[PATCH 1/2] Add 'gcc.target/nvptx/softstack-decl-1.c', 'gcc.target/nvptx/uniform-simt-decl-1.c'

[PATCH 1/2] Add 'gcc.target/nvptx/softstack-decl-1.c', 'gcc.target/nvptx/uniform-simt-decl-1.c'

[Thomas Schwinge]

... to document the status quo re implicit (via 'need_softstack_decl',
'need_unisimt_decl') and explicit declarations of '__nvptx_stacks',
'__nvptx_uni'.

        gcc/testsuite/
        * gcc.target/nvptx/softstack-decl-1.c: New.
        * gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
---
 .../gcc.target/nvptx/softstack-decl-1.c       | 20 +++++++++++++
 .../gcc.target/nvptx/uniform-simt-decl-1.c    | 29 +++++++++++++++++++
 2 files changed, 49 insertions(+)
 create mode 100644 gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
 create mode 100644 gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c

diff --git a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
new file mode 100644
index 00000000000..c502eacc1b3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options {-save-temps -O0 -msoft-stack} } */
+
+extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));
+
+void *f()
+{
+  /* Implicit '__nvptx_stacks' usage for frame; per 'init_softstack_frame':
+     { dg-final { scan-assembler-times {mov\.u64 %fstmp2, __nvptx_stacks;} 1 } }
+  */
+  void *stack_array[123];
+  /* Explicit '__nvptx_stacks' usage.  */
+  stack_array[5] = __nvptx_stacks[0];
+  return stack_array[5];
+}
+
+/* The implicit (via 'need_softstack_decl') and explicit declarations of
+   '__nvptx_stacks' are both emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
+*/
diff --git a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
new file mode 100644
index 00000000000..486456ab243
--- /dev/null
+++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
@@ -0,0 +1,29 @@
+/* { dg-do compile } */
+/* { dg-options {-save-temps -O0 -muniform-simt} } */
+
+extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));
+
+enum memmodel
+{
+  MEMMODEL_RELAXED = 0,
+};
+
+int a = 0;
+
+int f (void)
+{
+  /* Explicit '__nvptx_uni' usage.  */
+  __builtin_printf("%u\n", __nvptx_uni[0]);
+
+  /* Implicit '__nvptx_uni' usage; per 'nvptx_init_unisimt_predicate':
+     { dg-final { scan-assembler-times {mov\.u64 %r[0-9]+, __nvptx_uni;} 1 } }
+  */
+  int expected = 1;
+  return __atomic_compare_exchange_n (&a, &expected, 0, 0, MEMMODEL_RELAXED,
+                                     MEMMODEL_RELAXED);
+}
+
+/* The implicit (via 'need_unisimt_decl') and explicit declarations of
+   '__nvptx_uni' are both emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
+*/
--
2.25.1

-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[PATCH 2/2] nvptx: Prevent emitting duplicate declarations for '__nvptx_stacks', '__nvptx_uni'

[Thomas Schwinge]

As I have reported to Nvidia in 2022-12-01 'NVIDIA Incident Report (3891704):
ptxas: Duplicate declaration error: "cannot be resolved by a '.static'"',
'ptxas' has an inscrutable error mode for duplicate declarations:

    ptxas softstack-decl-1.o, line 11; error   : '.extern' variable '__nvptx_stacks' cannot be resolved by a '.static'
    ptxas fatal   : Ptx assembly aborted due to errors
    nvptx-as: ptxas returned 255 exit status

    ptxas uniform-simt-decl-1.o, line 12; error   : '.extern' variable '__nvptx_uni' cannot be resolved by a '.static'
    ptxas fatal   : Ptx assembly aborted due to errors
    nvptx-as: ptxas returned 255 exit status

This is inscrutable, because (a) what is "cannot be resolved by a '.static'"
supposed to tell me (there is no '.static' in PTX?), and (b) why arent't
repeated declaration just verified to match the first, but otherwise a no-op
(like in other programming languages)?

