IRC log of #maemo-ssu for Tuesday, 2012-07-10

*** BCMM has joined #maemo-ssu00:11
*** lizardo has quit IRC00:21
*** _ade_ has quit IRC00:33
*** luf has quit IRC00:38
amiconnBtw, what will happen right after installing thumb stuff (with a kernel including the workaround) before reboot?00:39
amiconnAt that time, thumb compiled binaries are already in place, but the running kernel doesn't have the workaround...00:39
kerioamiconn: that's why you first install and run the kernel, and *then* the rest of cssu-thumb00:42
kerioat least, that's what i'd do00:43
kerioanyway, upgrading mp-* makes you reboot anyway00:43
*** Vlad_on_the_road has quit IRC00:53
amiconnFirst installing the thumb kernel, and then rebooting before installing other thumb stuff makes sense, but does ham enforce this?00:56
kerioi doubt it00:57
kerioit's not like "the thumb kernel" makes sense00:57
amiconnI know I have to reboot after upgrading mp-*, but that happens after the upgrade finished (and *may* start newly installed binaries)00:57
amiconnYeah, kernel with thumb errata workaround in place00:58
kerioi don't think HAM is equipped to deal with what's almost like a change in architecture00:58
amiconnImo the workaround should be added to all kernels if possible (power and stock), and be always be enabled if this is ever going to be published to a repo with 'stable' in its name01:01
keriofreemangordon: wait, why is thumb microB faster than the ARM one? isn't it supposed to be slower?01:04
tadzik I assume it's swapping far less often01:05
*** andre__ has quit IRC01:06
DocScrutinizer05amiconn: you probably haven't understood the implications of what this thumb-erratum-workaround does01:08
amiconnI think I did. It flushes the branch cache on context switch01:09
DocScrutinizer05yep01:09
DocScrutinizer05which will slow down things, for everybody01:09
kerioD:01:09
keriomadness01:09
amiconnYes, by how many permille?01:09
keriowell, it's worth it if you're using thumb2 binaries01:10
amiconnYou'll notice when running a benchmark, but otherwise?01:10
DocScrutinizer05up to you to come up with numbers, it's you who demands the patch should go into any available kernel and thus users have no more choice about to use it or not01:10
amiconnI think the speed gain by fewer cache misses and less swapping is much higher than the little speed loss01:10
kerioDocScrutinizer05: well, the fix is enabled by default in KP5101:11
DocScrutinizer05"I think" is a weak argument01:11
amiconnReal-world speed gain, that is01:11
keriobut nothing prevents us from making it disabled by default in the future omap101:11
DocScrutinizer05sure, kerio. That's why I suggested to get that switch to enable/disable it01:12
kerioamiconn: well, it would only give you a slowdown if you don't install thumb2 binaries01:12
amiconnI'd rather accept the speed loss for everybody that create a situation where someone might end up with an unbootable system01:13
amiconns/that/than/01:13
infobotamiconn meant: I'd rather accept the speed loss for everybody than create a situation where someone might end up with an unbootable system01:13
DocScrutinizer05mhm01:13
kerioamiconn: that's stupid01:13
keriothe n900 is slow enough as it is01:13
DocScrutinizer05so why are you thinking about forcing a new kernel on everybody then, even on those users who think they don't want to take that risk of having an unbootable system due to something they don't want to use?01:14
amiconnYou may not have read my entire sentence. I said "..if this is ever going to be published to a repo with 'stable' in its name"01:15
amiconn"this" referring to "thumb compiled binaries, of course01:15
DocScrutinizer05Imo the workaround should be added to all kernels if possible (power and stock),01:15
DocScrutinizer05aaah, ok01:16
DocScrutinizer05no, thumb compiled binaries probably won't go to any repo with a plain "stable" without "thumb" in the name, for that very reason01:17
DocScrutinizer05a lot of users don't see this as an improvement of stability of the system01:17
DocScrutinizer05and so they might prefer to stay with a plain ARM system01:17
kerioDocScrutinizer05: but freemangordon said it's ballzy fast!01:19
DocScrutinizer05quite a number of *very* experienced developers still doubts stability of thumb on N900 at large01:19
keriofor some value of ballzy01:19
DocScrutinizer05kerio: I for one am fine with current speed of my N90001:19
