Saturday, 2018-12-15

deicide-can i change wallpaper?02:16
*** feodoran_ is now known as feodoran02:39
tortoisedochow di how07:58
tortoisedocmal : who could I ask about this commit
tortoisedocmal : it appears to break the sem_open08:15
tortoisedocin particular, the redirection towards the session_dir of /dev/shm/.. entries08:15
tortoisedocI don't know how serious this is08:15
attahHow can i make the pulley menu blink like in the email app?13:24
oltortoisedoc (IRC): In contrary, it fixes sem_open, not breaks it.14:27
*** feodoran is now known as Guest7057514:45
*** feodoran_ is now known as feodoran14:45
kimmoliattah: simple. busy. docs
attahcool, thanks16:58
attahdunno how i  missed that16:58
attahNow i'm battling with columns inside ListItem, or switchable menus... not sure which is more usable16:59
attahbleh.. i can't seem to have two menus, but i can at least hide items in a big momma menu depending on the click that activated it17:10
*** SpeedEvil is now known as Guest7306818:54
deicide-can i change wallpaper?18:57
kimmolideicide-: it is called ambience, from settinsg. or create your own.19:01
attahAnyone with Philips Hue that wants to give feedback?19:08
deicide-do i have to create my own?19:14
deicide-cant i just change the picture19:14
attahYou can "create your own" from just a picture, with no further actions required19:14
tortoisedocol: not for me, a simple code like this one fails22:26
maltortoisedoc: you can check the maintainers repo in mer-core to who maintains scratchbox22:31
tortoisedocmal : thanks22:32
tortoisedocso the usual suspects ;)22:33
olmal (IRC): That particular change we're discussing was mine, not maintainer's.22:33
maltortoisedoc: what exactly are you trying to do when you encounter the issue you have?22:35
oltortoisedoc (IRC): Do you actually have "/dev/shm" in your environment?22:37
tortoisedocmal : llvm tests rely on some crappy python multithread abstraction (multiprocessing), which run tests in parallel i suppose22:37
tortoisedocinternally it uses semaphores22:37
tortoisedocthis does not work under sb2 (for some reason, above snippet gives an idea, and a hint is maybe to be found in the commit)22:38
tortoisedocol: yes /dev/shm is in the target22:38
tortoisedocbut it's empyt22:38
tortoisedocthe semaphore is created in the host's /dev/sh22:38
olNo. Do you have "/dev/shm" on your *host*?22:39
tortoisedoc(as its supposed to be done)22:39
olIn your SDK, I mean. Do you run it in VirtualBox?22:39
tortoisedocthe creation works, the problem is the stat'ing after the creation (which redirects to /tmp)22:39
tortoisedocol : no22:39
tortoisedocol : platform sdk, *not* vbox22:39
olSo, what is the problem?22:40
tortoisedocol : as said in the snippet : oserror 2 when run via python (multiprocessing) or errno 2 via c22:40
tortoisedoc(python internally uses dynlib c for access to the same sem_open)22:41
tortoisedochow to fix this eludes me as my knowledge in sb2 is practically unexistent22:41
olWhat causes this error?22:41
tortoisedocol: a simple sem_create in a c snippet (like this one will do22:42
tortoisedoci *think* the sb2 should symlink the /tmp entry to the host's /dev/shm/sem-xxx22:43
olI mean, it should work. Yes, sb2 rules redirect temporary file creation to "/tmp". Why does it return ENOENT? Can it be a problem with your "/tmp"?22:43
tortoisedocbut i dont know22:43
tortoisedocol : have you tried the snipped?22:44
tortoisedocit might be my env being broken22:44
olI'm trying to start my Sailfish SDK.22:44
tortoisedoccant be sure unless someone trying22:44
tortoisedocone doesnt simply boot sfsdk :)22:45
olI have errno = 22.22:52
tortoisedocol: if I use strace on the a.out I get this for example:22:55
tortoisedocreadlink("/tmp/sb2-tone-20181215-225427.q13812/tmp/sem.pMmXOl", 0xffc27feb, 4096) = -1 ENOENT (No such file or directory) r22:55
tortoisedocso the path is there but something does a readlink internally22:55
tortoisedoc...except this will fail on a real file22:55
