From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-417978-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 83135 invoked by alias); 26 Dec 2015 01:57:31 -0000
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
Received: (qmail 82999 invoked by uid 89); 26 Dec 2015 01:57:30 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=3.7 required=5.0 tests=AWL,BAYES_99,KAM_LAZY_DOMAIN_SECURITY,KAM_LOTSOFHASH,RP_MATCHES_RCVD autolearn=no version=3.3.2 spammy=88, tarm, clearing, s18
X-HELO: foss.arm.com
Received: from foss.arm.com (HELO foss.arm.com) (217.140.101.70) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Sat, 26 Dec 2015 01:57:24 +0000
Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.72.51.249])	by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 628DF49;	Fri, 25 Dec 2015 17:56:55 -0800 (PST)
Received: from SHAWIN202 (unknown [10.164.6.62])	by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id DDBE53F308;	Fri, 25 Dec 2015 17:57:20 -0800 (PST)
From: "Thomas Preud'homme" <thomas.preudhomme@foss.arm.com>
To: <gcc-patches@gcc.gnu.org>,	"Richard Earnshaw" <Richard.Earnshaw@arm.com>,	"Ramana Radhakrishnan" <ramana.radhakrishnan@arm.com>,	"Kyrylo Tkachov" <kyrylo.tkachov@arm.com>
Subject: [RFC][PATCH, ARM 7/8] ARMv8-M Security Extension's cmse_nonsecure_call: use __gnu_cmse_nonsecure_call
Date: Sat, 26 Dec 2015 01:57:00 -0000
Message-ID: <001b01d13f80$c222f650$4668e2f0$@foss.arm.com>
MIME-Version: 1.0
Content-Type: text/plain;	charset="utf-8"
Content-Transfer-Encoding: quoted-printable
X-SW-Source: 2015-12/txt/msg02154.txt.bz2

[Sending on behalf of Andre Vieira]

Hello,

This patch extends support for the ARMv8-M Security Extensions 'cmse_nonsec=
ure_call' to use a new library function '__gnu_cmse_nonsecure_call'. This l=
ibrary function is responsible for (without using r0-r3 or d0-d7):
1) saving and clearing all callee-saved registers using the secure stack
2) clearing the LSB of the address passed in r4 and using blxns to 'jump' t=
o it
3) clearing ASPR, including the 'ge bits' if DSP is enabled
4) clearing FPSCR if using non-soft float-abi
5) restoring callee-saved registers.

The decisions whether to include DSP 'ge bits' clearing and floating point =
registers (single/double precision) all depends on the multilib used.

See Section 5.5 of ARM=C2=AEv8-M Security Extensions (http://infocenter.arm=
.com/help/topic/com.arm.doc.ecm0359818/index.html).

*** gcc/ChangeLog ***

*** gcc/ChangeLog ***
2015-10-27  Andre Vieira        <andre.simoesdiasvieira@arm.com>
            Thomas Preud'homme  <thomas.preudhomme@arm.com>

        * gcc/config/arm/arm.c (detect_cmse_nonsecure_call): New.
          (cmse_nonsecure_call_clear_caller_saved): New.
        * gcc/config/arm/arm-protos.h (detect_cmse_nonsecure_call): New.
        * gcc/config/arm/arm.md (call): Handle cmse_nonsecure_entry.
          (call_value): Likewise.
          (nonsecure_call_internal): New.
          (nonsecure_call_value_internal): New.
        * gcc/config/arm/thumb1.md (*nonsecure_call_reg_thumb1_v5): New.
          (*nonsecure_call_value_reg_thumb1_v5): New.
        * gcc/config/arm/thumb2.md (*nonsecure_call_reg_thumb2): New.
          (*nonsecure_call_value_reg_thumb2): New.
        * gcc/config/arm/unspecs.md (UNSPEC_NONSECURE_MEM): New.
        * libgcc/config/arm/cmse_nonsecure_call.S: New.
	* libgcc/config/arm/t-arm: Compile cmse_nonsecure_call.S


*** gcc/testsuite/ChangeLog ***
2015-10-27  Andre Vieira        <andre.simoesdiasvieira@arm.com>
            Thomas Preud'homme  <thomas.preudhomme@arm.com>

        * gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-11.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-13.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-6.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-13.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-7.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-8.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-13.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-7.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-8.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-13.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-7.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-8.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-7.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-8.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-13.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-7.c: New.
	* gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-8.c: New.


