Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2021-04-12

00:08:08***Saving seen data "./dancer.seen"
00:20:26 Quit kadoban (*.net *.split)
00:20:26 Quit edhelas (*.net *.split)
00:20:26 Quit michaelni (*.net *.split)
00:20:26 Quit shrizza (*.net *.split)
00:26:48 Join michaelni [0] (
00:26:48 Join shrizza [0] (
00:41:24 Join kadoban [0] (kadobanemp@gateway/shell/
01:21:08 Quit copper (Ping timeout: 252 seconds)
01:21:15 Join copper [0] (~copper@unaffiliated/copper)
01:26:38 Quit ac_laptop (Ping timeout: 268 seconds)
02:08:12***Saving seen data "./dancer.seen"
02:36:25braewoodsspeachy: i see. well it'd help if we had one...
02:39:58 Join Rower- [0] (~Rower@
02:42:18 Quit Rower (Ping timeout: 240 seconds)
02:49:02braewoodsi suspect we're not going to be testing either mpio port
02:54:13braewoodsgevaerts: you got any of the harder to find targets in your collection? mpio hd200, hd300? iriver H110 or H115?
02:54:37braewoodsif you do, repairing those may be our best option for testing them
02:54:53braewoodsor for that matter either of the m:robe ports, 100 or 500
03:05:10 Quit S|h|a|w|n (Read error: Connection reset by peer)
03:32:44 Join TheSphinX_ [0] (
03:33:43 Quit TheSphinX^ (Ping timeout: 260 seconds)
03:33:43 Nick TheSphinX_ is now known as TheSphinX^ (
03:59:04 Join edhelas [0] (9d94237298@2a03:4000:51:f44:4e1:2ff:fe00:4257)
04:08:13***Saving seen data "./dancer.seen"
04:10:58gevaertsbraewoods: I have the m:robes (assuming their batteries haven't effectively killed them). What exactly do we need tested?
04:17:15braewoodsgevaerts: speachy wanted to test that our builds are working basically
04:17:25braewoodsit appears someone tested the 100 already
04:17:38braewoodsthe 500 is having issues they say
04:17:57braewoodsso we'd need a unit to find out what the issue is
04:19:43gevaertsSomeone else has a working 500 still? :)
04:20:13braewoodsthey are available on ebay
04:20:21braewoodswhy do we want to pay $80 just to test this?
04:20:37braewoodsespecially one that doesn't appear to have many users
04:20:57gevaertsI mean, I have one here. I'll start by seeing if it still boots before thinking about testing
04:21:09braewoodsyou live in Europe if I recall?
04:21:34braewoodsit'd be nice if we had someone with time in Europe to troubleshoot it
04:21:46braewoodsassuming it even still works well enough to test
04:22:11braewoodsbut i was surprised to see how big the m:robe 500 is
04:22:18braewoodsit looks like a modern tablet
04:22:20gevaertsIt's a weird device
04:23:02braewoodsis it touch screen only?
04:23:26gevaertsIt is
04:23:31braewoodsi thought so
04:23:47braewoodshow'd we ever manage to support it? i thought we didn't work well with TS
04:24:11gevaertsIt never got out of "unstable"
04:24:30braewoodsi see
04:24:35braewoodsweird in any case
04:26:28braewoodsi guess speachy mainly wants to ensure no serious regressions
04:26:36gevaertsok, the build that's on there still works
04:26:44gevaertsThat's a start
04:27:09braewoodsok i'd get in touch with speachy then, since i think he's in a better position to debug it
04:27:43braewoodsi just wish we could find the rare H110 or H115
04:27:51braewoodsto test the new bootloader and such
04:27:55braewoodsit should work since it works in the H120+
04:28:06gevaertsYes, those are the dangerous ones...
04:28:18braewoodsthe rare units?
04:28:44gevaertsWell, those flashed bootloader ones without easy recovery
04:29:02braewoodsi actually found one but maybe just luck
04:29:26braewoodsit only works if the OF is still present and you have an original HDD installed
04:29:41braewoodsi used it to save my bacon when i was debuggin the H320 bootloader
04:29:58braewoodsthank you whoever wrote the reset cookie to try to boot the OF if boot hang occurred
04:31:49braewoodsgevaerts: you can update your H120s and H320s if you want. the latest bootloader has been out for awhile and works.
