IRC log of #maemo-ssu for Monday, 2016-02-08

Palifreemangordon, see luf messages ^^^00:02
freemangordonluf_: hi!00:03
freemangordonluf_: re your EDS patch - I tried to review it, but failed :). I simply have no idea what it fixes00:04
luf_freemangordon: it fixes 1 mem leak + 2 possible00:08
luf_nsSBCSGroupProber.cpp - it's called and it allocates new string00:09
freemangordonluf_: hmm, wait, there are new commits to EDS, I was thinking it is about "Remove lines with empty value from VCARD string (except required fields)"00:09
luf_it frees the memory only if the string len > 0.00:09
luf_https://github.com/community-ssu/evolution-data-server/pull/100:10
freemangordonoh, sorry, my bad, I was under the impression you have commit rights00:10
luf_I already commited it as I saw no response for quite long time :)00:10
luf_I have commit rights from pali but I like the idea someone take a look if I don't make some stupid mistake.00:11
luf_:)00:11
*** jonwil has joined #maemo-ssu00:11
freemangordonluf_: yeah, sure, it is just that I though it is already commited and saw "Remove lines with empty value from VCARD string (except required fields)"00:12
freemangordonluf_: what is PRUint32 type? int &?00:13
luf_freemangordon: where do you see it?00:14
freemangordonhttps://github.com/community-ssu/evolution-data-server/blob/9ce54231c5a94dad92561c4b7a3c2ef2a24a9eb5/libuchardet/src/nsSBCSGroupProber.cpp#L15200:15
luf_freemangordon: it's out parameter for FilterWithoutEnglishLettersFilterWithoutEnglishLetters00:16
freemangordonyes, that is why I am asking00:17
freemangordonanyway, does FilterWithoutEnglishLetters calls new char[0] for zero newLen1?00:17
freemangordonI guess I should clone the code and look deeper00:17
freemangordonhmm, seems like yes00:19
luf_Hmmm it seems like strange problem I overlooked.00:20
luf_PRBool nsCharSetProber::FilterWithoutEnglishLetters(const char* aBuf, PRUint32 aLen, char** newBuf, PRUint32& newLen)00:20
luf_But this part I haven't touched.00:21
luf_Ot it's some weird C++ syntax?00:23
luf_Ok, PRUint32& is reference (so it can change value inside the func)00:23
luf_So it's not a problem00:24
luf_freemangordon: you scared me :D :D00:25
*** xes has joined #maemo-ssu00:29
freemangordonluf_: sorry, didn't mean to do it :)00:32
freemangordonhowever, those 2 commits look fine on first look00:34
freemangordonluf_: tomorrow I'll try to find time to build EDS and curl and push it in cssu-devel00:35
freemangordonluf_: but don;t expect me to do a peer review on upstream fixes :)00:36
freemangordon(CVEs that is)00:36
*** hashcore has quit IRC00:36
luf_Don't put EDS into cssu-devel.00:37
luf_freemangordon: I have few changes prepared but I lend my HF unit ...00:38
luf_freemangordon: I'll commit curl if you're ok with it. I just took patches from wheezy.00:39
useretailhey guys, i noticed that calendar app doesn't show birthdays in day's details when month-view is selected00:41
useretailonly when week selected you can see the actual birthday00:42
freemangordonuseretail: how's that related to cssu?00:42
freemangordonluf_: yes, commit those curl patches00:42
useretailno idea, i;m just reporting00:42
freemangordonluf_: ok, will wait for EDS00:42
useretail*i'm00:42
useretailfreemangordon: ok, where i have to report?00:44
luf_freemangordon: I will commit it and I'll upload it to cssu-devel in few days. Thanks!00:45
*** xes_ has joined #maemo-ssu00:45
*** xes has quit IRC00:46
useretailactually i have CSSU 21.2011.38-1Tmaemo11+thumb0 testing00:46
luf_useretail: calendar frontend is not part of cssu ...00:55
luf_useretail: maybe I'm wrong but I expect it's closed source from Nokia :(00:56
luf_useretail: https://wiki.maemo.org/Fremantle_closed_packages01:00
jonwilyeah its closed source :(01:01
*** xes_ has quit IRC01:06
luf_freemangordon: https://github.com/community-ssu/libpng/pull/101:21
luf_freemangordon: it seems to me that patches doesn't conflict with your changes and neon patch.01:21