diff --git a/gcc/config/arm/arm-protos.h b/gcc/config/arm/arm-protos.h
index 9ee8c333046d9a5beeeeb0487f7b710a5aff42d2..694ee02f534019a5fc9377757f3=
269dfe6ccfbc0 100644
--- a/gcc/config/arm/arm-protos.h
+++ b/gcc/config/arm/arm-protos.h
@@ -132,6 +132,7 @@ extern int arm_const_double_inline_cost (rtx);
 extern bool arm_const_double_by_parts (rtx);
 extern bool arm_const_double_by_immediates (rtx);
 extern void arm_emit_call_insn (rtx, rtx, bool);
+bool detect_cmse_nonsecure_call (tree);
 extern const char *output_call (rtx *);
 void arm_emit_movpair (rtx, rtx);
 extern const char *output_mov_long_double_arm_from_arm (rtx *);
diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c
index 4b4eea88cbec8e04d5b92210f0af2440ce6fb6e4..320f7b447501047a59ceef4f7de=
d2dadc2088664 100644
--- a/gcc/config/arm/arm.c
+++ b/gcc/config/arm/arm.c
@@ -17403,6 +17403,129 @@ note_invalid_constants (rtx_insn *insn, HOST_WIDE=
_INT address, int do_pushes)
   return;
 }
=20
+/* Saves callee saved registers, clears callee saved registers and caller =
saved
+   registers not used to pass arguments before a cmse_nonsecure_call.  And
+   restores the callee saved registers after.  */
+
+static void
+cmse_nonsecure_call_clear_caller_saved (void)
+{
+  basic_block bb;
+
+  FOR_EACH_BB_FN (bb, cfun)
+    {
+      rtx_insn *insn;
+
+      FOR_BB_INSNS (bb, insn)
+	{
+	  uint64_t to_clear_mask, float_mask;
+	  rtx_insn *seq;
+	  rtx pat, call, unspec, link, reg, cleared_reg, tmp;
+	  unsigned int regno, maxregno;
+	  rtx address;
+
+	  if (!NONDEBUG_INSN_P (insn))
+	    continue;
+
+	  if (!CALL_P (insn))
+	    continue;
+
+	  pat =3D PATTERN (insn);
+	  gcc_assert (GET_CODE (pat) =3D=3D PARALLEL && XVECLEN (pat, 0) > 0);
+	  call =3D XVECEXP (pat, 0, 0);
+
+	  /* Get the real call RTX if the insn sets a value, ie. returns.  */
+	  if (GET_CODE (call) =3D=3D SET)
+	      call =3D SET_SRC (call);
+
+	  /* Check if it is a cmse_nonsecure_call.  */
+	  unspec =3D XEXP (call, 0);
+	  if (GET_CODE (unspec) !=3D UNSPEC
+	      || XINT (unspec, 1) !=3D UNSPEC_NONSECURE_MEM)
+	    continue;
+
+	  /* Determine the caller-saved registers we need to clear.  */
+	  to_clear_mask =3D (1LL << (NUM_ARG_REGS)) - 1;
+	  maxregno =3D NUM_ARG_REGS - 1;
+	  if (TARGET_HARD_FLOAT && TARGET_VFP)
+	    {
+	      float_mask =3D (1LL << (D7_VFP_REGNUM + 1)) - 1;
+	      float_mask &=3D ~((1LL << FIRST_VFP_REGNUM) - 1);
+	      to_clear_mask |=3D float_mask;
+	      maxregno =3D D7_VFP_REGNUM;
+	    }
+
+	  /* Make sure the register used to hold the function address is not
+	     cleared.  */
+	  address =3D RTVEC_ELT (XVEC (unspec, 0), 0);
+	  gcc_assert (MEM_P (address));
+	  gcc_assert (REG_P (XEXP (address, 0)));
+	  to_clear_mask &=3D ~(1LL << REGNO (XEXP (address, 0)));
+
+	  /* Set basic block of call insn so that df rescan is performed on
+	     insns inserted here.  */
+	  set_block_for_insn (insn, bb);
+	  df_set_flags (DF_DEFER_INSN_RESCAN);
+	  start_sequence ();
+
+	  /* Make sure the scheduler doesn't schedule other insns beyond here.  */
+	  emit_insn (gen_blockage ());
+
+	  /* Get the list of registers used by the function call.  */
+	  link =3D CALL_INSN_FUNCTION_USAGE (insn);
+	  for (; link !=3D NULL; link =3D XEXP (link, 1))
+	    {
+	      pat =3D XEXP (link, 0);
+	      if (GET_CODE (pat) !=3D USE)
+		continue;
+
+	      reg =3D XEXP (pat, 0);
+	      to_clear_mask &=3D ~(1LL << REGNO (reg));
+	      if (ARM_NUM_REGS (GET_MODE (reg)) > 1)
+		to_clear_mask &=3D ~(1LL << (REGNO (reg) + 1));
+	    }
+
+	  /* We use right shift and left shift to clear the LSB of the address
+	     we jump to instead of using bic, to avoid having to use an extra
+	     register on thumb1.  */
+	  cleared_reg =3D XEXP (address, 0);
+	  tmp =3D gen_rtx_LSHIFTRT (SImode, cleared_reg, const1_rtx);
+	  emit_insn (gen_rtx_SET (cleared_reg, tmp));
+	  tmp =3D gen_rtx_ASHIFT (SImode, cleared_reg, const1_rtx);
+	  emit_insn (gen_rtx_SET (cleared_reg, tmp));
+
+	  /* Clearing all registers that leak before doing a non-secure
+	     call.  */
+	  for (regno =3D 0; regno <=3D maxregno; regno++)
+	    {
+	      if (!(to_clear_mask & (1LL << regno)))
+		continue;
+
+	      /* If regno is a vfp register, even and its successor is also to
+		 be cleared, use vmov.  */
+	      if (IS_VFP_REGNUM (regno))
+		{
+		  if (TARGET_VFP_DOUBLE
+		      && VFP_REGNO_OK_FOR_DOUBLE (regno)
