| 1 |
On 09/05/18 19:18, Martin Vaeth wrote: |
| 2 |
> As mentioned, I wonder why gcc/clang do not yet support this |
| 3 |
> horribly slow but spectre-safe option. It can't be that hard to |
| 4 |
> implement in the actual code-producing back-end. |
| 5 |
|
| 6 |
Given the response by the gcc team to security people complaining that |
| 7 |
gcc was optimising out security-sensitive code (namely, a two-fingered |
| 8 |
salute near enough), I doubt the gcc team would have any interest in |
| 9 |
optimisations that SLOWED DOWN the resultant code. |
| 10 |
|
| 11 |
I suspect that might be one of the forces driving the kernel towards |
| 12 |
CLANG - a development team that is not obsessed with performance at the |
| 13 |
expense of breaking any code that uses undefined features. |
| 14 |
Unfortunately, when dealing with hardware, one is forced to rely on |
| 15 |
undefined features. A strong point of C, until the compiler decides to |
| 16 |
go "rogue" on you ... |
| 17 |
|
| 18 |
Cheers, |
| 19 |
Wol |
Archives