Messages in this thread | | | From | Linus Torvalds <> | Date | Thu, 5 Apr 2018 19:02:11 -0700 | Subject | Re: [GIT PULL] x86/build changes for v4.17 |
| |
On Thu, Apr 5, 2018 at 3:51 PM, James Y Knight <jyknight@google.com> wrote: > > Unfortunately, that behavior is required by the standard, it's not up to > compiler optimization to change.
I actually mis-read your example - in your case it obviously does pass the array itself down to the call, and yes, it obviously needs to be allocated.
I had a test-case at one point where gcc avoided the stack allocation entirely for a regular array, but not for a VLA (of the same constant size) because the VLA logic is apparently different enough - even when the size of the array is a compile-time constant.
We had that issue because we had a lot of trouble coming up with a "max()" macro that was still an I-C-E (and we had a number of array sizes that used "max()"). So all the array sizes were compile-time constants, they just weren't traditional C arrays.
But now I can't recreate the thing. Maybe I had screwed up in my test-case somehow.
Linus
| |