8do they not use a common interface for LCDs?
13:27:24pamauryyeah but LCD itself is completely different, LCDs are complicated beats, you need to send commands to them and those commands change with every lcd
13:27:55duo8huh
13:28:10duo8i guess it didn't help that the lcd on mine is literally no name
13:28:20pamauryalso installation is not always easy, if the devices provides an easy way to install a firmware, but if the device runs linux, once may need to patch uboot, doing so in a "we don't redistriute binaries without code" way might be a lot of work
13:28:25duo8is this how it is for all embedded devices?
13:28:30duo8(the lcd i mean)
13:28:32pamauryyes
13:28:39pamaurymost LCD in rockbox are no name
13:28:56pamauryactually an LCD is made up of two parts: controller + panel
13:29:24pamauryusually given the sequence of command in the original firmware, we can guess what the controller is (or at least is close to), that helps with the process
13:29:31duo8i guess laptops had it better
13:29:34pamaurybut some we have no clue, we just do what the OF does
13:30:04 Join idonob [0] (~Owner@S010610c37b922980.vs.shawcable.net)
13:30:21duo8so there's the LCD, what else need to be changed?
13:30:39pamauryembedded systems are completely different from laptops though
13:30:50duo8yeah i guess
13:31:03pamaurybutton driver, usually it's no too much work though
13:31:13pamauryinstallation as I said, can be a tricky point
13:32:15duo8the recent ingenic devices do use uboot afaik
13:32:29pamaurythey use uboot only if they run linux
13:32:45duo8yeah but the recent ones all do it seems
13:32:56duo8X1 gen2, all those X1000 devices
13:34:16duo8what about the battery?
13:34:36pamaurybattery is easy
13:35:00 Quit girafe2 (Read error: Connection reset by peer)
13:35:24duo8there's also the PMU which seems different on some devices
13:42:49pamauryyeah but it's probably not a huge deal
13:42:52pamauryhopefully
14:00
14:07:21 Quit parchd (Ping timeout: 240 seconds)
14:18:21 Quit gluytium (Ping timeout: 264 seconds)
14:25:55 Join gluytium [0] (U2FsdGVkX1@ma.sdf.org)
14:44:47***Saving seen data "./dancer.seen"
14:49:30 Join parchd [0] (~parchd@unaffiliated/parchd)
16:00
16:01:27 Quit parchd (Ping timeout: 248 seconds)
16:02:47 Quit MrZeus1 (Ping timeout: 255 seconds)
16:09:43 Join cc___ [0] (~ac@2001:910:113f:1:6a05:caff:fe1c:1627)
16:18:43 Quit jhMikeS (Ping timeout: 276 seconds)
16:19:51 Join furrywolf [0] (~randyg@72-57-75-247.pools.spcsdns.net)
16:38:20 Join JanC_ [0] (~janc@lugwv/member/JanC)
16:39:35 Nick JanC is now known as Guest57312 (~janc@lugwv/member/JanC)
16:39:35 Quit Guest57312 (Killed (hitchcock.freenode.net (Nickname regained by services)))
16:39:35 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
16:44:51***Saving seen data "./dancer.seen"
17:00
17:06:03 Quit fishbulb (Ping timeout: 245 seconds)
17:27:12 Join alexweissman [0] (~alexweiss@c-68-51-123-75.hsd1.in.comcast.net)
17:28:17 Join Bilgus_ [0] (~Bilgus@gateway/tor-sasl/bilgus)
17:31:09 Quit Bilgus (Ping timeout: 240 seconds)
17:55:25 Join _mt_ [0] (~MT@2601:482:4402:7b60:5c71:55a1:a0d1:1f60)
18:00
18:22:46 Nick __builtin is now known as builtin (~xray@rockbox/developer/builtin)
18:44:53***Saving seen data "./dancer.seen"
19:00
19:09:47 Join Mihail [0] (25d4bc5c@gateway/web/cgi-irc/kiwiirc.com/ip.37.212.188.92)
19:10:14Mihailpamaury: I want rework powermgmt-ascodec.c: I want to replace ascodec_endofch() on detecting by voltage (as we currently do for recharge detection). As you can see from furrywolf's case - we can't fully trust EoC. So detecting by voltage should by safe and flexible - we can easily set any voltages (for your g#1412). Is it acceptable?
19:10:16fs-bluebot_3Gerrit review #1412 at http://gerrit.rockbox.org/r/1412 : 3WIP: Implement final charge voltage selection, enable it on ascodec basec chargers. by Amaury Pouly
19:14:59 Join Strife89 [0] (~quassel@adsl-98-67-57-203.mcn.bellsouth.net)
19:15:08builtinit works!!
19:15:18*builtin can draw a 100x100 rectangle!
19:15:37builtinwith a bunch of artifacts, of course
19:15:40 Nick builtin is now known as __builtin (~xray@rockbox/developer/builtin)
19:16:27pamauryMihail: detecting by voltage is not very accurate though, for Li-Ion, only current draw can reliable detect end of charge
19:18:00pamaury*reliably
19:19:12Strife89__builtin: Which target(s) are you testing on, by the way? (Curious.)
