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 2017-10-11

00:05:03 Quit petur (Remote host closed the connection)
00:05:34 Quit mikroflops (Ping timeout: 240 seconds)
00:06:06Bilgusi think it still showed the same
00:10:32 Join mikroflops [0] (~yogurt@
00:16:16man_in_shacklsusb says "Sandisk Corp." and block device reports as " UNDEF "
00:16:28man_in_shackthe UNDEF is proably an indicator of death
00:31:12Bilguswhat does sudo fdisk -l && lsblk report?
00:34:42 Quit ender` (Quit: In the beginning, there was nothing. And God said, 'Let there be Light.' And there was still nothing, but you could see a bit better.)
00:41:27Bilgusoh sorry just saw you already did
00:43:53 Join jhMikeS [0] (
00:51:12 Quit man_in_shack (Ping timeout: 248 seconds)
01:04:32 Quit pamaury (Ping timeout: 248 seconds)
01:04:53 Quit Tony_ (Ping timeout: 260 seconds)
01:09:58 Join man_in_shack [0] (~chat@unaffiliated/man-in-shack/x-4279753)
01:17:33 Join Tony_ [0] (ae3638f0@gateway/web/freenode/ip.
01:28:45 Quit krabador (Quit: Leaving)
01:47:29***Saving seen data "./dancer.seen"
01:59:13 Quit Ruhan (Quit: Connection closed for inactivity)
02:33:04 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
02:36:14 Quit JdGordon_ (Ping timeout: 248 seconds)
02:48:02 Join The_Prospector [0] (~The_Prosp@unaffiliated/cornman)
03:03:56 Quit dys (Ping timeout: 264 seconds)
03:13:14 Quit Aldem (Read error: Connection reset by peer)
03:18:06 Quit prof_wolfff (Ping timeout: 260 seconds)
03:19:15 Join prof_wolfff [0] (
03:19:23 Join Aldem [0] (~Aldem@unaffiliated/aldem)
03:30:26__builtinpamaury (logs): yes
03:30:40__builtinI see, so ARM926 wouldn't support it then
03:30:46__builtintoo bad :(
03:31:02__builtinat least I can set the A bit and root out the crashes
03:38:40 Quit _meg (Ping timeout: 255 seconds)
03:39:22 Join _meg [0] (~notsure@
03:47:14 Join saratoga [0] (123e11e0@gateway/web/freenode/ip.
03:47:33***Saving seen data "./dancer.seen"
03:47:35saratogaBilgus: I get 250 MHz needed for real time decode of my AAC-HE test file
03:47:57saratogai tried unboosted, its still running :)
03:49:19Bilguson the fuze+?
03:50:17saratogadamn it, plugin has this bug where it doesn't show the speed if the plugin finishes with the screen off
03:50:24saratogaso i lost the results for the unboosted test
03:52:01saratogaaac in general is really slow on the fuze
03:52:12saratogamaybe slow RAM?
03:53:50Bilgusyeah its weird that locking the frequency even at 64 mhz and it works properly so IDK whats up but i'll look into it more
03:55:41saratogaif you have the test plugins configured, double check that the speed is correct for unboosted
03:56:01saratogalong select the file and run it with boost enabled, and with the boost disabled (test plugin can do both)
03:56:11saratogashould make it obvious if something is wrong
03:59:02 Quit man_in_shack (Read error: Connection reset by peer)
04:00:09 Join man_in_shack [0] (~chat@unaffiliated/man-in-shack/x-4279753)
04:01:51 Quit CrashBash-Kun (Read error: Connection reset by peer)
04:03:26saratogastrange that it is so slow given the 16KB cache and the ARM9e core
04:19:53 Join advcomp2019__ [0] (~advcomp20@unaffiliated/advcomp2019)
04:22:34 Quit advcomp2019_ (Ping timeout: 240 seconds)
04:43:51 Quit scorche|sh (Ping timeout: 260 seconds)
04:45:44 Join scorche|sh [0] (
04:45:45 Quit scorche|sh (Changing host)
04:45:45 Join scorche|sh [0] (~scorche@rockbox/administrator/scorche)
05:05:37 Quit Strife89 (Ping timeout: 255 seconds)
05:06:06 Join Strife89 [0] (
05:47:34***Saving seen data "./dancer.seen"
06:06:07 Quit TheSeven (Ping timeout: 258 seconds)
06:06:31 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:27:31 Quit TheSeven (Ping timeout: 246 seconds)
06:27:43 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
07:47:38***Saving seen data "./dancer.seen"
08:28:03 Join ender` [0] (
08:44:57man_in_shackrockbox to the rescue!