RaimuFWIW I haven't seen slowdown with the fix-inc'd kernel.01:20
DocScrutinizer05yeah, expected01:20
DocScrutinizer05the overhead during context switching is negligible _under normal conditions_01:20
* amiconn will very likely migrate to thumb01:20
RaimuBut no, just looking at that depressing tmo thread it's not for general use.01:20
amiconnNot so much because of the speed gain, but because of the reduced memory footprint01:21
kerioRaimu: the depressing tmo thread is caused by morons01:21
kerioand by the fact that it's TMO01:21
DocScrutinizer05amiconn: sane rationale01:21
Raimukerio: Well, yes, morons run this world.01:21
amiconnItt seems the OOM killer kicks in quite often on the N90001:21
DocScrutinizer05duh! really?01:22
* amiconn should check the logs01:22
RaimuDocScrutinizer05: What sort of conditions are the abnormal ones?01:22
DocScrutinizer05should happen no earlier than after it ate all your swap01:22
DocScrutinizer05~dict abnormal01:22
infobotDictionary 'abnormal' (1 of 5): aberrant, abominable, absurd, amorphous, anomalistic, anomalous, atrocious, atypical, bereft of reason, brainsick, crackbrained, cracked, crank, crankish, cranky, crazed, crazy, criminal, crotchety, daft, delinquent, deluded, demented, deprived of reason, deranged, deviant, deviative, different, disgraceful, disoriented, disproportionate, distraught, divergent, dotty, eccentric, erratic, evil, exceptional, fey, ...01:22
RaimuDocScrutinizer05: Haw, I was thinking more about an example of a situation where the switching would do something really unexpected.01:23
DocScrutinizer05do you suppose I expect the unexpected? :-D01:23
Raimu:D01:24
RaimuFair enough.01:24
RaimuYou do have sage powers.01:24
DocScrutinizer05I *could* figure something related to unusual IRQ load, created by whatever, e.g. accelerometer01:25
DocScrutinizer05exactly the situations where you don't like to see additional overhead slowing down system response time01:25
DocScrutinizer05but that's just cloudy visions01:26
DocScrutinizer05range of context switches per second (when really active system, not in idle) may vary by a factor 5001:27
DocScrutinizer05usual benchmarks are completely unsiuted for testing any impact of additional overhead on context switching01:28
DocScrutinizer05as context switches are mostly related to interactive actions at large01:29
keriomisread that as "unsluted"01:29
keriohm, is there a way to only apply the fix selectively, on the thumb2 processes?01:31
freemangordonno01:31
freemangordonbecause errata hits no matter ARM or thumb01:31
DocScrutinizer05that's not how this workaround works01:31
DocScrutinizer05exactly01:31
keriohm, perhaps beginning to apply the fix once the first thumb2 binary is started?01:32
freemangordonwhy?01:32
DocScrutinizer05kerio: simply forget it01:32
kerioso there's no risk of running thumb2 binaries without it, but there won't be any slowdown on a system that doesn't run any thumb2 binary01:32
RaimuThere's the "disable this" switch, already.01:33
RaimuIsn't there?01:33
freemangordonyep01:33
DocScrutinizer05Raimu: yes, I suggested that01:33
RaimuI remember.01:33
DocScrutinizer05but it's not meant to get used "dynamically"01:33
freemangordonwell, actually you can enable/disable the workaround whenever you like. but I don't see much use of that01:34
DocScrutinizer05and after all afaik there's no flag in ELF header about "I'm a thumb binary"01:34
kerioi see01:34
kerioif freemangordon actually delivers an omap1 with the errata workaround, should it be enabled by default, or disabled by default?01:35
DocScrutinizer05enabled of course01:35
kerios/omap1/omap1, ABI-compatible with the stock kernel,/01:35
infobotkerio meant: if freemangordon actually delivers an omap1, ABI-compatible with the stock kernel, with the errata workaround, should it be enabled by default, or disabled by default?01:35
kerioDocScrutinizer05: but slowdowns and stuff, for normal people!01:35
DocScrutinizer05well, for the stock replacement "clean" kernel it probably should be disabled01:36
freemangordonkerio: benchmark that slowdown, if we have 1mS slowdown per context switch, how that affects "normal people"01:36
keriofreemangordon: idk, ask doc01:37
DocScrutinizer05but generally if you want to bootup a thumb system you don't want to bother about the first few binaries to be ARM until you get around to enable the workaround01:37
