--- Log for 17.09.121 Server: silver.libera.chat Channel: #rockbox --- Nick: rb-logbot Version: Dancer V4.16 Started: 14 days and 12 hours ago 00.35.29 Quit smithjd (Ping timeout: 268 seconds) 01.43.24 *** Saving seen data "./dancer.seen" 02.47.38 Join lebellium [0] (~lebellium@pop.92-184-118-123.mobile.abo.orange.fr) 03.43.28 *** No seen item changed, no save performed. 04.10.27 Quit ufdm (Ping timeout: 240 seconds) 04.19.33 Join mendel_munkis [0] (~mendel_mu@ool-ae2cb218.dyn.optonline.net) 04.20.10 Quit munkis (Read error: Connection reset by peer) 04.25.07 Join lebellium_ [0] (~lebellium@pop.92-184-118-123.mobile.abo.orange.fr) 04.27.58 Quit lebellium (Ping timeout: 268 seconds) 04.29.10 Quit lebellium_ (Ping timeout: 240 seconds) 05.16.02 Join munkis [0] (~mendel_mu@ool-ae2cb218.dyn.optonline.net) 05.17.42 Quit mendel_munkis (Ping timeout: 268 seconds) 05.21.43 Join mendel_munkis [0] (~mendel_mu@ool-ae2cb218.dyn.optonline.net) 05.22.45 Quit munkis (Ping timeout: 260 seconds) 05.24.49 Join lebellium [0] (~lebellium@pop.92-184-97-231.mobile.abo.orange.fr) 05.35.15 Quit lebellium (Quit: Leaving) 05.43.31 *** Saving seen data "./dancer.seen" 05.55.32 Quit mendel_munkis (Ping timeout: 268 seconds) 06.19.35 Join amachronic [0] (~amachroni@user/amachronic) 06.36.45 # sigh... it seems I'm suffering from the dreaded cache problems with USB. 06.37.49 # it seems mass storage keeps hitting stuff in the cache which does not belong there, for no apparent reason 06.48.20 # at least i've managed to uncover a public datasheet which documents the dwc2 usb controller: 06.48.23 # https://www.silabs.com/documents/public/reference-manuals/efm32hg-rm.pdf 06.49.08 # from the sounds of it, the way we're handling setup packets in the usb driver is semi-broken. 06.51.46 # the way we're doing it in the common usb code is pretty confusing and error prone, which doesn't help either. 06.54.06 # so, i'm no closer to getting DMA working but I /think/ the problem is down to the setup packet handling and cache issues. 06.55.38 # Sorry if this is too offtopic for this channel... but just for the lulz ;) https://twitter.com/letshugbro/status/1438110445613760513 07.04.34 Join ufdm [0] (~ufdm@c-73-164-63-214.hsd1.mn.comcast.net) 07.18.59 Quit amachronic (Quit: amachronic) 07.43.32 *** Saving seen data "./dancer.seen" 07.47.23 Join massiveH [0] (~massiveH@ool-18e4e82f.dyn.optonline.net) 09.05.10 Quit massiveH (Quit: Leaving) 09.43.33 *** Saving seen data "./dancer.seen" 09.59.22 # fuzev1 usb storage mode is broken in master 10.51.30 Join smithjd [0] (~enderw@node-1w7jra28yypte4b1rejm4lqiw.ipv6.telus.net) 10.58.56 Join smithjd_ [0] (~enderw@d199-126-230-79.abhsia.telus.net) 10.59.20 Quit smithjd (Ping timeout: 268 seconds) 11.09.41 Join ZincAlloy [0] (~Adium@2a02:8108:943f:d824:8ce4:b2e5:939d:7244) 11.43.36 *** Saving seen data "./dancer.seen" 11.50.38 # <_bilgus> paulcarroty, define broken? like it won't connect with usb? 11.51.20 # <_bilgus> does it work in 3.15? 