public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc/devel/c++-modules] [nvptx] Fix array dimension in nvptx_assemble_decl_begin
@ 2020-08-27 18:07 Nathan Sidwell
  0 siblings, 0 replies; only message in thread
From: Nathan Sidwell @ 2020-08-27 18:07 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:b9c7fe59f9f66ecc091e215c826ecd1a04d032dc

commit b9c7fe59f9f66ecc091e215c826ecd1a04d032dc
Author: Tom de Vries <tdevries@suse.de>
Date:   Tue Aug 11 18:20:58 2020 +0200

    [nvptx] Fix array dimension in nvptx_assemble_decl_begin
    
    When compiling test-case builtin-object-size-21.c, cc1 emits:
    ...
      .visible .global .align 1 .u32 xm3_3[-2305843009213693951] =
    ...
    for:
    ...
    struct Ax_m3 { char a[PTRDIFF_MAX - 3], ax[]; };
    
    struct Ax_m3 xm3_3 = { { 0 }, { 1, 2, 3 } };
    ...
    
    Fix this by:
    - changing the printing format for unsigned HOST_WIDE_INT init_frag.remaining
      to HOST_WIDE_INT_PRINT_UNSIGNED
    - changing the type of local variable elt_size in nvptx_assemble_decl_begin
      to unsigned HOST_WIDE_INT.
    such that we have:
    ...
      .visible .global .align 1 .u32 xm3_3[2305843009213693952] =
    ...
    where 2305843009213693952 == 0x2000000000000000, so the array is claiming
    0x8000000000000000 bytes, which is one more than PTRDIFF_MAX.  This is due
    to using .u32 instead of .u8, so strictly speaking we should downgrade to
    using .u8 in this case, but that corner-case problem doesn't look urgent
    enough to fix in this commit.
    
    Build on nvptx, tested with make check-gcc.
    
    gcc/ChangeLog:
    
            * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Make elt_size an
            unsigned HOST_WIDE_INT.  Print init_frag.remaining using
            HOST_WIDE_INT_PRINT_UNSIGNED.

Diff:
---
 gcc/config/nvptx/nvptx.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/gcc/config/nvptx/nvptx.c b/gcc/config/nvptx/nvptx.c
index cf53a921e5b..39d0275493a 100644
--- a/gcc/config/nvptx/nvptx.c
+++ b/gcc/config/nvptx/nvptx.c
@@ -2202,7 +2202,7 @@ nvptx_assemble_decl_begin (FILE *file, const char *name, const char *section,
     /* Neither vector nor complex types can contain the other.  */
     type = TREE_TYPE (type);
 
-  unsigned elt_size = int_size_in_bytes (type);
+  unsigned HOST_WIDE_INT elt_size = int_size_in_bytes (type);
 
   /* Largest mode we're prepared to accept.  For BLKmode types we
      don't know if it'll contain pointer constants, so have to choose
@@ -2232,7 +2232,7 @@ nvptx_assemble_decl_begin (FILE *file, const char *name, const char *section,
   if (size)
     /* We make everything an array, to simplify any initialization
        emission.  */
-    fprintf (file, "[" HOST_WIDE_INT_PRINT_DEC "]", init_frag.remaining);
+    fprintf (file, "[" HOST_WIDE_INT_PRINT_UNSIGNED "]", init_frag.remaining);
   else if (atype)
     fprintf (file, "[]");
 }


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2020-08-27 18:07 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-27 18:07 [gcc/devel/c++-modules] [nvptx] Fix array dimension in nvptx_assemble_decl_begin Nathan Sidwell

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).