+		      && to_clear_mask & (1LL << (regno + 1)))
+		    emit_move_insn (gen_rtx_REG (DFmode, regno++),
+				    CONST1_RTX (DFmode));
+		  else
+		    emit_move_insn (gen_rtx_REG (SFmode, regno),
+				    CONST1_RTX (SFmode));
+		}
+	      else
+		emit_move_insn (gen_rtx_REG (SImode, regno), cleared_reg);
+	    }
+
+	  seq =3D get_insns ();
+	  end_sequence ();
+	  emit_insn_before (seq, insn);
+
+	}
+    }
+}
+
 /* Rewrite move insn into subtract of 0 if the condition codes will
    be useful in next conditional jump insn.  */
=20
@@ -17698,6 +17821,8 @@ arm_reorg (void)
   HOST_WIDE_INT address =3D 0;
   Mfix * fix;
=20
+  if (use_cmse)
+    cmse_nonsecure_call_clear_caller_saved ();
   if (TARGET_THUMB1)
     thumb1_reorg ();
   else if (TARGET_THUMB2)
@@ -18073,6 +18198,23 @@ vfp_emit_fstmd (int base_reg, int count)
   return count * 8;
 }
=20
+/* Returns true if -mcmse has been passed and the function pointed to by '=
addr'
+   has the cmse_nonsecure_call attribute and returns false otherwise.  */
+
+bool
+detect_cmse_nonsecure_call (tree addr)
+{
+  if (!addr)
+    return FALSE;
+
+  tree fntype =3D TREE_TYPE (addr);
+  if (use_cmse && lookup_attribute ("cmse_nonsecure_call",
+				    TYPE_ATTRIBUTES (fntype)))
+    return TRUE;
+  return FALSE;
+}
+
+
 /* Emit a call instruction with pattern PAT.  ADDR is the address of
    the call target.  */
=20
diff --git a/gcc/config/arm/arm.md b/gcc/config/arm/arm.md
index 5ac9b7096e51355d75a6197b21b6e26e1031350e..ec7912b5348739a3d645ba196ca=
0f97e9e590091 100644
--- a/gcc/config/arm/arm.md
+++ b/gcc/config/arm/arm.md
@@ -7614,6 +7614,7 @@
   "
   {
     rtx callee, pat;
+    tree addr =3D MEM_EXPR (operands[0]);
=20=20=20=20=20
     /* In an untyped call, we can get NULL for operand 2.  */
     if (operands[2] =3D=3D NULL_RTX)
@@ -7628,8 +7629,17 @@
 	: !REG_P (callee))
       XEXP (operands[0], 0) =3D force_reg (Pmode, callee);
=20
-    pat =3D gen_call_internal (operands[0], operands[1], operands[2]);
-    arm_emit_call_insn (pat, XEXP (operands[0], 0), false);
+    if (detect_cmse_nonsecure_call (addr))
+      {
+	pat =3D gen_nonsecure_call_internal (operands[0], operands[1],
+					   operands[2]);
+	emit_call_insn (pat);
+      }
+    else
+      {
+	pat =3D gen_call_internal (operands[0], operands[1], operands[2]);
+	arm_emit_call_insn (pat, XEXP (operands[0], 0), false);
+      }
     DONE;
   }"
 )
@@ -7640,6 +7650,24 @@
 	      (use (match_operand 2 "" ""))
 	      (clobber (reg:SI LR_REGNUM))])])
=20
+(define_expand "nonsecure_call_internal"
+  [(parallel [(call (unspec:SI [(match_operand 0 "memory_operand" "")]
+			       UNSPEC_NONSECURE_MEM)
+		    (match_operand 1 "general_operand" ""))
+	      (use (match_operand 2 "" ""))
+	      (clobber (reg:SI LR_REGNUM))
+	      (clobber (reg:SI 4))])]
+  ""
+  "
+  {
+    rtx tmp;
+    tmp =3D copy_to_suggested_reg (XEXP (operands[0], 0),
+				 gen_rtx_REG (SImode, 4),
+				 SImode);
+
+    operands[0] =3D replace_equiv_address (operands[0], tmp);
+  }")
+
 (define_insn "*call_reg_armv5"
   [(call (mem:SI (match_operand:SI 0 "s_register_operand" "r"))
          (match_operand 1 "" ""))
@@ -7675,6 +7703,7 @@
   "
   {
     rtx pat, callee;
+    tree addr =3D MEM_EXPR (operands[1]);
=20=20=20=20=20
     /* In an untyped call, we can get NULL for operand 2.  */
     if (operands[3] =3D=3D 0)
@@ -7689,9 +7718,18 @@
 	: !REG_P (callee))
       XEXP (operands[1], 0) =3D force_reg (Pmode, callee);
=20
-    pat =3D gen_call_value_internal (operands[0], operands[1],
-				   operands[2], operands[3]);
-    arm_emit_call_insn (pat, XEXP (operands[1], 0), false);
+    if (detect_cmse_nonsecure_call (addr))
+      {
+	pat =3D gen_nonsecure_call_value_internal (operands[0], operands[1],
+						 operands[2], operands[3]);
+	emit_call_insn (pat);
+      }
+    else
+      {
+	pat =3D gen_call_value_internal (operands[0], operands[1],
+				       operands[2], operands[3]);
+	arm_emit_call_insn (pat, XEXP (operands[1], 0), false);
+      }
     DONE;
   }"
 )
@@ -7703,6 +7741,25 @@
 	      (use (match_operand 3 "" ""))
 	      (clobber (reg:SI LR_REGNUM))])])
=20
+(define_expand "nonsecure_call_value_internal"
+  [(parallel [(set (match_operand       0 "" "")
+		   (call (unspec:SI [(match_operand 1 "memory_operand" "")]
+				    UNSPEC_NONSECURE_MEM)
+			 (match_operand 2 "general_operand" "")))
+	      (use (match_operand 3 "" ""))
