configure: Don't assume an aligned stack on clang on windows

Message ID 1520894304-6249-1-git-send-email-martin@martin.st
State Committed
Commit 847190ebd99ffd57dc89bd568a33bf2d5c424129
Headers show
Series
  • configure: Don't assume an aligned stack on clang on windows
Related show

Commit Message

Martin Storsjö March 12, 2018, 10:38 p.m.
If we'd enable a 16 byte aligned stack, clang/llvm would also assume
that alignment everywhere and produce code that strictly requires it.
That would require adding realignment (via attribute_align_arg) on every
single public library function or enable -mstackrealign (which does the
same on every single function).

Also relatedly; the parameter currently tested (-mllvm
-stack-alignment=16) hasn't actually been supported for quite some
time; current clang versions use -mstack-alignment=16 for the same.
Actually testing for that parameter would be a different change
though, since it has a real risk of changing behaviour on any other
platform where clang is used.
---
 configure | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

Comments

Luca Barbato March 12, 2018, 11:09 p.m. | #1
On 12/03/2018 23:38, Martin Storsjö wrote:
> If we'd enable a 16 byte aligned stack, clang/llvm would also assume
> that alignment everywhere and produce code that strictly requires it.
> That would require adding realignment (via attribute_align_arg) on every
> single public library function or enable -mstackrealign (which does the
> same on every single function).
> 
> Also relatedly; the parameter currently tested (-mllvm
> -stack-alignment=16) hasn't actually been supported for quite some
> time; current clang versions use -mstack-alignment=16 for the same.
> Actually testing for that parameter would be a different change
> though, since it has a real risk of changing behaviour on any other
> platform where clang is used.
> ---
>   configure | 11 ++++++++++-
>   1 file changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/configure b/configure
> index b91be32..7042635 100755
> --- a/configure
> +++ b/configure
> @@ -4955,7 +4955,16 @@ elif enabled gcc; then
>   elif enabled llvm_gcc; then
>       check_cflags -mllvm -stack-alignment=16
>   elif enabled clang; then
> -    check_cflags -mllvm -stack-alignment=16
> +    if [ "$target_os" = "mingw32" -o "$target_os" = "win32" ] && enabled x86_32; then
> +        # Clang doesn't support maintaining alignment without assuming the
> +        # same alignment in every function. If 16 byte alignment would be
> +        # enabled, one would also have to either add attribute_align_arg on
> +        # every single entry point into the libraries or enable -mstackrealign
> +        # (doing stack realignment in every single function).
> +        disable aligned_stack
> +    else
> +        check_cflags -mllvm -stack-alignment=16
> +    fi
>       check_cflags -Qunused-arguments
>       check_cflags -Werror=implicit-function-declaration
>       check_cflags -Werror=missing-prototypes
> 

Ok.

Patch

diff --git a/configure b/configure
index b91be32..7042635 100755
--- a/configure
+++ b/configure
@@ -4955,7 +4955,16 @@  elif enabled gcc; then
 elif enabled llvm_gcc; then
     check_cflags -mllvm -stack-alignment=16
 elif enabled clang; then
-    check_cflags -mllvm -stack-alignment=16
+    if [ "$target_os" = "mingw32" -o "$target_os" = "win32" ] && enabled x86_32; then
+        # Clang doesn't support maintaining alignment without assuming the
+        # same alignment in every function. If 16 byte alignment would be
+        # enabled, one would also have to either add attribute_align_arg on
+        # every single entry point into the libraries or enable -mstackrealign
+        # (doing stack realignment in every single function).
+        disable aligned_stack
+    else
+        check_cflags -mllvm -stack-alignment=16
+    fi
     check_cflags -Qunused-arguments
     check_cflags -Werror=implicit-function-declaration
     check_cflags -Werror=missing-prototypes