04:32:04braewoodsit fixes CF compatibility.
04:34:04gevaertsok, I can confirm that the current build doesn't boot on mr500
04:34:51braewoodsspeachy can troubleshoot the rest
04:35:04braewoodsfirst thing we should check is if the problem is similar to the gigabeat S one we patched
04:35:09gevaertsI'll try to bisect soon
04:35:21braewoodstry from before the toolchain last year
04:35:29braewoodsthen if it boots check the stack usage of threads
04:35:46braewoodsthe gigabeat S failed to boot because of stack overflow
04:35:54braewoodsmay as well check it here just to be sure
04:35:57braewoodsit's easy to check if you can boot
04:35:59 Join vmx [0] (
04:36:47braewoodsit's likely either a toolchain issue, lcd issue, or both
04:39:52gevaertsOh, right... toolchains were bumped "recently", right?
04:41:04braewoodsbilgus also redid LCDs which broke some stuff
04:41:16braewoodshard to say which is at fault here
04:41:39braewoodsaround october is when these hit
04:41:43braewoodsof last year
04:41:48gevaertsI think I'll try the current revision with old gcc to start witj
04:41:56braewoodsthat's not advisable
04:42:05braewoodswe started using new GCC features
04:42:11gevaertsI'll see :)
04:42:20braewoodsi presume it'll fail
04:42:24braewoodsbut ok
04:42:38gevaertsI mean, it depends on where the failures are
04:42:41braewoodsyou can also try dialing back optimizations
04:42:51braewoodsthe gigabeat S worked on -O1 but not -Os
04:43:05gevaertsWell, I don't have the new gcc yet, so might as well try this first before upgrading that
04:43:08braewoodsindeed it does
04:43:17braewoodsi'd start from a revision before GCC bump
04:43:22braewoodsjust to be sure that works
04:43:31braewoodsthe build you had is probably really old
04:43:57gevaertsthe gigabeat S fix is in a device-specific thread stack, so that fix isn't copyable
04:44:03braewoodsi know
04:44:08braewoodsbut it could be a similar problem here
04:44:21braewoodsbut no way to know until you start finding out where it first broke
04:45:16gevaertsoh, right, the reset pin on this thing needs opening the case...
04:45:36braewoods... wow that's stupid
04:45:47braewoodsdepending on where it was
04:45:54braewoodsi might just drill a hole in the case to acces it
04:46:55gevaertsworth considering :)
04:47:51braewoodsyea if i was developing fori t
04:48:03braewoodsi'd consider it if the reset thing was easy to access
04:48:17braewoodsmight install a button just to short the pins when i need to reset
04:49:15gevaertsOr didn't it have one? Oh well, it's open, I can disconnect the battery :)
04:51:12gevaertsYes, running 7f4c696-120226... That's not a new build exactly :)
04:58:14braewoods_bilgus: it seems like the fixes you applied boosted the H10's write speed by about 50%
04:59:31 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
04:59:49braewoodsi was getting like 5MB/s
04:59:52braewoodsnow it's like 7.5MB
05:00:05 Quit ccf-100 (Quit: Idle for 30+ days)
05:57:40_bilgusbraewoods, thats a pretty good boost
06:08:17***Saving seen data "./dancer.seen"
06:12:23 Quit Acou_Bass (Ping timeout: 260 seconds)
06:21:30 Join Acou_Bass [0] (
06:32:54gevaertsLooking for boot-breaking bugs does have the advantage that testing doesn't take too much time :)
06:42:37speachygevaerts: it's the only tms320 target that even made it to the "unstable" list, with daily builds and everything.
06:44:25speachyIt's debatable whether or not it's worth even trying to fix this but if not, there's a pile of other targets that are equally undeserving
06:44:41gevaertsI should have a breaking revision soon. It's definitely from before the gcc update
06:45:21gevaertsIt's probably worth having a look at, but given its unstable status it shouldn't be a factor for a release decision
06:45:59speachywell, it's obviously been the first report of anyone using a build less than a year old on this thing.
06:48:59speachyall known issues so far are on unstable targets, or aren't regressions (eg the SSD issues on the ipods)
06:49:46speachyspeaking of which, seems the ipod6g still has SSD issues. probably because it has its own ATA driver and also powers down the ATA hardware/interface.
06:54:01gevaertsOh, bah. I wish we'd looked earlier...