19:19:22__builtinStrife89: ipod6g
19:19:27Strife89Ahhh
19:19:36__builtinit's still buggy as hell
19:20:02Strife89Here I am imagining trying to do that on my targets, which all have tiny, low-res screens. (Clip Zip, Clip+, c200v1)
19:28:04__builtinahh, I think I know what the bug is
19:28:30__builtinTLSF isn't getting reset between runs
19:34:37__builtinhmm, not sure why, but everything's being drawn at half height too
19:39:33__builtinheh, dividing the stride by 2 fixes everything :)
19:46:58__builtintime to write some input drivers
19:53:25 Quit pamaury (Remote host closed the connection)
19:53:52 Join girafe [0] (~girafe@LFbn-1-11729-221.w2-7.abo.wanadoo.fr)
19:54:46furrywolfanother bug... I can't browse the last 10% or so of my playlist.
19:55:35 Join jhMikeS [0] (~jethead71@d192-24-173-177.try.wideopenwest.com)
19:55:55furrywolfAs I scroll down towards the end of the playlist, first acceleration stops working (that is, spinning the wheel faster no longer scrolls more than one line), then the time to scroll each line becomes stupid, until it takes several seconds per track.
19:56:14furrywolfplaylist is only ~3900 tracks.
19:58:46furrywolfcharge voltage is constant for the last, to ass-pull a number, 20% of the charge. it can't be used to detect end of charge.
20:00
20:02:33Mihailfurrywolf: I think we can add detection broken EoC - if we have EoC at low voltage we should switch it off and detect end of charge by voltage
20:04:09Bilgus_I think that is asking for trouble, what exactly is wrong with EoC?
20:06:04furrywolfMihail says my player sets the endofch flag when it's not done charging. I still think it's more likely to be some sequencing issue with turning the charger on/off or something.
20:06:24__builtinStrife89: feel free to try g#1521
20:06:26fs-bluebot_3Gerrit review #1521 at http://gerrit.rockbox.org/r/1521 : 3[WIP] Port of SDL 1.2.15 by Franklin Wei
20:06:33furrywolfthe playlist bug is... wtf? how can a bug as bad as not being able to use the bottom 10% of your playlist not fixed yet? lol
20:06:45__builtinif it uploads...
20:07:08Bilgus_IIRC there is a configurable EoC voltage flag
20:07:55furrywolfthat sets the actual charge output voltage.
20:08:19furrywolfat least if I remember the datasheet right.
20:08:27furrywolfand I hope we're setting it based on reverse-engineering, not randomly. :)
20:09:25Bilgus_https://github.com/Rockbox/rockbox/blob/1f8ea9fe27313228e5df67ce6447830b5c30e5e3/firmware/target/arm/as3525/ascodec-as3525.c#L312
20:09:42Bilgus_that also sets up the eoc threshold
20:12:27furrywolfI don't see a setting for the eoc threshold
20:12:38MihailI can't reproduce problem with EoC on my players and we have only one report about this problem, so it looks like hardware problem
20:13:32*furrywolf thinks ungodly slow interrupt handling is the problem
20:13:55furrywolfdoing an i2c transfer in the isr is just... ugh
20:15:49Bilgus_VCHG_trick Charger Endpoint Voltage (trickle
20:15:50Bilgus_charge)
20:15:50Bilgus_BVDD<=3V, CHG_IN = 4.4V 0.70*
20:15:50DBUGEnqueued KICK Bilgus_
20:15:50Bilgus_CHG_IN
20:15:50Bilgus_0.72*
20:15:51***Alert Mode level 1
20:15:51Bilgus_CHG_IN
20:15:53Bilgus_0.74*
20:15:57Bilgus_CHG_IN
20:15:59Bilgus_V
20:16:58furrywolfBilgus_: that's not something we can set, as far as I can see...
20:17:10Bilgus_so if charge volts is 3.7v trickle chg is 2.59 - 2.73v
20:17:22Bilgus_no its strickly a %
20:17:23furrywolfno
20:17:37furrywolfyou are mis-interpreting what those numbers are, and do. :)
20:17:58furrywolfCHG_IN is the voltage applied to the charger pin - USB bus voltage, wall charger, whatever.
20:18:21furrywolfand the trickle charge is the very initial charge of a completely drained battery, usually not even used, not the final finishing charge.
20:18:46furrywolfthat's just saying what it will trickle charge an over-drained battery to before switching to the main charge cycle
20:20:13Bilgus_BVDD<=3V, CHG_IN = 4.4V yep you are right
20:20:40furrywolfthe final charge is constant voltage. to detect end of charge, it looks for when the charge current drops below a specific threshold, which is Ichg_off, and is a percentage of the programmer charge current.
20:21:08furrywolfprogrammed
20:22:12furrywolftypical is 10%. so if you program it for 150ma charging, it declares the battery charged (and you should turn off the charger) when the charge current drops below 15ma.
