From 18d90d75195fb56834360c42a4f06318afa3cccc Mon Sep 17 00:00:00 2001
From: Dave Rodgman <dave.rodgman@arm.com>
Date: Fri, 19 Jan 2024 14:08:04 +0000
Subject: [PATCH] Make mbedtls_xor always inline

Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
---
 library/common.h        | 18 +++++++++++++++++-
 library/platform_util.c |  9 ---------
 2 files changed, 17 insertions(+), 10 deletions(-)

diff --git a/library/common.h b/library/common.h
index faefd64ea..760dff49e 100644
--- a/library/common.h
+++ b/library/common.h
@@ -158,6 +158,12 @@ static inline const unsigned char *mbedtls_buffer_offset_const(
     return p == NULL ? NULL : p + n;
 }
 
+/* Always inline mbedtls_xor for similar reasons as mbedtls_xor_no_simd. */
+#if defined(__IAR_SYSTEMS_ICC__)
+#pragma inline = forced
+#elif defined(__GNUC__)
+__attribute__((always_inline))
+#endif
 /**
  * Perform a fast block XOR operation, such that
  * r[i] = a[i] ^ b[i] where 0 <= i < n
@@ -169,7 +175,10 @@ static inline const unsigned char *mbedtls_buffer_offset_const(
  * \param   b Pointer to input (buffer of at least \p n bytes)
  * \param   n Number of bytes to process.
  */
-inline void mbedtls_xor(unsigned char *r, const unsigned char *a, const unsigned char *b, size_t n)
+static inline void mbedtls_xor(unsigned char *r,
+                               const unsigned char *a,
+                               const unsigned char *b,
+                               size_t n)
 {
     size_t i = 0;
 #if defined(MBEDTLS_EFFICIENT_UNALIGNED_ACCESS)
@@ -209,6 +218,13 @@ inline void mbedtls_xor(unsigned char *r, const unsigned char *a, const unsigned
     }
 }
 
+/* Always inline mbedtls_xor_no_simd as we see significant perf regressions when it does not get
+ * inlined (e.g., observed about 3x perf difference in gcm_mult_largetable with gcc 7 - 12) */
+#if defined(__IAR_SYSTEMS_ICC__)
+#pragma inline = forced
+#elif defined(__GNUC__)
+__attribute__((always_inline))
+#endif
 /**
  * Perform a fast block XOR operation, such that
  * r[i] = a[i] ^ b[i] where 0 <= i < n
diff --git a/library/platform_util.c b/library/platform_util.c
index e79fc5c8e..9f5dcb874 100644
--- a/library/platform_util.c
+++ b/library/platform_util.c
@@ -217,15 +217,6 @@ struct tm *mbedtls_platform_gmtime_r(const mbedtls_time_t *tt,
 void (*mbedtls_test_hook_test_fail)(const char *, int, const char *);
 #endif /* MBEDTLS_TEST_HOOKS */
 
-/*
- * Provide external definitions of some inline functions so that the compiler
- * has the option to not inline them
- */
-extern inline void mbedtls_xor(unsigned char *r,
-                               const unsigned char *a,
-                               const unsigned char *b,
-                               size_t n);
-
 #if defined(MBEDTLS_HAVE_TIME) && !defined(MBEDTLS_PLATFORM_MS_TIME_ALT)
 
 #include <time.h>