| 1 |
Has anyone on the list started experimenting with building a bionic |
| 2 |
libc cross-toolchain yet with gcc-4.6 [1]? |
| 3 |
|
| 4 |
... seems fairly interesting. |
| 5 |
|
| 6 |
I just finished a small project that put together a fairly minimal |
| 7 |
bionic rootfs with nothing but the libc, libstdc++, libm, jamvm, |
| 8 |
gnu-classpath, dropbear, & busybox. I was surprised at how well things |
| 9 |
actually worked (both arm and x86_64). |
| 10 |
|
| 11 |
Surely there will continue to be some regressions. For example, bionic |
| 12 |
traps SIGUSR1 for debugging purposes - so any other binary that uses |
| 13 |
that for signalling will need to change it to SIGUSR2. Patching the |
| 14 |
build system was fairly straight forward. |
| 15 |
|
| 16 |
I think doing a cross-toolchain with crossdev might even be do-able... |
| 17 |
just apply a couple of patches to android's build/ tree, build libc, |
| 18 |
libm, libstdc++. |
| 19 |
|
| 20 |
It would be freaking sweet to just be able to 'emerge' libraries for |
| 21 |
Android instead of going through the often painful process of |
| 22 |
retooling it for Android.mk. It would an accomplishment to bootstrap |
| 23 |
gcc, that's for sure. |
| 24 |
|
| 25 |
armv7a-neon-linux-bionic-emerge world |
| 26 |
|
| 27 |
;-) |
| 28 |
|
| 29 |
Cheers, |
| 30 |
|
| 31 |
C |
| 32 |
|
| 33 |
[1] http://www.phoronix.com/scan.php?page=news_item&px=OTI1NQ |
Archives