20:23:49Bilgus_setting vChg though sets that constant voltage though
20:24:03furrywolfyes
20:24:25furrywolfand it needs to be set based on either the battery cell datasheet, or reverse-engineering the stock firmware. it is _not_ something you can screw with.
20:25:24furrywolfknowing that voltage won't let you know when it's done charging, though. when it hits the constant voltage point, it will still be drawing the full charge current, and the battery might only be 70-80% charged. then you wait for the charge current to decrease, and declare it charged when it drops below a certain threshold.
20:25:52***Alert Mode OFF
20:26:08furrywolfthe charger should be doing all of this for you... mihail says my charger is broken, but I doubt it.
20:26:16Bilgus_i'm pretty sure we set Ichg as .5c or less
20:26:32furrywolfsince the stock firmware, and the non-overly-slow i2c code, both work fine.
20:27:28Bilgus_yes thats my point if you start screwing with EoC rather than relying on the built-in (SAFE) you are asking for trouble
20:29:31Bilgus_and actually there is a patch up on gerrit that allows you to go under the set Vchg #g1412
20:29:32fs-bluebot_3Gerrit review #1412 at http://gerrit.rockbox.org/r/1412 : 3WIP: Implement final charge voltage selection, enable it on ascodec basec chargers. by Amaury Pouly
20:30:46 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:6189:5124:b028:c993)
20:31:00Bilgus_pamaury and I set that one up to allow you to undercharge a battery read < the max Vchg (which came from the batt data sheet)
20:32:02furrywolfundercharging gives you longer battery life at the expense of shorter runtime
20:32:18Bilgus_yep
20:33:20Bilgus_doesn't take into account DoD but hey its a start
20:33:47furrywolfany ideas about the playlist speed bug? it's bad enough I'm surprised it exists, but it does...
20:34:58 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
20:35:32lebelliumpamaury: I just bought a NW-A25. I needed so spend some money \o/
20:35:49pamauryoh wow, that's petty money, how much ?
20:36:10lebelliumhttp://www.ebay.fr/itm/332098977109?_trksid=p2060353.m2748.l2649&ssPageName=STRK%3AMEBIDX%3AIT
20:36:38lebelliumthere is a PHA-1A DAC/AMP with
20:36:44Bilgus_have you tried GS>Display>Scrolling and screwing around with the settings to see if it changes anything?
20:36:49pamaurywow, that's not even expensive !!
20:37:25Mihailfurrywolf: your player do hundreds INT_AUDIO calls on USB insert: how it related to i2c? All other players do one call on insert and two on disconnect.
20:37:43lebelliumpamaury: yeah, next one is A30 :D
20:37:57furrywolfMihail: because you do an i2c transfer _in the isr_, which is a very slow thing, and should not be done in an isr. :)
20:38:38Mihailwhy all other work?
20:39:05furrywolfdunno. why does the stock firmware work? why does the async version work?
20:39:25Mihailold code was even slower
20:39:54Mihailare you sure it charge right?
20:40:11furrywolfare you sure? the old code looks like it handles an interrupt many, many times faster, since all the isr does is stick it in a queue, not actually go and do the transfer then.
20:40:28furrywolfthe bars in the battery graphic increase, until it's full...
20:40:53Bilgus_the old code might be slower but it has a way to switch tasks
20:40:56jhMikeSthe new driver won't work in an isr without always disabling interrupts for the entire transfer
20:41:11Bilgus_and that sucks
20:41:20Mihailit stop INT_AUDIO until all read will be done
20:41:40furrywolfno, it stops all interrupts, and blocks the entire system while it waits for an i2c transfer to complete.
20:41:51pamaurylebellium: that one will kill your wallet ;)
20:41:59furrywolfboth be initiated and completed
20:42:13jhMikeSMihail: no, because if a thread is using it, INT_AUDIO still happens and can reenter the functions
20:42:14lebelliumpamaury: I decided that in 2017 I'll spend 2/3 of my wage into Sony players. Sounds reasonable
20:42:16lebellium\o/
20:42:22pamauryhaha
20:42:37pamauryor buy it 80€, put rockbox, change destination, sell it 200€ ;)
20:42:44 Quit pamaury (Remote host closed the connection)
20:43:13CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
20:43:13*furrywolf really doesn't think going and fetching something off an i2c bus is appropriate behavior for an isr.
20:43:48jhMikeSfurrywolf: you mean in bulk?
20:44:38MihailjhMikeS: old code: void INT_AUDIO(void) { VIC_INT_EN_CLEAR = INTERRUPT_AUDIO;
20:44:57***Saving seen data "./dancer.seen"
20:45:07jhMikeSMihail: that's within the ISR to ack the interrupt
20:45:26furrywolfjhMikeS: the register that contains the charger, usb, etc flags isn't memory-mapped... it's off in an accessory hanging off the i2c bus. so to determine what triggered the interrupt, it has to initiate and complete a transfer over the i2c bus.