        gcc/
        * config/nvptx/nvptx.cc (nvptx_assemble_undefined_decl): Notice
        '__nvptx_stacks', '__nvptx_uni' declarations.
        (nvptx_file_end): Don't emit duplicate declarations for those.
        gcc/testsuite/
        * gcc.target/nvptx/softstack-decl-1.c: Make 'dg-do assemble',
        adjust.
        * gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
---
 gcc/config/nvptx/nvptx.cc                          | 14 ++++++++++++--
 gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c  |  8 ++++----
 .../gcc.target/nvptx/uniform-simt-decl-1.c         |  8 ++++----
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/gcc/config/nvptx/nvptx.cc b/gcc/config/nvptx/nvptx.cc
index 8e49dd9c647..b93a253ab31 100644
--- a/gcc/config/nvptx/nvptx.cc
+++ b/gcc/config/nvptx/nvptx.cc
@@ -180,9 +180,11 @@ static GTY(()) tree global_lock_var;

 /* True if any function references __nvptx_stacks.  */
 static bool need_softstack_decl;
+static bool have_softstack_decl;

 /* True if any function references __nvptx_uni.  */
 static bool need_unisimt_decl;
+static bool have_unisimt_decl;

 static int nvptx_mach_max_workers ();

@@ -2571,6 +2573,13 @@ nvptx_assemble_undefined_decl (FILE *file, const char *name, const_tree decl)
                             TREE_TYPE (decl), size ? tree_to_shwi (size) : 0,
                             DECL_ALIGN (decl), true);
   nvptx_assemble_decl_end ();
+
+  static tree softstack_id = get_identifier ("__nvptx_stacks");
+  static tree unisimt_id = get_identifier ("__nvptx_uni");
+  if (DECL_NAME (decl) == softstack_id)
+    have_softstack_decl = true;
+  else if (DECL_NAME (decl) == unisimt_id)
+    have_unisimt_decl = true;
 }

 /* Output a pattern for a move instruction.  */
@@ -6002,7 +6011,7 @@ nvptx_file_end (void)
     write_shared_buffer (asm_out_file, gang_private_shared_sym,
                         gang_private_shared_align, gang_private_shared_size);

-  if (need_softstack_decl)
+  if (need_softstack_decl && !have_softstack_decl)
     {
       write_var_marker (asm_out_file, false, true, "__nvptx_stacks");
       /* 32 is the maximum number of warps in a block.  Even though it's an
@@ -6011,7 +6020,8 @@ nvptx_file_end (void)
       fprintf (asm_out_file, ".extern .shared .u%d __nvptx_stacks[32];\n",
               POINTER_SIZE);
     }
-  if (need_unisimt_decl)
+
+  if (need_unisimt_decl && !have_unisimt_decl)
     {
       write_var_marker (asm_out_file, false, true, "__nvptx_uni");
       fprintf (asm_out_file, ".extern .shared .u32 __nvptx_uni[32];\n");
diff --git a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
index c502eacc1b3..2415f6adb1f 100644
--- a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
+++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile } */
+/* { dg-do assemble } */
 /* { dg-options {-save-temps -O0 -msoft-stack} } */

 extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));
@@ -14,7 +14,7 @@ void *f()
   return stack_array[5];
 }

-/* The implicit (via 'need_softstack_decl') and explicit declarations of
-   '__nvptx_stacks' are both emitted:
-   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
+/* Of the implicit (via 'need_softstack_decl') and explicit declarations of
+   '__nvptx_stacks', only one is emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 1 } }
 */
diff --git a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
index 486456ab243..5a975bdb269 100644
--- a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
+++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile } */
+/* { dg-do assemble } */
 /* { dg-options {-save-temps -O0 -muniform-simt} } */

 extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));
@@ -23,7 +23,7 @@ int f (void)
                                      MEMMODEL_RELAXED);
 }

-/* The implicit (via 'need_unisimt_decl') and explicit declarations of
-   '__nvptx_uni' are both emitted:
-   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
+/* Of the implicit (via 'need_unisimt_decl') and explicit declarations of
+   '__nvptx_uni', only one is emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 1 } }
 */
--
2.25.1

-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[PING] [PATCH 2/2] nvptx: Prevent emitting duplicate declarations for '__nvptx_stacks', '__nvptx_uni'

[Thomas Schwinge]

Hi!