06:54:05gevaertsI found again :)
06:55:17gevaertsThat concludes that the new toolchain doesn't have the problem I think?
06:56:08gevaertsspeachy: you may remember that better. I'm a bit confused after reading those logs
06:56:13speachyI suppose so. But I obviously completely forgot about that too
06:56:32speachymust have been from when I still had both toolchains sitting around
06:56:59gevaertsBut the report is that the current build doesn't work, so it presumably broke again after that?
06:57:22speachyso that seems like some sort of linking problem
06:58:01gevaertsWell, time to upgrade toolchains then :)
06:58:39speachythe problematic commit was never reverted.
06:59:25speachy(and that also explains the provenence of that build cereal_eater was using. it's something I built for him..)
06:59:46gevaertsNo, but the toolchain was upgraded, so if that fixed it the first time either something was shuffled again to break things the annoying way, or something else broke
07:02:27*speachy nods.
07:03:46speachyre-reading this log makes me wonder if the -1 build I supplied (new toolchain, then-git master) was actually booted up.
07:04:49braewoodsit may seem weird but i build my test builds on a server with ECC RAM
07:04:57braewoodsi'm a bit paranoid about consistency
07:05:00braewoodswith this kind of stuff
07:12:03speachyon my systems with ECC, I average less than one detected error a year
07:13:19speachyI suspect most of our builders aren't equipped with ECC. Though 2 of my 3 are.
07:13:25speachy(and the main server)
07:13:58speachyI take that back, all 3 are; the third is unbuffered ECC.
07:16:27braewoodsunlikely but i like it
07:16:55braewoodsi personally don't want to chase a phantom issue that was due to a flipped bit or something
07:18:08braewoodsspeachy: you know what i find funny on ebay?
07:18:26braewoodspeople selling stuff at inflated prices and then seeing that for the most part only the modestly priced ones sell
07:18:40braewoodsdo they not care about holding inventory indefinitely?
07:18:51*braewoods boggles
07:19:34speachydepends on how much they paid for it. and, especially after a certian point, if you need something specific, you're going to NEED it at any price.
07:20:54braewoodsi see these old Dell mp3 players
07:21:07braewoodsprobably some PP derivative
07:23:26 Part edhelas
07:23:40braewoodsthey still provide the OF
07:23:43braewoodslet's check the firmware
07:24:14 Join edhelas [0] (9d94237298@2a03:4000:51:f44:4e1:2ff:fe00:4257)
07:27:30braewoodsnot a set of files...
07:27:34braewoodssome flash program?
07:33:28braewoodsseems someone already did research on it
07:35:24braewoodspossible port here but probably not worth it at this point
07:53:42gevaertsspeachy: if I revert b0de98ad and build current master it boots, so still the same issue. I have no idea why there was a working build in that debug session last year...
07:53:56gevaerts(with a fresh toolchain)
07:54:59braewoodsthe gba sp of mp3 players
07:56:47braewoodsgevaerts: sounds like we need a port specific workaround
07:56:55braewoodssince nothing else seems to be bothered by it
07:57:30braewoodsreminds me of when i was debugging H320 bootloader
07:57:39braewoodsit was an LCD optimization that broke it
07:58:09braewoodssince i'm not an expert on the port i ended up reintroducing the original code but only enabled for bootloader builds
07:58:26braewoodssince i didn't know how else to fix it
08:08:21***Saving seen data "./dancer.seen"
08:08:35 Quit pamaury (Ping timeout: 260 seconds)
08:09:09 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
08:14:31gevaerts fixes the issue for me, while still having working thumb builds (which is what b0de98ad fixed). The effect is that -lfirmware is now in the link cmdline twice
08:14:53gevaertsOf course all of this is magic, so I don't know if it breaks anything else now...
08:15:11speachyhmm, I never did like that thumb-cc hack
08:16:13gevaertsMy worry isn't thumb so much as knowing that link order broke the mr500 so I'm not sure changing link order again won't break something else
08:16:44speachyas Isaid in that earlier log all of this is a canary.. the RealProblem(tm) is something else
08:17:13speachymaybe some sort of symbol collision
08:17:28speachyso the ordering of linked libraries matters
08:19:11braewoodsspeachy: is it time to move our python scripts to 3?