08:46:27man_in_shackmy dad has given me an old android phone and rockbox works nicely on it :D
08:46:34man_in_shack << means i don't need to buy these now :P
08:49:48alexbobpsansa fuzes are awesome though :P rockbox for android is a different experience
08:49:51alexbobpbut long as you're happy!
08:51:21 Join wodz [0] (
08:51:36man_in_shackyah they awesome but also cost more than $0
08:51:38 Join petur [0] (~petur@
08:51:38 Quit petur (Changing host)
08:51:38 Join petur [0] (~petur@rockbox/developer/petur)
08:51:54man_in_shackalso bluetooth audio is nice
08:52:03man_in_shackone fewer cable to run
09:02:48 Join JanC_ [0] (~janc@lugwv/member/JanC)
09:03:43wodzone more battery to charge :P
09:04:05 Nick JanC is now known as Guest86327 (~janc@lugwv/member/JanC)
09:04:05 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
09:05:11 Quit Guest86327 (Ping timeout: 260 seconds)
09:05:33man_in_shackwodz: not really. running a dedicated music player in my car one way or another
09:05:48man_in_shackhas been sansa clip+ but the flash in it is dead
09:46:13 Join p3tur [0] (~petur@rockbox/developer/petur)
09:46:13 Quit petur (Read error: Connection reset by peer)
09:47:40***Saving seen data "./dancer.seen"
09:54:53 Nick p3tur is now known as petur (~petur@rockbox/developer/petur)
10:03:09 Quit tchan (Quit: WeeChat 1.6)
10:23:05 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
10:29:52 Quit mc2739 (Ping timeout: 248 seconds)
10:29:54 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:35:43 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
10:40:51 Quit Moarc (Ping timeout: 260 seconds)
10:42:12 Join Moarc [0] (
10:49:33 Quit pamaury (Ping timeout: 264 seconds)
10:51:59 Join p3tur [0] (~petur@rockbox/developer/petur)
10:54:43 Quit petur (Ping timeout: 255 seconds)
11:19:41 Nick p3tur is now known as petur (~petur@rockbox/developer/petur)
11:23:29 Join PimpiN8 [0] (~textual@2a02:a454:38ea:1:ed48:8e3f:8a7d:94dc)
11:35:41 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:47:41***Saving seen data "./dancer.seen"
12:20:12 Join CH23 [0] (4dfa0218@gateway/web/freenode/ip.
12:20:46CH23is the clip+ recovery mode run from the internal flash, or from the SOC?
12:26:19CH23man_in_shack: are you willing to sell me your broken clip+ ? i have some things i want to test
12:28:28wodzCH23: from SoC rom AFAIK
12:29:08CH23wodz: just spoke with dongs, yeah it is.
12:29:41CH23dongs, the person. not the body part.
12:29:46 Quit yosafbridge (Quit: Leaving)
12:30:36wodzCH23: consult pamaury, he tried disassembling recovery mode on amsv2
12:31:32CH23i have most of the info i need; now i just need a matching NAND chip, and broken device.
12:31:45 Join robertd1 [0] (~root@
12:33:48wodzAnyway I am pretty sure recover mode runs from rom and not from nand (there might be 2 recovery modes actually). You can force amsv2 device into recovery mode by shorting nand address lines and creating flash fault. How would it load recovery image then?