RaimuThe only reason not to disable it by default on clean kernel would be to pre-empt the "oh shit we released a thumb2 binary" hassle01:37
freemangordonker: trust me, we had lots of discussions with doc ;)01:37
freemangordonkerio: ^^^01:37
DocScrutinizer05freemangordon: 1ms / context switch? will bring system to a grinding halt!01:38
freemangordonfor the last 8 or so months01:38
kerioi hope he meant 1µs01:38
kerioor something01:38
freemangordonwhy? we have HZ set to 12801:38
DocScrutinizer05freemangordon: context switches are largely unrelated to HZ01:38
freemangordonand have in mind this does not kick in for IRQ/FIQ ;)01:39
freemangordonworkaround is only for userland processes AIUI01:39
DocScrutinizer05but for any function that runs into wait01:39
freemangordonDocScrutinizer05: well. I have NFC what is the impact, what I know is that it is not visible01:40
DocScrutinizer05HZ is the *minimum* taskswitching frequency on a system with >=2 ready processes01:40
freemangordonok, ok, it was a bad example01:41
DocScrutinizer05freemangordon: I'm tending to agree01:41
DocScrutinizer05I never had any actual concerns about BPB flushing overhead01:41
DocScrutinizer05the debate been about why we'd want to force it on *all* N900 users01:42
freemangordonRaimu: BTW I am uploading thumb microbengine in the repo, just gimme a couple of minutes to check if it works. Will have to do apt-get install microb-engine though, I just don;t have time to play with mp- thing now :)01:42
DocScrutinizer05and I mentioned to amiconn that there are valid reasons why some users might not want to have it01:42
freemangordonDocScrutinizer05: I would, we know there are thumb-compiled binaries distributed by Nokia01:42
Raimufreemangordon: Oh! Let me know when up.01:43
freemangordonok01:43
freemangordonDocScrutinizer05: there is at least one. And noone knows what is the actual amount of such mistakenly build binaries01:44
DocScrutinizer05freemangordon: yeah, if that gets verified it is the bummer of the year, regarding "Nokia fecked it up"01:44
freemangordonDocScrutinizer05: you can bet your ass on that01:45
chem|sthail I hit a BT bug01:45
freemangordonI bet mine too ;)01:45
chem|stafter removing a client from list of devices my BT refuses to work01:45
chem|stblank name shown bluetoothd segfaulting etc01:46
DocScrutinizer05I'd actually tend to claim THEN we either need to deliver ARM builds for all those POS, or actually ship kernel with thumb-fix01:46
freemangordonchem|st: :(01:46
chem|stexactly!01:46
DocScrutinizer05hi chem|st01:46
*** MrPingu has quit IRC01:47
chem|sttmo says flashing fixed it... <- don't like01:47
chem|stany other ideas?01:47
chem|sthi DocScrutinizer0501:47
freemangordonchem|st: yeah, file a bug and wait merlin1991 to return from greece01:47
DocScrutinizer05err, I guess "reboot" wouldn't earn me a "thanks" ;-)01:47
chem|stDocScrutinizer05: you owe me a beer now...01:48
chem|st:)01:48
DocScrutinizer05;-P01:48
keriochem|st: restore your backupmenu backup01:48
freemangordonchem|st: also you may want to backup your /var/lib/bluetooth (so we can reproduce the bugger) and after that delete it01:48
DocScrutinizer05wait, what exactly went awry?01:48
freemangordonchem|st: lemme check the exact location01:48
* DocScrutinizer05 ponders reflashing BT firmware01:49
chem|stDocScrutinizer05: that was my last-resort idea01:49
freemangordonchem|st: yeah, delete everything under /var/lib/bluetooth01:49
freemangordonand then do:01:49
freemangordonstop bluetoothd01:50
freemangordonstart bluetoothd01:50
freemangordonwait, first stop bluetoothd and then delete the directory01:50
freemangordonchem|st: ^^^01:50
DocScrutinizer05hehe01:50
DocScrutinizer05better sequence01:51
freemangordonyeah, bluetoothd flushes the files when stopped01:51
DocScrutinizer05I'd also mv instead rm01:51
freemangordon <freemangordon> chem|st: also you may want to backup your /var/lib/bluetooth01:51
freemangordonDocScrutinizer05: :P01:51
DocScrutinizer05:nod:01:51
DocScrutinizer05chem|st: weird idea and probably nonsense: remove main battery01:52
chem|stDocScrutinizer05: that works for gsm modem and I tried that01:53
DocScrutinizer05thought as much :-S01:53
chem|stfreemangordon: should I try to make it broke again when I have it working again?01:53
