[gcc r14-1782] arm: Extend -mtp= arguments

[gcc r14-1782] arm: Extend -mtp= arguments

[Kyrylo Tkachov]

https://gcc.gnu.org/g:cca8d9e5be2ffffd6ef7276234e1097763b15e19

commit r14-1782-gcca8d9e5be2ffffd6ef7276234e1097763b15e19
Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Date:   Tue Jun 13 10:17:24 2023 +0100

    arm: Extend -mtp= arguments
    
    After discussing the -mtp= option with Arm's LLVM developers we'd like to extend
    the functionality of the option somewhat.
    There are actually 3 system registers that can be accessed for the thread pointer
    in aarch32: tpidrurw, tpidruro, tpidrprw.  They are all read through the CP15 co-processor
    mechanism. The current -mtp=cp15 option reads the tpidruro register.
    This patch extends -mtp to allow for the above three explicit tpidr names and
    keeps -mtp=cp15 as an alias of -mtp=tpidruro for backwards compatibility.
    
    Bootstrapped and tested on arm-none-linux-gnueabihf.
    
    gcc/ChangeLog:
    
            * config/arm/arm-opts.h (enum arm_tp_type): Remove TP_CP15.
            Add TP_TPIDRURW, TP_TPIDRURO, TP_TPIDRPRW values.
            * config/arm/arm-protos.h (arm_output_load_tpidr): Declare prototype.
            * config/arm/arm.cc (arm_option_reconfigure_globals): Replace TP_CP15
            with TP_TPIDRURO.
            (arm_output_load_tpidr): Define.
            * config/arm/arm.h (TARGET_HARD_TP): Define in terms of TARGET_SOFT_TP.
            * config/arm/arm.md (load_tp_hard): Call arm_output_load_tpidr to output
            assembly.
            (reload_tp_hard): Likewise.
            * config/arm/arm.opt (tpidrurw, tpidruro, tpidrprw): New values for
            arm_tp_type.
            * doc/invoke.texi (Arm Options, mtp): Document new values.
    
    gcc/testsuite/ChangeLog:
    
            * gcc.target/arm/mtp.c: New test.
            * gcc.target/arm/mtp_1.c: New test.
            * gcc.target/arm/mtp_2.c: New test.
            * gcc.target/arm/mtp_3.c: New test.
            * gcc.target/arm/mtp_4.c: New test.

Diff:
---
 gcc/config/arm/arm-opts.h            |  4 +++-
 gcc/config/arm/arm-protos.h          |  1 +
 gcc/config/arm/arm.cc                | 32 +++++++++++++++++++++++++++++++-
 gcc/config/arm/arm.h                 |  2 +-
 gcc/config/arm/arm.md                |  4 ++--
 gcc/config/arm/arm.opt               | 11 ++++++++++-
 gcc/doc/invoke.texi                  | 15 +++++++++------
 gcc/testsuite/gcc.target/arm/mtp.c   |  8 ++++++++
 gcc/testsuite/gcc.target/arm/mtp_1.c |  7 +++++++
 gcc/testsuite/gcc.target/arm/mtp_2.c |  7 +++++++
 gcc/testsuite/gcc.target/arm/mtp_3.c |  7 +++++++
 gcc/testsuite/gcc.target/arm/mtp_4.c |  7 +++++++
 12 files changed, 93 insertions(+), 12 deletions(-)

diff --git a/gcc/config/arm/arm-opts.h b/gcc/config/arm/arm-opts.h
index 9964fd2dbb5..174dbc505e0 100644
--- a/gcc/config/arm/arm-opts.h
+++ b/gcc/config/arm/arm-opts.h
@@ -61,7 +61,9 @@ enum float_abi_type
 enum arm_tp_type {
   TP_AUTO,
   TP_SOFT,
-  TP_CP15
+  TP_TPIDRURW,
+  TP_TPIDRURO,
+  TP_TPIDRPRW
 };
 