11.51.41 # _bilgus only charging. working fine in 3.15 11.55.16 # <_bilgus> oi lots of USB stuff since then too :/ 11.57.14 # <_bilgus> might even be the usb prompt messing up if charging still works 11.58.14 Quit smithjd_ (Ping timeout: 260 seconds) 12.02.41 Join LambdaCalculus37 [0] (~LambdaCal@191.96.185.143) 12.13.25 Join amachronic [0] (~amachroni@user/amachronic) 12.15.44 # paulcarroty: not sure if the fuze has it, but does it work if you turn off USB HID? 12.23.56 # nope, nothing changes 12.31.10 # have you tried bisecting it? 12.55.25 # https://github.com/Rockbox/rockbox/commit/0400a391d3138800752c4301070c65707043034d https://github.com/Rockbox/rockbox/commit/d8d37ffdb8bed39df5d021afa9e09bf087b45d0d 12.56.40 # maybe it happened after these commits, can't find nothing fuze-related 13.21.21 Quit amachronic (Quit: amachronic) 13.43.38 *** Saving seen data "./dancer.seen" 14.27.04 Quit jschwart (Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) 14.32.30 Join jschwart [0] (~quassel@2001:985:2c6e:0:b00b:32ff:fe28:5567) 14.45.11 Quit jschwart (Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) 14.47.25 Join jschwart [0] (~quassel@2001:985:2c6e:0:b00b:32ff:fe28:5567) 14.48.27 Quit jschwart (Client Quit) 15.02.29 Join jschwart [0] (~quassel@2001:985:2c6e:0:b00b:32ff:fe28:5567) 15.02.35 # paulcarroty: just confirming that the usb mode stuff is not set to prompt. 15.14.22 Join S|h|a|w|n [0] (~shawn156@user/shawn/x-4432647) 15.20.15 # in other news, things are staged for the forums to move. 15.37.37 # tested all usb options, also loaded sansa firmware for port&cable check 15.42.31 Join johnb2 [0] (~johnb2@p5b3afbae.dip0.t-ipconnect.de) 15.43.39 *** Saving seen data "./dancer.seen" 15.45.02 # e09df1ce5b-210403 is already broken for me. 15.46.31 # That is the version I had on my fuze. I thought it might be my usb ports, but with 3.15 it's fine. 15.46.49 # USB 2.0 or 3.0 port? 15.47.00 # 2.0 15.47.33 # I have a windows laptop. It fails to recognize the player. 15.47.54 # pops up the error message 15.51.18 # well, fine is relative: wiht 3.15 the player is recognized, but trying to copy many files, the transfer aborts somewhere in the middle. 15.51.37 # also a fuzev1? 15.51.48 # yes, v1 15.51.54 Join smithjd_ [0] (~enderw@d199-126-230-79.abhsia.telus.net) 15.54.18 # hmm. The only sansa device I have here is an AMSv2 and seems to work okay 15.58.43 # AMSv2 is fine (my clip+), but my e200v2 (AMSv1) seems to have the same problem: not being recognized. 16.05.13 Quit smithjd_ (Ping timeout: 265 seconds) 16.05.38 # ok, so it's isolated to AMSv1 targets, that does help a lot 16.06.15 # hmm. I actually do have a fuzev1 in my pile of broken dreams.. it ought to work okay for this purpose 16.06.58 # (don't actually need working controls to test USB...) 16.10.51 # okay to try some bisected builds? 16.11.29 # https://www.shaftnet.org/~pizza/fuzev1-test1.zip 16.16.14 # This works 16.18.40 # ok. next up: https://www.shaftnet.org/~pizza/fuzev1-test2.zip 16.21.32 # works,too. 16.22.49 # the battery on my v1 is too drained to do anything with yet 16.23.10 # ok, that's roughly 2020-10-03 :) 16.23.47 # so down to a mere 6 month window 16.24.30 # ok, #3: https://www.shaftnet.org/~pizza/fuzev1-test3.zip 16.25.31 # (heh, my v1 has 8ecbcad-121226 :) 16.27.30 # That is broken 16.28.56 # #4: https://www.shaftnet.org/~pizza/fuzev1-test4.zip 16.30.39 Join smithjd_ [0] (~enderw@d199-126-230-79.abhsia.telus.net) 16.31.22 # broken 16.31.47 # narrows it down to about 3 weeks in October 2020 16.32.51 # next up: https://www.shaftnet.org/~pizza/fuzev1-test5.zip 16.33.07 Quit smithjd_ (Read error: Connection reset by peer) 16.33.13 # ... I wonder if this is fallout from the GCC 4.9 migration 16.34.07 # (or more likely, the switch to -Os by default) 16.35.10 Join smithjd_ [0] (~enderw@2001:56a:f947:bf00:352f:c44e:58c4:c588) 16.35.20 # broken 16.37.20 # 2next up: https://www.shaftnet.org/~pizza/fuzev1-test6.zip 16.38.20 Quit asabas (Quit: Relay server offline) 16.40.16 # works 16.40.37 # it's all but guaranteed to be the -Os change. 