crosstool-ng/packages/gcc/7.5.0/0025-aarch64-Add-bytes_below_hard_fp-to-frame-info.patch

From e97c93de7ceadad75774f897424a50ece3215129 Mon Sep 17 00:00:00 2001
From: Richard Sandiford <richard.sandiford@arm.com>
Date: Fri, 16 Jun 2023 16:55:12 +0100
Subject: [PATCH 25/30] aarch64: Add bytes_below_hard_fp to frame info

The frame layout code currently hard-codes the assumption that
the number of bytes below the saved registers is equal to the
size of the outgoing arguments.  This patch abstracts that
value into a new field of aarch64_frame.

gcc/
	* config/aarch64/aarch64.h (aarch64_frame::bytes_below_hard_fp): New
	field.
	* config/aarch64/aarch64.c (aarch64_layout_frame): Initialize it,
	and use it instead of crtl->outgoing_args_size.
	(aarch64_get_separate_components): Use bytes_below_hard_fp instead
	of outgoing_args_size.
	(aarch64_process_components): Likewise.
---
 gcc/config/aarch64/aarch64.c | 40 +++++++++++++++++++++---------------
 gcc/config/aarch64/aarch64.h |  6 +++++-
 2 files changed, 28 insertions(+), 18 deletions(-)

diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
index bb4f6020e5df..12df8bb783c7 100644
--- a/gcc/config/aarch64/aarch64.c
+++ b/gcc/config/aarch64/aarch64.c
@@ -2826,6 +2826,8 @@ aarch64_layout_frame (void)
 	last_fp_reg = regno;
       }
 
+  cfun->machine->frame.bytes_below_hard_fp = crtl->outgoing_args_size;
+
   if (frame_pointer_needed)
     {
       /* FP and LR are placed in the linkage record.  */
@@ -2885,7 +2887,7 @@ aarch64_layout_frame (void)
 
   cfun->machine->frame.frame_size
     = ROUND_UP (cfun->machine->frame.hard_fp_offset
-		+ crtl->outgoing_args_size,
+		+ cfun->machine->frame.bytes_below_hard_fp,
 		STACK_BOUNDARY / BITS_PER_UNIT);
 