luf_freemangordon: Do I remember correctly that microb should be recompiled with patched libpng?01:23
luf_I accepted curl pull request with wheezy patches.01:25
luf_I'm going to test curl + libpng on my day to day N900 (not only dev)01:25
*** M4rtinK has quit IRC01:32
*** luf_ has quit IRC01:57
*** NishanthMenon has quit IRC02:33
*** Pali has quit IRC02:33
*** Pali_ has joined #maemo-ssu02:33
*** jonwil has quit IRC02:33
*** NishanthMenon has joined #maemo-ssu02:34
*** NishanthMenon has joined #maemo-ssu02:34
*** Pali_ has quit IRC03:06
*** jonwil has joined #maemo-ssu04:05
*** LauRoman has joined #maemo-ssu04:42
*** LauRoman|Phone has quit IRC04:57
*** pagurus has quit IRC07:11
*** ravelo has quit IRC07:30
*** liteIRC__ has joined #maemo-ssu07:30
*** liteIRC__ is now known as liteIRC___07:31
*** liteIRC___ is now known as ravelo07:31
*** liteIRC__ has joined #maemo-ssu07:44
*** ravelo has quit IRC07:44
*** liteIRC__ is now known as liteIRC___07:44
*** liteIRC___ is now known as ravelo07:44
*** ravelo has quit IRC07:47
*** liteIRC__ has joined #maemo-ssu07:47
*** liteIRC__ is now known as liteIRC___07:47
*** liteIRC___ is now known as ravelo07:47
*** sparetire has quit IRC08:00
*** xes has joined #maemo-ssu08:35
*** amiconn has quit IRC09:00
*** amiconn has joined #maemo-ssu09:00
*** ravelo has quit IRC11:51
*** RedM has quit IRC12:19
*** RedW has joined #maemo-ssu12:23
*** hashcore has joined #maemo-ssu12:37
*** pagurus has joined #maemo-ssu13:11
*** LauRoman has quit IRC13:43
*** hashcore has quit IRC14:11
*** DocScrutinizer05 has joined #maemo-ssu14:39
DocScrutinizer05IroN900:~# od -A x -x /usr/lib/mce/modules/libfilter-brightness-als.so|grep 0086a0; md5sum /usr/lib/mce/modules/libfilter-brightness-als.so14:59
DocScrutinizer050086a0 0005 0000 0005 0000 0000 0000 0000 000014:59
DocScrutinizer059db7e3c959bd5b47c7fcfb6ebc1ca764  /usr/lib/mce/modules/libfilter-brightness-als.so14:59
DocScrutinizer05http://mg.pov.lt/maemo-ssu-irclog/%23maemo-ssu.2013-11-23.log.html#t2013-11-23T06:20:4814:59
DocScrutinizer05https://talk.maemo.org/showthread.php?p=1388368#post138836814:59
DocScrutinizer05IroN900:~# apt-cache policy mce15:01
DocScrutinizer05mce:15:01
DocScrutinizer05  Installed: 1.8.127.3+0m515:01
DocScrutinizer05IroN900:~# apt-cache policy mp-fremantle-community-pr15:03
DocScrutinizer05mp-fremantle-community-pr:15:03
DocScrutinizer05  Installed: 21.2011.38-1Tmaemo1115:03
*** LauRoman has joined #maemo-ssu15:40
DocScrutinizer05IroN900:~# od -A x -tx1 /usr/lib/mce/modules/libfilter-brightness-als.so|grep '0086a0 05 00 00 00 05 00 00 00 00 00 00 00' && sed -i -E 's/\x05\x00\x00\x00\x05\x00\x00\x00\x00\x00\x00\x00/\x05\x00\x00\x00\x05\x00\x00\x00\x05\x00\x00\x00/' /usr/lib/mce/modules/libfilter-brightness-als.so&&od -A x -tx1 /usr/lib/mce/modules/libfilter-brightness-als.so|grep '0086a0'16:09
DocScrutinizer050086a0 05 00 00 00 05 00 00 00 00 00 00 00 00 00 00 0016:09
DocScrutinizer050086a0 05 00 00 00 05 00 00 00 05 00 00 00 00 00 00 0016:09
DocScrutinizer05https://talk.maemo.org/showthread.php?p=1388368#post1388368 updated16:14
jonwilhttp://talk.maemo.org/showthread.php?p=1498178#post149817816:21
*** sparetire has joined #maemo-ssu16:39
*** jonwil has quit IRC16:52
*** hashcore has joined #maemo-ssu17:12
*** jonwil has joined #maemo-ssu17:17
*** jonwil has quit IRC17:21
*** NishanthMenon has quit IRC17:30
*** NishanthMenon has joined #maemo-ssu17:31
*** hashcor has joined #maemo-ssu17:35
*** hashcor is now known as help17:39
*** help is now known as hashcor17:40
*** hashcor has quit IRC17:40
*** hashcor has joined #maemo-ssu17:41
*** hashcor has quit IRC17:43
*** Sicelo009N has joined #maemo-ssu18:10
DocScrutinizer05dbus on maemo is flawed, massively. We need an update which will fix many of the randomly happening erratic behavior of the whole system. See https://lists.freedesktop.org/archives/dbus/2008-March/009526.html  >> At which point the badness happens: The daemon (apparently) immediately closes the connection, oblivious to the fact that there is critical data in it's socket buffer as-yet un-read.<<18:53
