| *** zbenjamin_ is now known as zbenjamin | 02:34 | |
| *** frinring_ is now known as frinring | 07:02 | |
| r0kk3rz | well, you must be in a non euro timezone :P | 08:42 |
|---|---|---|
| *** pketo_ is now known as pketo | 09:54 | |
| *** chem|st_ is now known as chem|st | 12:37 | |
| tortoisedoc | ol : what was the main thought behind mounting /tmp into tmpfs? | 15:16 |
| tortoisedoc | (sb2 temp that is | 15:16 |
| tortoisedoc | ) | 15:16 |
| deicide- | euro time zone just night owl | 15:52 |
| deicide- | r0kk3rz | 15:52 |
| ol | tortoisedoc (IRC): What's the problem with "/tmp"? | 16:19 |
| *** Venemo_j_ is now known as Venemo_j | 17:03 | |
| *** Colgate_ is now known as Colgate | 18:08 | |
| tortoisedoc | ol : on both mer sdk sb2 i486 and vm the sem_open does not work | 18:13 |
| tortoisedoc | I am not sure if it's due to the fact the sem.* file is created with a glibc-internal function (__libc_open), which is not hooked by sb2 | 18:14 |
| tortoisedoc | I was wondering if the /tmp as tmpfs (as /dev/shm is tmpfs too) was supposed to be an attempt at a fix for this | 18:15 |
| tortoisedoc | but probably not | 18:15 |
| tortoisedoc | if this is the case, a patch for glibc might be necessary to migrate from __libc_open to standard open | 18:16 |
| ol | tortoisedoc (IRC): The problem is because of __libc_open(). | 18:18 |
| tortoisedoc | ol : so a patch is the only solution | 18:21 |
| tortoisedoc | this glibc is a complete spaghetti accident | 18:22 |
| ol | Multiple solutions are possible. | 18:22 |
| tortoisedoc | ol : tell me more | 18:22 |
| tortoisedoc | the tmpfs is a bit problematic, cause it would need to be mounted in the session dir | 18:22 |
| tortoisedoc | (somewhere) | 18:22 |
| ol | A gate for __libc_open() can br added to sb2. However, I'm not sure whether this can lead to unintended vonsequences in other parts of glibc. | 18:23 |
| tortoisedoc | ol : hmm im afraid the __libc_open alias is marked as hidden | 18:24 |
| tortoisedoc | is that not limiting sb2's hooking capabilities? | 18:24 |
| tortoisedoc | and yes you are right, its a private function so quite sensible to tweak with | 18:24 |
| ol | A gate for sem_open() can be added ro sb2 that does essentially the same as sem_open() in libpthread. This can also be problematic in case inner workings of libpthread change in new version of glibc (or non-glibc libc used inside sb2 session). | 18:25 |
| ol | __libc_open() can't be hidden because it's called from other libraries. Its private nature is just hinted by "__" in the beginning of its name, but this hint is for humans, not dynamic linker. | 18:26 |
| tortoisedoc | ol : aha ok (glibc n00b here btw9 | 18:27 |
| tortoisedoc | ) | 18:27 |
| tortoisedoc | hmm | 18:27 |
| tortoisedoc | ol : is patching glibc in the specific sem_open a big deal? | 18:27 |
| ol | Yes, it is. | 18:28 |
| ol | You can't patch glibc. | 18:28 |
| tortoisedoc | i beg your pardon, I can't but someone else has done it, and quite proficently? :D | 18:28 |
| ol | Nobody will use patched glibc. | 18:28 |
| tortoisedoc | ok thats a better argument | 18:29 |
| tortoisedoc | so maybe __libc_open gate is a good starting point | 18:29 |
| ol | Although dangerous one. | 18:29 |
| tortoisedoc | nothing is simple, indeed | 18:29 |
| tortoisedoc | ol : how can I "easily" add such a gate? | 18:31 |
| ol | You have to find all occurences where __libc_oprn() is used in all libraries that are parts of glibc suite (except libc itself) and make sure that all these uses are safe to be replaced by a gate. And then you have to pray that no new uses of __libc_open() are introduced in the future. | 18:32 |
| ol | There is no easy way. | 18:32 |
| tortoisedoc | I'd like to get back to that glibc patching | 18:33 |
| tortoisedoc | ol : https://git.merproject.org/mer-core/glibc | 18:33 |
| tortoisedoc | there's plenty of patches on the glibc in mer it seems | 18:33 |
| tortoisedoc | i'd assume from this that patching glibc is a "supported" way of working in mer? | 18:33 |
| ol | Ask those who maintain this package. | 18:35 |
| *** svartoyg_afk is now known as svartoyg | 18:35 | |
| tortoisedoc | ol : yes I think that's the best idea | 18:35 |
| tortoisedoc | thanks | 18:35 |
| ol | You'd better patch your application than patching entire glibc. | 18:36 |
| tortoisedoc | my application is unpatchable :( | 18:37 |
| tortoisedoc | but yes I agree that would be the right approach in the "normal" universe; | 18:37 |
| tortoisedoc | except I don't know what I could do in order to make POSIX sem_open to work from my user-land application | 18:38 |
| *** svartoyg is now known as svartoyg_afk | 19:49 | |
| taixzo | Is there a patch to re-enable closing apps by swiping down from the top, in sfos 3? | 20:10 |
| mal | taixzo: swipe near the sides from top | 20:17 |
| taixzo | oh, cool! | 20:19 |
| yorambh | hello all | 22:06 |
| yorambh | anybody in mood to help with problem | 22:07 |
| yorambh | ? | 22:07 |
| *** ahjolinna_ is now known as ahjolinna | 23:47 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!