Re: [PATCH, GCC/ARM, 7/10] Clear all VFP regs inline in hardfloat nscall functions

Tue, 12 Nov 2019 02:20:25 -0800 


On 10/23/19 10:26 AM, Mihail Ionescu wrote:

[PATCH, GCC/ARM, 7/10] Clear all VFP regs inline in hardfloat nscall 
functions


Hi,

=== Context ===

This patch is part of a patch series to add support for Armv8.1-M
Mainline Security Extensions architecture. Its purpose is to generate
inline instructions to save, clear and restore callee-saved VFP
registers before doing a call to a function with the cmse_nonsecure_call
attribute.

=== Patch description ===

The patch is fairly straightforward in its approach and consist of the
following 3 logical changes:
- abstract the number of floating-point register to clear in
  max_fp_regno
- use max_fp_regno to decide how many registers to clear so that the
  same code works for Armv8-M and Armv8.1-M Mainline
- emit vpush and vpop instruction respectively before and after a
  nonsecure call

Note that as in the patch to clear GPRs inline, debug information has to
be disabled for VPUSH and VPOP due to VPOP adding CFA adjustment note
for SP when R7 is sometimes used as CFA.

ChangeLog entries are as follows:

*** gcc/ChangeLog ***

2019-10-23  Mihail-Calin Ionescu <mihail.ione...@arm.com>
2019-10-23  Thomas Preud'homme <thomas.preudho...@arm.com>

        * config/arm/arm.c (vfp_emit_fstmd): Declare early.
        (arm_emit_vfp_multi_reg_pop): Likewise.

        (cmse_nonsecure_call_inline_register_clear): Abstract number 
of VFP
        registers to clear in max_fp_regno.  Emit VPUSH and VPOP to 
save and

        restore callee-saved VFP registers.

*** gcc/testsuite/ChangeLog ***

2019-10-23  Mihail-Calin Ionescu <mihail.ione...@arm.com>
2019-10-23  Thomas Preud'homme <thomas.preudho...@arm.com>


        * gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-13.c: Add 
check for

        VPUSH and VPOP and update expectation for VSCCLRM.
        * gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-7.c: Likewise.
        * gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-8.c: Likewise.
        * gcc.target/arm/cmse/mainline/8_1m/hard/cmse-13.c: Likewise.
        * gcc.target/arm/cmse/mainline/8_1m/hard/cmse-7.c: Likewise.
        * gcc.target/arm/cmse/mainline/8_1m/hard/cmse-8.c: Likewise.

Testing: bootstrapped on arm-linux-gnueabihf and testsuite shows no
regression.

Is this ok for trunk?


Ok.

Thanks,

Kyrill



Best regards,

Mihail



###############     Attachment also inlined for ease of reply    
###############



diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c

index 
c24996897eb21c641914326f7064a26bbb363411..bcc86d50a10f11d9672258442089a0aa5c450b2f 
100644

--- a/gcc/config/arm/arm.c
+++ b/gcc/config/arm/arm.c

@@ -188,6 +188,8 @@ static void emit_constant_insn (rtx cond, rtx 
pattern);

 static rtx_insn *emit_set_insn (rtx, rtx);
 static rtx emit_multi_reg_push (unsigned long, unsigned long);
 static void arm_emit_multi_reg_pop (unsigned long);
+static int vfp_emit_fstmd (int, int);
+static void arm_emit_vfp_multi_reg_pop (int, int, rtx);
 static int arm_arg_partial_bytes (cumulative_args_t,
                                   const function_arg_info &);

 static rtx arm_function_arg (cumulative_args_t, const 
function_arg_info &);

@@ -17834,8 +17836,10 @@ cmse_nonsecure_call_inline_register_clear (void)
           unsigned address_regnum, regno;
           unsigned max_int_regno =
             clear_callee_saved ? IP_REGNUM : LAST_ARG_REGNUM;
+         unsigned max_fp_regno =
+           TARGET_HAVE_FPCTX_CMSE ? LAST_VFP_REGNUM : D7_VFP_REGNUM;
           unsigned maxregno =
-           TARGET_HARD_FLOAT_ABI ? D7_VFP_REGNUM : max_int_regno;
+           TARGET_HARD_FLOAT_ABI ? max_fp_regno : max_int_regno;
           auto_sbitmap to_clear_bitmap (maxregno + 1);
           rtx_insn *seq;
           rtx pat, call, unspec, clearing_reg, ip_reg, shift;
@@ -17883,7 +17887,7 @@ cmse_nonsecure_call_inline_register_clear (void)

               bitmap_clear (float_bitmap);
               bitmap_set_range (float_bitmap, FIRST_VFP_REGNUM,
-                               D7_VFP_REGNUM - FIRST_VFP_REGNUM + 1);
+                               max_fp_regno - FIRST_VFP_REGNUM + 1);

               bitmap_ior (to_clear_bitmap, to_clear_bitmap, 
float_bitmap);

             }

@@ -17960,6 +17964,16 @@ cmse_nonsecure_call_inline_register_clear (void)
               /* Disable frame debug info in push because it needs to be
                  disabled for pop (see below).  */
               RTX_FRAME_RELATED_P (push_insn) = 0;
+
+             /* Save VFP callee-saved registers.  */
+             if (TARGET_HARD_FLOAT_ABI)
+               {
+                 vfp_emit_fstmd (D7_VFP_REGNUM + 1,
+                                 (max_fp_regno - D7_VFP_REGNUM) / 2);

+                 /* Disable frame debug info in push because it needs 
to be

+                    disabled for vpop (see below).  */
+                 RTX_FRAME_RELATED_P (get_last_insn ()) = 0;
+               }
             }