20:45:44jhMikeSnvm
20:45:56jhMikeSfurrywolf: indeed
20:45:56furrywolfthe old code queued a "we should go read that register at some point", the new code blocks the system and goes and reads it in the isr.
20:46:36jhMikeSMihail: was thinking
20:47:21jhMikeSMihail: no, that's right, it masks INT_AUDIO until it's serviced. threads using the driver do not, so INT_AUDIO can disrupt them
20:47:40furrywolfMihail: how about making the isr just set a flag and return, and have a thread that looks for the flag being set, and then fetches the register, performs appropriate actions, and clears the flag?
20:48:09jhMikeSfurrywolf: i don't understand the argument against it. it is NOT complicated code it's simple and I understand the reason for it. go look at linux or something if you want complicated.
20:48:09 Join TheLemonMan [0] (~root@irssi/staff/TheLemonMan)
20:48:26furrywolfjhMikeS: it's not complicated. but it's ungodly awfully slow.
20:48:28jhMikeSfurrywolf: it's sort of how I wrote every single driver for imx31
20:48:51 Join naleo_ [0] (~naleo@unaffiliated/naleo)
20:48:52jhMikeSfurrywolf: there will be lags because of other threads running in between, sure
20:48:53furrywolfISRs should be _fast_. they should do as little as possible. slow things, like i2c transfers, shouldn't happen in them.
20:49:46Mihailin any case I want move i2c reading from INT_AUDIO
20:49:54jhMikeSfurrywolf: I think I misunderstood. yes, doing the whole thing stuck in an ISR occupies everything for a long time
20:50:11funmanfurrywolf: how much time does the i2c transfer takes?
20:50:41Bilgus_up to 1/2 second i think
20:51:05jhMikeSMihail: everything was fine before. now you're going to introduce all sorts of gymnastics to avoid something that wasn't a problem.
20:51:10furrywolffunman: a lot. it runs at a 400khz clock, and needs to send a request to the accessory and wait for the data to come back from it. and with it in the isr, with interrupts disabled, it blocks the whole system to do it.
20:51:56Mihailfunman: 190 us for reading
20:52:03Bilgus_see I think we should go back to interrupt based code and fix the issue there, it is the proper way to do it
20:53:10jhMikeSBilgus_: I was running my v2 with interrupt-based code once I fixed the issue with the frequency switching code being called from a tick task. I just needed a less absurd minimum voltage.
20:53:24furrywolfthe interrupt based code has the advantage of actually working on my player, so I can't argue with that. :)
20:53:26 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
20:53:43*jhMikeS isn't sure where these voltage values are taken from and why anyone presumes they're okay. they seem really low.
20:54:07 Quit dfkt (Read error: Connection reset by peer)
20:54:53funmanmaybe they are optimized for several-years-old-players :)
20:55:06jhMikeSfurrywolf: right now it's all dead because I put asserts in to nuke things if blocking functions are called from ISR
20:55:35jhMikeSfunman: chips require less voltage as they age?
20:55:47jhMikeSTIL
20:56:07furrywolflol
20:57:03MihailjhMikeS: https://www.rockbox.org/irc/log-20170121#16:14:34
20:57:57jhMikeSfurrywolf: I'm tweaking the ISR one so it only inits and posts to semaphores if the transfer actually demands it. it takes some overhead out.
21:00
21:00:59jhMikeSMihail: yeah, Fuze v2 needs increase I'd say. Mine does to have everything work properly and i2c interrupts seem a bit more demanding.
21:01:08Bilgus_saratoga said each player was tested for minimum working voltage and then was set at a thresh above that
21:01:45Bilgus_^ yeah that
21:02:13jhMikeSstill guesswork with nothing but a bunch of anecdotes.
21:02:21jhMikeShow much does this voltage drop actually help?
21:02:33Bilgus_oh on runtime? no clue
21:03:09jhMikeSso.....if it's worth doing it better help runtime imo
21:03:33Bilgus_only a whole lot of testing is going to tell you that
21:03:37jhMikeSI don't mean 5 or 10 minutes or something
21:04:23MihailCVDD1 scaling add 50-70%
21:04:58jhMikeSMihail: I guess that's worthwhile
21:05:27Bilgus_on what players
21:05:35MihailAMSv2
21:05:40Bilgus_I get 12h on OF something like 16 on RB
21:06:02Mihail23-26 hours on clip zip
21:06:46MihailI also work on CVDD2 scaling - should by > 30 hours
21:07:57Bilgus_Any way in Bootloader to do some level testing and set it dynamically?
21:09:52MihailBilgus_: what you mean?
21:09:56jhMikeSMihail: is the 50-70% just voltage scaling compared to scaling only frequency or is it both compared to always running maximum frequency?
21:10:07 Quit _mt_ (Ping timeout: 255 seconds)
21:10:33Bilgus_like lower it till it fails to make a benchmark and set it back one level
21:10:34 Quit naleo_ (Ping timeout: 276 seconds)
21:12:27MihailjhMikeS: we need frequency scaling to set low voltage. Frequency scaling don't and much runtime by it self.