tortoisedoc(which the file in question seems to be22:55
tortoisedoc /tmp/sb2-tone-20181215-225427.q13812/tmp/sem.pMmXOl )22:56
tortoisedocthis is just my theory here22:56
olCan you run your shell in sb2? Then in this shell:22:59
ol* check whether you have "/tmp" and "/dev/shm";22:59
ol* create a file in "/dev/shm", check whether this file appeared in both "/dev/shm" and "/tmp";22:59
ol* link this file to another name in "/dev/shm" (hard link, not symlink), check whether this new name appeared in "/dev/shm" and "/tmp".22:59
tortoisedocol : if you try the python sample here, the creation will except and the removal of the sem wont happen so you can actually see the file in the sdk's /tmp23:00
tortoisedocok wait23:00
olNo, this is too complex. Let's try to debug a simple test case.23:00
tortoisedocin sb2 -> ls /dev is empty23:01
tortoisedocol : ^23:04
olThis is your problem.23:04
tortoisedochow can I solve it?23:04
tortoisedocor at least debug it23:05
olIn my case everuthing works after I added one more line before sem_open() call:23:05
olerrno = 0;23:05
olWithout this line subsequent printf was reporting whatever garbage that was left in errno before sem_open().23:06
olRun "ls /dev" under sb2 with "-L noise" option. Read resulting log file.23:07
tortoisedocol : not for me, still errno 223:07
tortoisedocol : ok ls -L noise gives me /dev stuff plus this : ls: cannot access noise: No such file or directory23:08
olDoes your host have "/dev" in the right place?23:08
olNo, "-L noise" is option for sb2, not ls.23:08
tortoisedocok wait23:08
olThe file is quite big. You'll be interesting in rewriting "/dev". Go to the end of the file and scroll up.23:09
tortoisedocthe host's dev is a basic /dev23:09
tortoisedocol I get "no match" pretty much everywhere23:10
tortoisedocfor instance23:10
tortoisedoc1544915326.618 (9)ls[14814]ruletree_test_path_match '/usr/share/locale' (17), '/dev' (4) => -1 (no match)[pathmapping/paths_ruletree_mapping.c:113]23:10
olIgnore that. Concentrate on "/dev", not other paths.23:11
olAt least, we've just figures out that my patch is not to blame. You have something else wrong in your setup.23:11
tortoisedocyes so it seems :)23:12
olThis line just ways: "/dev" did not match "/usr/share/locale", so rule for mapping "/usr/share/locale" does not apply.23:12
olThis line just says: "/dev" did not match "/usr/share/locale", so rule for mapping "/usr/share/locale" does not apply.23:13
olFind successful matches.23:13
tortoisedocol :
tortoisedocthis excludes all the "no match"23:15
olSo, it maps "/dev" to "/dev", as expected.23:16
tortoisedocshouldnt that be /parentroot/dev?23:17
olDo egrep 'return(ed|s)' on this file/23:17
olAre you in chroot?23:18
tortoisedocin sb2 only23:18
olAn you see that "/dev" is mapped to "/dev".23:18
olWhat does "ls /dev" (without sb2) show you in the same command line as you run sb2?23:19
olIs it also empty?23:19
tortoisedocnot anymore23:20
tortoisedocnew terminal -> fixed23:20
olAre you sure you were not in chroot?23:22
tortoisedocsame terminal than before23:22
tortoisedocsame problem23:22
tortoisedocnew terminal23:22
tortoisedocno problem :D23:22
tortoisedocive been tweaking things around a while now23:23
tortoisedocmaybe ill reboot first23:23
tortoisedocthen we can cotinue (there seems to be no /dev/shm too)23:23
olI remember, Mer SDK was using chroot (and I contributed a change to use it inside unshare to bind mount "/dev", "/sys" and everything else into chroot in a private namespace so you don't need to unmount it when you exit it).23:23
olWhat does "ls /dev" show in old and new terminal?23:24
olIs output different?23:24
olAnd are you on the same host?23:24
tortoisedocol :yes same host23:24
tortoisedocdifferent terminals23:25
olBut no "/dev/shm" in just one terminal?23:25
tortoisedocterminal 1 ) ls /dev -> empty23:25
tortoisedocterminal 2 ) ls /dev -> full, but no shm23:25
olThis means that you are in different namespaces.23:25