 /* Which TLS scheme to use.  */
diff --git a/gcc/config/arm/arm-protos.h b/gcc/config/arm/arm-protos.h
index 61fcd671437..7d73c66a15d 100644
--- a/gcc/config/arm/arm-protos.h
+++ b/gcc/config/arm/arm-protos.h
@@ -182,6 +182,7 @@ extern bool arm_is_long_call_p (tree);
 extern int    arm_emit_vector_const (FILE *, rtx);
 extern void arm_emit_fp16_const (rtx c);
 extern const char * arm_output_load_gr (rtx *);
+extern const char * arm_output_load_tpidr (rtx, bool);
 extern const char *vfp_output_vstmd (rtx *);
 extern void arm_output_multireg_pop (rtx *, bool, rtx, bool, bool);
 extern void arm_set_return_address (rtx, rtx);
diff --git a/gcc/config/arm/arm.cc b/gcc/config/arm/arm.cc
index c3e731b8982..38f0839de1c 100644
--- a/gcc/config/arm/arm.cc
+++ b/gcc/config/arm/arm.cc
@@ -3927,7 +3927,7 @@ arm_option_reconfigure_globals (void)
   if (target_thread_pointer == TP_AUTO)
     {
       if (arm_arch6k && !TARGET_THUMB1)
-	target_thread_pointer = TP_CP15;
+	target_thread_pointer = TP_TPIDRURO;
       else
 	target_thread_pointer = TP_SOFT;
     }
@@ -34648,4 +34648,34 @@ arm_get_mask_mode (machine_mode mode)
   return default_get_mask_mode (mode);
 }
 
+/* Output assembly to read the thread pointer from the appropriate TPIDR
+   register into DEST.  If PRED_P also emit the %? that can be used to
+   output the predication code.  */
+
+const char *
+arm_output_load_tpidr (rtx dst, bool pred_p)
+{
+  char buf[64];
+  int tpidr_coproc_num = -1;
+  switch (target_thread_pointer)
+    {
+    case TP_TPIDRURW:
+      tpidr_coproc_num = 2;
+      break;
+    case TP_TPIDRURO:
+      tpidr_coproc_num = 3;
+      break;
+    case TP_TPIDRPRW:
+      tpidr_coproc_num = 4;
+      break;
+    default:
+      gcc_unreachable ();
+    }
+  snprintf (buf, sizeof (buf),
+	    "mrc%s\tp15, 0, %%0, c13, c0, %d\t@ load_tp_hard",
+	    pred_p ? "%?" : "", tpidr_coproc_num);
+  output_asm_insn (buf, &dst);
+  return "";
+}
+
 #include "gt-arm.h"
diff --git a/gcc/config/arm/arm.h b/gcc/config/arm/arm.h
index 7d40b8b7e00..4f54530adcb 100644
--- a/gcc/config/arm/arm.h
+++ b/gcc/config/arm/arm.h
@@ -152,8 +152,8 @@ emission of floating point pcs attributes.  */
 #define TARGET_AAPCS_BASED \
     (arm_abi != ARM_ABI_APCS && arm_abi != ARM_ABI_ATPCS)
 
-#define TARGET_HARD_TP			(target_thread_pointer == TP_CP15)
 #define TARGET_SOFT_TP			(target_thread_pointer == TP_SOFT)
+#define TARGET_HARD_TP			!TARGET_SOFT_TP
 #define TARGET_GNU2_TLS			(target_tls_dialect == TLS_GNU2)
 
 /* Only 16-bit thumb code.  */
diff --git a/gcc/config/arm/arm.md b/gcc/config/arm/arm.md
index 2c7249f0193..2ac97232ffd 100644
--- a/gcc/config/arm/arm.md
+++ b/gcc/config/arm/arm.md
@@ -12275,7 +12275,7 @@
   [(set (match_operand:SI 0 "register_operand" "=r")
 	(unspec:SI [(const_int 0)] UNSPEC_TLS))]
   "TARGET_HARD_TP"
-  "mrc%?\\tp15, 0, %0, c13, c0, 3\\t@ load_tp_hard"
+  "* return arm_output_load_tpidr (operands[0], true);"
   [(set_attr "predicable" "yes")
    (set_attr "type" "mrs")]
 )
@@ -12285,7 +12285,7 @@
   [(set (match_operand:SI 0 "register_operand" "=r")
 	(unspec_volatile:SI [(const_int 0)] VUNSPEC_MRC))]
   "TARGET_HARD_TP"
-  "mrc\\tp15, 0, %0, c13, c0, 3\\t@ reload_tp_hard"
+  "* return arm_output_load_tpidr (operands[0], false);"
   [(set_attr "type" "mrs")]
 )
 
diff --git a/gcc/config/arm/arm.opt b/gcc/config/arm/arm.opt
index 3a49b51ece0..88299dabc3a 100644
--- a/gcc/config/arm/arm.opt
+++ b/gcc/config/arm/arm.opt
@@ -230,7 +230,16 @@ EnumValue
 Enum(arm_tp_type) String(auto) Value(TP_AUTO)
 
