1 |
There has already been som effort in this area. I think the conclusion |
2 |
was that the syntax of /proc/cpuinfo was much to inconsistent to be |
3 |
usable. |
4 |
|
5 |
A patch to change the syntax of cpuinfo is probably the right solution. |
6 |
Just pull the supported archs from the compiler compiling the kernel and |
7 |
put a suitable arch in cpuinfo. |
8 |
Hint: gcc-3.2/gcc/config/i386/i386.c |
9 |
|
10 |
-John |
11 |
|
12 |
> > > 4. It would be really nice if we had a tool to automate the detection of |
13 |
> > > the build architecture (CHOST, -march & -mcpu) based on the info from |
14 |
> > > /proc/cpu to help the newbie with the installation procedure. |
15 |
> > |
16 |
> > considering the *large* number of /proc/cpuinfo and the fact that one set |
17 |
> does |
18 |
> > not map properly onto the other set, this is impossible to a degree |
19 |
> > (-march/-mcpu) |
20 |
> > |
21 |
> > in terms of CHOST, simply picking the right stage tarball would be enough |
22 |
> ... |
23 |
> > most people shouldnt have to change it |
24 |
> |
25 |
> Atleast for x86 there's a fairly small number of /proc/cpuinfo layouts. I |
26 |
> cannot speak for other architectures though. And its all in the kernel |
27 |
> sources, imho its fairly simple to write an automated detection system. |
28 |
> |
29 |
> CHOST will probably be changed by all those using stage1. |