freemangordonchem|st: just try what i said, that should solve your segfaulting bluetoothd. though you will have to re-pair all of your devices01:54
DocScrutinizer05I'd have thought bluetoothd "flashes" firmware (if any) on each powerup of BT chip01:54
DocScrutinizer05freemangordon: indeed01:54
freemangordonchem|st: yep, that is why I said to backup /var/lib/bluetooth, so we will be able to reproduce01:54
DocScrutinizer05freemangordon: I bet you're right on that01:54
DocScrutinizer05probably one of the paired device files contains crap01:55
chem|stDocScrutinizer05: +101:56
freemangordonRaimu: your are good to apt-get update01:56
freemangordonapt-get install microb-engine microb-engine-common01:56
freemangordonhmm, or you may try with apt-get upgrade, not sure about dependencies01:57
freemangordonRaimu: ^^^01:57
keriofreemangordon: won't that horrendously break metapackages that depend on exact versions? D:01:57
freemangordonno01:57
keriok01:58
keriowait, so why won't just "upgrade" work?01:58
freemangordonbecause I don't know what repos has he enabled01:58
kerioi see01:58
freemangordonnot that Raimu ia a nood, but just in case01:58
freemangordon*noob01:58
kerioanyway, i'm waiting for Pali to push KP51 to -devel and then i'll probably try this thumb thingy01:58
DocScrutinizer05Nokia legacy: dependcies a maga pile of mess01:58
DocScrutinizer05mega even01:59
kerionight, everyone01:59
chem|stfreemangordon: sdptool browse did show some audio services while bt was not working02:00
RaimuHehe. I'll manage.02:01
chem|stand for sure not the same as I see now when turned on^^02:01
freemangordonchem|st: maybe it is not using bluetoothd at all, but kernel, i.e, socket stuff02:01
chem|stusually there is no sdptool browse output if device is off02:02
chem|sttry yourself02:02
freemangordonRaimu: please tell me what apt-get reports for "xxx les bytes will be used"02:03
freemangordonchem|st: which device?02:03
chem|stbt02:03
freemangordonyou mean to stop bluetoothd and to run discovery using sdptool?02:04
chem|styes02:04
freemangordonchem|st: not now :)02:04
chem|stI had some weird stuff showing, like the output of line 10 to 20 (literaly)02:05
chem|stand not 1 to 4002:05
freemangordonI was supposed to be in the bed by now ;)02:05
chem|stfreemangordon: me too...02:05
chem|stas always02:05
freemangordon:D02:05
freemangordoncheers02:05
chem|stthanks for the var lib advice02:06
freemangordondid it help?02:06
freemangordonchem|st: ^^^02:07
freemangordonRaimu: are you updating? As I am wating your feedback before making an announcement :)02:11
chem|stfreemangordon: yeah worked02:13
chem|stseems like the profile got borked02:13
freemangordon:)02:14
freemangordonglad to hear02:15
*** BCMM has quit IRC02:15
Raimufreemangordon: backupped, now installing microb-engine and *-common02:15
RaimuAnd a quick reboot...02:16
Raimufreemangordon: Yes, works.02:19
RaimuWay too quick to comment on usage, but everything runs. Announce away, I guess.02:19
freemangordonnice :)02:19
freemangordonhow is the memory usage?02:19
freemangordonBTW: [01:57] <freemangordon> Raimu: please tell me what apt-get reports for "xxx les bytes will be used"02:20
*** trbs has quit IRC02:22
RaimuOh, crap. I didn't see that line. Where's the apt log?02:22
freemangordonthere is no :(02:23
RaimuI'm sorry. :(02:23
freemangordonNP02:23
RaimuSome other user has to fill you in on that one.02:23
freemangordon:) yeah02:26
RaimuOh, this thing definitely isn't slower.02:40
freemangordonisn't slower? WTF man, it is FAST :P02:41
RaimuThat's what I was saying. :D02:42
freemangordon:D02:42
*** LaoLang_cool has joined #maemo-ssu02:43
*** M4rtinK has quit IRC02:43
*** taziff1 has quit IRC02:57
*** taziff has joined #maemo-ssu02:59
*** gri has quit IRC03:27
*** gri has joined #maemo-ssu03:29
*** nox- has quit IRC03:31
*** arcean_ has quit IRC03:41
*** LaoLang_coo_ has joined #maemo-ssu04:18
*** LaoLang_cool has quit IRC04:22
*** LaoLang_coo_ has quit IRC04:26
*** LaoLang_cool has joined #maemo-ssu04:27
*** gregoa has quit IRC04:37
*** Sc0rpius has quit IRC04:37
*** Scorpius has joined #maemo-ssu04:37
*** gregoa has joined #maemo-ssu04:44
*** amiconn_ has joined #maemo-ssu05:10
*** amiconn has quit IRC05:10
*** amiconn_ is now known as amiconn05:10
*** taziff1 has joined #maemo-ssu06:16
*** taziff has quit IRC06:18
*** kerio has quit IRC07:10
*** taziff has joined #maemo-ssu07:10
*** taziff1 has quit IRC07:12
*** kerio has joined #maemo-ssu07:14
*** taziff has quit IRC07:33
*** taziff has joined #maemo-ssu07:34
*** MohammadAG has quit IRC07:38
*** xnt14 has quit IRC07:38
*** xnt14 has joined #maemo-ssu07:53
*** mirandir has joined #maemo-ssu08:39
*** mirandir has left #maemo-ssu08:39
*** Pali has joined #maemo-ssu08:58
* amiconn would be interested in browsermark results from thumb microb10:23
