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 2022-11-27

00:28:51chris_salways encouraging to see even you make mistakes ;)
00:33:45chris_sstring handling in C generally seems pretty insane to me
00:37:27 Quit massiveH (Quit: Leaving)
00:37:28speachywhat's really fun is that some CPU arches have "native" C stringhandling instrctions.
00:38:21_bilgusI make tons of mistakes
00:38:59_bilguscould you test g#4846 and see if it fixes the issue
00:39:02rb-bluebotGerrit review #4846 at https://gerrit.rockbox.org/r/c/rockbox/+/4846 : bookmark.c fix off by one error in filename generator by William Wilgus
00:42:17_bilgusI've found it quite challenging to get rid of those static buffers its hard building strings piecemeal
00:42:58speachyI think I've identified a few more fields we can use to guess at SSD support, if the adapter passes through the identify info unmolested.
00:43:34_bilgustalk about some happy iPod users
00:44:37speachyprobably should just buy some hardware myself. :/
00:45:12speachyhuh, my toolchain build succeeded this time. wonder if it works.
00:46:13chris_s_bilgus: still happening...
00:49:42chris_si see no rhyme or reason to when it works and when it doesn't though
00:50:49_bilguscould you give me the string it created in the bmark file?
00:51:13_bilgusI'm generating the name based on that so it might give me a clue
00:51:49chris_s>3;7;846577;0;3900;0;0;10000;10000;/Playlists/God is in the House.m3u;/<microSD0>/Music/Art Tatum/God is in the House/08 Knockin' Myself Out.m4a
00:52:14chris_sfilename for the bookmark is God is in the House.m3u.bmark
00:53:20_bilgusso still knocing off the 8? then
00:53:44_bilgusoh but it did it in the bmark file too
00:57:33chris_ssometimes more..
01:00
01:02:59_bilgushmm it gets the string from playlist_get_name()
01:04:00speachyanyone have an arm target handy to run a smoke test build for me?
01:04:28speachy(all of mine are a ~3h drive away)
01:04:42chris_si have an ipod handy
01:04:54speachywhich one?
01:05:16speachyI just built a mini2g image with only one new warning. :)
01:05:16chris_soh yeah.. the iPod video
01:05:56speachyok, comipling.
01:10:25speachychris_s: https://www.shaftnet.org/~pizza/ipodviode-test.zip
01:10:32speachy(typo is "correct")
01:15:25speachyit's about an 8.5KB smaller main binary than the last dev build with 4.9.4, apparently.
01:16:05speachyer, 6.5KB
01:16:11speachyditto on RAM usage.
01:18:12_bilgusok the only other thing I see is I might be screwing up strlcat by feeding it the current buffer position rather than letting it traverse the string
01:18:24chris_sspeachy: sounds good – all seems well so far. Boots up and plays music... anything specific I should be testing?
01:18:43_bilguschris_s when you get a chance could you try 4846 again
01:18:54speachynothing in particular; the fact that it runs at all is a massive step.
01:19:39speachythis means that this toolchain bump is ready for wider testing.
01:19:44_bilgussorry 4848
01:20:05rb-bluebotBuild Server message: New build round started. Revision 65db4acabe, 303 builds, 7 clients.
01:21:24speachychris_s: The GCC 4.9.4 (and -Os) migration was pretty rocky for the ipods. several major issues were uncovered.
01:22:53_bilgusI'm sure there is more of that hiding waiting for discovery but ASAN and UBSAN have helped uncover a bunch of stuff
01:25:25speachy7.1K shrink on the clipplus, huh.
01:25:57chris_s_bilgus: same thing....
01:27:31speachyg4850 has my latest attempts to make mSATA SSDs happy. total guesswork.
01:27:34_bilgushmm still the same thing in the .bmark file too?
01:28:05_bilgusat this point might be chasing a bug elsewhere
01:28:24chris_s>3;4;122499;0;4598;0;0;10000;10000;/Playlists/The Tallest Man On Eart;/<microSD0>/LQ/The Tallest Man On Earth/Shallow Grave/05 Where Do My Bluebird Fly.mp3
01:28:27chris_syeah
01:28:57chris_sreverting 88ecaf2 seems to fix it thoigh
01:31:08_bilgusthen my math is off or something
01:31:58speachygonna crash finally.
01:32:48_bilgusnn speachy
01:37:41rb-bluebotBuild Server message: Build round completed after 1056 seconds.