           /* Clear caller-saved registers that leak before doing a 
non-secure

@@ -17974,9 +17988,25 @@ cmse_nonsecure_call_inline_register_clear (void)

           if (TARGET_HAVE_FPCTX_CMSE)
             {
-             rtx_insn *next, *pop_insn, *after = insn;
+             rtx_insn *next, *last, *pop_insn, *after = insn;

               start_sequence ();
+
+             /* Restore VFP callee-saved registers.  */
+             if (TARGET_HARD_FLOAT_ABI)
+               {
+                 int nb_callee_saved_vfp_regs =
+                   (max_fp_regno - D7_VFP_REGNUM) / 2;
+                 arm_emit_vfp_multi_reg_pop (D7_VFP_REGNUM + 1,
+ nb_callee_saved_vfp_regs,
+ stack_pointer_rtx);

+                 /* Disable frame debug info in vpop because the SP 
adjustment

+                    is made using a CFA adjustment note while CFA used is

+                    sometimes R7.  This then causes an assert failure 
in the

+                    CFI note creation code.  */
+                 RTX_FRAME_RELATED_P (get_last_insn ()) = 0;
+               }
+
               arm_emit_multi_reg_pop (callee_saved_mask);
               pop_insn = get_last_insn ();

@@ -17993,13 +18023,15 @@ cmse_nonsecure_call_inline_register_clear (void)
                  not reliable.  */
               RTX_FRAME_RELATED_P (pop_insn) = 0;

+             seq = get_insns ();
+             last = get_last_insn ();
               end_sequence ();

-             emit_insn_after (pop_insn, after);
+             emit_insn_after (seq, after);


               /* Skip pop we have just inserted after nonsecure call, 
we know

                  it does not contain a nonsecure call.  */
-             insn = pop_insn;
+             insn = last;
             }
         }
     }

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-13.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-13.c
index 
67ced090163bc4b40c09b00305d49d21084f7fb8..e759db24fac27be2de664ad4faced7bb5d3be84a 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-13.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-13.c
@@ -9,12 +9,14 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

 /* { dg-final { scan-assembler-not "vmov\.f32\ts0, #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 "vscclrm\t\{s1, VPR\}" } } */
-/* { dg-final { scan-assembler "vscclrm\t\{s4-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s4-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-7.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-7.c
index 
a6d9d14b63be8a226242512e36ba313849a3d35a..9df7b5de3ad64e719fc97d7a67a4893398e87b32 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-7.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-7.c
@@ -9,8 +9,10 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

-/* { dg-final { scan-assembler "vscclrm\t\{s0-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s0-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-8.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-8.c
index 
ff4763529c7d280193d99ef695ce426038b7ceb9..b36ff73cb9b46d4d3a91f854c022787dc000d9c3 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-8.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard-sp/cmse-8.c
@@ -9,10 +9,12 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

 /* { dg-final { scan-assembler-not "vmov\.f32\ts0, #1\.0" } } */
 /* { dg-final { scan-assembler-not "vmov\.f32\ts1, #1\.0" } } */
-/* { dg-final { scan-assembler "vscclrm\t\{s2-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s2-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-13.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-13.c
index 
952010baee36cc404f1a85252c3bf9da7a6c3ce1..72493f032355828bdf48b84dc98d8102becd1b36 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-13.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-13.c
@@ -9,6 +9,7 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

 /* { dg-final { scan-assembler-not "vmov\.f32\ts0, #1\.0" } } */
 /* { dg-final { scan-assembler-not "vmov\.f64\td0, #1\.0" } } */
@@ -16,7 +17,8 @@
 /* { dg-final { scan-assembler-not "vmov\.f32\ts2, #1\.0" } } */
 /* { dg-final { scan-assembler-not "vmov\.f32\ts3, #1\.0" } } */
 /* { dg-final { scan-assembler "vscclrm\t\{s1, VPR\}" } } */
-/* { dg-final { scan-assembler "vscclrm\t\{s4-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s4-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-7.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-7.c
index 
ec1284117f8937026b263ba1bd623e9942527877..112ed78d6f172be31d4a4f06840d32a90cdf3dd4 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-7.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-7.c
@@ -9,8 +9,10 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

-/* { dg-final { scan-assembler "vscclrm\t\{s0-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s0-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */

diff --git 
a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-8.c 
b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-8.c
index 
d70a58a4c84c5d042f4efefbf36316d5401f770c..f48e8a0a020cebed8c9fa124d352f047442979d6 
100644

--- a/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-8.c
+++ b/gcc/testsuite/gcc.target/arm/cmse/mainline/8_1m/hard/cmse-8.c
@@ -9,9 +9,11 @@
 /* { dg-final { scan-assembler "lsrs\tr4, r4, #1" } } */
 /* { dg-final { scan-assembler "lsls\tr4, r4, #1" } } */

 /* { dg-final { scan-assembler "push\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */
+/* { dg-final { scan-assembler "vpush.64\t\{d8, d9, d10, d11, d12, 
d13, d14, d15\}" } } */
 /* { dg-final { scan-assembler "clrm\t\{r0, r1, r2, r3, r5, r6, r7, 
r8, r9, r10, fp, ip, APSR\}" } } */

 /* { dg-final { scan-assembler-not "vmov\.f64\td0, #1\.0" } } */
-/* { dg-final { scan-assembler "vscclrm\t\{s2-s15, VPR\}" } } */
+/* { dg-final { scan-assembler "vscclrm\t\{s2-s31, VPR\}" } } */
+/* { dg-final { scan-assembler "vldm\tsp!, \{d8-d15\}" } } */

 /* { dg-final { scan-assembler "pop\t\{r4, r5, r6, r7, r8, r9, r10, 
fp\}" } } */


 /* Now we check that we use the correct intrinsic to call. */















    











					Reply via email to