21:12:44Mihail*add much
21:13:10lebelliumpamaury: oh and actually, this A25 is my 100th player! Finally!
21:13:18pamaurywow, a hundred
21:13:26pamauryyou definitely don't use most of them ;)
21:15:53__builtindaaang, lebellium
21:16:13*__builtin is riding at 2 functioning players
21:18:12__builtinand around ~5 dead ones
21:18:21lebelliumpamaury: Indeed. I can't use all of them even if I wanted. The ones I use the most are still the Clip+ and Clip Zip. But those days I use my E585 and A15 quite much!
21:19:20pamaurythat means we need rockbox on them ;)
21:21:28lebelliumsure
21:21:46lebelliumthere is much expectation as you already saw with the dest_tool
21:25:36 Quit Strife89 (Ping timeout: 256 seconds)
21:26:30lebellium__builtin: that's not a good ratio! I only have 7 dead ones among the 100
21:28:32 Join parchd [0] (~parchd@unaffiliated/parchd)
21:28:41furrywolfhow does battery charge current and usb negotiated current interact?
21:30:01furrywolfthat is, is it trying to pull 200ma without first negotiating for it?
21:31:19Bilgus_I imagine it just doesn't
21:32:34furrywolfhrmm...
21:34:22funmanjhMikeS: didn't know what we were talking about.. i am a bit out of the loop these days :)
21:35:29Bilgus_and in that case maybe it would be a good idea to set 100ma on data connection on the players that allow it
21:35:44furrywolfMihail: on your next version, add debug output for VBUS and CHG_IN off the adc, please. :)
21:36:18Bilgus_though it might negotiate I'm not familiar enough with the usb part to tell you
21:36:39Mihailfurrywolf: you can check it in debug menu
21:36:52furrywolfnot while the device is frozen I can't. lol
21:37:28furrywolfalthough I could flash your irq-based one and try then
21:38:28Bilgus_ah do you think the device is kicking your USB port?
21:38:46Mihailfurrywolf: or v3.13
21:39:02furrywolfwhat debug option shows VBUS and CHG_IN?
21:39:20Bilgus_View IO ports
21:39:49furrywolfthat only shows gpio, nothing from the adc
21:40:03Bilgus_hit the down button
21:40:08pamauryfurrywolf: Bilgus_on players that support it, it only draws what the host allows
21:40:18pamauryat least it is supposed to
21:40:26__builtinhmm... this is not goot
21:40:28__builtin*good
21:40:40pamaurybut not that many port implement it I think
21:40:48__builtinsomething's seriously wrong with the filesystem on my build at least
21:40:48 Join _mt_ [0] (~MT@2601:482:4402:7b60:5c71:55a1:a0d1:1f60)
21:41:20furrywolfhah! vbus is dipping down to 4.3, chg_in down to 4.1.
21:42:02furrywolf4.1 into a charger trying to charge a battery to 4.2v might cause spurious endofch interrupts.
21:42:08 Quit parchd (Ping timeout: 240 seconds)
21:44:12furrywolfI should try to find something with a 500ma port and see if it behaves differently.
21:46:36furrywolfalthough it's probably not the problem, since when the battery is actually mostly charged, it wouldn't pull enough current to cause an issue.
21:47:57furrywolfoh well, it was worth thinking about.
21:51:03 Join saratoga [0] (c036de1c@gateway/web/freenode/ip.192.54.222.28)
21:51:09furrywolflatest devel insta-panics on usb connect
21:51:16saratogavoltage scaling seems to make a large difference on AMSv2, I suspect its a combination of the cpu using less power and the SOC peripherals draining less
21:51:22jhMikeSfurrywolf: :)
21:51:40furrywolf"Incorrect CPU mode in mutex_lock (0x13 != 01F)
21:51:59jhMikeScrap, I forgot an 'x'?
21:52:09furrywolfmore likely I typoed it. :P
21:52:14saratogafrequency scaling alone make relatively little difference on all AMS (v1 and v2), i think probably the rest of the SOC consumes a sizable fraction of total power
21:52:18pamauryisn't it normal since the code still mutex_lock in ISR ?
21:52:27saratogai think on v1 adding it in gave maybe an extra 2 hours
21:52:29furrywolfyes, the x is there. my typo.
21:52:37jhMikeSwhew
21:52:51furrywolfnow go fix your bug instead of complaining about my typing. :P
21:53:01jhMikeSit's not my bug
21:53:33pamauryit's Mihail's bug
21:53:35*jhMikeS just makes the kernel code do stuff
21:54:18furrywolf2 hours on a player with 12-hour life is not a trivial improvement.
21:54:39saratogai think the problem with voltage scaling on the fuzev2 is that 90% of people have clip players, so we estimated the fuzev2 voltages from just a couple players and a safety margin
21:55:13saratogahard to estimate a distribution from a few samples
21:55:25*jhMikeS has the third Fuze v2 in existence
21:55:28furrywolfwhere does the automatic installer save its cached downloads?