Ping.


Grüße
 Thomas


On 2022-12-19T21:40:07+0100, I wrote:
> As I have reported to Nvidia in 2022-12-01 'NVIDIA Incident Report (3891704):
> ptxas: Duplicate declaration error: "cannot be resolved by a '.static'"',
> 'ptxas' has an inscrutable error mode for duplicate declarations:
>
>     ptxas softstack-decl-1.o, line 11; error   : '.extern' variable '__nvptx_stacks' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
>     ptxas uniform-simt-decl-1.o, line 12; error   : '.extern' variable '__nvptx_uni' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
> This is inscrutable, because (a) what is "cannot be resolved by a '.static'"
> supposed to tell me (there is no '.static' in PTX?), and (b) why arent't
> repeated declaration just verified to match the first, but otherwise a no-op
> (like in other programming languages)?
>
>         gcc/
>         * config/nvptx/nvptx.cc (nvptx_assemble_undefined_decl): Notice
>         '__nvptx_stacks', '__nvptx_uni' declarations.
>         (nvptx_file_end): Don't emit duplicate declarations for those.
>         gcc/testsuite/
>         * gcc.target/nvptx/softstack-decl-1.c: Make 'dg-do assemble',
>         adjust.
>         * gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
> ---
>  gcc/config/nvptx/nvptx.cc                          | 14 ++++++++++++--
>  gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c  |  8 ++++----
>  .../gcc.target/nvptx/uniform-simt-decl-1.c         |  8 ++++----
>  3 files changed, 20 insertions(+), 10 deletions(-)
>
> diff --git a/gcc/config/nvptx/nvptx.cc b/gcc/config/nvptx/nvptx.cc
> index 8e49dd9c647..b93a253ab31 100644
> --- a/gcc/config/nvptx/nvptx.cc
> +++ b/gcc/config/nvptx/nvptx.cc
> @@ -180,9 +180,11 @@ static GTY(()) tree global_lock_var;
>
>  /* True if any function references __nvptx_stacks.  */
>  static bool need_softstack_decl;
> +static bool have_softstack_decl;
>
>  /* True if any function references __nvptx_uni.  */
>  static bool need_unisimt_decl;
> +static bool have_unisimt_decl;
>
>  static int nvptx_mach_max_workers ();
>
> @@ -2571,6 +2573,13 @@ nvptx_assemble_undefined_decl (FILE *file, const char *name, const_tree decl)
>                              TREE_TYPE (decl), size ? tree_to_shwi (size) : 0,
>                              DECL_ALIGN (decl), true);
>    nvptx_assemble_decl_end ();
> +
> +  static tree softstack_id = get_identifier ("__nvptx_stacks");
> +  static tree unisimt_id = get_identifier ("__nvptx_uni");
> +  if (DECL_NAME (decl) == softstack_id)
> +    have_softstack_decl = true;
> +  else if (DECL_NAME (decl) == unisimt_id)
> +    have_unisimt_decl = true;
>  }
>
>  /* Output a pattern for a move instruction.  */
> @@ -6002,7 +6011,7 @@ nvptx_file_end (void)
>      write_shared_buffer (asm_out_file, gang_private_shared_sym,
>                          gang_private_shared_align, gang_private_shared_size);
>
> -  if (need_softstack_decl)
> +  if (need_softstack_decl && !have_softstack_decl)
>      {
>        write_var_marker (asm_out_file, false, true, "__nvptx_stacks");
>        /* 32 is the maximum number of warps in a block.  Even though it's an
> @@ -6011,7 +6020,8 @@ nvptx_file_end (void)
>        fprintf (asm_out_file, ".extern .shared .u%d __nvptx_stacks[32];\n",
>                POINTER_SIZE);
>      }
> -  if (need_unisimt_decl)
> +
> +  if (need_unisimt_decl && !have_unisimt_decl)
>      {
>        write_var_marker (asm_out_file, false, true, "__nvptx_uni");
>        fprintf (asm_out_file, ".extern .shared .u32 __nvptx_uni[32];\n");
> diff --git a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> index c502eacc1b3..2415f6adb1f 100644
> --- a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> @@ -1,4 +1,4 @@
> -/* { dg-do compile } */
> +/* { dg-do assemble } */
>  /* { dg-options {-save-temps -O0 -msoft-stack} } */
>
>  extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));
> @@ -14,7 +14,7 @@ void *f()
>    return stack_array[5];
>  }
>
> -/* The implicit (via 'need_softstack_decl') and explicit declarations of
> -   '__nvptx_stacks' are both emitted:
> -   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
> +/* Of the implicit (via 'need_softstack_decl') and explicit declarations of
> +   '__nvptx_stacks', only one is emitted:
> +   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 1 } }
>  */
> diff --git a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> index 486456ab243..5a975bdb269 100644
> --- a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> @@ -1,4 +1,4 @@
> -/* { dg-do compile } */
> +/* { dg-do assemble } */
>  /* { dg-options {-save-temps -O0 -muniform-simt} } */
>
>  extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));
> @@ -23,7 +23,7 @@ int f (void)
>                                       MEMMODEL_RELAXED);
>  }
>
> -/* The implicit (via 'need_unisimt_decl') and explicit declarations of
> -   '__nvptx_uni' are both emitted:
> -   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
> +/* Of the implicit (via 'need_unisimt_decl') and explicit declarations of
> +   '__nvptx_uni', only one is emitted:
> +   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 1 } }
>  */
> --
> 2.25.1
-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[og12] Add 'gcc.target/nvptx/softstack-decl-1.c', 'gcc.target/nvptx/uniform-simt-decl-1.c'