12:33:58 Join yosafbridge [0] (
12:34:48CH23as i said, it runs from SOC
12:54:20 Quit PimpiN8 (Quit: My MacBook has gone to sleep. ZZZzzz…)
12:54:52 Join PimpiN8 [0] (~textual@2a02:a454:38ea:1:ed48:8e3f:8a7d:94dc)
12:54:57 Quit PimpiN8 (Client Quit)
13:01:04 Quit robertd1 (Remote host closed the connection)
13:11:11Bilgussaratoga pamaury I did the test codec for boost/unboosted at normal, locked 64MHz and Locked 454Mhz the results jive with the observations still not sure what it means though,52018.msg240757.html#msg240757
13:17:44man_in_shack<CH23> man_in_shack: are you willing to sell me your broken clip+ ? i have some things i want to test << sure
13:18:02man_in_shackwhat country you in?
13:18:06 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
13:18:48 Quit TheLemonMan (Client Quit)
13:20:31CH23man_in_shack: netherlands
13:20:51man_in_shackaustralia here
13:20:55CH23oh boy
13:21:08CH23that's gonna be heaps in shipping cost
13:21:14man_in_shackcould be
13:24:41 Join robertd1 [0] (~root@
13:26:35man_in_shackau$21.50 shipping
13:28:59CH23and how much do you want for the device itself?
13:36:34pamauryCH23: wodz: I once tried to disassemble the ROM from amsv2 but it's very complicated, I never finished. My impression though was that the ROM supports two modes, one mode where the flash is accessible but does not contain valid code (I guess that's the 900MB partition thing) and one mode where the NAND is not accessible (the 30MB probably-ram-disk)
13:37:45pamaurybut the thing that is not clear to me, and the code is too complicated to follow, is what would happen if you had a valid flash but not formatted. On a related question, we still don't know what exactly is responsible for the flash FTL and if it's possible to recreate/format it.
13:39:06CH23previously i pulled a full disk image from my 4gb clip+, when in recovery though?
13:39:23Bilgusbut that doesn't have any of the nand stuff
13:40:14BilgusI have those images btw up on the forum and a build with them that allows you to access the 'hidden' part of the drive without recovery mode
13:41:04Bilgussee basically the sansa OF and rockbox block the first 0xF000 sectors from you to protect the firmware
13:41:28Bilgusits not special in any way except not being a valid partition
13:41:50CH23and that's not writable in recovery
13:41:58Bilgussure it is
13:42:22CH23isn't that all you need to make it work then?
13:42:23Bilgusits writable in those special builds that I unblocked it as well
13:42:44BilgusIt doesn't have anything needed for the nand init
13:43:35CH23i'm a bit confused. the recovery is 'directed' from the SOC rom, then gives access to the NAND, right?
13:44:12pamauryit gives you access to the NAND only if the NAND is accessible
13:44:31pamaurywhen the NAND is dead, it seems to switch to this 30MB ram disk
13:44:47pamaurywhich we don't know how to use afaik?
13:44:53CH23so if you were to completely blank out the NAND, would it still give access?
13:45:06Bilguslike basically what recovery is for is if you screw the image up and the device no longer boots because bad code but hardware is still working fine
13:45:45Bilgusit seems the 30MB soesn't persist writes so not sure if its even a ram disk
13:46:09CH23i want man_in_shack's device to see if i can transplant an equal NAND chip onto it, then via recovery DD an image to it
13:46:54Bilgusbut I could take my device that works now erase the whole drive and it would be a brick.. put the device in 'recovery' write back the image reboot and all would work
13:47:16Bilgusthe equal nand chip is the issue
13:47:23CH23maybe not
13:47:43***Saving seen data "./dancer.seen"
13:47:44CH23the same flash chip was used in the sandisk cruzer 4gb and 8gb drives
13:47:44Bilgusmaybe pigs can fly with enough thrust
13:47:54CH23and some others
13:48:33BilgusI for one hope thats the case but I'm not sure how likely it is but in theory yes that should work
13:50:27Bilgusspeaking of johnb is going to try brute forcing an image to his readonly clip+ this weekend
13:50:28pamauryBilgus: we don't really know what this "ram disk" is for, that's the main problem, and I never was able to figure it out
13:50:32pamaurybut I could have another try
13:50:36CH23from what dongs told me, it seems that these sandisk NAND chips aren't writable with those normal NAND flashing modules
13:50:49Bilgusit doesn't accept writes either so maybe some init that missing
13:50:50CH23that's why i had the idea of doing it via recovery
13:51:34BilgusCH23 dongs seems far more versed in NAND technology I can't say one way or another
13:51:44CH23my readonly clip+ decided that readonly sucks and it's writable again
13:51:50pamaurywell one thing we don't really know is whether those NAND are really NANDS
13:51:57pamauryor so SD with NAND "interface"
13:52:06CH23according to the datasheet they seem to be
13:52:29Bilgusbtw CH23 you said you found that datasheet could you upload it somewhere
13:52:58pamauryhow far dongs went in his exploration of the NAND? Can he read/send command to the NAND chip?
