diff options
| author | Sumit Semwal <sumit.semwal@linaro.org> | 2016-11-09 01:38:33 +0530 |
|---|---|---|
| committer | Sumit Semwal <sumit.semwal@linaro.org> | 2016-12-20 22:19:27 +0530 |
| commit | 9bf3c0bec59b16a068d9c5457696265de590717d (patch) | |
| tree | cc278051100344671f7807dde971fea2af201fc9 | |
| parent | 00556abdf94819c7fa3680247ff4a74bd3898ed7 (diff) | |
DO-NOT-UPSTREAM: arm64,llvm: prfm
Fix for prfm error on AArch64
Inspired by 4ddb5e09c741: ("DO-NOT-UPSTREAM arm64, LLVMLinux: prfm")
Upstream Commit: cd5e10bdf379:
("arm64: prefetch: don't provide spin_lock_prefetch with LSE") updates
spin_lock_prefetch() with a prfm instruction, which suffers from the
same issue mentioned in the 'inspired by' commit above.
This is visible as a build time error with clang:
./arch/arm64/include/asm/processor.h:185:8: error:
invalid operand in inline asm: 'prfm pstl1strm, ${0:a}'
"prfm pstl1strm, %a0",
This patch works around the same way as in the LLVMLinux commit.
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
| -rw-r--r-- | arch/arm64/include/asm/processor.h | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 8142505bac7e..ed7b2f4ff788 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -182,8 +182,8 @@ static inline void prefetchw(const void *ptr) static inline void spin_lock_prefetch(const void *ptr) { asm volatile(ARM64_LSE_ATOMIC_INSN( - "prfm pstl1strm, %a0", - "nop") : : "p" (ptr)); + "prfm pstl1strm, [%x0]", + "nop") : : "r" (ptr)); } #define HAVE_ARCH_PICK_MMAP_LAYOUT |