DocScrutinizer05I only suspect this is the root cause, the symptom is >>needs --print-reply  to make `dbus-send --system --print-reply --dest=com.nokia.mce /com/nokia/mce/request com.nokia.mce.request.req_vibrator_pattern_activate string:PatternChatAndEmail` work. Remove --print-reply and it simply fails<<18:54
*** Pali has joined #maemo-ssu18:55
DocScrutinizer05https://bugs.freedesktop.org/show_bug.cgi?id=896#c26   https://bugzilla.redhat.com/show_bug.cgi?id=477964#c1118:56
povbotBug 896: missing instructions in point 4 of installing scratchbox in gregale/INSTALL.txt18:56
povbotBug 477964: was not found.18:56
DocScrutinizer05https://cgit.freedesktop.org/dbus/dbus/commit/?id=87ddff6b24d9b9d4bba225c33890db25022d8cbe18:57
DocScrutinizer05note that this bug will affect *all* clients that send a dbus message and then immediately terminate19:02
DocScrutinizer05the message simply gets lost19:02
*** futpib has joined #maemo-ssu19:04
*** merlin1991 has quit IRC19:13
*** merlin1991 has joined #maemo-ssu19:15
DocScrutinizer05I don't know if the more generic async problem in dbus we faced back when we did FSO-vala is related -  >>fsogsmd: Waiting for async. dbus server support in Vala.<< issue in http://wiki.openmoko.org/wiki/OpenmokoFramework/Status_Update_7#DBus_APIs19:27
DocScrutinizer05I just recall this was a bug in dbus that really affected all async dbus-send() calls19:29
*** RedW has quit IRC19:37
*** RedW has joined #maemo-ssu20:00
*** Sicelo009N has quit IRC20:51
*** pagurus has quit IRC21:05
*** Sicelo009N has joined #maemo-ssu21:13
*** eugene_ has joined #maemo-ssu21:20
Paligitorious is back! only git repositories in R/O mode, but at least something for Maemo to not loose git repos...21:22
Palihttps://gitorious.org/community-ssu/hildon-application-manager21:22
bencohwoohoo21:23
bencohat last21:23
Palihttps://gitorious.org/maemo-af/ke-recv21:23
Palioriginal URLs working21:24
bencohI suggest we move everything that's still there :)21:24
bencoh(and not already hosted on some other platform)21:24
Palimaemo infra has backups...21:25
bencohyeah, but they're not public21:25
Palimerlin restored some gitorious repos when I needed them21:25
Paliit is just for emergency21:25
*** eugene_ has quit IRC21:28
*** eugene_ has joined #maemo-ssu21:29
*** arcean has joined #maemo-ssu22:07
DocScrutinizer05Pali: would you want to have a look into dbus and see if we have https://cgit.freedesktop.org/dbus/dbus/commit/?id=87ddff6b24d9b9d4bba225c33890db25022d8cbe ?22:10
DocScrutinizer05Pali: it's a pretty bad bug that has impact on complete system22:11
PaliDocScrutinizer05: do we need to backport that commit to maemo?22:12
Paliit is from 2009, is not already included?22:12
DocScrutinizer05obviously not, see --print-reply22:13
DocScrutinizer05http://mg.pov.lt/maemo-ssu-irclog/latest.log.html#t2016-02-08T18:53:2122:13
freemangordonPali: please backup felipe's repos on gitorious22:15
*** eugene__ has joined #maemo-ssu22:24
*** eugene_ has quit IRC22:26
*** futpib has quit IRC22:43
PaliI have it for a long time22:43
PaliDocScrutinizer05: patch apply fine on maemo dbus library!22:44
Paliso it is definitely not part of Maemo22:44
DocScrutinizer05:-) ... and :-D22:46
DocScrutinizer05we really *should* get this into all maemo systems ASAP22:46
Palimerlin1991: we need new urgent CSSU update22:47
Palimerlin199: see ^^^22:47
*** eugene__ has quit IRC22:51
DocScrutinizer05dang, this feels good to have squished a bug22:56
DocScrutinizer05particularly such a nasty one that potentially caused bazillion random erratic behaviors in system without leaving much of a trace22:57
bencoh"urgent"... it's been lying there for years ;)22:57
bencohbut we haven't even tested the fixed version yet :)22:58
freemangordonyep, I wouldn't say it is urgent if has been here for years without obvious effects. also, please someone to provide a test case, the same we did for pselect() bug22:59
Sicelo009N:)23:00