tortoisedoccan I list the namespace?23:26
olWhat does "cat /proc/self/mounts" show on both of them? What's tje difference?23:26
tortoisedocon the term 1 it hangs23:26
tortoisedocon term 2, a full list of entries23:27
tortoisedocoff to reboot23:27
tortoisedoclets see23:30
tortoisedocchroot -> ls /dev ok23:31
tortoisedochas /dev/shm23:31
olWhy do you need chroot?23:31
olIs this setup supported now?23:31
tortoisedocim not needing it, just telling you that in chroot all seems to be ok23:31
tortoisedoc(as in it that chroot seems to see /dev )23:32
olIf you don't need chroot, why do you use it?23:32
tortoisedoccan I get into sb2 without chroot?23:32
tortoisedocI do23:32
tortoisedocsfossdk -> sb223:32
olSailfish SDK does not use chroot.23:32
olIt uses VirtualBox though, but it's needed for creating unified environment between various Linux distributions and Windows.23:33
tortoisedocwell sfossdk is only the name i gave to the env bootstrapping ( i think it originally was mersdk)23:34
olWhat bootstrapping?23:34
olJust run official Sailfish SDK.23:34
tortoisedocin virtual box you mean?23:35
tortoisedocthat wont rid me of sb223:36
olYes. But it will remove complexity of chroot. And this is supported way.23:39
olYou're lucky that I was here and I was reading your question because your reference to my commit triggered my attention. Otherwise, I would just ignore because I was doing something else.23:41
tortoisedocill have gazillions of other problems, tho (including disk space)23:41
olWith official SDK you can get official support.23:41
tortoisedocso hm23:41
tortoisedoc /dev/shm is not in sb223:41
olAnd it was not obvoius that you use chroot.23:41
tortoisedocno thats true and my fault :)23:42
olSo what? It's mapped.23:42
tortoisedocjust to reply your previous question :)23:42
olWhy do you need a file in target root if you can't even access it from sb2?23:42
tortoisedocwhat do you mean by target root?23:43
tortoisedocand by file you mean /dev/shm?23:43
olYou can't access $target_root/dev/shm from sb2 because sb2 maps "/dev/shm" to "/dev/shm" for stat and to "$SESSIONDIR/tmp" for reading/writing files.23:44
tortoisedocokay thats clearer :)23:44
olSo, you don't need this file. And I presume, $target_root/dev is empty for similar reason.23:45
tortoisedocol : are the $SESSION/tmp files symlinks or real files?23:45
olWhatever you create there.23:46
tortoisedoc(real as in dummy)23:46
tortoisedocI dont but sb2 does create my /dev/shm/sem-xxx there23:46
olIt's your temporary "/tmp" for session.23:46
tortoisedocso when I do sem_opne23:46
tortoisedocI get /dev/shm/sem-xxx in my host23:46
tortoisedocand $SESSION/tmp/sem-xxx23:47
olIt's mapped to your session's temporary "/tmp", so it's only there.23:47
tortoisedocso my session's /tmp is the same tmpfs as my host's /dev/shm?23:48
olFunctionally it's the same, but it's implemented by removing temporary session dir when your session ends.23:48
tortoisedocol : ok but physically its a different tmpfs partition?23:49
olAs session dir is created in your host's "/tmp", it's whatever your "/tmp" is.23:49
olNo, it's a directory in your host's "/tmp".23:50
tortoisedocah ok23:50
tortoisedocpardon my n00bness here :)23:50
olJust read rewrite rules, it's all there,.23:50
tortoisedocso to recap, all is fine with /dev/shm23:51
tortoisedocalbeit my script still doesnt work23:51
tortoisedocor better23:51
tortoisedocmy test.c23:51
olDo you already have a file in your emulated "/dev/shm"?23:53
olWhat is errno now?23:54
olDid you forget to assign 0 to errno before calling sem_open()?23:54
tortoisedocno added23:54
tortoisedocsame result23:54
olerrno = 2?23:55
tortoisedoc(HELLO 2)23:55
olDo you have "/dev/shm" in the same terminal?23:55
olI mean, does it exist?23:55
olIs "/dev" or "/dev/shm" empty?23:57
tortoisedoc /dev/shm23:57
tortoisedoc /dev is populated23:57
olIs "ls /dev" in sb2 output the same as "ls /dev" without sb2 in the same terminal?23:58

Generated by 2.17.1 by Marius Gedminas - find it at!