16.41.37 # https://www.shaftnet.org/~pizza/fuzev1-test7.zip 16.43.01 # (remaining things in the gap are some hosted port work, h1x0 bootloader fixes, and the toolchain bump 16.44.01 # works 16.46.46 # https://www.shaftnet.org/~pizza/fuzev1-test8.zip 16.49.20 # works 16.52.20 # nearly no differences between v1 and v2 as far as the USB code is concerned... 16.52.30 # but there are some 16.52.57 # https://www.shaftnet.org/~pizza/fuzev1-test9.zip 16.56.36 Quit LambdaCalculus37 (Ping timeout: 268 seconds) 16.57.36 Quit ZincAlloy (Quit: Leaving.) 16.59.14 Quit johnb2 (Ping timeout: 260 seconds) 16.59.28 Join johnb2 [0] (~johnb2@p5b3afbae.dip0.t-ipconnect.de) 16.59.57 # works 17.02.00 # https://www.shaftnet.org/~pizza/fuzev1-test10.zip 17.02.07 # (This I expect to fail) 17.04.39 Join LambdaCalculus37 [0] (~LambdaCal@191.96.185.143) 17.05.20 # correct, it fails. 17.06.02 # https://www.shaftnet.org/~pizza/fuzev1-test10a.zip 17.06.25 Join asaba [0] (~asabas@103.113.159.250) 17.06.38 # (identical to the previous one, but backs off the optimization flag change) 17.09.02 # works 17.11.11 # okay. 17.11.58 # can you compile for e200 v2, too, for completeness? 17.11.59 # Back onto master. Let's try this: https://www.shaftnet.org/~pizza/fuzev1-test11.zip 17.14.35 # (this uses the same usb static buffer logic for v1 as v2 uses) 17.17.16 # it fails 17.21.18 # ok. so much to the easy fix. 17.22.11 # https://www.shaftnet.org/~pizza/e200v2-noopt.zip 17.22.13 # that should work 17.23.23 # https://www.shaftnet.org/~pizza/fuzev1-noopt.zip 17.23.33 # both are current git master with the optimization flags changed 17.24.37 # delving into _what_ is going wrong will take more work that I probably won't be able to get further into tonight. 17.26.15 # e200 works 17.27.06 # Alright, thanks so far. 17.27.22 # Then I will quit for today. 17.28.20 Quit johnb2 (Quit: Nettalk6 - www.ntalk.de) 17.32.41 Quit LambdaCalculus37 (Quit: Quit) 17.43.40 *** Saving seen data "./dancer.seen" 18.22.42 # <_bilgus> ill throw up a coverity build maybe we will get lucky 19.06.07 # @bilgus: New comments on the Database patches. A bug was fixed in the other patch and all three fields have now been tested to work. 19.08.08 # I tried to make canonicalartist into a virtual tag but Rockbox wasn't having any of that. I kept getting a "Database not ready" error when attempting to use the entry. 19.13.26 Nick smithjd_ is now known as smithjd (~enderw@2001:56a:f947:bf00:352f:c44e:58c4:c588) 19.43.42 *** No seen item changed, no save performed. 19.45.39 # <_bilgus> smithjd, I don't think you chnaged anything in that second patch set 19.47.33 # <_bilgus> I think the backend uses artist or fallsback to album artist might have the order reversed 19.48.01 # <_bilgus> I think I discovered this when I was messing with pictureflow 19.48.12 # <_bilgus> so might be specific to that as well 19.51.47 # _bilgus: could you do a coverity run with the fuzev1 or some other amsv1 target? 