DocScrutinizer05Pali: there's a point in the source where it gets a  SIGPIPE on read and then (in unpatched) quits, or (in patch) ignores the SIGPIPE. I suggest to have a logger to syslog there shoing a message in syslog each time the situation occurs and would have caused trouble in the unpatched recently used version23:00
DocScrutinizer05freemangordon: testcase >>needs --print-reply  to make `dbus-send --system --print-reply --dest=com.nokia.mce /com/nokia/mce/request com.nokia.mce.request.req_vibrator_pattern_activate string:PatternChatAndEmail` work. Remove --print-reply and it simply fails<<23:01
DocScrutinizer05totally obvious and simple testcase23:01
freemangordonDocScrutinizer05: I am not convinced the bug is that critical, otherwise syslog would have been full of "SIGPIPE killed process" messages23:03
freemangordonnot saying it should not be fixed23:03
DocScrutinizer05no, the process isn't killed by SIGPIPE23:03
DocScrutinizer05dbus-daemon simply aborts the read of message from sender23:03
DocScrutinizer05See https://lists.freedesktop.org/archives/dbus/2008-March/009526.html  >> At which point the badness happens: The daemon (apparently) immediately closes the connection, oblivious to the fact that there is critical data in it's socket buffer as-yet un-read.<<23:04
freemangordonso teh sequence is - open dbus connection, send message, close immediately, correct?23:04
DocScrutinizer05the sender process already regularly quit anyway, and dbus-daemon drops the message and nothing will indicate something bad happened, except the fact that the message "goes to dev/null"23:05
DocScrutinizer05yes23:05
DocScrutinizer05100% correct23:06
freemangordondo we have any other example, besides dbus-send?23:07
DocScrutinizer05dunno23:07
DocScrutinizer05how would we possibly know23:07
freemangordonalso, is this for both sync and async messages?23:07
DocScrutinizer05there's no indication at all except the missing message23:07
DocScrutinizer05seems for both, though I haven't looked into it23:07
freemangordonI guess for async messages only23:08
DocScrutinizer05possible, yes23:08
freemangordonbecause for sync message, you're blocked so cannot close the connection23:08
DocScrutinizer05blocked until when?23:08
freemangordonuntil return from dbus function call, iirc23:08
DocScrutinizer05yeah, but what does function call accomplish?23:09
freemangordonlemme check23:09
DocScrutinizer05does it guarantee the buffer got read completely by daemon, before it returns?23:09
DocScrutinizer05I even suspect there's more bugs to the async stuff than only this one23:10
DocScrutinizer05not sure about that though23:10
DocScrutinizer05still trying to find out23:10
freemangordonhmm, read a bit https://dbus.freedesktop.org/doc/api/html/group__DBusConnection.html#gae1cb64f4cf550949b23fd3a756b2f7d023:11
DocScrutinizer05http://logs.nslu2-linux.org/livelogs/openmoko-cdevel.txt23:11
freemangordonDocScrutinizer05: it seems this behaviour is by design, not a bug23:12
freemangordonso I really wonder what was "fixed" in this commit23:13
DocScrutinizer05it definitely is a bug, see behavior of dbus-send23:13
freemangordonDocScrutinizer05: readt the ^^^ link, it clearly says you should call dbus_connection_flush() if you're going to close the connection and exit23:14
DocScrutinizer05well, then it's a bug in dbus-send and a bazillion lib*dbus*23:16
*** luf_ has joined #maemo-ssu23:16
DocScrutinizer05in my design book the dbus_connection_flush() belongs *inside* the dbus_connection_close()23:17
DocScrutinizer05maybe there are arguments to not have it that way23:18
freemangordonDocScrutinizer05: maybe someone should look at dbus-send code, if that flush() is not called, then it is definitely a bug in dbus-send.23:18
DocScrutinizer05well, we need to check a twenty dozen other executables and libs too23:19
freemangordonDocScrutinizer05: imagine a crash in an multithreaded client app, when only half of the data was written23:19