[Thomas Schwinge]

[-- Attachment #1: Type: text/plain, Size: 745 bytes --]

Hi!

On 2022-12-19T21:40:06+0100, Thomas Schwinge <thomas@codesourcery.com> wrote:
> ... to document the status quo re implicit (via 'need_softstack_decl',
> 'need_unisimt_decl') and explicit declarations of '__nvptx_stacks',
> '__nvptx_uni'.

For now pushed to devel/omp/gcc-12 branch in
commit 703ebfdb483fdade316ceb003a0d57ca132a090b
"Add 'gcc.target/nvptx/softstack-decl-1.c', 'gcc.target/nvptx/uniform-simt-decl-1.c'",
see attached.


Grüße
 Thomas


-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Add-gcc.target-nvptx-softstack-decl-1.c-gcc.target-n.patch --]
[-- Type: text/x-diff, Size: 3363 bytes --]

From 703ebfdb483fdade316ceb003a0d57ca132a090b Mon Sep 17 00:00:00 2001
From: Thomas Schwinge <thomas@codesourcery.com>
Date: Mon, 19 Dec 2022 17:10:52 +0100
Subject: [PATCH] Add 'gcc.target/nvptx/softstack-decl-1.c',
 'gcc.target/nvptx/uniform-simt-decl-1.c'

... to document the status quo re implicit (via 'need_softstack_decl',
'need_unisimt_decl') and explicit declarations of '__nvptx_stacks',
'__nvptx_uni'.

	gcc/testsuite/
	* gcc.target/nvptx/softstack-decl-1.c: New.
	* gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
---
 gcc/testsuite/ChangeLog.omp                   |  3 ++
 .../gcc.target/nvptx/softstack-decl-1.c       | 20 +++++++++++++
 .../gcc.target/nvptx/uniform-simt-decl-1.c    | 29 +++++++++++++++++++
 3 files changed, 52 insertions(+)
 create mode 100644 gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
 create mode 100644 gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c

diff --git a/gcc/testsuite/ChangeLog.omp b/gcc/testsuite/ChangeLog.omp
index 7339bf41482..5b3d9fe416b 100644
--- a/gcc/testsuite/ChangeLog.omp
+++ b/gcc/testsuite/ChangeLog.omp
@@ -1,5 +1,8 @@
 2023-01-20  Thomas Schwinge  <thomas@codesourcery.com>
 