keriodammit Pali, hurry :(10:25
Palikerio, hi10:25
keriohi!10:26
freemangordonRaimu: http://talk.maemo.org/showpost.php?p=1235019&postcount=32510:28
kerioPali: where's my KP51 in -devel? :C10:29
freemangordonkerio: no KP51 for you :P10:29
DocScrutinizer05kerio: where's your report about testing KP51 from http://talk.maemo.org/showpost.php?p=1234313&postcount=44510:50
*** freemangordon has quit IRC10:56
*** wmarone has quit IRC11:21
*** wmarone has joined #maemo-ssu11:32
chem|stmoo12:19
chem|stafter I had my bt back last night I tried to pair it with my tv, no luck, with my desktop I only got filetransfer working12:20
*** arcean has joined #maemo-ssu12:29
*** ekze has joined #maemo-ssu12:41
*** Pali has quit IRC12:52
*** M4rtinK has joined #maemo-ssu13:16
*** arcean has quit IRC13:25
*** Pali has joined #maemo-ssu13:40
*** taziff has quit IRC14:04
*** taziff has joined #maemo-ssu14:04
kerioDocScrutinizer05: there isn't one14:07
keriobecause KP51 is not in the repos!14:07
DocScrutinizer05kerio: we won't put it into any repo until you have tested it ;-P14:09
kerio:(14:09
keriook14:09
kerioi tested it14:09
kerioeverything works14:09
PaliDocScrutinizer05, see bq27x patch14:09
DocScrutinizer05pling, another 10 credits off your jar14:10
DocScrutinizer05Pali: thanks14:10
kerioten? D:14:10
PaliDocScrutinizer05, you can add support for this sysfs entry to bq.sh script14:10
kerioanyway, is the kernel battery module better than bme?14:11
DocScrutinizer05Pali: I'll look into it14:11
DocScrutinizer05kerio: no14:11
Palikerio, battery module only report some informations14:11
DocScrutinizer05there's nothing as awesome as bme on God's great earth14:12
Palicharger module is for charging14:12
PaliDocScrutinizer05, I do not agree! UEFI secure boot is better then any BME...14:13
DocScrutinizer05hehe14:13
*** arcean has joined #maemo-ssu14:13
Palialso TPM for UEFI is better then some aegis's TPM :D14:14
kerioPali: is that information enough for advanced power monitor, though?14:15
kerioit doesn't necessarily use BME14:15
Palikerio, no idea what is advanced ...14:16
kerioit's a better battery monitor, it can use different "sources" of information14:16
kerioit replaces the battery icon in the status area and in the status menu14:16
Palibut charger + battery module provides all informations except real battery tempereture and real battery capacity14:17
Paliif you calibrate battery chip then reported capacity will be usefull too14:18
kerioso, bme is not needed then?14:18
kerioneat14:18
Palikerio, yes14:19
PaliI wrote needed kernel drivers14:19
Paliand freemangordon writing user space libs14:19
kerioand DocScrutinizer05 preventing you from adding that to cssu because some third party application could be relying on BME14:20
DocScrutinizer05STFU14:21
kerioDocScrutinizer05: <314:21
DocScrutinizer05you're close to either an /ignore or a +q14:23
DocScrutinizer05it really doesn't help your reputation to spread BS about me14:24
keriojoke's on you, i have no reputation14:25
kerio...that made me sad14:25
DocScrutinizer05if you're too ignorant to understand what CSSU founders had in mind, and you don't get it this isn't tzhe leetness competition distro but a extremely conservative bugfix deployment channel, you're really pathetic14:26
kerioanyway, still mad about the estel thing?14:26
kerio(from the KP50 thread)14:26
kerioi doubt most people even got the reference14:29
Lava_Croftyou're just ignorant! (c) Micheal Jackson14:30
* Lava_Croft runs14:30
kerio>micheal?14:30
Lava_Croftoh my a typo, im sorry aspell14:30
DocScrutinizer05Pali: nevertheless, it's a valid topic to think how we'd eventually deploy/offer the whole zoo of bq27x00.ko bq2415x.ko hlad-addon-bme_replacement.so, some tweaked initscripts and whatnot else14:36
DocScrutinizer05to make this work we need to switch all at once14:37
keriowell, you have a repo, you can provide updated versions for all the packages that hold files that you're going to change14:37