01:37:42rb-bluebotBuild Server message: Revision 65db4acabe result: All green
01:43:24_bilguschris_s well sure enough when I passed back the strings you posted it did the same thing WTH
01:44:56_bilguswell with the added 'h' ofc
01:45:52chris_sglad you're able to reproduce it at least
01:47:14_bilgusffs I see what it is
01:47:46***Saving seen data "./dancer.seen"
01:47:52_bilgususing bmsz instead of bufsz to playlist_get_name
01:48:22_bilgusIDIOT in stimpy voice
01:49:37chris_s:o
01:52:17_bilgusok can you try that one it should be good now
01:56:53chris_ssure enough... looks good :D
01:57:56_bilguslol
01:58:10_bilgusim renameing bmsz to something longer
01:58:18chris_sha
01:58:53_bilgusanyway thanks for your testing
01:59:08chris_sno problem, thanks for keeping at it
02:00
02:05:34rb-bluebotBuild Server message: New build round started. Revision 3f4e55a872, 303 builds, 7 clients.
02:23:59rb-bluebotBuild Server message: Build round completed after 1104 seconds.
02:24:15rb-bluebotBuild Server message: Revision 3f4e55a872 result: All green
02:37:48_bilgushmm I do see another bug though sometimes playlist_get_resume_info() doesn't return the proper index
02:38:30_bilgusI noticed it first on a playlist with only 1 item the bookmark recorded 3 for the index but it was index 0
02:39:09_bilgusthen I tried a playlist with 2 items and sometimes it returns 0 even though i'm playing song at index 1
03:00
03:47:48***No seen item changed, no save performed.
04:00
04:05:43 Join advcomp2019_ [0] (~advcomp20@user/advcomp2019)
04:08:56 Quit advcomp2019 (Ping timeout: 256 seconds)
04:13:56 Join fourHZ [0] (~fourHZ@92-52-40-121.dynamic.orange.sk)
04:17:21 Join lebellium [0] (~lebellium@2a01cb040109a60001a53c527cc10874.ipv6.abo.wanadoo.fr)
05:00
05:35:44 Quit CH23_M (Ping timeout: 260 seconds)
05:38:39 Join Pokey [0] (~pokey@spikeyCactus/hoosky)
05:39:02 Quit Pokey (Client Quit)
05:40:00 Join CH23_M [0] (~CH23@revspace/participant/ch23)
05:43:06 Join Pokey [0] (~pokey@spikeyCactus/hoosky)
05:47:49***Saving seen data "./dancer.seen"
06:00
06:08:23 Quit fourHZ (Ping timeout: 260 seconds)
07:00
07:01:10 Join fourHZ [0] (~fourHZ@92-52-40-121.dynamic.orange.sk)
07:28:04 Quit hactar|ant (Ping timeout: 260 seconds)
07:28:30 Quit chris_s (Ping timeout: 265 seconds)
07:35:34 Join hactar|ant [0] (~zem@c-24-21-103-100.hsd1.or.comcast.net)
07:47:52***Saving seen data "./dancer.seen"
09:00
09:41:16 Quit Pokey (Quit: Hecc! My server must have died!)
09:45:21 Join Pokey [0] (~pokey@spikeyCactus/hoosky)
09:47:53***Saving seen data "./dancer.seen"
10:00
10:13:27_bilguswell scratch that after adding debugging and recompiling I can't repro
10:13:44_bilgusand going back I can't repro either
10:14:01speachyso... swamp gas reflecting off a weather baloon?
10:14:17_bilgusexactly
10:15:10_bilgusI do get weird stuff like that on ocassion just replacing the binary only in an install even though nothing else has changed
10:15:29_bilgusseg faults and whatnot
10:15:46_bilgusvoodoo or something
10:17:13speachyI guess one of us "core" devleopers should get one of these ipod ssd thingeys.
10:17:47speachyand/or kick out another release so we can stop getting bug reports about stuff fixed two years ago
10:18:59_bilguswe have one I don't remember who has it
10:19:29_bilgusno thats the sd adapter
10:19:39_bilgustheres a ssd now?
10:19:50speachyzif->mSATA
10:19:59speachyso you can plug in a generic mSATA ssd.
10:20:08_bilgusnice
10:20:23speachyvast overkill for performance (and likda power hungry but still better than the stock spinning rust)
10:20:42_bilgusso same issue as the iflash then just new weird behavior?
