asalkeld | Hi another cross compile question: uClibc does not have malloc_usable_size() | 00:04 |
---|---|---|
asalkeld | can we do with out it? | 00:05 |
Keybuk | it's probably only used by the test suite | 00:13 |
asalkeld | is there an easy way to disable the test suite | 00:14 |
asalkeld | (from building) | 00:14 |
Keybuk | it won't build until you run "make check" | 00:19 |
asalkeld | with a "make" I get: | 00:21 |
asalkeld | ../nih-dbus/.libs/libnih-dbus.a(alloc.o): In function `nih_alloc_size': | 00:22 |
asalkeld | /home/anguss/work/main/buildroot/build_powerpc_8540/upstart-0.6.3/nih/alloc.c:743: undefined reference to `malloc_usable_size' | 00:22 |
Keybuk | yeah that's code in the core to support the test suite | 00:23 |
Keybuk | you'll have to patch that out | 00:23 |
asalkeld | so I have put a "AC_CHECK_FUNC(malloc_usable_size, [with_malloc_usable_size=yes])" | 00:23 |
asalkeld | in configure.ac | 00:23 |
asalkeld | etc ... | 00:23 |
Keybuk | that won't do anything unless you patch the code to check for that | 00:23 |
asalkeld | yip, done that | 00:24 |
asalkeld | so all is good | 00:24 |
asalkeld | but, I just wanted to make sure that it would not get called | 00:24 |
Keybuk | can't think of a reason it would be | 00:24 |
asalkeld | (which you answered) | 00:24 |
asalkeld | I'll send a patch | 00:25 |
Keybuk | patch won't be accepted I'm afraid | 00:25 |
asalkeld | Can you suggest a way to get around nil-dbus-tool getting built with the cross CC | 00:25 |
asalkeld | why (about the patch) | 00:26 |
Keybuk | Upstart is only intended to work with glibc | 00:26 |
Keybuk | if you want to run it with something else, that's fine - feel free to post the patch to the ML | 00:27 |
Keybuk | or even make a "uclibc" bzr branch with any necessary patches applied | 00:27 |
Keybuk | (that others can merge to keep up to date) | 00:27 |
Keybuk | but I'm keeping the core #ifdef free | 00:27 |
asalkeld | that might be a bit tricky | 00:28 |
asalkeld | does nil-dbus-tool do anything dbus version dependant? | 00:29 |
Keybuk | no | 00:29 |
Keybuk | it just generates C code | 00:29 |
asalkeld | could I build upstart on my host then copy the generated c files to my cross code tree | 00:30 |
Keybuk | yes | 00:30 |
asalkeld | and have a configure option to not run or build the tool | 00:30 |
Keybuk | I tried for a while to actually have the sources included in the tarballs | 00:31 |
asalkeld | would you accept a patch like that | 00:31 |
Keybuk | but gave up with that | 00:31 |
Keybuk | nope | 00:31 |
Keybuk | oh | 00:31 |
Keybuk | depends | 00:31 |
Keybuk | if you can get the files included in the source, I probably would | 00:31 |
asalkeld | like: configure --disable-nil-dbus-tool | 00:32 |
Keybuk | no, again, it's necessary | 00:32 |
asalkeld | I worry about the generated c files | 00:32 |
Keybuk | nih-dbus-tool is actually being split out anyway | 00:32 |
Keybuk | I have most of the code ready to make libnih a separate module | 00:32 |
Keybuk | so you'll just need it as a build-dependency on the build machine | 00:32 |
Keybuk | so that solves that particular problem | 00:33 |
asalkeld | ok, that sounds better | 00:33 |
asalkeld | are you going to release a version soon with this? | 00:33 |
Keybuk | yes | 00:33 |
asalkeld | cool | 00:33 |
Keybuk | probably this or next week | 00:33 |
asalkeld | ok, I'll wait | 00:33 |
asalkeld | thanks for your help | 00:34 |
Keybuk | no probs | 00:36 |
ion | keybuk: Btw, how about something like #define NIH_MUST(_e) ({ typeof (_e) __ret; for (int __i = 0; ! (__ret = (_e)); __i++) if (i >= 10) log_an_error ("...%s...", #_e); __ret; }) | 00:40 |
ion | Of course, the logger shouldn’t use NIH_MUST to alloc stuff. :-P | 00:42 |
Keybuk | ion: how will that help? | 00:47 |
=== pocek_ is now known as pocek | ||
=== pocek_ is now known as pocek | ||
=== robbiew1 is now known as robbiew | ||
=== sadmac_h1me is now known as sadmac_home |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!