[libc-commits] [PATCH] D136595: [libc] Switch to new implementation of mem* functions

[Guillaume Chatelet via Phabricator via libc-commits]

gchatelet added a subscriber: sivachandra.
gchatelet added inline comments.


================
Comment at: libc/src/string/memory_utils/memcpy_implementations.h:127
+#elif defined(LLVM_LIBC_ARCH_ARM)
+  return inline_memcpy_embedded_tiny(dst, src, count);
 #else
----------------
gchatelet wrote:
> courbet wrote:
> > why is `ARM` always `embedded_tiny` ? Shouldn't that be orthogonal ? I could imagine some tiny aarch64 processors...
> Fair enough, for now this is not customizable but it is expected to be in the future.
> 
> There has been some discussions on [[ https://discourse.llvm.org/t/llvm-libc-and-embedded-systems-follow-up-from-llvm-embedded-toolchains-working-group/63509 | discourse ]] but we need to better define what would be a reasonable default implementation.
> 
> Also it is unclear whether we want to customize //per function// or for the library as a whole
> 
> I will hold this patch until I'm clear whether this will impact our users or not.
@sivachandra confirmed that `ARM` support is minimal for now and that we don't have active users . `aarch64` is solely targeted at ARM servers. Of course this can change but I think the patch is safe to land as it is now. Let's refine the needs and iterate in subsequent patches.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136595/new/

https://reviews.llvm.org/D136595