+	      (clobber (reg:SI LR_REGNUM))
+	      (clobber (reg:SI 4))])]
+  ""
+  "
+  {
+    rtx tmp;
+    tmp =3D copy_to_suggested_reg (XEXP (operands[1], 0),
+				 gen_rtx_REG (SImode, 4),
+				 SImode);
+
+    operands[1] =3D replace_equiv_address (operands[1], tmp);
+  }")
+
 (define_insn "*call_value_reg_armv5"
   [(set (match_operand 0 "" "")
         (call (mem:SI (match_operand:SI 1 "s_register_operand" "r"))
diff --git a/gcc/config/arm/thumb1.md b/gcc/config/arm/thumb1.md
index cdcf397b0cab0ea7f246ad830ad28b07fed71d8b..984d68938c25363ce6989902570=
9b208aac228fb 100644
--- a/gcc/config/arm/thumb1.md
+++ b/gcc/config/arm/thumb1.md
@@ -1619,6 +1619,19 @@
    (set_attr "type" "call")]
 )
=20
+(define_insn "*nonsecure_call_reg_thumb1_v5"
+  [(call (unspec:SI [(mem:SI (match_operand:SI 0 "register_operand" "l*r")=
)]
+		    UNSPEC_NONSECURE_MEM)
+	 (match_operand 1 "" ""))
+   (use (match_operand 2 "" ""))
+   (clobber (reg:SI LR_REGNUM))
+   (clobber (match_dup 0))]
+  "TARGET_THUMB1 && use_cmse && !SIBLING_CALL_P (insn)"
+  "bl\\t__gnu_cmse_nonsecure_call"
+  [(set_attr "length" "4")
+   (set_attr "type" "call")]
+)
+
 (define_insn "*call_reg_thumb1"
   [(call (mem:SI (match_operand:SI 0 "register_operand" "l*r"))
 	 (match_operand 1 "" ""))
@@ -1651,6 +1664,21 @@
    (set_attr "type" "call")]
 )
=20
+(define_insn "*nonsecure_call_value_reg_thumb1_v5"
+  [(set (match_operand 0 "" "")
+	(call (unspec:SI
+	       [(mem:SI (match_operand:SI 1 "register_operand" "l*r"))]
+	       UNSPEC_NONSECURE_MEM)
+	      (match_operand 2 "" "")))
+   (use (match_operand 3 "" ""))
+   (clobber (reg:SI LR_REGNUM))
+   (clobber (match_dup 1))]
+  "TARGET_THUMB1 && use_cmse"
+  "bl\\t__gnu_cmse_nonsecure_call"
+  [(set_attr "length" "4")
+   (set_attr "type" "call")]
+)
+
 (define_insn "*call_value_reg_thumb1"
   [(set (match_operand 0 "" "")
 	(call (mem:SI (match_operand:SI 1 "register_operand" "l*r"))
diff --git a/gcc/config/arm/thumb2.md b/gcc/config/arm/thumb2.md
index 0c07df91fe8844a7e7437ef5b7f00f74815d63f4..6a47c5dfe9b0bf24f503b462cc2=
96ba0af9c5b6b 100644
--- a/gcc/config/arm/thumb2.md
+++ b/gcc/config/arm/thumb2.md
@@ -581,6 +581,19 @@
   [(set_attr "type" "call")]
 )
=20
+(define_insn "*nonsecure_call_reg_thumb2"
+  [(call (unspec:SI [(mem:SI (match_operand:SI 0 "s_register_operand" "r")=
)]
+		    UNSPEC_NONSECURE_MEM)
+	 (match_operand 1 "" ""))
+   (use (match_operand 2 "" ""))
+   (clobber (reg:SI LR_REGNUM))
+   (clobber (match_dup 0))]
+  "TARGET_THUMB2 && use_cmse"
+  "bl\\t__gnu_cmse_nonsecure_call"
+  [(set_attr "length" "4")
+   (set_attr "type" "call")]
+)
+
 (define_insn "*call_value_reg_thumb2"
   [(set (match_operand 0 "" "")
 	(call (mem:SI (match_operand:SI 1 "register_operand" "l*r"))
@@ -592,6 +605,21 @@
   [(set_attr "type" "call")]
 )
=20
+(define_insn "*nonsecure_call_value_reg_thumb2"
+  [(set (match_operand 0 "" "")
+	(call
+	 (unspec:SI [(mem:SI (match_operand:SI 1 "register_operand" "l*r"))]
+		    UNSPEC_NONSECURE_MEM)
+	 (match_operand 2 "" "")))
+   (use (match_operand 3 "" ""))
+   (clobber (reg:SI LR_REGNUM))
+   (clobber (match_dup 1))]
+  "TARGET_THUMB2 && use_cmse"
+  "bl\t__gnu_cmse_nonsecure_call"
+  [(set_attr "length" "4")
+   (set_attr "type" "call")]
+)
+
 (define_insn "*thumb2_indirect_jump"
   [(set (pc)
 	(match_operand:SI 0 "register_operand" "l*r"))]
diff --git a/gcc/config/arm/unspecs.md b/gcc/config/arm/unspecs.md
index 67acafd075fb515a848fbe968a0183e4673ab0cd..7fce8d99a4b0e26030ebc7a05f6=
889f205c4cb5c 100644
--- a/gcc/config/arm/unspecs.md
+++ b/gcc/config/arm/unspecs.md
@@ -84,6 +84,8 @@
   UNSPEC_VRINTA         ; Represent a float to integral float rounding
                         ; towards nearest, ties away from zero.
   UNSPEC_PROBE_STACK    ; Probe stack memory reference
+  UNSPEC_NONSECURE_MEM	; Represent non-secure memory in ARMv8-M with
+			; security extension
 ])
=20
 (define_c_enum "unspec" [
diff --git a/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-11.c b/gcc/tes=
tsuite/gcc.target/arm/cmse/baseline/cmse-11.c
new file mode 100644
index 0000000000000000000000000000000000000000..3007409ad88b9d2312ece4d55f3=
695787a1aa566
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-11.c
@@ -0,0 +1,22 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_base_ok } */
+/* { dg-add-options arm_arch_v8m_base } */
+/* { dg-options "-mcmse" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (int);