DocScrutinizer05since I don't think other devels implemented stuff that wasn't even in dbus-send which is sort of the reference implementation23:19
freemangordonso, you'll have some random partial message23:21
freemangordonDocScrutinizer05: to make it clear - I don;t say we should not pick the patch, only that it is not that *urgent* :)23:22
DocScrutinizer05a message is atomic aiui23:22
freemangordonisn;t that unix socket?23:22
freemangordonor even tcp socket?23:22
DocScrutinizer05tcp prolly23:22
freemangordonno way it is atomic23:23
freemangordonrouters can (and do) whatever segmentation fits them23:23
freemangordonkeep in mind you can call remote dbus23:23
DocScrutinizer05I don't like to dive into the mud of dbus code to find out what could possibly go wrong23:23
DocScrutinizer05https://lists.freedesktop.org/archives/dbus/2008-March/009526.html sounds like a clear race condition to me23:24
DocScrutinizer05dbus-daemon not servicing the postbox for outbound letters when it can't deliver an inbound letter23:26
DocScrutinizer05dbus-send which is the reference implementation supposed to be written by the inventors of dbus is not working like supposed. It's pretty likely that all other apps that were written based on same knowledge show a similar behavior23:28
DocScrutinizer05we either can fix all other apps or we fix dbus-daemon23:29
freemangordonoh oh, I didn;t mean to say there is no bug, but that this fix should not be treated as "urgent" but go through the normal cssu devel-testing-stable cycle23:29
DocScrutinizer05I never said "urgent", did I?23:30
DocScrutinizer05I said "ASAP"23:30
DocScrutinizer05which of course implies regulr CSSU testing and all23:30
freemangordonthis means "for yesterday" in my book :)23:30
DocScrutinizer05it's not an emergency fix23:31
DocScrutinizer05:-)23:31
DocScrutinizer05and I wouldn't want to see inferior system stability getting introduced by this patch23:31
DocScrutinizer05first we even could not change the normal behavior but simply add a syslog line whenever such premature abort of message transmission occurs, including the message content and the from: and to:23:33
freemangordon:nod:23:34
DocScrutinizer05this would give us an idea how often actual problems are introduced by this bug23:34
freemangordonagree23:34
bencohDocScrutinizer05: have you tried the vibrate req with patched dbus?23:34
DocScrutinizer05when we also apply the new behavior in dbus-daemon then we still get that syslog and can look if in contect of the log time anything odd happened from the new behavior23:35
DocScrutinizer05bencoh: I have no patched dbus23:35
freemangordonanyway, /mes is going to have some sleep, night23:35
DocScrutinizer05n823:35
luf_freemangordon: Do you want to take a look at libpng pull request (patches from wheezy)?23:35
luf_freemangordon: enjoy the night ;)23:36
*** arcean has quit IRC23:43
*** hashcore has quit IRC23:45
PaliDocScrutinizer08: now I checked again dbus source and patch is there23:51
Paliin debian/patches folder23:51
Palipatches are applied *after* starting dpkg-buildpackage23:51
Paliby debian/rules23:51
bencohindeed23:51
Paliso no action is needed23:51
Paliwe already have patch in maemo23:52
PaliI was confused, because I downloaded that upstream patch and it applied on unpacked tarball23:53
Palihttps://gitorious.org/maemo-af/dbus23:53
Palihttps://gitorious.org/maemo-af/dbus?p=maemo-af:dbus.git;a=commit;h=119515a9844cb3139514f4f0aaa179141febe2de23:53
Palihttps://gitorious.org/maemo-af/dbus?p=maemo-af:dbus.git;a=commitdiff;h=119515a9844cb3139514f4f0aaa179141febe2de;hp=f3d6699a5d4615c527c4bac6290eb3591af4944023:54
DocScrutinizer05then we have another yet unidentified bug23:54
DocScrutinizer05:-/23:54
Palifreemangordon, merlin1991: no action needed, patch is already in maemo dbus version23:55
DocScrutinizer05yet the bug exists23:57
DocScrutinizer05we should compare dbus-send of 2008 with recent dbus-send then23:58
DocScrutinizer05or do a strace and see why dbus-send fails when no --print-reply23:59

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