1 |
On 07/01/2013 04:48 PM, heroxbd wrote: |
2 |
> Dear Guys and Gals, |
3 |
> |
4 |
> Gentoo on Android is about running Gentoo natively in a directory prefix |
5 |
> on Android devices[1] in parallel with native Android, mentored by Luca. |
6 |
> |
7 |
> Progress report for the past 3 days, |
8 |
> |
9 |
> 1. gcc specs hack is not that interesting as it first looks. |
10 |
> - parsing or generating specs script is no fun. |
11 |
> The best parser/generator is gcc itself |
12 |
> - appending simple specs (suches -dynamic-linker=/xx/xx.so) feel |
13 |
> not clever, appending --with-spec to configure either. |
14 |
> |
15 |
> Therefore, I decide not to use specs hack as the main building |
16 |
> block. At the same time, it can be handy to do manual tuning |
17 |
> with it. |
18 |
|
19 |
That must be reconsidered later to understand what's the problem and if |
20 |
upstream can help on that. |
21 |
|
22 |
> Relying on binutils wrapper is no good either. The solution |
23 |
> finally falls on a mixed one between sysroot and native paths |
24 |
> methods (ref. table in [2]). Where I make use of eprefixify of |
25 |
> prefix.eclass to modify location of dynamic linker for *run |
26 |
> time* and sysroot for *compile time*. |
27 |
|
28 |
It isn't perfect, but as long it gives result I can live with that. |
29 |
|
30 |
> 2. Prefix/libc armhf works rock solid. It is now being "emerge -e |
31 |
> @world" with the help of many distcc cross compilers. I have to |
32 |
> leave my cell phone home for a constant power supply and |
33 |
> ethernet connection (yes! USB ethernet adapter) though. |
34 |
|
35 |
Make sure it is near a heatsink (AC or such). |
36 |
|
37 |
> The bug is related to VDSO. A full description of the problem is |
38 |
> on stackoverflow[3]. I'd be very appreciated for any hints. |
39 |
|
40 |
uclibc or musl might or might not help? |
41 |
|
42 |
lu |