+	* gcc.target/nvptx/softstack-decl-1.c: New.
+	* gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
+
 	* gcc.target/nvptx/nvptx.exp
 	(check_effective_target_default_ptx_isa_version_at_least_6_0):
 	New.
diff --git a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
new file mode 100644
index 00000000000..c502eacc1b3
--- /dev/null
+++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
@@ -0,0 +1,20 @@
+/* { dg-do compile } */
+/* { dg-options {-save-temps -O0 -msoft-stack} } */
+
+extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));
+
+void *f()
+{
+  /* Implicit '__nvptx_stacks' usage for frame; per 'init_softstack_frame':
+     { dg-final { scan-assembler-times {mov\.u64 %fstmp2, __nvptx_stacks;} 1 } }
+  */
+  void *stack_array[123];
+  /* Explicit '__nvptx_stacks' usage.  */ 
+  stack_array[5] = __nvptx_stacks[0];
+  return stack_array[5];
+}
+
+/* The implicit (via 'need_softstack_decl') and explicit declarations of
+   '__nvptx_stacks' are both emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
+*/
diff --git a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
new file mode 100644
index 00000000000..486456ab243
--- /dev/null
+++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
@@ -0,0 +1,29 @@
+/* { dg-do compile } */
+/* { dg-options {-save-temps -O0 -muniform-simt} } */
+
+extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));
+
+enum memmodel
+{
+  MEMMODEL_RELAXED = 0,
+};
+
+int a = 0;
+
+int f (void)
+{
+  /* Explicit '__nvptx_uni' usage.  */
+  __builtin_printf("%u\n", __nvptx_uni[0]);
+
+  /* Implicit '__nvptx_uni' usage; per 'nvptx_init_unisimt_predicate':
+     { dg-final { scan-assembler-times {mov\.u64 %r[0-9]+, __nvptx_uni;} 1 } }
+  */
+  int expected = 1;
+  return __atomic_compare_exchange_n (&a, &expected, 0, 0, MEMMODEL_RELAXED,
+				      MEMMODEL_RELAXED);
+}
+
+/* The implicit (via 'need_unisimt_decl') and explicit declarations of
+   '__nvptx_uni' are both emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
+*/
-- 
2.25.1

[og12] nvptx: Prevent emitting duplicate declarations for '__nvptx_stacks', '__nvptx_uni'

[Thomas Schwinge]

[-- Attachment #1: Type: text/plain, Size: 1551 bytes --]

Hi!

On 2022-12-19T21:40:07+0100, Thomas Schwinge <thomas@codesourcery.com> wrote:
> As I have reported to Nvidia in 2022-12-01 'NVIDIA Incident Report (3891704):
> ptxas: Duplicate declaration error: "cannot be resolved by a '.static'"',
> 'ptxas' has an inscrutable error mode for duplicate declarations:
>
>     ptxas softstack-decl-1.o, line 11; error   : '.extern' variable '__nvptx_stacks' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
>     ptxas uniform-simt-decl-1.o, line 12; error   : '.extern' variable '__nvptx_uni' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
> This is inscrutable, because (a) what is "cannot be resolved by a '.static'"
> supposed to tell me (there is no '.static' in PTX?), and (b) why arent't
> repeated declaration just verified to match the first, but otherwise a no-op
> (like in other programming languages)?

For now pushed to devel/omp/gcc-12 branch in
commit ea52f1ca16870e4228f8044588b1bf958d4723b0
"nvptx: Prevent emitting duplicate declarations for '__nvptx_stacks', '__nvptx_uni'",
see attached.