13:53:16Bilgusalso if its writable again why not put the multiboot bootloader on it and never touch the internal memory again?
13:53:24CH23 Bilgus we've been here before ;)
13:53:59CH23also i already put the multiboot on it, because within rockbox it was always writable, so i copied it from the µSD card
13:55:35pamaurywhat does lsusb shows on recovery? There is the UNDEF with 30 MB, and the other one?
13:55:44BilgusI'm saying the actual bootloader?
13:56:14Bilgusalso I can't download that data sheet I'm sure most of that has to do with my lack of chinese
13:56:27CH23oh hold on i have a mega link
13:57:37*pamaury starts looking at the ROM again
13:58:31CH23this is for the SDTNPNAHEM-008G
14:03:22 Quit jhMikeS (Ping timeout: 258 seconds)
14:05:09 Quit deevious (Ping timeout: 255 seconds)
14:05:21 Join deevious [0] (~Thunderbi@
14:13:03BilgusI always knew NAND was a scary thing to trust your data to but wow
14:13:11man_in_shackuery CH23
14:17:01CH23i had one SSD fail on me, flash memory is scary stuff
14:19:30wodzI was scarred when I discovered that perfectly working nand chip reported 4 bits of error in some cells (still corrected by BCH).
14:28:29man_in_shackso are there docs on writing themes?
14:30:11 Join jhMikeS [0] (
14:34:24CH23 man_in_shack there's this
14:34:46 Quit jhMikeS (Ping timeout: 255 seconds)
14:37:48Bilgusman_in_shack, do be aware while mostly correct there are some errors in there
14:38:37man_in_shackthanks :)
14:38:39Bilgusso if you just can't something to work it might not be you
14:39:28man_in_shackso next, is there some sort of emulator for testing themes? :D
14:40:51man_in_shackahha, right at the bottom of the CustomWPS page">
14:41:37BilgusNot sure how up to date those are but I've not heard any complaints
14:42:06BilgusIf you have linux we can give you more recent builds
14:45:35 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
14:47:18man_in_shackwell it's for a 320x480 android target ... :P
14:48:50pixelmahmm, reminds me of my phone, but that one has a too "new" Android version for Rockbox
14:49:39*man_in_shack lords his 2.3.7 cyanogenmod over pixelma
14:49:46 Join amayer [0] (
14:50:54man_in_shackit's so old it doesn't do google 2-phase auth
14:51:03man_in_shackso i can't get to play store :P
14:51:10Bilgusyou could use a sim for any target thats 320x240 with a touchscreen
14:51:40man_in_shackexecpt that's half the height :P
14:52:52Bilgussorry I ignored the 480 conveniently?
14:53:02man_in_shackyes, yes you did
14:53:16Bilgusfor that matter grab a bluetooth kb and dev on the device
14:53:29pixelmayou can build a sim for any size RaaA target, I might even have one of the exact resolution from 2 years ago or so (probably cross-compiled for Windows but can't check right now)
14:53:47man_in_shackew windows
14:54:13pixelmaback when my phone could still run Rockbox
14:54:20Bilgusoh yeah you do have linux duh hey you could just build your own
14:58:50CH23about that i never got the simulator to work for me
15:00:38man_in_shackso is there a way to just build the simulator?