+
+int
+foo (int a)
+{
+  return bar (bar (a + 1));
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "movs\tr1, r4" } } */
+/* { dg-final { scan-assembler "movs\tr2, r4" } } */
+/* { dg-final { scan-assembler "movs\tr3, r4" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-13.c b/gcc/tes=
tsuite/gcc.target/arm/cmse/baseline/cmse-13.c
new file mode 100644
index 0000000000000000000000000000000000000000..f2b931be5912b421f8427535ded=
3298883bec639
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-13.c
@@ -0,0 +1,25 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_base_ok } */
+/* { dg-add-options arm_arch_v8m_base } */
+/* { dg-options "-mcmse" } */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (float, double);
+
+int
+foo (int a)
+{
+  return bar (1.0f, 2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "movs\tr0, r4" } } */
+/* { dg-final { scan-assembler "\n\tmovs\tr1, r4" } } */
+/* { dg-final { scan-assembler-not "\n\tmovs\tr2, r4\n\tmovs\tr3, r4" } } =
*/
+/* { dg-final { scan-assembler-not "vmov" } } */
+/* { dg-final { scan-assembler-not "vmsr" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-6.c b/gcc/test=
suite/gcc.target/arm/cmse/baseline/cmse-6.c
new file mode 100644
index 0000000000000000000000000000000000000000..95da045690ae170a10194913448=
2e821f27fe3aa
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/baseline/cmse-6.c
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_base_ok } */
+/* { dg-add-options arm_arch_v8m_base } */
+/* { dg-options "-mcmse" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Remember dont clear r0 and r1, because we are passing the double parame=
ter
+ * for bar in them.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "movs\tr2, r4" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/cmse-9.c b/gcc/testsuite/gcc=
.target/arm/cmse/cmse-9.c
index 2846ce6cc203f1810977656bdc09ff86419b7007..1dd82b5cee735025054e517d1bb=
d9c6ab1ad462a 100644
--- a/gcc/testsuite/gcc.target/arm/cmse/cmse-9.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/cmse-9.c
@@ -3,11 +3,19 @@
 /* { dg-skip-if "Testing exclusion of -mcmse" { arm-*-* } { "-mcmse" } { "=
" } }  */
=20
=20
+void __attribute__ ((cmse_nonsecure_call)) (*bar) (int);
+typedef void __attribute__ ((cmse_nonsecure_call)) baz (int);
+
 int __attribute__ ((cmse_nonsecure_entry))
-foo (int a)
+foo (int a, baz b)
 {
+  bar (a);
+  b (a);
   return a + 1;
 }
=20
+/* { dg-final { scan-assembler-not "bxns" } } */
+/* { dg-final { scan-assembler-not "blxns" } } */
+/* { dg-final { scan-assembler-not "bl\t__gnu_cmse_nonsecure_call" } } */
 /* { dg-final { scan-assembler "foo:" } } */
 /* { dg-final { scan-assembler-not "__acle_se_foo:" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-13.c b=
/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-13.c
new file mode 100644
index 0000000000000000000000000000000000000000..285fa854b3d61c4d856e9be863e=
e73a87f1b11a9
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-13.c
@@ -0,0 +1,43 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing double precision" {*-*-*} {"-mfpu=
=3Dfpv[4-5]-d16"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-sp-d16" }  */
+
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (float, double);
+
+int
+foo (int a)
+{
+  return bar (3.0f, 2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts2, #1\.0" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts7, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts8, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts9, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts10, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts11, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts12, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts13, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts14, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts15, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-7.c b/=
gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-7.c
new file mode 100644
index 0000000000000000000000000000000000000000..8f56aaf2a97e740a77473cab245=
614b6fa696c04
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-7.c
@@ -0,0 +1,42 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing double precision" {*-*-*} {"-mfpu=
=3Dfpv[4-5]-d16"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-sp-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (void);