   cfun->machine->frame.locals_offset = cfun->machine->frame.saved_varargs_size;
@@ -2904,32 +2906,33 @@ aarch64_layout_frame (void)
   if (cfun->machine->frame.saved_regs_size == 0)
     cfun->machine->frame.initial_adjust = cfun->machine->frame.frame_size;
   else if (cfun->machine->frame.frame_size < max_push_offset
-	   && crtl->outgoing_args_size == 0)
+	   && cfun->machine->frame.bytes_below_hard_fp == 0)
     {
-      /* Simple, small frame with no outgoing arguments:
+      /* Simple, small frame with no data below the saved registers.
 	 stp reg1, reg2, [sp, -frame_size]!
 	 stp reg3, reg4, [sp, 16]  */
       cfun->machine->frame.callee_adjust = cfun->machine->frame.frame_size;
     }
-  else if ((crtl->outgoing_args_size
+  else if ((cfun->machine->frame.bytes_below_hard_fp
 	    + cfun->machine->frame.saved_regs_size < 512)
 	   && !(cfun->calls_alloca
 		&& cfun->machine->frame.hard_fp_offset < max_push_offset))
     {
-      /* Frame with small outgoing arguments:
+      /* Frame with small area below the saved registers:
 	 sub sp, sp, frame_size
-	 stp reg1, reg2, [sp, outgoing_args_size]
-	 stp reg3, reg4, [sp, outgoing_args_size + 16]  */
+	 stp reg1, reg2, [sp, bytes_below_hard_fp]
+	 stp reg3, reg4, [sp, bytes_below_hard_fp + 16]  */
       cfun->machine->frame.initial_adjust = cfun->machine->frame.frame_size;
       cfun->machine->frame.callee_offset
 	= cfun->machine->frame.frame_size - cfun->machine->frame.hard_fp_offset;
     }
   else if (cfun->machine->frame.hard_fp_offset < max_push_offset)
     {
-      /* Frame with large outgoing arguments but a small local area:
+      /* Frame with large area below the saved registers, but with a
+	 small area above:
 	 stp reg1, reg2, [sp, -hard_fp_offset]!
 	 stp reg3, reg4, [sp, 16]
-	 sub sp, sp, outgoing_args_size  */
+	 sub sp, sp, bytes_below_hard_fp  */
       cfun->machine->frame.callee_adjust = cfun->machine->frame.hard_fp_offset;
       cfun->machine->frame.final_adjust
 	= cfun->machine->frame.frame_size - cfun->machine->frame.callee_adjust;
@@ -2945,17 +2948,19 @@ aarch64_layout_frame (void)
       cfun->machine->frame.callee_adjust = varargs_and_saved_regs_size;
       cfun->machine->frame.final_adjust
 	= cfun->machine->frame.frame_size - cfun->machine->frame.callee_adjust;
+      cfun->machine->frame.bytes_below_hard_fp
+	= cfun->machine->frame.final_adjust;
       cfun->machine->frame.hard_fp_offset = cfun->machine->frame.callee_adjust;
       cfun->machine->frame.locals_offset = cfun->machine->frame.hard_fp_offset;
     }
   else
     {
-      /* Frame with large local area and outgoing arguments using frame pointer:
+      /* General case:
 	 sub sp, sp, hard_fp_offset
 	 stp x29, x30, [sp, 0]
 	 add x29, sp, 0
 	 stp reg3, reg4, [sp, 16]
-	 sub sp, sp, outgoing_args_size  */
+	 sub sp, sp, bytes_below_hard_fp  */
       cfun->machine->frame.initial_adjust = cfun->machine->frame.hard_fp_offset;
       cfun->machine->frame.final_adjust
 	= cfun->machine->frame.frame_size - cfun->machine->frame.initial_adjust;
@@ -3313,9 +3318,11 @@ aarch64_get_separate_components (void)
     if (aarch64_register_saved_on_entry (regno))
       {
 	HOST_WIDE_INT offset = cfun->machine->frame.reg_offset[regno];
+
+	/* Get the offset relative to the register we'll use.  */
 	if (!frame_pointer_needed)
-	  offset += cfun->machine->frame.frame_size
-		    - cfun->machine->frame.hard_fp_offset;
+	  offset += cfun->machine->frame.bytes_below_hard_fp;
+
 	/* Check that we can access the stack slot of the register with one
 	   direct load with no adjustments needed.  */
 	if (offset_12bit_unsigned_scaled_p (DImode, offset))
@@ -3418,8 +3425,8 @@ aarch64_process_components (sbitmap components, bool prologue_p)
       rtx reg = gen_rtx_REG (mode, regno);
       HOST_WIDE_INT offset = cfun->machine->frame.reg_offset[regno];
       if (!frame_pointer_needed)
-	offset += cfun->machine->frame.frame_size
-		  - cfun->machine->frame.hard_fp_offset;
+	offset += cfun->machine->frame.bytes_below_hard_fp;
+
       rtx addr = plus_constant (Pmode, ptr_reg, offset);
       rtx mem = gen_frame_mem (mode, addr);
 
@@ -3460,8 +3467,7 @@ aarch64_process_components (sbitmap components, bool prologue_p)
       /* REGNO2 can be saved/restored in a pair with REGNO.  */
       rtx reg2 = gen_rtx_REG (mode, regno2);
       if (!frame_pointer_needed)
-	offset2 += cfun->machine->frame.frame_size
-		  - cfun->machine->frame.hard_fp_offset;
+	offset2 += cfun->machine->frame.bytes_below_hard_fp;
       rtx addr2 = plus_constant (Pmode, ptr_reg, offset2);
       rtx mem2 = gen_frame_mem (mode, addr2);
       rtx set2 = prologue_p ? gen_rtx_SET (mem2, reg2)
diff --git a/gcc/config/aarch64/aarch64.h b/gcc/config/aarch64/aarch64.h
index ddf833ebfe85..6dc6ef2b989b 100644
--- a/gcc/config/aarch64/aarch64.h
+++ b/gcc/config/aarch64/aarch64.h
@@ -558,9 +558,13 @@ struct GTY (()) aarch64_frame
   HOST_WIDE_INT saved_varargs_size;
 
   /* The size of the saved callee-save int/FP registers.  */
-
   HOST_WIDE_INT saved_regs_size;
 
+  /* The number of bytes between the bottom of the static frame (the bottom
+     of the outgoing arguments) and the hard frame pointer.  This value is
+     always a multiple of STACK_BOUNDARY.  */
+  HOST_WIDE_INT bytes_below_hard_fp;
+
   /* Offset from the base of the frame (incomming SP) to the
      top of the locals area.  This value is always a multiple of
      STACK_BOUNDARY.  */
-- 
2.42.0