15:01:43man_in_shackoh target=sdlapp type=sim
15:02:21man_in_shacki was trying target=android :)
15:02:58CH23i believe target should be a number
15:03:16man_in_shackit also allows the string name
15:03:27CH23oh didn't know that
15:03:28man_in_shackwheeeeee lots of warnings :D
15:03:38man_in_shackCH23: read the help more :P
15:03:45CH23i get 3 error messages
15:04:19CH23:/usr/bin/ld: cannot find -lXrandr
15:04:26CH23:/usr/bin/ld: cannot find -lXrendr
15:04:33CH23collect2: error: ld returned 1 exit status
15:04:34man_in_shackthat's cos you're missing libs :)
15:04:54CH23apparently but i have xrandr
15:06:09CH23so i'm not sure what libraries those would be
15:08:01 Join dys [0] (~dys@2a01:598:a084:182f:eab1:fcff:fe36:4b0b)
15:08:25 Quit sielicki (Quit: WeeChat 1.9.1)
15:08:51 Join sielicki [0] (~sielicki@unaffiliated/n1cky)
15:10:56man_in_shackyou need libxrandr-dev
15:11:01man_in_shackand libxrender-dev
15:14:31 Quit sielicki (Quit: WeeChat 1.9.1)
15:14:59 Join sielicki [0] (~sielicki@unaffiliated/n1cky)
15:18:25CH23thanks, now it works.
15:19:18 Quit sielicki (Client Quit)
15:20:05 Join sielicki [0] (~sielicki@unaffiliated/n1cky)
15:25:39 Quit sielicki (Quit: WeeChat 1.9.1)
15:27:36 Join sielicki [0] (~sielicki@unaffiliated/n1cky)
15:46:53pamaurylooking at the amsv2 bootloader code, I suspect there is a pin to select whether it boots from internal SD or microsd
15:47:01pamauryI think it's XPB[5]
15:47:18pamaurybut I'm not entirely sure
15:47:46***Saving seen data "./dancer.seen"
15:49:07pamaurymmh, or maybe not, I see that in rockbox code this is used to determine the "variant"
15:49:40man_in_shackgot simulator running ok now
15:52:00pamauryI need to disassemble more code
15:58:56 Quit wodz (Ping timeout: 248 seconds)
16:06:56man_in_shackexcept i built it wrong :D
16:07:09man_in_shackno disassemble number pamaury!
16:11:07*pamaury cannot parse this sentence
16:12:59CH23syntax error
16:15:30 Quit CH23 (Quit: Page closed)
16:23:01 Quit Aldem (Remote host closed the connection)
16:23:30 Join Aldem [0] (~Aldem@unaffiliated/aldem)
16:34:46 Join krabador [0] (~krabador@unaffiliated/krabador)
16:54:24 Join Ruhan [0] (uid76353@gateway/web/
17:02:55 Quit prof_wolfff (Ping timeout: 248 seconds)
17:16:02 Quit petur (Quit: Connection reset by beer)
17:24:52 Quit krabador (Remote host closed the connection)
17:31:34 Quit prg318 (Quit: ZNC 1.6.5 -
17:34:07 Join prg318 [0] (~prg@deadcodersociety/prg318)
17:37:34 Quit Huntereb (Ping timeout: 248 seconds)
17:41:38Bilgusok I'm pretty sure it isn't the fuze+ causing the problems with the AAC-HE file
17:42:42Bilgusalso pamaury for future reference you can't change the CPU speeds in system-target.h the tables in system-imx233.h need to be changed
17:43:40BilgusI'm pretty sure the AAC-HE bug is either in the init of the buffer or something to do with watermark
17:47:48***Saving seen data "./dancer.seen"
17:49:06 Quit dys (Ping timeout: 240 seconds)
17:49:06pamauryBilgus: I wrote the code so I'm pretty damn sure one can ;)
17:50:57pamauryBilgus: imx233_set_cpu_frequency looks up the given cpu frequency in the table and simply applies what is listed. imx233_set_cpu_frequency is called only with CPUFREQ_DEFAULT_{SLEEP,NORMAL,MAX}
17:51:16Bilgusjust experienced it first hand I set all defines to IMX233_CPUFREQ_64_MHz in system-target, then it looks up the setting by key in system-imx233 and 454 MHZ is the first one looked at, it then matches and returns the settings for 454 MHZ
17:51:19pamaurysorry CPUFREQ_{DEFAULT,SLEEP,MAX}
17:51:50pamauryhuh, unless you changed something, there is no way IMX233_CPUFREQ_454_MHz =IMX233_CPUFREQ_64_MHz
17:54:23Bilgusthats exactly what I did lol
17:55:29BilgusI didn't say you were wron I said for future reference
17:59:03Bilgusnow the next question is where CPUFREQ_MAX is actually used
18:00:35pamaurywhat did you do?