10:20:52speachybut $/GB it's still the cheapest way to go for large sizes.
10:21:12speachyit's similar-ish. By appearances writs are really unreliable
10:21:30speachyI suspect it's the same underlying problem (ie power being turned off before things are ready for it)
10:21:52speachyI don't know if mSATA is truly meant to have the power completely turned off in the same way FWIW.
10:22:17_bilgusIt probably isn't seems like it'd want to keep state
10:22:59speachydunno. I need to see what the ATA IDENTIFY info comes up with and I can go from there
10:23:27speachy...something similar showing up on the iHP-3xx too.
10:23:32speachysame ATa driver
10:24:36 Join steelbas [0] (~steelbas@185.65.134.179)
10:24:38speachybut with an unknown chinesium SSD that I think is also a mSATA adapter
10:24:40_bilguswell being that iPods seem to be our core users at this point.. guess there are a bunch secondhand still alive
10:25:25speachyipods will be aroun long after everything else has bitrotten, alas.
10:28:17paulcarrotyipods are the best choice for easy storage and battery upgrade
10:29:38speachywell, the "classic" ipods are. the newer flash-based ons have mostly already failed.
10:29:50 Quit steelbas (Quit: Leaving)
10:31:34speachyI'm going to reach out to some of those "reburbished ipod" sellers in china and see if they'll tell me what type of SSD they're putting inside.
10:32:38paulcarrotysansa/iriver updrade is painful, sadly, at least without 3D printer for custom case.
10:35:59speachyit's gonna cost about $120 just for a working ipod and the msata adapter, and more for the ssd; might as well buy one that already has the work done.
10:36:14speachyI _might_ have an old mSATA SSD or two lying around
10:37:17speachyLOL, the response is "Hello, dear. SSD type."
10:39:03paulcarrotytell them "I wanna buy a lot, 100+, and need a free sample" :D
10:41:34speachysome of the sellers list it as "SD mod" but most don't.
10:41:51speachyto get exactly what we want I'll probably have to do the work myself, joy.
10:48:03speachylooks like about $150 all-in, including a noname 256GB mSATA SSD. for a 5th gen 30gb.
10:52:45speachyand I still need to fix the X3's USB stuff. :D
10:53:06speachy(...speaking of a mostly irrelevant player that nobody van get any more...)
11:00
11:03:11 Quit speachy (Quit: WeeChat 3.6)
11:04:44 Quit Pokey (Ping timeout: 260 seconds)
11:07:28 Join Pokey [0] (~pokey@spikeyCactus/hoosky)
11:28:44_bilgusmight post to the forum about someone donating one or money for one
11:28:58_bilgusif not I have $20 towards the cause
11:29:40_bilgusI can't see me ever using one but I can't see you paying out of pocket to do the work too
11:33:22hactar|antmsata is basically dead right? in favour of m.2 sata/pcie
11:35:11_bilguschris_s (logs) your recent Return to opened playlists −− how does reopen_last_playlist do with the two paths into display_playlists() (catalog_view_playlists & catalog_add_to_a_playlist) wouldn't it cause it to pull you into the current playlist rather than to the menu for selecting a new one?
11:40:02_bilgusanother issue id that it pulls you directly to the WPS when you try to go back to the playlist catalog after its already playing
11:41:09_bilgusthat is probably existing
11:44:20_bilgusah nevermind I see its done on the view conditional
11:47:55***Saving seen data "./dancer.seen"
11:56:14 Join mink [0] (~mink@125.215.164.109.static.wline.lns.sme.cust.swisscom.ch)
11:56:33 Quit CH23_M (Read error: Connection reset by peer)
11:57:24 Join CH23_M [0] (~CH23@revspace/participant/ch23)
11:58:27 Quit mink (Remote host closed the connection)
12:00
12:01:59 Quit CH23_M (Ping timeout: 264 seconds)
12:05:52 Join CH23_M [0] (~CH23@revspace/participant/ch23)
12:21:06 Quit CH23_M (Read error: Connection reset by peer)
12:22:02 Join CH23_M [0] (~CH23@revspace/participant/ch23)
12:26:28rb-bluebotBuild Server message: New build round started. Revision 0c7394e39a, 303 builds, 7 clients.