21:55:34saratogai tried for a long time to find one
21:55:45saratogai think i bought 3 or 4 fuzes over the years and they're all v1s
21:55:55saratogai don't think sandisk made very many of them
21:56:18saratogamaybe why no one has noticed that the voltage is too low
21:56:40Bilgus_johnb2 has a fuzev2 as well
21:56:52furrywolfminimum voltage could also be random - some individual chips will work better at lower voltage than others.
21:57:05Bilgus_^true
21:57:07saratogaby the way, amsv1 frequency scaling is still disabled due to random crashes which are probably fixed by now by subsequent improvements :)
21:57:15saratogayes there will be a distribution
21:57:19furrywolfand I doubt sandisk is going to disclose their minimum voltage binning practices. :)
21:57:27*jhMikeS did some revisions to the interrupt I2C driver and now the fuzev2 seems to be running at the low voltage with no immediate issue
21:57:27saratogai doubt they know
21:57:33Bilgus_not at this point at least
21:57:56furrywolfjhMikeS: now convince Mihail to switch back to the interrupt i2c driver. :)
21:58:02saratogathey made these chips over a long period of time as well, who knows if theyy're even all identical
21:58:14jhMikeSfurrywolf: why convince him when I can pull rank?
21:58:28furrywolfthat's a form of convincing. blackmail would work too.
21:59:07*jhMikeS guesses he's RB old fogey since he came here in 2006 (iirc)
21:59:17saratogathen again i guess its on a 130 nm process (?) so probably the equipment was 10 years old at the time and not changing too much
21:59:20furrywolfthe non-interrupt one is a lot cleaner... but doing i2c bus transfers inside an isr just rubs me the wrong way.
21:59:43pamaurymy guess is that the OF uses a very safe and possibly very suboptimal voltage
22:00
22:00:03 Join Senji [0] (~Senji@85.187.103.250)
22:00:08jhMikeSfurrywolf: me too, besides not letting other threads run concurrently on a pretty quick CPU
22:00:16furrywolfdunno, battery life is a big advertising point. they might push it more than usual.
22:00:41 Quit Senji_ (Read error: Connection reset by peer)
22:01:26furrywolfI haven't looked at the simulator... does it actually emulate the hardware, or is it just a replacement api that sticks everything in a window and ignores actual hardware calls?
22:01:45pamauryfurrywolf: it replaces api
22:01:45saratogadoesn't emulate anything
22:02:05pamaurybasically it replaces the "os part" of rockbox
22:02:17saratogasomeone tried writing an emulator for PP, it could half boot actually
22:02:32saratogabut its a lot of work to get everything exactly right
22:02:40pamaurythe problem with emulators is that they are a lot of work, and that doesn't even help figuring out when you don't know
22:02:41jhMikeSfurrywolf: I made the driver a bit more "regular" and killed some overhead. also, fixed a neglected sem wait
22:02:55furrywolftoo bad we don't have a hardware emulator for reverse-engineering the stock firmware.
22:03:06jhMikeSpamaury: that e200 emulator actually helped uncover some good stuff though
22:03:15saratogai am sure the manufacturer specs extremely conservative voltages so that they have the option of shipping lower binnned parts in the future
22:03:21Bilgus_UGH I hate the time settings that use string choice talk about a convoluted way to set times
22:03:23saratogaremember the write the firmware before the parts actually ship
22:03:34furrywolfwatch in real time as the stock firmware twiddles the speed and voltage bits
22:03:40saratogawe have the advantage of developing for a platform that shipped long ago, so we know what the hardware actually needs
22:04:05jhMikeSfurrywolf: yep, fuzev2 seems happy with voltages in HEAD running interrupt driver
22:04:38furrywolfcompletely off-topic question: the time I've spent this morning not playing with my mp3 player has been spent trying to organize my living room... how the heck is it now WORSE? lol
22:04:43*furrywolf should work harder and irc less
22:05:03jhMikeSfurrywolf: I think it was just a couple driver bugs + the action.c bug causing the issue.
22:05:14saratogai should get back to work as well
22:06:33Bilgus_anyone know if we have a ms_to_ticks and ticks_to_ms function somewhere else? , I've placed min in misc.c
22:06:44Bilgus_mine*
22:07:41furrywolfdoes view i/o ports intentionally or accidentally disable usb?
22:08:20jhMikeSfurrywolf: the debug screens tend not to ack USB
22:10:25 Quit _mt_ (Ping timeout: 255 seconds)
22:11:19furrywolfusb seems to behave less poorly when connected to a device with 500ma ports. I think part of the issue may be spurious endofch interrupts due to usb voltage sag, due to turning the charger on full power prior to negotiation.
22:11:46furrywolfand the spurious interrupts cause the slow-as-fuck interrupt handler to consume too much cpu.