18:01:11pamauryBilgus: firmware/system.c
18:04:24 Join Halamix2 [0] (~Halamix2@unaffiliated/halamix2)
18:13:01 Join Huntereb [0] (
18:15:53 Join PimpiN8 [0] (
18:18:58Bilgusok so I'm back to not knowing WTF is the problem my clipzip plays the file fine it must be this fuze+
18:27:17 Join krabador [0] (~krabador@unaffiliated/krabador)
18:31:18 Quit Huntereb (Ping timeout: 258 seconds)
18:43:29 Quit Halamix2 (Quit: Leaving)
18:46:13 Join Huntereb [0] (
18:49:19 Quit Huntereb (Client Quit)
18:50:08 Join Huntereb [0] (
18:51:46 Quit robertd1 (Ping timeout: 240 seconds)
19:03:23 Quit Huntereb (Read error: Connection reset by peer)
19:03:42 Join Huntereb [0] (
19:04:37 Join petur [0] (
19:04:39 Quit petur (Changing host)
19:04:39 Join petur [0] (~petur@rockbox/developer/petur)
19:09:52BilgusPamaury I fixed it, moved cancel_cpu_boost(); from before the queue wait to after it and it works fine now the question is how much more time will devices spend boosted with that change or is there some other way to fix it like maybe checking how often the buffer needs filled in a time period
19:09:53 Quit Huntereb (Read error: Connection reset by peer)
19:12:48Bilgusmaybe change the queue wait to queue_wait_w_tmo and set it to a hz/2 or so
19:23:29 Join Huntereb [0] (
19:29:01 Quit Huntereb (Read error: Connection reset by peer)
19:33:27 Join Huntereb [0] (
19:36:37 Quit Huntereb (Read error: Connection reset by peer)
19:39:56 Join Huntereb [0] (
19:42:39 Quit michaelni (Ping timeout: 240 seconds)
19:47:49***Saving seen data "./dancer.seen"
19:50:44 Quit Huntereb (Read error: Connection reset by peer)
19:52:45 Join dys [0] (~dys@
19:54:33 Join Huntereb [0] (
19:56:10 Join michaelni [0] (
20:03:26 Quit Huntereb (Ping timeout: 240 seconds)
20:09:46 Join Huntereb [0] (
20:09:55 Join prof_wolfff [0] (
20:13:02 Join lebellium [0] (
20:13:20Bilgusok Fixed the issue g#1693
20:13:22fs-bluebotGerrit review #1693 at : Move cpu_boost for codec_thread & Fix AAC-HE Fuze+ (others?) by William Wilgus
20:14:22 Quit Huntereb (Ping timeout: 248 seconds)
20:19:21 Join Huntereb [0] (
20:29:44 Quit Huntereb (Read error: Connection reset by peer)
20:33:32 Join Huntereb [0] (
20:43:33 Quit _meg (Ping timeout: 240 seconds)
20:44:14 Quit pamaury (Ping timeout: 246 seconds)
20:46:30 Join _meg [0] (~notsure@
20:53:47 Quit PimpiN8 (Quit: My MacBook has gone to sleep. ZZZzzz…)
20:57:43 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 56.0/20170926190823])
20:57:45BilgusJhMikeS I see you've done quite a bit of work in codec_thread.c does this ^ look sane to you??