12:36:39_bilgusthe only thing thats a little off about the patch in the end is that if you were in a playlist lets call it 'A' and added to playlist 'B' then backed out to the catalog screen it would now have 'B' selected and take you to the selected item index you were using in 'A"
12:37:29_bilgusI did check if that index was out of range for B then it selects index 0 so meh quirky but not bad
12:40:51 Join Malinux [0] (~malin@2001:4641:4dfa::12c:c4a7)
12:43:57 Quit Pokey (Quit: Hecc! My server must have died!)
12:45:09 Join Pokey [0] (~pokey@spikeyCactus/hoosky)
12:55:54rb-bluebotBuild Server message: Build round completed after 1765 seconds.
12:55:58rb-bluebotBuild Server message: Revision 0c7394e39a result: 2 errors 0 warnings
13:00
13:00:16_bilguswell just got a case of it selecting the wrong index again so far its always one off either -1 or +1 but this time it was in invalid index so it corrupted the resume state
13:01:31_bilgusit then had playlist control invalid error it makes me wonder if its that infamous bug rearing its ugly head or something I or we've brought upon ourselves :)
13:02:52_bilgusoh no scratch that it wasn't an invalid index hmm so might be the whole current_playlist state getting corrupted
13:21:26 Join mink [0] (~mink@178.197.232.222)
13:22:13 Quit mink (Read error: Connection reset by peer)
13:35:38_bilguswow went back to a build from 10-2020 and the lag is striking versus the lack thereof from today
13:47:58***Saving seen data "./dancer.seen"
14:00
14:01:36_bilguswell it happens before 88ecaf2b8c so I guess I need to go on a bisect adventure
14:13:14_bilgusI'll try to get to that this eve or this week if anyone wants to play along just set the bookmark options to [Yes][Bookmark on stop, update on stop, Maintain a lict of recent] and using Playlist Catalog create an /all playlist then choose a song and stop it [Bookmark Created] finally go to Recent Bookmarks and choose the latest bookmark then rinse repeat I usually get it to fail with Playlist buffer Full by the 5th time of choosing the
14:13:14_bilguslatest/stopping/choosing the latest
14:14:08_bilgusmight have to do with the number of songs (I have 11000 in there and a max playlist size of 32000)
14:38:17 Join speachy [0] (~speachy@tunnel658658-pt.tunnel.tserv12.mia1.ipv6.he.net)
14:38:17 Quit speachy (Changing host)
14:38:17 Join speachy [0] (~speachy@rockbox/developer/speachy)
14:38:17Mode"#rockbox +v speachy" by ChanServ (ChanServ@services.libera.chat)
14:38:42speachy_bilgus: there's money in the slush fund, just don't like spending that much unliaterally.
14:39:40speachyhactar|ant: it's obsolete but there's still a huge deployed base of non-mvme-capable systems.
14:41:06speachysweet! USB on the X3 is still not working but at least it's enumerating now.
14:41:31speachyfinally rewrote the ctrl stuff for the new API
14:47:17speachyhuh, usb hid works but not mass storage.
14:48:02hactar|anti meant both pcie and sata
14:48:08hactar|antboth can use the m.2 connector
14:48:39speachymsata uses the m.2 connector
14:48:58speachykeyed differently, naturally.
14:49:59speachyOK, it's DMA causing the problems.
15:00
15:06:00 Quit speachy (Quit: WeeChat 3.6)
15:25:23 Join chris_s [0] (~chris_s@ip-095-223-073-220.um35.pools.vodafone-ip.de)
15:25:57chris_s_bilgus: re the "Return to opened playlists" patch:
15:26:05chris_s1) Trying to figure out what you meant by this. Could you explain a bit more?
15:26:12chris_s"it pulls you directly to the WPS when you try to go back to the playlist catalog after its already playing" ?
15:26:20chris_s2) By selecting a playlist different from the one previously selected in the "Add to Playlist" dialog, in theory, it should reset the selected playlist *and* the selected index in that playlist (to 0), but please let me know if you're seeing something else.
15:26:32chris_sThis side effect is a bit quirky, but it's basically existing behavior.
15:26:39chris_sI guess it has the advantage that after you've added something to a playlist, you can go the Playlist Catalogue and immediately see *that* playlist selected.
15:26:46chris_sIt may be worth it to change it so that a selection in the dialog doesn't affect the regular menu.
15:46:49chris_sI still want to add a way (which the other patch is for that's still a WIP – in terms of the manual needing to be updated) to be able to return from the WPS to the Playlist Catalogue directly, i.e. basically (at least optionally) treat it, like Files and the database, as just another "browser".