+
+int
+foo (int a)
+{
+  return bar () + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts7, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts8, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts9, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts10, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts11, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts12, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts13, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts14, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts15, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-8.c b/=
gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-8.c
new file mode 100644
index 0000000000000000000000000000000000000000..749f8a7278cb4ea9fd4462887ce=
7f37b57f83248
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard-sp/cmse-8.c
@@ -0,0 +1,41 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing double precision" {*-*-*} {"-mfpu=
=3Dfpv[4-5]-d16"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-sp-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts7, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts8, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts9, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts10, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts11, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts12, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts13, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts14, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts15, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-13.c b/gc=
c/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-13.c
new file mode 100644
index 0000000000000000000000000000000000000000..4ad5068aac263f1c2edb17b8707=
fc9d1c0af6959
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-13.c
@@ -0,0 +1,35 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-d16" }  */
+
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (float, double);
+
+int
+foo (int a)
+{
+  return bar (3.0f, 2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler-not "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-7.c b/gcc=
/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-7.c
new file mode 100644
index 0000000000000000000000000000000000000000..b18259eaef6da1933ef2d58d63b=
97d4bb4e450f4
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-7.c
@@ -0,0 +1,34 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (void);
+
+int
+foo (int a)
+{
+  return bar () + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-8.c b/gcc=
/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-8.c
new file mode 100644
index 0000000000000000000000000000000000000000..182ddf9540374f30ae3694259f1=
edb62161455e1
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/hard/cmse-8.c
@@ -0,0 +1,33 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dhard -mfpu=3Dfpv5-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov\.f64\td0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-13.c b/gc=
c/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-13.c
new file mode 100644
index 0000000000000000000000000000000000000000..0e097549e63c26cd5f1a46e1728=
bca10046a1c6f
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-13.c
@@ -0,0 +1,26 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-options "-mcmse" } */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (float, double);
+
+int
+foo (int a)
+{
+  return bar (1.0f, 2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler-not "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler-not "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov" } } */
+/* { dg-final { scan-assembler-not "vmsr" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-7.c b/gcc=
/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-7.c
new file mode 100644
index 0000000000000000000000000000000000000000..e33568400ef830abaa019b5c272=
7a9f3b5d5704b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-7.c
@@ -0,0 +1,27 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dhard" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoft" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (void);
+
+int
+foo (int a)
+{
+  return bar () + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov" } } */
+/* { dg-final { scan-assembler-not "vmsr" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-8.c b/gcc=