Grüße
 Thomas


-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-nvptx-Prevent-emitting-duplicate-declarations-for-__.patch --]
[-- Type: text/x-diff, Size: 6569 bytes --]

From ea52f1ca16870e4228f8044588b1bf958d4723b0 Mon Sep 17 00:00:00 2001
From: Thomas Schwinge <thomas@codesourcery.com>
Date: Mon, 19 Dec 2022 17:19:19 +0100
Subject: [PATCH] nvptx: Prevent emitting duplicate declarations for
 '__nvptx_stacks', '__nvptx_uni'

As I have reported to Nvidia in 2022-12-01 'NVIDIA Incident Report (3891704):
ptxas: Duplicate declaration error: "cannot be resolved by a '.static'"',
'ptxas' has an inscrutable error mode for duplicate declarations:

    ptxas softstack-decl-1.o, line 11; error   : '.extern' variable '__nvptx_stacks' cannot be resolved by a '.static'
    ptxas fatal   : Ptx assembly aborted due to errors
    nvptx-as: ptxas returned 255 exit status

    ptxas uniform-simt-decl-1.o, line 12; error   : '.extern' variable '__nvptx_uni' cannot be resolved by a '.static'
    ptxas fatal   : Ptx assembly aborted due to errors
    nvptx-as: ptxas returned 255 exit status

This is inscrutable, because (a) what is "cannot be resolved by a '.static'"
supposed to tell me (there is no '.static' in PTX?), and (b) why arent't
repeated declaration just verified to match the first, but otherwise a no-op
(like in other programming languages)?

	gcc/
	* config/nvptx/nvptx.cc (nvptx_assemble_undefined_decl): Notice
	'__nvptx_stacks', '__nvptx_uni' declarations.
	(nvptx_file_end): Don't emit duplicate declarations for those.
	gcc/testsuite/
	* gcc.target/nvptx/softstack-decl-1.c: Make 'dg-do assemble',
	adjust.
	* gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
---
 gcc/ChangeLog.omp                                  |  4 ++++
 gcc/config/nvptx/nvptx.cc                          | 14 ++++++++++++--
 gcc/testsuite/ChangeLog.omp                        |  4 ++++
 gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c  |  8 ++++----
 .../gcc.target/nvptx/uniform-simt-decl-1.c         |  8 ++++----
 5 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/gcc/ChangeLog.omp b/gcc/ChangeLog.omp
index 382cd5c80c2..127b450644b 100644
--- a/gcc/ChangeLog.omp
+++ b/gcc/ChangeLog.omp
@@ -1,5 +1,9 @@
 2023-01-20  Thomas Schwinge  <thomas@codesourcery.com>
 
+	* config/nvptx/nvptx.cc (nvptx_assemble_undefined_decl): Notice
+	'__nvptx_stacks', '__nvptx_uni' declarations.
+	(nvptx_file_end): Don't emit duplicate declarations for those.
+
 	* config/nvptx/nvptx.md (nvptx_uniform_warp_check): Make fit for
 	non-full-warp execution.
 
diff --git a/gcc/config/nvptx/nvptx.cc b/gcc/config/nvptx/nvptx.cc
index da735cf82ff..9c284ed5b01 100644
--- a/gcc/config/nvptx/nvptx.cc
+++ b/gcc/config/nvptx/nvptx.cc
@@ -181,9 +181,11 @@ static GTY(()) tree global_lock_var;
 
 /* True if any function references __nvptx_stacks.  */
 static bool need_softstack_decl;
+static bool have_softstack_decl;
 
 /* True if any function references __nvptx_uni.  */
 static bool need_unisimt_decl;
+static bool have_unisimt_decl;
 
 static int nvptx_mach_max_workers ();
 
@@ -2572,6 +2574,13 @@ nvptx_assemble_undefined_decl (FILE *file, const char *name, const_tree decl)
 			     TREE_TYPE (decl), size ? tree_to_shwi (size) : 0,
 			     DECL_ALIGN (decl), true);
   nvptx_assemble_decl_end ();
+
+  static tree softstack_id = get_identifier ("__nvptx_stacks");
+  static tree unisimt_id = get_identifier ("__nvptx_uni");
+  if (DECL_NAME (decl) == softstack_id)
+    have_softstack_decl = true;
+  else if (DECL_NAME (decl) == unisimt_id)
+    have_unisimt_decl = true;
 }
 
 /* Output a pattern for a move instruction.  */
@@ -6052,7 +6061,7 @@ nvptx_file_end (void)
     write_shared_buffer (asm_out_file, gang_private_shared_sym,
 			 gang_private_shared_align, gang_private_shared_size);
 