08:19:26braewoodspython2 is being phased out
08:19:29speachyeverything in the normal build path is python3 clean
08:19:48braewoodsi see
08:19:50speachyI retired server-side python2-specific stuff
08:20:21braewoodsseems my H10 got scratched somehow during shipment or w/e
08:20:31braewoodsi'll try buffing it out of the metal
08:26:58speachyhuh, we only use that thumb script on the original clip, m200v4 and c200v2. Ie onr remaining 2mb targets.
08:27:12speachyand all it ultimately does is try compiling with and without thumb, to see which is smaller.
08:32:14gevaertsA naive search (objdump -t and then look for g or F) finds one "duplicate" symbol between libfirmware and libunwarminder, data_abort_handler, which is weak in libfirmware.a so I would expect the later one to be used anyway?
08:33:20gevaertsAnd I'd hope we're not hitting a data abort anyway
08:34:24gevaertsSo my guess is it's something else again
08:47:21_bilgusbraewoods probably more likely its time on my desk
08:49:17_bilgusI vaguely remember doing it for the clip+ as well?
08:50:41 Join Saijin_Naib [0] (
08:53:15speachy_bilgus: hmm?
08:59:30braewoods_bilgus: it just looked like someone took a razor blade to it
08:59:59braewoodsoh well, the problem is solved so small price to pay
09:00:37braewoodshuh seems the PPs that load entirely from their drives...
09:00:45braewoodsdon't even need the OF the way we pack it up
09:00:52braewoodsonly the rom bootloader remains OF
09:01:05 Join massiveH [0] (
09:01:25braewoods_bilgus: i was just thinking though, if not for the iPod using PP so heavily, i doubt we'd have any support for it
09:01:48braewoodswe have no cpu datasheet
09:02:03braewoodsif not for the ipods being such a big deal back in the day
09:02:15braewoodswe probably would have been unable to progress with it
09:02:50braewoodswe've had to make guesses about how it works just to get what we have
09:03:06braewoodswhich makes it hard to support some targets fully depending on how they choose to use IO pins
09:08:53_bilgusno I didn't use any metal to open it up I assume it arrived sealed with packing tape on the box and duct tape on the player bubble wrap?
09:09:47braewoods_bilgus: yea
09:09:56braewoods_bilgus: in any case, don
09:09:59braewoodsdon't worry about it
09:10:03braewoodsi don't really know how it happened
09:10:53braewoodsthe issue is solved and that's what matters
09:11:09braewoodsi can always buy a better one if i cared that much
09:11:37_bilgusthe ipods are still a big deal for now
09:11:51_bilgusstill no data sheets
09:12:10braewoodswhich is funny considering the PPs we care about are no longer really used
09:12:20braewoodsso what does holding onto them datasheets do for them
09:13:22 Join chris_s [0] (
09:16:33chris_sMaybe it's just my filter bubble, but I feel like iPods (and possibly Rockbox) are having at least a little bit of a revival. It's all relative of course. Pretty much everybody is streaming...or listening to vinyl I guess ;)
09:17:09braewoodshahaha streaming...
09:17:14*braewoods goes into a tunnel
09:17:20braewoods"Why did my audio cut out!?"
09:17:39braewoodsrockbox still has a niche
09:17:46braewoodsstreaming isn't going to work everywhere
09:21:05braewoodsplus i don't want to waste my cellular data on audio streaming given how they price it in the US
09:21:54chris_stbf, you can download tracks using Wifi on Spotify/Apple Music for offline listening, but yeah, there's some advantages still to "owning" your music...
09:23:19braewoodssame reason chrome books aren't going to replace PCs for everyone
09:23:33braewoodsor similar
09:24:30 Join ats_ [0] (
09:24:49 Quit atsampson (Ping timeout: 258 seconds)
09:30:18braewoodsvery interesting
09:30:34braewoodsthe iriver h100 remote also works in a CD/MP3 player hybrid
09:30:36braewoodsiriver slimx
09:47:46 Quit chris_s (Quit: Ping timeout (120 seconds))
10:08:23***Saving seen data "./dancer.seen"
10:22:29 Quit massiveH (Quit: Leaving)
10:39:56 Quit akaWolf (Read error: Connection reset by peer)
10:40:18 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
10:46:22 Nick ats_ is now known as ats (
10:50:16 Quit akaWolf (Read error: Connection reset by peer)
10:51:02 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
12:08:26***Saving seen data "./dancer.seen"
12:09:46 Join ZincAlloy [0] (
12:56:12 Join Soap [0] (~Soap@rockbox/staff/soap)
13:17:01 Join lebellium [0] (
13:42:18 Quit vmx (Quit: Leaving)
14:06:40 Join S|h|a|w|n [0] (~shawn156@unaffiliated/shawn156)
14:08:27***Saving seen data "./dancer.seen"
14:19:15fs-bluebotBuild Server message: New build round started. Revision b6fce99046, 298 builds, 10 clients.