/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-8.c
new file mode 100644
index 0000000000000000000000000000000000000000..024a12e0a414730eb1cbebfca85=
74d548238ab28
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/soft/cmse-8.c
@@ -0,0 +1,26 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dhard" -mfloat-abi=3Dsoftfp } {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoft" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler-not "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler-not "vmov" } } */
+/* { dg-final { scan-assembler-not "vmsr" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-7.c =
b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-7.c
new file mode 100644
index 0000000000000000000000000000000000000000..0cf1a68267ee1b3b0207e149e17=
b3d3fbcc57e40
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-7.c
@@ -0,0 +1,42 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dhard } {""} } */
+/* { dg-skip-if "Skip these if testing double precision" {*-*-*} {"-mfpu=
=3Dfpv[4-5]-d16"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoftfp -mfpu=3Dfpv5-sp-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (void);
+
+int
+foo (int a)
+{
+  return bar () + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts7, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts8, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts9, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts10, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts11, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts12, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts13, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts14, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts15, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-8.c =
b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-8.c
new file mode 100644
index 0000000000000000000000000000000000000000..e9be53a42f1d62929327a531dda=
9544bd4c2df62
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp-sp/cmse-8.c
@@ -0,0 +1,41 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dhard } {""} } */
+/* { dg-skip-if "Skip these if testing double precision" {*-*-*} {"-mfpu=
=3Dfpv[4-5]-d16"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoftfp -mfpu=3Dfpv5-sp-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler-not "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts7, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts8, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts9, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts10, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts11, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts12, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts13, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts14, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f32\ts15, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-13.c b/=
gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-13.c
new file mode 100644
index 0000000000000000000000000000000000000000..515fd22b9da2216d3f27ed4b053=
5dc3ab88e0eea
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-13.c
@@ -0,0 +1,34 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dhard } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoftfp -mfpu=3Dfpv5-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (float, double);
+
+int
+foo (int a)
+{
+  return bar (1.0f, 2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "\n\tmov\tr1, r4" } } */
+/* { dg-final { scan-assembler-not "\n\tmov\tr2, r4\n\tmov\tr3, r4" } } */
+
+/* { dg-final { scan-assembler "vmov\.f64\td0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-7.c b/g=
cc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-7.c
new file mode 100644
index 0000000000000000000000000000000000000000..fbe3f6d4fd10700269995cde4aa=
b0f105165c56d
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-7.c
@@ -0,0 +1,34 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dhard } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoftfp -mfpu=3Dfpv5-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (void);
+
+int
+foo (int a)
+{
+  return bar () + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
+
diff --git a/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-8.c b/g=
cc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-8.c