-  if (need_softstack_decl)
+  if (need_softstack_decl && !have_softstack_decl)
     {
       write_var_marker (asm_out_file, false, true, "__nvptx_stacks");
       /* 32 is the maximum number of warps in a block.  Even though it's an
@@ -6061,7 +6070,8 @@ nvptx_file_end (void)
       fprintf (asm_out_file, ".extern .shared .u%d __nvptx_stacks[32];\n",
 	       POINTER_SIZE);
     }
-  if (need_unisimt_decl)
+
+  if (need_unisimt_decl && !have_unisimt_decl)
     {
       write_var_marker (asm_out_file, false, true, "__nvptx_uni");
       fprintf (asm_out_file, ".extern .shared .u32 __nvptx_uni[32];\n");
diff --git a/gcc/testsuite/ChangeLog.omp b/gcc/testsuite/ChangeLog.omp
index 5b3d9fe416b..c942c34dc70 100644
--- a/gcc/testsuite/ChangeLog.omp
+++ b/gcc/testsuite/ChangeLog.omp
@@ -1,5 +1,9 @@
 2023-01-20  Thomas Schwinge  <thomas@codesourcery.com>
 
+	* gcc.target/nvptx/softstack-decl-1.c: Make 'dg-do assemble',
+	adjust.
+	* gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
+
 	* gcc.target/nvptx/softstack-decl-1.c: New.
 	* gcc.target/nvptx/uniform-simt-decl-1.c: Likewise.
 
diff --git a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
index c502eacc1b3..2415f6adb1f 100644
--- a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
+++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile } */
+/* { dg-do assemble } */
 /* { dg-options {-save-temps -O0 -msoft-stack} } */
 
 extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));
@@ -14,7 +14,7 @@ void *f()
   return stack_array[5];
 }
 
-/* The implicit (via 'need_softstack_decl') and explicit declarations of
-   '__nvptx_stacks' are both emitted:
-   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
+/* Of the implicit (via 'need_softstack_decl') and explicit declarations of
+   '__nvptx_stacks', only one is emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 1 } }
 */
diff --git a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
index 486456ab243..5a975bdb269 100644
--- a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
+++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
@@ -1,4 +1,4 @@
-/* { dg-do compile } */
+/* { dg-do assemble } */
 /* { dg-options {-save-temps -O0 -muniform-simt} } */
 
 extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));
@@ -23,7 +23,7 @@ int f (void)
 				      MEMMODEL_RELAXED);
 }
 
-/* The implicit (via 'need_unisimt_decl') and explicit declarations of
-   '__nvptx_uni' are both emitted:
-   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
+/* Of the implicit (via 'need_unisimt_decl') and explicit declarations of
+   '__nvptx_uni', only one is emitted:
+   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 1 } }
 */
-- 
2.25.1

Re: [PATCH 2/2] nvptx: Prevent emitting duplicate declarations for '__nvptx_stacks', '__nvptx_uni'

[Thomas Schwinge]

Hi!

On 2022-12-19T21:40:07+0100, Thomas Schwinge <thomas@codesourcery.com> wrote:
> As I have reported to Nvidia in 2022-12-01 'NVIDIA Incident Report (3891704):
> ptxas: Duplicate declaration error: "cannot be resolved by a '.static'"',
> 'ptxas' has an inscrutable error mode for duplicate declarations:
>
>     ptxas softstack-decl-1.o, line 11; error   : '.extern' variable '__nvptx_stacks' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
>     ptxas uniform-simt-decl-1.o, line 12; error   : '.extern' variable '__nvptx_uni' cannot be resolved by a '.static'
>     ptxas fatal   : Ptx assembly aborted due to errors
>     nvptx-as: ptxas returned 255 exit status
>
> This is inscrutable, because (a) what is "cannot be resolved by a '.static'"
> supposed to tell me (there is no '.static' in PTX?), and (b) why arent't
> repeated declaration just verified to match the first, but otherwise a no-op
> (like in other programming languages)?