22:14:38furrywolfalso, when you hold down the power switch while on the i/o ports screen, it seems to update about ten times too fast. lol
22:15:32furrywolfspecifically, it does not freeze when connected to either a wall usb charger, or to my snap-on modis. I didn't check to see if it enumerated on the modis, since wince is annoying to do anything with, and it's pretty locked down.
22:16:30pamauryfurrywolf: that's probably the effect of gui boost
22:18:54Mihailpamaury: debug menu update immediately on any button press, or 10 times per second if no press
22:19:14pamauryah
22:19:43 Join _mt_ [0] (~MT@2601:482:4402:7b60:5c71:55a1:a0d1:1f60)
22:19:47 Quit StaticAmbience (Ping timeout: 260 seconds)
22:20:18furrywolfit looks like the charger is smart enough to deal with low usb power, but it generates extra interrupts while doing so.
22:22:34 Quit robertd1 (Ping timeout: 240 seconds)
22:24:31 Join robertd1 [0] (~root@186-90-12-124.genericrev.cantv.net)
22:27:21girafehello, is there some little player that is already in sell (at little cost, like sansa clip +) that support rockbox nowadays ?
22:30:11lebelliumno
22:30:32lebelliumyour best option is still finding a second-handed Clip+ or Clip Zip
22:33:05 Join fishbulb [0] (~fishbulb@unaffiliated/fishbulb)
22:34:36__builtinor hope that the agptek "ROCKER" is real
22:34:43__builtinand actually runs rockbox
22:35:18lebelliumreal for sure
22:35:22lebelliumthe latter, no :P
22:35:37 Quit _mt_ (Ping timeout: 255 seconds)
22:44:59***Saving seen data "./dancer.seen"
22:46:17TorCGuest44936: Not surprised. Remember, I'd edited it to take (some) advantage of the zip's taller screen and added a couple lines.
22:50:59 Join naleo_ [0] (~naleo@unaffiliated/naleo)
22:51:35 Join MrZeus1 [0] (~MrZeus@2a02:c7f:7025:ed00:d10f:861d:c41:1227)
22:53:56Mihailpamaury: I want move ascodec_read() from INT_AUDIO to power thread as you say, but we have one target (m200v4) without charging. We don't have dev builds for it and wiki say "port has received little attention and is extremely buggy" so I not sure is it work at all. What should I do?
22:54:54saratogadon't worry about the m200v4
22:55:08saratogathe player reboots if you try to change the volume, it isn't really usable
22:55:25saratogahardly any were ever made so it is unlikely anyone will want to work on it again
22:56:14Mihailok, thanks!
22:57:48lebelliumthat sounds like a challenge. I must find this rare m200v4 :P
22:58:05saratogamost of the m200s out there are v3s IIRC
22:59:22 Join Strife89 [0] (~quassel@adsl-98-80-185-9.mcn.bellsouth.net)
23:00
23:00:03 Join semitones [0] (~znc@unaffiliated/semitones)
23:00:16pamauryMihail: why not still go for the async interrupt version ? jhMikeS says he as a cleaned ip version that works
23:01:24jhMikeSg1522
23:01:25fs-bluebot_3Gerrit review #1522 at http://gerrit.rockbox.org/r/1522 : 3AMS: Gloriously return ascodec to interrupt-based I2C2 driver by Michael Sevakis
23:09:44Mihailwe should carefully test it - if no problems - I agree
23:10:54jhMikeShave at it. so far it's played through one album for about an hour and it's on the second
23:11:27girafelebellium : the problem is second-handed clips will probably start to have problem with these old batteries
23:11:44jhMikeSvolates are 1.1875/0.8750
23:12:27MihailjhMikeS: you should try connect/disconnect USB 20-40 times
23:13:49jhMikeSI forgot to plug USB. when I do there's something still trying to set CPU frequency from in ISR somewhere.
23:14:04__builtinthere we go, input works now :D
23:14:32jhMikeSwhere the hell is that coming from?
23:15:01 Quit alexweissman (Remote host closed the connection)
23:15:34Mihailfirmware/target/arm/as3525/usb-as3525.c ?
23:15:58Mihailusb_enable()
23:16:28Bilgus_girafe, the batteries are very easy to change
23:16:46Bilgus_hardest part was opening the player
23:17:06jhMikeSMihail: why is that being called from and ISR?
23:17:26saratogais there a good source of clip batteries?
23:17:40MihailI don't know
23:17:47jhMikeSit shouldn't be
23:18:32Bilgus_china?
23:19:16Bilgus_I found one on ebay US but it was a bit expensive imo $10
23:19:52Bilgus_If i needed like say 10 I'd probably look for the source and wait a few months (what I assume this guy did)
23:22:44lebelliumClip have soldered batteries, no?
23:23:12lebelliumReplacing anything soldered is not what I would call "easy" for a normal user
23:23:40Bilgus_yes soldered
23:24:08lebelliuma normal user doesn't have a soldering iron at home
23:24:12Bilgus_you can get low temp solder and do it with a bic lighter and a nail if need be
23:24:14pamauryjhMikeS: usb_enable is called from usb thread usually no ?