 EnumValue
-Enum(arm_tp_type) String(cp15) Value(TP_CP15)
+Enum(arm_tp_type) String(tpidrurw) Value(TP_TPIDRURW)
+
+EnumValue
+Enum(arm_tp_type) String(cp15) Value(TP_TPIDRURO)
+
+EnumValue
+Enum(arm_tp_type) String(tpidruro) Value(TP_TPIDRURO)
+
+EnumValue
+Enum(arm_tp_type) String(tpidrprw) Value(TP_TPIDRPRW)
 
 mtpcs-frame
 Target Mask(TPCS_FRAME)
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index 54c377fd815..8fa3f9fae01 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -22645,12 +22645,15 @@ by default.
 
 @opindex mtp
 @item -mtp=@var{name}
-Specify the access model for the thread local storage pointer.  The valid
-models are @samp{soft}, which generates calls to @code{__aeabi_read_tp},
-@samp{cp15}, which fetches the thread pointer from @code{cp15} directly
-(supported in the arm6k architecture), and @samp{auto}, which uses the
-best available method for the selected processor.  The default setting is
-@samp{auto}.
+Specify the access model for the thread local storage pointer.  The model
+@samp{soft} generates calls to @code{__aeabi_read_tp}.  Other accepted
+models are @samp{tpidrurw}, @samp{tpidruro} and @samp{tpidrprw} which fetch
+the thread pointer from the corresponding system register directly
+(supported from the arm6k architecture and later).  These system registers
+are accessed through the CP15 co-processor interface and the argument
+@samp{cp15} is also accepted as a convenience alias of @samp{tpidruro}.
+The argument @samp{auto} uses the best available method for the selected
+processor.  The default setting is @samp{auto}.
 
 @opindex mtls-dialect
 @item -mtls-dialect=@var{dialect}
diff --git a/gcc/testsuite/gcc.target/arm/mtp.c b/gcc/testsuite/gcc.target/arm/mtp.c
new file mode 100644
index 00000000000..d994c377fd3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/mtp.c
@@ -0,0 +1,8 @@
+__thread int i;
+
+int
+foo (void)
+{
+  return i;
+}
+
diff --git a/gcc/testsuite/gcc.target/arm/mtp_1.c b/gcc/testsuite/gcc.target/arm/mtp_1.c
new file mode 100644
index 00000000000..678d27d9234
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/mtp_1.c
@@ -0,0 +1,7 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target tls_native } */
+/* { dg-options "-O -mtp=cp15" } */
+
+#include "mtp.c"
+
+/* { dg-final { scan-assembler-times {mrc\tp15, 0, r3, c13, c0, 3} 1 } } */
diff --git a/gcc/testsuite/gcc.target/arm/mtp_2.c b/gcc/testsuite/gcc.target/arm/mtp_2.c
new file mode 100644
index 00000000000..bcb308f2637
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/mtp_2.c
@@ -0,0 +1,7 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target tls_native } */
+/* { dg-options "-O -mtp=tpidrprw" } */
+
+#include "mtp.c"
+
+/* { dg-final { scan-assembler-times {mrc\tp15, 0, r3, c13, c0, 4} 1 } } */
diff --git a/gcc/testsuite/gcc.target/arm/mtp_3.c b/gcc/testsuite/gcc.target/arm/mtp_3.c
new file mode 100644
index 00000000000..7d5cea3cab6
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/mtp_3.c
@@ -0,0 +1,7 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target tls_native } */
+/* { dg-options "-O -mtp=tpidruro" } */
+
+#include "mtp.c"
+
+/* { dg-final { scan-assembler-times {mrc\tp15, 0, r3, c13, c0, 3} 1 } } */
diff --git a/gcc/testsuite/gcc.target/arm/mtp_4.c b/gcc/testsuite/gcc.target/arm/mtp_4.c
new file mode 100644
index 00000000000..068078df84e
--- /dev/null
+++ b/gcc/testsuite/gcc.target/arm/mtp_4.c
@@ -0,0 +1,7 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target tls_native } */
+/* { dg-options "-O -mtp=tpidrurw" } */
+
+#include "mtp.c"
+
+/* { dg-final { scan-assembler-times {mrc\tp15, 0, r3, c13, c0, 2} 1 } } */