arch: arm: cortex_m: make reading tls pointer faster on v7m and v8m.main

Encoding T3 allows for an offset of up to 12bits in size allowing for a
single instruction instead of 3.

Signed-off-by: Wilfried Chauveau <[email protected]>

Author: ithinuel

arch/arm/core/cortex_m/swap_helper.S

@@ -159,9 +159,13 @@ SECTION_FUNC(TEXT, z_arm_pendsv)
 
 #if defined(CONFIG_THREAD_LOCAL_STORAGE)
     /* Grab the TLS pointer */
+#if defined(CONFIG_ARMV6_M_ARMV8_M_BASELINE)
     ldr r4, =_thread_offset_to_tls
     adds r4, r2, r4
     ldr r0, [r4]
+#else
+    ldr r0, [r2, #_thread_offset_to_tls]
+#endif
 
     /* For Cortex-M, store TLS pointer in a global variable,
      * as it lacks the process ID or thread ID register