21:05:58 Join jhMikeS [0] (
21:19:36 Quit alexweissman ()
21:33:21 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
21:34:09 Join TheLemonMan [0] (~lemonboy@irssi/staff/TheLemonMan)
21:36:39 Quit dys (Ping timeout: 240 seconds)
21:43:45 Join kugel_ [0] (~kugel@rockbox/developer/kugel)
21:43:59 Quit kugel (Ping timeout: 248 seconds)
21:47:53***Saving seen data "./dancer.seen"
22:10:09 Quit Huntereb (Read error: Connection reset by peer)
22:13:30BilgusNope I spoke too soon It still happens just happens to have a specific scenario to make it happen you have to start the song and not press any buttons
22:16:45 Join Huntereb [0] (
22:19:27saratogapressing any buttons will enable boost
22:19:41saratogato trouble shoot this you may want to disable the GUI boost feature if you haven't
22:19:46saratogamakes things more reproducible
22:20:45saratogaoh does it not have it? it felt like it did when i was using the UI
22:21:17BilgusIt does
22:21:51Bilgusstarting to piss me off
22:23:44 Quit Huntereb (Ping timeout: 255 seconds)
22:26:15 Join Huntereb [0] (
22:28:19Bilguslol and it doesn't do it with a debug build
22:28:36BilgusI wonder if its some kind of a race condition
22:33:02*pamaury now remembers why the amsv2 rom is absolutely horrible to disassemble
22:36:32 Quit Huntereb (Quit: See ya!)
22:37:21 Join Huntereb [0] (
22:41:51 Join wodz [0] (
22:41:58wodzpamaury: why?
22:42:30pamauryjust about everything is an object
22:42:33pamaurylike everything
22:43:46pamaurya buffer? It's an object. A queue? It's vector (object) of buffers (object). An OS queue? It's queue (object) with a semaphore (object)
22:44:10wodzc++ to the extreme?
22:44:12pamauryNow I let you imagine what a usb driver can be. I've got at least 10 levels of objects imbrications, and I'm not done
22:44:30pamauryyeah, very extreme
22:44:44pamaurywith vtables everywhere
22:45:27pamauryoh also they do things like: A USB queue? It's a OS queue, but let's subclass it just in case, even though we don't actually change it
22:45:57pamauryThey even have an object to represent the hardware registers of USB ?!
22:46:20pamauryIt's too hard to do USB_REG_BLA = x; no no no, USB_REG::get()->write_reg(BLA, x)
22:46:50pamauryprobably there is going to be an allocation in this call for good measure
22:47:21wodzthats the type of code when PC programmer starts to do embedded with c++
22:47:39wodzlets abstract everything attitude
22:48:10pamauryI swear they even abstract thing that are clearly going to be unique
22:48:41wodzmaybe they did this as RE countermeasure
22:48:44pamauryI'm sure the main code is:
22:48:44pamaury(new BootCode())->boot()
22:49:27wodzI've heard of one product where firmware was intentionally rewritten in c++ to make RE much harder
22:50:09pamaurywell I guess that's successful. But I wonder if it's worth it. I suspect the C++ code is also impossible to follow
22:52:20*wodz is seriously pissed off by the lack of documentation and comments in qemu
23:07:24saratogastill don't understand why the fuze+ is so much slower than the same arm core in a fuzev2
23:11:38pamauryis it? is it the cpu or the ram?
23:14:16Bilgusso are you able to reproduce the AAC-HE bug on your fuze+ saratoga? you power on, set screen timeout to on, go to files, choose file and don't press a button and it skips within 5 secs?
23:14:42Bilgusif its just my device i'm fine with leaving it be at this point
23:22:15 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
23:25:27 Quit bray90820 ()
23:26:40saratogapamaury: not sure what the cause is, but codecs need much higher boost frequency
23:26:51saratogaBilgus: can i use any AACHE file?
23:27:09Bilgusidk maybe?