Since my report, this had its 'Status changed [...] to "Closed - Fixed"'
(2023-01-28), with comment:

| [...] fix should be available in a later release.
| The compiler was modified to allow duplicate declaration of extern symbol. You will not see an error for this case.
| The documentation is also being changed to reflect this new change.

I've not yet verified the CUDA/'ptxas'-level fix, but I suggest to
retract my GCC-level proposed change:

> --- a/gcc/config/nvptx/nvptx.cc
> +++ b/gcc/config/nvptx/nvptx.cc

> +static bool have_softstack_decl;

> +static bool have_unisimt_decl;

> @@ -2571,6 +2573,13 @@ nvptx_assemble_undefined_decl (FILE *file, const char *name, const_tree decl)
>                              TREE_TYPE (decl), size ? tree_to_shwi (size) : 0,
>                              DECL_ALIGN (decl), true);
>    nvptx_assemble_decl_end ();
> +
> +  static tree softstack_id = get_identifier ("__nvptx_stacks");
> +  static tree unisimt_id = get_identifier ("__nvptx_uni");
> +  if (DECL_NAME (decl) == softstack_id)
> +    have_softstack_decl = true;
> +  else if (DECL_NAME (decl) == unisimt_id)
> +    have_unisimt_decl = true;
>  }

> @@ -6002,7 +6011,7 @@ nvptx_file_end (void)
>      write_shared_buffer (asm_out_file, gang_private_shared_sym,
>                          gang_private_shared_align, gang_private_shared_size);
>
> -  if (need_softstack_decl)
> +  if (need_softstack_decl && !have_softstack_decl)
>      {
>        write_var_marker (asm_out_file, false, true, "__nvptx_stacks");
>        /* 32 is the maximum number of warps in a block.  Even though it's an
> @@ -6011,7 +6020,8 @@ nvptx_file_end (void)
>        fprintf (asm_out_file, ".extern .shared .u%d __nvptx_stacks[32];\n",
>                POINTER_SIZE);
>      }
> -  if (need_unisimt_decl)
> +
> +  if (need_unisimt_decl && !have_unisimt_decl)
>      {
>        write_var_marker (asm_out_file, false, true, "__nvptx_uni");
>        fprintf (asm_out_file, ".extern .shared .u32 __nvptx_uni[32];\n");

..., and suggest that we instead fix up duplicate declarations in the
nvptx-tools 'as', and once GCC depends on a nvptx-tools version with that
addressed, we still change the test cases from "compile" to "assemble" as
proposed:

> --- a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> @@ -1,4 +1,4 @@
> -/* { dg-do compile } */
> +/* { dg-do assemble } */
>  /* { dg-options {-save-temps -O0 -msoft-stack} } */
>
>  extern void *__nvptx_stacks[32] __attribute__((shared,nocommon));

> --- a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> @@ -1,4 +1,4 @@
> -/* { dg-do compile } */
> +/* { dg-do assemble } */
>  /* { dg-options {-save-temps -O0 -muniform-simt} } */
>
>  extern unsigned __nvptx_uni[32] __attribute__((shared,nocommon));

..., but (obviously) without the following changes:

> --- a/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/softstack-decl-1.c

> -/* The implicit (via 'need_softstack_decl') and explicit declarations of
> -   '__nvptx_stacks' are both emitted:
> -   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 2 } }
> +/* Of the implicit (via 'need_softstack_decl') and explicit declarations of
> +   '__nvptx_stacks', only one is emitted:
> +   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_stacks\[32\];} 1 } }

> --- a/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c
> +++ b/gcc/testsuite/gcc.target/nvptx/uniform-simt-decl-1.c

> -/* The implicit (via 'need_unisimt_decl') and explicit declarations of
> -   '__nvptx_uni' are both emitted:
> -   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 2 } }
> +/* Of the implicit (via 'need_unisimt_decl') and explicit declarations of
> +   '__nvptx_uni', only one is emitted:
> +   { dg-final { scan-assembler-times {(?n)\.extern .* __nvptx_uni\[32\];} 1 } }


Grüße
 Thomas
-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955