keriomake each one depend by version on the other, so there can't be any mismatch14:38
DocScrutinizer05not that a proper set of dependencies wouldn't catch most potential headache, but we need to think about it14:38
kerioDocScrutinizer05: do you want to completely remove bme, or just to disable it?14:38
kerios/just to/just/14:39
infobotkerio meant: DocScrutinizer05: do you want to completely remove bme, or just disable it?14:39
DocScrutinizer05I'll never ever remove anything unless we need the storage space14:39
PaliDocScrutinizer05, yes everything must be replaces at once. so I'm waiting for userspace libs14:39
DocScrutinizer05:nod:14:39
DocScrutinizer05Pali: will this become an "app" in the end? in extras-devel?14:39
DocScrutinizer05or in CSSU-extras optional?14:40
Paliand init scripts - only bme must be deleted and everyting which depends (in upstart/init) must be fixed14:40
kerioDocScrutinizer05: changing system files from another package is a Bad Idea™14:40
Palifirst it must work :D then we can start talking where will we put application :D14:40
DocScrutinizer05Pali: (initscripts) yes, I know14:40
DocScrutinizer05Pali: :-D14:41
Paliwe have n900 qemu image, so we can remove bme first here14:41
Paliand if maemo image in qemu will work, then we can do it on real n900 too14:41
DocScrutinizer05Pali: I guess some hal config needs tweaking too?14:42
DocScrutinizer05or will we replace hald-addon-bme14:42
DocScrutinizer05ß14:42
DocScrutinizer05?14:42
Palifreemangordon, wrote hald-addon-bme which use bq kernel drivers14:42
DocScrutinizer05is it meant to replace the original one, even same name and all?14:43
Paliand also wrote libbmeipc which is used by more maemo apps for battery temperature14:43
Palireplace original14:43
DocScrutinizer05k14:43
Palisame name, same properties14:43
DocScrutinizer05this means we need to ship it via cssu extras then14:43
Paliwe can create one deb package which dpkg-divert all these binaries14:44
DocScrutinizer05hmm14:44
Paliand package will unpack our open source14:44
Palithen it can go to extras (depends only on kernel-power or kernel-feature-***)14:45
DocScrutinizer05:nod:14:45
Palibut now in kp51 is /sys/ entry for speed of connected usb device, no need to dmesg | grep ....14:45
PaliI will write some status menu aplet for host mode14:46
DocScrutinizer05hey, kernel-feature-*** ? :-)you forgot to define a proper feature for it?14:46
Palikernel-feature-battery is for bq27x00_battery and kernel-feature-charger is for bq2415x_charger driver14:46
kerioit's kernel-feature-charger it hink14:46
kerioheh14:46
Palialready in kp5114:46
kerio*think14:46
PaliDocScrutinizer05, kp51 is prepaired for all14:47
DocScrutinizer05Pali: regarding this applet: I'd appreciate it was a generic applet via cmdline, so it can get invoked from h-e-n script.14:47
Paliyes, I'm thinging about aplet which will only call some shell script14:48
DocScrutinizer05aaah this way round14:48
DocScrutinizer05an applet with buttton in menu14:48
PaliI want to switch it via xterm too :D14:48
PaliI think blueled already wrote some applet in python14:49
DocScrutinizer05yeah, there is some stuff I lost track of14:49
DocScrutinizer05seems it needs some consolidation and refactoring/update for h-e-n et al14:50
Palido you have link?14:50
DocScrutinizer05putting your awesome work in kernel together with any applet, and speed autonegotiation and udev stuff14:51
DocScrutinizer05nope, sorry14:51
DocScrutinizer05no link14:51
DocScrutinizer05I lost track of all the h-e-n related stuff last 9 months14:51
keriough, my status menu is already full enough as it is :c14:52
DocScrutinizer05kerio: aiui it will *replace* the standard battery icon/button14:52
kerioDocScrutinizer05: but it's already replaced, here D:14:53
keriomeh, i'll just remove advanced-power-monitor, it's kinda bad anyway14:53
Paliudev stuff? for what?14:53
kerioit's not really intuitive to add usb host mode to the battery icon, but i suppose that it makes sense14:53
DocScrutinizer05Pali: usual linux way to mount stuff14:53
Palike-recv working fine14:54
DocScrutinizer05or rather, load drivers for attached new devices14:54
kerioDocScrutinizer05: oh, the reason there's a "enumerate" button in h-e-n?14:54
Paliloading usb drivers working too14:54
DocScrutinizer05ke-recv maybe working fine for usual storage stuff14:54
PaliI'm using my rj45 <--> usb without any problem or loading any driver14:54