14:33:03fs-bluebotBuild Server message: Build round completed after 828 seconds.
14:33:05fs-bluebotBuild Server message: Revision b6fce99046 result: All green
14:49:47 Join MrZeus__ [0] (~MrZeus@2a02:c7f:a0aa:4400:49fc:7eea:1fa3:615)
15:04:30 Join MrZeus [0] (~MrZeus@2a02:c7f:a0aa:4400:49fc:7eea:1fa3:615)
15:05:19 Quit MrZeus__ (Ping timeout: 260 seconds)
15:05:49 Quit dweeber (Ping timeout: 260 seconds)
15:07:43 Join dweeber [0] (
15:27:29 Join MrZeus_ [0] (~MrZeus@
15:27:43 Quit MrZeus (Ping timeout: 260 seconds)
15:39:51 Join lebellium_ [0] (
15:40:16 Quit lebellium (Ping timeout: 252 seconds)
15:58:23 Quit speachy (Quit: WeeChat 3.0.1)
16:08:28***Saving seen data "./dancer.seen"
17:43:14 Join ac_laptop [0] (~ac_laptop@
17:55:06fs-bluebotBuild Server message: New build round started. Revision c0a49d9bdf, 298 builds, 10 clients.
18:00:58 Quit lebellium_ (Quit: Leaving)
18:06:24 Join cockroach [0] (~blattodea@pdpc/supporter/active/cockroach)
18:07:37 Quit ZincAlloy (Quit: Leaving.)
18:08:28fs-bluebotBuild Server message: Build round completed after 802 seconds.
18:08:30***Saving seen data "./dancer.seen"
18:08:30fs-bluebotBuild Server message: Revision c0a49d9bdf result: All green
18:12:52 Quit beencubed (Quit: Leaving)
19:03:36 Quit MrZeus_ (Read error: Connection reset by peer)
19:04:04 Join MrZeus_ [0] (~MrZeus@2a02:c7f:a0aa:4400:49fc:7eea:1fa3:615)
19:38:33 Quit pamaury (Ping timeout: 240 seconds)
19:56:13 Join beencubed [0] (~beencubed@
20:03:32 Quit ac_laptop (Ping timeout: 240 seconds)
20:08:33***Saving seen data "./dancer.seen"
20:29:55 Quit paulk-leonov (Ping timeout: 252 seconds)
20:30:41 Join paulk-leonov [0] (
20:36:11 Quit MrZeus_ (Ping timeout: 260 seconds)
20:36:46 Join ac_laptop [0] (~ac_laptop@
20:39:41 Quit paulk-leonov (Ping timeout: 240 seconds)
20:40:59 Join paulk-leonov [0] (
21:16:41 Quit ac_laptop (Ping timeout: 240 seconds)
21:23:34 Quit cockroach (Quit: leaving)
22:08:36***Saving seen data "./dancer.seen"
22:23:24 Quit Ckat (Ping timeout: 268 seconds)
22:29:05 Join Ckat [0] (~Ckat@xn--z7x.xn--6frz82g)
22:31:40 Join f1reflyylmao [0] (~f1refly@2a01:c23:905e:7e00:9b6c:a2fe:4a40:53af)
22:34:21 Quit f1refly (Ping timeout: 260 seconds)
22:34:21 Nick f1reflyylmao is now known as f1refly (~f1refly@2a01:c23:905e:7e00:9b6c:a2fe:4a40:53af)
22:52:22 Join ufdm_ [0] (
22:52:35 Quit ufdm (Ping timeout: 248 seconds)
23:14:44 Quit Saijin_Naib (Ping timeout: 258 seconds)
23:29:04 Join Stanley00 [0] (~stanley00@unaffiliated/stanley00)

Previous day | Next day