| 1 |
On 08/22/2011 01:41 PM, Alex Schuster wrote: |
| 2 |
>>> > > I don't think this is true - as long as the CHOST is identical, there |
| 3 |
>>> > > should be no problem. |
| 4 |
>> > |
| 5 |
>> > CHOST defines the arch (i686, amd64, arm ..) whilst CFLAGS control gcc |
| 6 |
>> > behavior and the binary code generation produced by compiler. in my case |
| 7 |
>> > core2quad (q8300) i'm using in the desktop supports sse4.1 instruction |
| 8 |
>> > set and notebook powered with core2duo (t7600) does not have that cpu |
| 9 |
>> > feature. having option `-msse4.1' set in CFLAGS at desktop side will |
| 10 |
>> > causes frequent compilation failures (initiated by distcc) or, in worst |
| 11 |
>> > case - arbitrary crashes at notebook when running binaries compiled in |
| 12 |
>> > distributed distcc environment |
| 13 |
> But how could the gcc on the desktop, called via distcc, even see the CFLAGS |
| 14 |
> in the desktops make.conf? This just cannot be. |
| 15 |
> I have been using march=athlon-xp on my desktop, and yet used it as distcc |
| 16 |
> server for my Pentium-III notebook for years. Worked fine. |
| 17 |
|
| 18 |
The arguments sent to gcc tell it what kind of object gcc is supposed to |
| 19 |
create. If gcc cannot create the object you request because it's not the |
| 20 |
right "kind" of gcc, implies that, within the same gcc version, it is |
| 21 |
unable to reproduce object files. |
Archives