arm: Consistently use proper interworking function returns

Use "bx lr", or "pop {lr}", which do proper mode switching
between thumb and arm modes. A plain "mov pc, lr" does not switch
from thumb mode to arm mode (while in arm mode, it does switch
mode for a thumb caller).

This is normally not an issue, as CONFIG_THUMB only is enabled if
the C compiler defaults to thumb; but stick to patterns that can
do mode switching if needed, for consistency.

Signed-off-by: Martin Storsjö <martin@martin.st>
(cherry picked from commit 77e6293735)

libswresample/arm/resample.S

@@ -30,7 +30,7 @@ function ff_resample_common_apply_filter_x4_float_neon, export=1
     vpadd.f32           d0, d0, d1                                     @ pair adding of the 4x32-bit accumulated values
     vpadd.f32           d0, d0, d0                                     @ pair adding of the 4x32-bit accumulator values
     vst1.32             {d0[0]}, [r0]                                  @ write accumulator
-    mov pc, lr
+    bx                  lr
 endfunc
 
 function ff_resample_common_apply_filter_x8_float_neon, export=1
@@ -46,7 +46,7 @@ function ff_resample_common_apply_filter_x8_float_neon, export=1
     vpadd.f32           d0, d0, d1                                     @ pair adding of the 4x32-bit accumulated values
     vpadd.f32           d0, d0, d0                                     @ pair adding of the 4x32-bit accumulator values
     vst1.32             {d0[0]}, [r0]                                  @ write accumulator
-    mov pc, lr
+    bx                  lr
 endfunc
 
 function ff_resample_common_apply_filter_x4_s16_neon, export=1
@@ -59,7 +59,7 @@ function ff_resample_common_apply_filter_x4_s16_neon, export=1
     vpadd.s32           d0, d0, d1                                     @ pair adding of the 4x32-bit accumulated values
     vpadd.s32           d0, d0, d0                                     @ pair adding of the 4x32-bit accumulator values
     vst1.32             {d0[0]}, [r0]                                  @ write accumulator
-    mov pc, lr
+    bx                  lr
 endfunc
 
 function ff_resample_common_apply_filter_x8_s16_neon, export=1
@@ -73,5 +73,5 @@ function ff_resample_common_apply_filter_x8_s16_neon, export=1
     vpadd.s32           d0, d0, d1                                     @ pair adding of the 4x32-bit accumulated values
     vpadd.s32           d0, d0, d0                                     @ pair adding of the 4x32-bit accumulator values
     vst1.32             {d0[0]}, [r0]                                  @ write accumulator
-    mov pc, lr
+    bx                  lr
 endfunc

libswscale/arm/hscale.S

@@ -65,6 +65,5 @@ function ff_hscale_8_to_15_neon, export=1
     subs                r2, #2                                         @ dstW -= 2
     bgt                 1b                                             @ loop until end of line
     vpop                {q4-q7}
-    pop                 {r4-r12, lr}
-    mov pc, lr
+    pop                 {r4-r12, pc}
 endfunc

libswscale/arm/output.S

@@ -73,6 +73,5 @@ function ff_yuv2planeX_8_neon, export=1
     subs                r4, r4, #8                                     @ dstW -= 8
     bgt                 2b                                             @ loop until width is consumed
     vpop                {q4-q7}
-    pop                 {r4-r12, lr}
-    mov                 pc, lr
+    pop                 {r4-r12, pc}
 endfunc

libswscale/arm/yuv2rgb_neon.S

@@ -262,8 +262,7 @@ function ff_\ifmt\()_to_\ofmt\()_neon, export=1
     increment_and_test_\ifmt
     bgt                 1b
     vpop                {q4-q7}
-    pop                 {r4-r12, lr}
-    mov                 pc, lr
+    pop                 {r4-r12, pc}
 endfunc
 .endm