23:25:05Bilgus_if you can't do that then get conductive epoxy and glue em but that stuff is $$
23:25:14lebelliumlast time I used a soldering iron was at secondary school, in "IT/technologies" class. That's probably the same for girafe who's french too
23:25:49Bilgus_biggest part to soldering is Flux and not burning yourself/item
23:26:19Bilgus_but I get where the normal user is afraid of it
23:26:19saratogayou can get a soldering iron for less than 10 on Amazon, making the battery more expensive
23:26:30saratogabut those little battery powered microsoldering irons might be better for something like this
23:26:50Bilgus_btw the place I bought the battery from now wants ~17 for it
23:27:03jhMikeSMihail: Looks like that might have been a stack issue
23:27:11Bilgus_@saratoga, those battery powered ones are awesome
23:27:22jhMikeSMihail: i just plugged and unplugged a bunch of times with no problem
23:29:31Bilgus_323036P is the part number, and the guy puts like 10 each time to keep the supply low
23:29:57Bilgus_(on the ebay listing)
23:30:56lebelliumwhat does "genuine" mean in the listing?
23:31:05lebelliumdoes it come from Sandisk or the same supplier?
23:31:08 Quit saratoga (Ping timeout: 260 seconds)
23:31:13Bilgus_Doubt it
23:31:17MihailjhMikeS: ok, I'll try check on my players. Before I have "Unhandled IRQ" panic time to time.
23:31:26Bilgus_It does look the part though
23:31:42Bilgus_BAk is the manufacturer
23:33:15Bilgus_I saved my original Clip+ Battery BAK 323036P 3F05 1.07Wh +3.7V 290mAh
23:33:48lebelliumso it's the same manufacturer?
23:34:02semitonesHello all −− I have a Clip+ that it seems the NAND may be dying −− it hung and I was waiting for the battery to run out to try reinstalling the bootloader. TorC was helping me
23:34:04Bilgus_Well who knows
23:34:10jhMikeSMihail: btw, there's a bug by boosting in usb_enable: it stays at highest even if only using it to charge
23:34:19semitonesI may need help scouring the forums for the version of the bootloader that will work.
23:34:27semitonesWHich I don't understand too well
23:35:05girafeyou're right lebellium :)
23:35:46girafei'm just sad to see there's no mean in 2017 to buy the same good hardware that few years ago
23:35:52 Join alexweissman [0] (~alexweiss@149-160-186-148.dhcp-bl.indiana.edu)
23:36:56semitonesI'm seeming to have trouble connecting to the forums −− does anybody know of a bootloader version that would allow me to recover the Clip+ from this state: where holding the power button down 30s doesn't work anymore
23:37:09lebelliumgirafe: SanDisk released the Clip Sport and Clip Jam which are not discontinued yet but the hardware is worse than the Clip+/Zip and not enough for Rockbox. SanDisk ruined it all!
23:37:12MihailjhMikeS: it not problem as we on charge :) We should do voltage rising before we try use USB
23:39:11Bilgus_semitones, the forum is down atm, but if the nand is already dead it is too late anyways
23:39:14jhMikeSmaybe. it's annoying though since I have to turn off that boost to have it scale when plugged
23:40:02 Quit alexweissman (Remote host closed the connection)
23:40:05furrywolfheh, kinda like linksys, where every new version had less flash and less ram than the previous version?
23:42:11 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
23:43:36Bilgus_dongs have you gotten your new NAND to test with yet?
23:43:58semitonesBilgus_, i don't know if the nand is dead - I'm not even sure what that is
23:44:43Bilgus_when you plug it to the pc does it show 32mb RAW partition?
23:44:48semitonesAll I could tell you is that just earlier this week, it was working, then it started having issues hanging, where the power off and on would fix it, and 2 days ago, that trick stopped working. So I've waited for the battery to run out completely to try the next troubleshooting step
23:45:11semitonesBilgus_, plugging it in before showed nothing. I'm about to plug it in now, except that TorC said to have the bootloader ready to install when I did
23:46:18Bilgus_then wait till the forums come back up
23:47:03girafelebellium : hard to understand why, would it not be costless for sandisk to sell again the same hardware rather than using a new one, even less powerful
23:47:55 Join alexweissman [0] (~alexweiss@149-160-186-148.dhcp-bl.indiana.edu)
23:53:34TorCsemitones: johnb2 has a ready-made one.
23:54:19semitonesperfect! I am having trouble finding it
23:54:39 Quit paulk-collins (Ping timeout: 240 seconds)
23:54:51TorCHe's not on IRC right now, but I think he reads the logs.
23:56:07TorCActually, check the logs. Try Dec 30th or so. I think he posted a link to download it separately from the forums around that time.
23:58:08 Join paulk-collins [0] (~paulk@gagarine.paulk.fr)
23:58:40Bilgus_8johnb3Regarding the modified boot loader: it was Mihail on the forums.

Previous day | Next day