| 1 |
On Tuesday, 5 December 2017 20:45:21 GMT Michael Orlitzky wrote: |
| 2 |
> On 12/05/2017 03:26 PM, Corbin wrote: |
| 3 |
> > In "packages" that throw out the "CFLAGS / CXXFLAGS" values in the |
| 4 |
> > end-users "make.conf" and substitute their own ... how will that be |
| 5 |
> > handled? |
| 6 |
> The GCC ebuilds all use toolchain.eclass which is incomprehensible to |
| 7 |
> me, but it looks like the default behavior for gcc-6.x is to pass |
| 8 |
> "--enable-default-pie" and "--enable-default-ssp" to the build process |
| 9 |
> of GCC itself. That changes the default behavior of GCC to (as the names |
| 10 |
> say) enable PIE and SSP by default. |
| 11 |
> |
| 12 |
> Consequently, if a package ignores your CFLAGS, the PIE/SSP should still |
| 13 |
> take effect, because GCC does them by default. Only a package that adds |
| 14 |
> its own -no-pie flag (for example) would cause problems. |
| 15 |
|
| 16 |
I just noticed chromium shows (pic) in brackets, which I assume it means |
| 17 |
forced. |
| 18 |
|
| 19 |
-- |
| 20 |
Regards, |
| 21 |
Mick |
Archives