DocScrutinizer05what about mouse(-pointer), keybd mapping for 104key14:54
DocScrutinizer05etc14:54
Paliudev will load it after detected14:54
Palifor keyboard is needed extkbd package which configure xinput (?)14:55
DocScrutinizer05yeah, a lot of stuff already working. Just saying it's another area where we need to consolidate into one nice package14:55
Palione metapackage is enought14:56
DocScrutinizer05:nod:14:56
PaliI already created metapackage for ipv614:56
Paliyesterday pushed to extras-devel14:56
Palinow configure icd2, avahi-daemon, mdns, resolv.conf, microb and dhcp-wide ipv614:57
DocScrutinizer05yay14:57
Palifrom quim we got permission to push closed icd2 ipv6 library from diablo to extras-devel-non-free14:58
Palidiablo library working fine14:58
DocScrutinizer05hail quim14:59
DocScrutinizer05I idly ponder if he's part of jolla14:59
Palisearching for blueled script on tmo...15:01
Palihttp://talk.maemo.org/showpost.php?p=1133260&postcount=92915:04
chem|stDocScrutinizer05: for the looks of now I bet it is nokia's planB and they had to do it this way to not brake any windows contracts15:08
DocScrutinizer05chem|st: I'm missing context15:09
chem|stjolla15:09
DocScrutinizer05aaah15:09
chem|stwinmo is down the pipe for ages now... they will never reach any sink ever again...15:10
DocScrutinizer05and recent announcement that lumia never will run win8... killed Nokia winphone plans for good15:11
chem|st+115:13
chem|stbad marketing15:13
Lava_Croftjolla isnt plan b15:14
Lava_Croftplan b isnt plan b either15:14
Lava_Croftgotta love the grapevine when it comes to out of context quotes15:15
DocScrutinizer05Pali: (blueled script) I haven't looked into the script, but the fundamental crux of auto speed in hostmode is: you need to power down vbus *and wait for it to discharge under session-valid threshold, before you can switch speed15:15
keriodo you *need* vbus for usb to work properly?15:16
kerioeven with a powered hub?15:16
DocScrutinizer05yes15:16
Lava_Crofti bet 99% of the people talking about plan b, havent even seen the interview in which a 'contingency plan' was mentioned15:16
*** Pali has quit IRC15:17
DocScrutinizer05controller 1707 will detect VBUS and switch musb-core statemachine accordingly15:17
DocScrutinizer05without vbus no hostmode15:17
DocScrutinizer05and *with* vbus no speed-switching15:18
DocScrutinizer05that's why in h-e-n you *first* select speed, *then* enable vbus15:18
DocScrutinizer05and you need vbus to query speed of attached device15:19
keriois there an actual difference between high/full/low speeds? can't you just go with high?15:19
kerioi mean, obviously there's a difference, but is there no backwards compatibility?15:19
DocScrutinizer05no, as low speed devices don't know high/full15:20
keriohow would you connect a low-speed *and* a high-speed device through a hub, then?15:20
DocScrutinizer05hub converts15:20
keriogreat, so we just need to put a tiny 1-port high-speed usb hub in the HEN-branded usb host mode adapter we're going to sell :P15:22
DocScrutinizer05would work, yes15:22
keriohow complicated would that be, actually?15:22
DocScrutinizer05not much15:22
*** Pali has joined #maemo-ssu15:23
keriomicrousb plug to n900, microusb port for wallcharger, one or two usb A ports for devices15:23
DocScrutinizer05:nod:15:23
keriocan you charge in vbus mode?15:23
DocScrutinizer05pretty convenient design15:23
DocScrutinizer05basically yes15:24
kerioin "fast" mode too?15:24
kerioor is it limited to .5A?15:25
DocScrutinizer05since we need an own dedicated process(script) to manage the charging during hostmode anyway, we can charge as fast as it goes15:26
keriothe wallcharger provides 1.1A, right?15:26
keriomaybe .5 to n900 and .5 to usb device?15:26
keriowith one port15:26
DocScrutinizer05yep15:26
keriodoes the n900 have to know how fast it's charging the battery?15:27
keriobecause having it self-regulate would be neat, too15:27
kerio1A if no usb device is drawing power, .5 with one and 0 with two15:27
DocScrutinizer05you need to set that in charger chip15:27
kerioi see15:27
DocScrutinizer05max_usb_current15:27
keriois the .2A restriction something software-controlled?15:28
kerioi mean, can you pull the full 1.1A from the wallcharger *and* push .5 to the usb devices?15:28
DocScrutinizer05no, it's as much as N900 can deliver15:28
kerioi see15:28
