[compiler-rt] [compiler-rt] Add support for big endian for Arm's __negdf2vfp (PR #127096)

[Victor Campos via llvm-commits]

https://github.com/vhscampos created https://github.com/llvm/llvm-project/pull/127096

In soft floating-point ABI, this function takes the double argument as a pair of registers r0 and r1.

The ordering of these two registers follow the endianness rules, therefore the register on which the bit flipping must happen depends on the endianness.

>From a4f2f191bd4a161e35223002afda733b87116716 Mon Sep 17 00:00:00 2001
From: Victor Campos <victor.campos at arm.com>
Date: Thu, 13 Feb 2025 17:00:15 +0000
Subject: [PATCH] [compiler-rt] Add support for big endian for Arm's
 __negdf2vfp

In soft floating-point ABI, this function takes the double argument as a
pair of registers r0 and r1.

The ordering of these two registers follow the endianness rules,
therefore the register on which the bit flipping must happen depends on
the endianness.
---
 compiler-rt/lib/builtins/arm/negdf2vfp.S | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/compiler-rt/lib/builtins/arm/negdf2vfp.S b/compiler-rt/lib/builtins/arm/negdf2vfp.S
index b7cf91877e38c..329c6de757f68 100644
--- a/compiler-rt/lib/builtins/arm/negdf2vfp.S
+++ b/compiler-rt/lib/builtins/arm/negdf2vfp.S
@@ -20,7 +20,11 @@ DEFINE_COMPILERRT_FUNCTION(__negdf2vfp)
 #if defined(COMPILER_RT_ARMHF_TARGET)
 	vneg.f64 d0, d0
 #else
-	eor	r1, r1, #-2147483648	// flip sign bit on double in r0/r1 pair
+#if _YUGA_BIG_ENDIAN
+	eor	r0, r0, #0x80000000	// flip sign bit on double in r0/r1 pair
+#else
+	eor	r1, r1, #0x80000000	// flip sign bit on double in r0/r1 pair
+#endif
 #endif
 	bx	lr
 END_COMPILERRT_FUNCTION(__negdf2vfp)