23:27:25Bilgusi can link you to the one i've been testing with
23:27:50 Quit amayer (Quit: Leaving)
23:28:08saratogamy fuze+ apparently just makes loud electronic static noises when powered on
23:28:18saratogai don't think i'd ever tried using it for music actually
23:29:02saratogasomething must be wrong with the headphone circuit
23:30:19saratogayeah sorry
23:30:23saratogai just used it for benchmarking really
23:33:40pamaurysaratoga: how do you compare? You mean that even at top frequency the fuzev2 is faster than the fuze+?
23:34:21saratogapamaury: test_codec for the same files gives much lower MHZ required for real-time decode
23:35:38pamauryis it an average frequency of boost/unboost?
23:35:50saratogayou can do it either fully boosted or fully unboosted
23:36:13saratogait decodes the file as fast as possible using either setting, then gives you the % realtime and the frequency required for 100% decode
23:36:49pamauryI see
23:36:49 Quit Soap_ (Read error: Connection reset by peer)
23:36:58pamauryis it possible that the core is actually different?
23:37:16pamauryie the core in amsv2 is not what we think it is?
23:37:36saratogaIIRC it was identified from the ARM version bits
23:37:45saratogabut in any case, it is some kind of ARM9E core
23:37:50saratogai don't think there is much variety there
23:38:04saratogai guess some difference in memory or cache is most likely
23:38:07pamaurycould it be the ram speed? That would be the obvious difference
23:39:29saratogai'm testing the files now to compare
23:40:54saratogaAAC-HE needs 73 MHz on AMSv2 vs. 260 MHz on Fuze+
23:41:25pamaurywow that's huge
23:41:39pamauryis that boosted?
23:41:42saratogayeah it seems like something isn't right
23:41:44saratogayeah boosted
23:42:06Bilgushuh when I ran it on my clipzip I got ~150
23:42:14saratogaboosted AMSv2 for AAC-HE-PS takes 92 MHz
23:42:26BilgusI'm pretty sure thats amsv2
23:42:27wodzwhat mem_test plugin says?
23:42:29saratogacompare to bilgus here:,52018.msg240757.html#msg240757
23:42:41pamauryit might be worth checking if the auto scaling function has a negative impact
23:42:42saratogaBilgus: interesting, what build?
23:43:32pamaurychanging imx233_clkctrl_enable_auto_slow(true) to imx233_clkctrl_enable_auto_slow(false) in system-imx233.c
23:43:56saratogapart of the difference is due to the lower clock on AMSv2 (38 MHz and 192 MHz), so memory latency will be lower
23:44:43saratogabut boosted AMSv2 (192 MHz) is still way faster per clock than unboosted imx233 (64 MHz)
23:45:02Bilgusits the multiboot build so like dev from july?
23:45:09saratogahuh, which file?
23:46:09saratogai got 384 MHz on fuze+ a couple months ago with this file:
23:46:21saratogaand 92 MHz just now
23:46:57saratogabuild on amsv2 is from last april
23:47:01 Quit krabador (Remote host closed the connection)
23:47:03saratogastock dev build
23:47:12Bilgusoh ok fuze+ vs amsv2
23:47:12saratogabut with test plugins enabled
23:47:40saratogawodz: i tried test_mem a while back but it didn't clear much up for me
23:47:56***Saving seen data "./dancer.seen"
23:48:23saratogai'm not sure if the plugin is very accurate
23:50:42*pamaury goes to bed
23:51:08Bilgusill try that test file nero on the fuze+ so we get some apples to apples comparison
23:51:53saratogatest_mem says that boosted fuze+ memory has similar read speed but double write speed as amsv2 boosted
23:51:56saratogafor whatever that is worth
23:53:13pamaurycould there be a difference due to iram?
23:53:19saratogayes, likely
23:53:26saratogahmm, not sure what test_mem tests
23:53:34pamaurythe fuze+ has basically no iram
23:53:59Bilgusok so first it skips the same as the other one
23:55:44Bilguspamaury could you do a test on your fuze+ when you get a chance I'd hate to be hunting a bug that was unique to my player
23:56:36saratogait tests the plugin buffer
23:57:41Bilgusok that file boosted decode 148.25 118.73% 383 MHZ
23:58:46BilgusRunning unboosted now

Previous day | Next day