new file mode 100644
index 0000000000000000000000000000000000000000..4ae1b2f90f6164b4252029fbdac=
2e9f820d0741c
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/softfp/cmse-8.c
@@ -0,0 +1,33 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target arm_arch_v8m_main_ok } */
+/* { dg-add-options arm_arch_v8m_main } */
+/* { dg-skip-if "Do not combine float-abi=3D hard | soft | softfp" {*-*-*}=
 {"-mfloat-abi=3Dsoft" -mfloat-abi=3Dhard } {""} } */
+/* { dg-skip-if "Skip these if testing single precision" {*-*-*} {"-mfpu=
=3D*-sp-*"} {""} } */
+/* { dg-options "-mcmse -mfloat-abi=3Dsoftfp -mfpu=3Dfpv5-d16" }  */
+
+int __attribute__ ((cmse_nonsecure_call)) (*bar) (double);
+
+int
+foo (int a)
+{
+  return bar (2.0) + a + 1;
+}
+
+/* Checks for saving and clearing prior to function call.  */
+/* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */
+/* { dg-final { scan-assembler-not "mov\tr0, r4" } } */
+/* { dg-final { scan-assembler-not "mov\tr1, r4" } } */
+/* { dg-final { scan-assembler "mov\tr2, r4" } } */
+/* { dg-final { scan-assembler "mov\tr3, r4" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td0, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td1, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td2, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td3, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td4, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td5, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td6, #1\.0" } } */
+/* { dg-final { scan-assembler "vmov\.f64\td7, #1\.0" } } */
+
+/* Now we check that we use the correct intrinsic to call.  */
+/* { dg-final { scan-assembler "bl\t__gnu_cmse_nonsecure_call" } } */
diff --git a/libgcc/config/arm/cmse_nonsecure_call.S b/libgcc/config/arm/cm=
se_nonsecure_call.S
new file mode 100644
index 0000000000000000000000000000000000000000..bdc140f5bbe87c6599db225b1b9=
b7bbc7d606710
--- /dev/null
+++ b/libgcc/config/arm/cmse_nonsecure_call.S
@@ -0,0 +1,87 @@
+.syntax unified
+.thumb
+__gnu_cmse_nonsecure_call:
+#if defined (__ARM_ARCH_8M_MAIN__)
+push	    {r5-r11,lr}
+mov	    r5, r4
+mov	    r6, r4
+mov	    r7, r4
+mov	    r8, r4
+mov	    r9, r4
+mov	    r10, r4
+mov	    r11, r4
+mov	    ip, r4
+#if !defined (__SOFTFP__)
+vpush.f64   {d8-d15}
+#if __ARM_FP & 0x04
+vmov.f32    s16, #1.00000
+vmov.f32    s17, #1.00000
+vmov.f32    s18, #1.00000
+vmov.f32    s19, #1.00000
+vmov.f32    s20, #1.00000
+vmov.f32    s21, #1.00000
+vmov.f32    s22, #1.00000
+vmov.f32    s23, #1.00000
+vmov.f32    s24, #1.00000
+vmov.f32    s25, #1.00000
+vmov.f32    s26, #1.00000
+vmov.f32    s27, #1.00000
+vmov.f32    s28, #1.00000
+vmov.f32    s29, #1.00000
+vmov.f32    s30, #1.00000
+vmov.f32    s31, #1.00000
+#elif __ARM_FP & 0x08
+vmov.f64    d8, #1.00000
+vmov.f64    d9, #1.00000
+vmov.f64    d10, #1.00000
+vmov.f64    d11, #1.00000
+vmov.f64    d12, #1.00000
+vmov.f64    d13, #1.00000
+vmov.f64    d14, #1.00000
+vmov.f64    d15, #1.00000
+#else
+#error "Half precision implementation not supported."
+#endif
+vmsr	    fpscr, r4
+#endif
+#if defined (__ARM_FEATURE_DSP)
+msr	    APSR_nzcvqg, r4
+#else
+msr	    APSR_nzcvq, r4
+#endif
+blxns	    r4
+#if !defined (__SOFTFP__)
+vpop.f64    {d8-d15}
+#endif
+pop	    {r5-r11}
+#elif defined (__ARM_ARCH_8M_BASE__)
+push	    {r5-r7}
+mov	    r5, r8
+mov	    r6, r9
+mov	    r7, r10
+push	    {r5-r7}
+mov	    r5, r11
+mov	    r6, lr
+push	    {r5,r6}
+mov	    r5, r4
+mov	    r6, r4
+mov	    r7, r4
+mov	    r8, r4
+mov	    r9, r4
+mov	    r10, r4
+mov	    r11, r4
+mov	    ip, r4
+msr	    APSR_nzcvq, r4
+blxns	    r4
+pop	    {r5, r6}
+mov	    lr, r6
+mov	    r11, r5
+pop	    {r5-r7}
+mov	    r10, r7
+mov	    r9, r6
+mov	    r8, r5
+pop	    {r5-r7}
+blx	    lr
+#else
+#error "This should only be used for armv8-m base- and mainline."
+#endif
diff --git a/libgcc/config/arm/t-arm b/libgcc/config/arm/t-arm
index e6024f38b0b01ab2d2921d6f8da5b522422d0581..9e85ac06b146feaa14ab83374a0=
156737b870310 100644
--- a/libgcc/config/arm/t-arm
+++ b/libgcc/config/arm/t-arm
@@ -8,8 +8,10 @@ CMSE_OPTS:=3D-mcmse
 endif
=20
 ifdef HAVE_CMSE
-libgcc-objects +=3D cmse.o
+libgcc-objects +=3D cmse.o cmse_nonsecure_call.o
=20
 cmse.o: $(srcdir)/config/arm/cmse.c
 	$(gcc_compile) -c $(CMSE_OPTS) $<
+cmse_nonsecure_call.o: $(srcdir)/config/arm/cmse_nonsecure_call.S
+		       $(gcc_compile) -c $<
 endif


We welcome any comment.

Cheers,

Andre