19.52.05 # <_bilgus> its already up 19.52.09 # awesome! 19.56.37 # @bilgus: I made the fix and rebased the branch. It came down to leaving the artist and albumartist fields in the sort define. 20.14.32 Quit braewoods (Quit: WeeChat 2.8) 20.14.53 Join braewoods [0] (~braewoods@user/braewoods) 20.17.11 Quit _bilgus (Quit: Leaving) 20.17.25 Join _bilgus [0] (~bilgus@162.154.213.134) 20.25.16 # <_bilgus> smithjd, ok 20.28.17 # pictureflow looks broken in master. I don't get any cover art and entries are interspersed with albums. 20.30.19 # i'm going to update the patchset for pictureflow and I think change the name of tag_canonicalartist to tag_virt_canonicalartist. 20.40.47 Join massiveH [0] (~massiveH@ool-18e4e82f.dyn.optonline.net) 20.49.43 # <_bilgus> pictureflow works for me at head 20.50.45 # on your device or in the sim? 20.50.57 # <_bilgus> sim 21.05.50 Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) 21.09.07 # <_bilgus> and on the device too 21.09.17 # <_bilgus> Fuze Plus @ Head 21.10.03 # <_bilgus> most likely your albums arent set up right 21.10.34 # <_bilgus> though I do see a few ones which I thought I had gotten rid of (making sane fallbacks) 21.10.56 # <_bilgus> cover art needs to be named specifically as well 21.11.13 # <_bilgus> such as cover100x100.bmp IIRC 21.11.23 # fiio m3k. make clean'ed, still no pictures. sim and @HEAD 21.11.57 # <_bilgus> cover_100x100.bmp 21.12.04 # embedded coverart here. 21.12.12 # <_bilgus> it doesn't use it 21.12.21 # <_bilgus> never worked 21.12.53 # <_bilgus> could it work yes but currently no 21.22.57 # OK, new patch series up. DB works as expected and PictureFlow doesn't segfault or anything. 21.23.59 # Looks like nowhere else used tag_albumartist. 21.39.29 # <_bilgus> I don't see anything that makes me go BAH but I'm not familar with this code and its a MF'ing large mess each time I jump in it 21.39.49 # <_bilgus> so that being said you ready to push it smithjd? 21.40.03 # yep. push away. 21.42.23 # <_bilgus> ok I think they should just go 21.43.43 *** Saving seen data "./dancer.seen" 21.44.52 # thanks again for the review! 21.45.30 # <_bilgus> np lets hope it works :) 21.45.48 Quit smithjd (Quit: Konversation terminated!) 21.46.04 # <_bilgus> IDK if it will just bunch them all together or do them separated 21.50.36 # <_bilgus> speachy the more I look at this the more it looks wrong https://scan3.coverity.com/reports.htm#v55173/p14259/fileInstanceId=59588457&defectInstanceId=10979166&mergedDefectId=356971 21.51.55 # <_bilgus> it is a 32 bit struct 22.03.39 # <_bilgus> huh it apparently works 22.18.27 Join dconrad [0] (~dconrad@208.38.228.17) 22.20.00 Quit S|h|a|w|n (Read error: Connection reset by peer) 22.20.09 Join S|h|a|w|n [0] (~shawn156@user/shawn/x-4432647) 22.33.29 # yeah, it can't quite grok the aliasing going on. 22.34.21 # it might need to be marked as explicitly volatile though? 22.42.35 # I have a USB sniffer here, might be useful to see what actually happens over the wire 22.42.42 # and work back from that 22.59.20 # <_bilgus> yeah timing sounds like something that would be quite apparent 23.17.26 # if anybody here feels like testing g#3777 yes/no keymapping on Q1, Sansa Connect, HM60x, HM801, ihifi, or MA, I haven't been able to test them on sim for various reasons 23.43.46 *** Saving seen data "./dancer.seen"