diff mbox series

[2/9] powerpc/64s: Disable pcrel code model on Clang

Message ID 20230426055848.402993-3-npiggin@gmail.com (mailing list archive)
State Accepted
Commit 169f8997968ab620d750d9a45e15c5288d498356
Headers show
Series powerpc: Build fixes | expand

Commit Message

Nicholas Piggin April 26, 2023, 5:58 a.m. UTC 
Clang has a bug that casues the pcrel code model not to be used when any of
-msoft-float, -mno-altivec, or -mno-vsx are set. Leaving these off causes
FP/vector instructions to be generated, causing crashes. So disable pcrel
for clang for now.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/Kconfig | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
diff mbox series

Patch

diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index 7a0f25a09759..261e9453b43c 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -8,7 +8,12 @@  config CC_HAS_PREFIXED
 	def_bool PPC64 && $(cc-option, -mcpu=power10 -mprefixed)
 
 config CC_HAS_PCREL
-	def_bool PPC64 && $(cc-option, -mcpu=power10 -mpcrel)
+	# Clang has a bug (https://github.com/llvm/llvm-project/issues/62372)
+	# where pcrel code is not generated if -msoft-float, -mno-altivec, or
+	# -mno-vsx options are also given. Without these options, fp/vec
+	# instructions are generated from regular kernel code. So Clang can't
+	# do pcrel yet.
+	def_bool PPC64 && CC_IS_GCC &&  $(cc-option, -mcpu=power10 -mpcrel)
 
 config 32BIT
 	bool