keriowell, it's weird it *can* do that, isn't it15:29
kerioi mean, it's not supposed to do any of this15:29
kerioor was host/otg mode supposed to be in and then scrapped?15:30
DocScrutinizer05yep, got scrapped late in development15:38
DocScrutinizer05the 200mA boostmode are an inherent property of the bq24150 charger chip which doesn't need a single additional component, so it's unclear if they even planned to use that15:41
DocScrutinizer05iirc they had the coice to delay rollout for yet another dozen of weeks, to get USBcert tests done which needed drivers etc that weren't ready. Or they binned USB-OTG and declared the N900 a non-OTG device, swapped micro-AB OTG receptacle for a micro-B peripheral-only type, and finally shipped it several weeks earlier15:44
kerio*dozen* of weeks? wtf15:47
DocScrutinizer05well, 3 months = 12 weeks15:48
DocScrutinizer05and it seems there's been a silicon erratum in TWL4030 regarding OTG in the early chips15:49
DocScrutinizer05so Nokia had to bite the lemon and decide they don't know for sure if it will be feasible, or when15:50
keriomeh, they could've gotten fmg to write a software workaround for them >:)15:50
DocScrutinizer05alas the rework introduced some flaws then which massively spoiled 'natural' hostmode15:50
DocScrutinizer05particularly the 1707 is missing a way/cmd to simulate ID grounding, and the pin to do that via hw been routed to GND instead of a GPIO of SoC15:53
*** MrPingu has joined #maemo-ssu15:53
kerioID grounding?15:53
DocScrutinizer05pulling ID pin of USB jack to GND15:54
DocScrutinizer05the flag that a A plug got plugged in and thus device should enter A-mode (aka hostmode)15:54
kerioi see15:54
kerioso that's why you have to enable it manually?15:54
kerioand isn't it strictly for usb otg?15:55
DocScrutinizer05yes15:55
kerioi mean, maybe they *couldn't* keep it sw-controlled15:55
keriobecause of certifications15:55
DocScrutinizer05no, the whole friggin mentorgrafix musb-core is controlled that way15:55
DocScrutinizer05certifications rarely require something to *not* be there15:56
DocScrutinizer05anyway they built the thing as if15:57
DocScrutinizer05even with a proper OTG kernel and a special plug or receptacle to simulate A-mode ID pin it wouldn't work15:58
kerioshame15:58
keriooh well, it's not a big deal to have to tap a button15:58
MrPinguoh that cert problem affects gmail too.16:05
MrPinguWell that's the risk of testing branch :P16:05
keriocert problem?16:06
MrPinguhttps of google sites don't work with the newest thumb-compiled microb-engine ;)16:12
keriobut they don't work *quickly*, right?16:13
keriofastest error messages ever16:13
keriodid fmg find the error yet?16:14
MrPinguSee here: http://talk.maemo.org/showthread.php?t=84829&page=3516:15
MrPinguYes, it fails quickly16:15
MrPingujust as it looks, it starts loading it displays that error ;)16:17
*** LaoLang_cool has quit IRC16:23
*** Estel_ has quit IRC16:25
*** taziff has quit IRC16:43
*** MrPingu has quit IRC16:45
*** MrPingu has joined #maemo-ssu16:46
*** DocScrutinizer has quit IRC16:49
*** DocScrutinizer has joined #maemo-ssu16:49
*** DocScrutinizer05 has quit IRC16:52
*** DocScrutinizer05 has joined #maemo-ssu16:52
*** MrPingu has quit IRC17:06
*** BCMM has joined #maemo-ssu17:08
*** MrPingu has joined #maemo-ssu17:14
*** kerio92 has joined #maemo-ssu18:21
*** kerio has quit IRC18:23
*** kerio92 is now known as kerio18:23
*** NIN101 has joined #maemo-ssu18:52
*** MrPingu has quit IRC19:43
*** MrPingu has joined #maemo-ssu19:47
*** andre__ has joined #maemo-ssu19:56
*** andre__ has joined #maemo-ssu19:56
*** Pali has quit IRC20:13
*** Pali has joined #maemo-ssu20:14
*** lizardo has joined #maemo-ssu20:39
*** Pali has quit IRC20:50
*** Pali has joined #maemo-ssu20:58
*** BCMM has quit IRC21:17
*** MrPingu has quit IRC21:31
*** MrPingu has joined #maemo-ssu21:32
*** MrPingu has quit IRC21:48
*** Pali has quit IRC21:53
*** MrPingu has joined #maemo-ssu21:55
*** MrPingu has quit IRC22:32
*** MrPingu has joined #maemo-ssu22:33
*** MrPingu has quit IRC22:38
*** MrPingu has joined #maemo-ssu22:40
*** luf has joined #maemo-ssu23:20
lufchem|st: ping23:20
luf~seen chem|st23:21
infobotchem|st is currently on #maemo #harmattan #maemo-ssu. Has said a total of 46 messages. Is idling for 1h 51m 42s, last said: 'can someone explain?'.23:21

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!