15:47:25chris_sg4627
15:48:02***Saving seen data "./dancer.seen"
15:50:42chris_sin a way, this is just a prerequisite for that, so that you can return from the WPS to the playlist you had selected
15:59:33 Join speachy [0] (~speachy@tunnel658658-pt.tunnel.tserv12.mia1.ipv6.he.net)
15:59:33 Quit speachy (Changing host)
15:59:33 Join speachy [0] (~speachy@rockbox/developer/speachy)
15:59:33Mode"#rockbox +v speachy" by ChanServ (ChanServ@services.libera.chat)
17:00
17:04:46 Quit Serke_ (Remote host closed the connection)
17:09:50 Join Serke [0] (Serke@176.212.168.111)
17:12:32 Quit reductum (Quit: WeeChat 2.8)
17:13:16 Join reductum [0] (~reductum@2603-8000-b400-8764-dea6-32ff-fe16-a622.res6.spectrum.com)
17:21:56 Join tomato [0] (~tomato@user/tomato)
17:32:00 Quit speachy (Quit: WeeChat 3.6)
17:48:05***Saving seen data "./dancer.seen"
18:00
18:08:37 Quit Serke (Remote host closed the connection)
18:47:43 Quit lebellium (Quit: Leaving)
19:00
19:37:33 Quit chris_s (Quit: Connection closed)
19:48:09***Saving seen data "./dancer.seen"
21:00
21:01:04 Quit tomato (Quit: The Lounge - https://thelounge.chat)
21:12:13 Join speachy [0] (~speachy@gw-tf.shaftnet.org)
21:12:14 Quit speachy (Changing host)
21:12:14 Join speachy [0] (~speachy@rockbox/developer/speachy)
21:12:14Mode"#rockbox +v speachy" by ChanServ (ChanServ@services.libera.chat)
21:12:53speachyOk, just spent $145.27 of the rockbox slush fund on a 30GB ipod 5.5g, an iflash msata v10, and a 256GB m.2 mSATA SSD. Should get everything by this coming weekend.
21:18:14_bilguschris_s I was referring to how it pulls you back into the WPS when you visited the playlist catalog but that appears to be connected to autoplay bookmarks and was the previous behavior
21:18:50_bilgusspeachy nice!
21:20:18hactar|antspeachy msata is bigger than m.2
21:20:23hactar|anthttps://www.titancomputers.com/v/vspfiles/assets/images/M.2_and_mSATA_SSDs_comparison.jpg
21:20:41hactar|antboth sata and nvme use the m.2 form factor
21:32:54_bilgusok so the bug behavior I observed is there 03 2022 further back I haven't reprod it yet but its painfully slow to do so
21:33:10 Join tomato [0] (~tomato@user/tomato)
21:48:12***Saving seen data "./dancer.seen"
22:00
22:07:27speachyI was careful to pick up the m.2 variant for this, as that's what the latest bug report is about.
22:08:05speachythe old mSATA drives I have lying around are the older form factor (iirc shared with mini-pci-express)
22:10:39_bilgusrepro with 05-2021
22:11:44_bilgusI think i'll try 3.15 next if that still shows I won't bother with a bisect
22:28:25speachythere's no shortage of old bugs
22:28:40speachyso there's no need to keep introducing new ones, eh?
22:33:41_bilgusoh? missed the memo :p
22:34:54_bilgusah crap I don't think 3.15 is multiboot aware
22:35:30_bilgusah just not with a custom folder
23:00
23:15:41_bilgusgod that was painful yes I can confirm it loses the index in 3.15 as well
23:16:13_bilgusit is amazing how slow 3.15 is compared to now
23:20:23speachy@!$#@!$ now I'm getting an apparently unrelated panic. I really should revert to the old toolchain just to be sure it's due to my code changes.
23:23:37_bilgusnow time to go searching for this bug wish I could get it in the sim in a more consistent manner
23:31:21speachytime for me to give up for the day.
23:31:26speachyg'nite
23:31:28 Quit speachy (Quit: WeeChat 3.6)
23:32:53_bilgusnn
23:37:16 Quit m01 (Quit: Konversation terminated.)
23:39:30 Join m01 [0] (~quassel@vps-b172b88b.vps.ovh.net)
23:42:28 Join othello7 [0] (~Thunderbi@pool-100-36-133-223.washdc.fios.verizon.net)
23:48:15***Saving seen data "./dancer.seen"

Previous day | Next day