Spender's mindset is even worse though.
Kernel can't be the one responsible for protection against this because it doesn't control stack setup. (Only for initial thread.)
-
-
KSTACKOVERFLOW / VMAP_STACK move the kernel stack to virtual memory where they get guard pages, it's not about userspace stack exhaustion.
-
Oh, misread context then. Sorry.
End of conversation
New conversation -
-
-
It's compiler's fault for not doing the same sort of stack probing they already do on windows targets.
-
I'm glad to see work starting to progress in Clang/LLVM again as a result
-
Note that it's once again a set of changes for Rust, which
@whitequark and@pcwalton revived. Clang would need to wire up a switch to it. -
Maybe -fstack-check (instead of it the very poor decision to make it a no-op...) or maybe a new switch breaking away from weird GCC choices.
-
Clang/LLVM already had stack probes for Windows which are unconditionally enabled but not wired up in a way that it could be used elsewhere.
-
Apart from a -fstack-check implementation in clang this also needs compiler-rt code, that's trickier because of large support matrix
-
I wouldn't be the one to write it, though feel free to tag me as a reviewer
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.