diff options
| author | Arnaud Charlet <charlet@gcc.gnu.org> | 2004-04-29 17:24:56 +0200 |
|---|---|---|
| committer | Arnaud Charlet <charlet@gcc.gnu.org> | 2004-04-29 17:24:56 +0200 |
| commit | d935a36e27c5a15e05cabdc6b6e9b6d68538a14f (patch) | |
| tree | 32aad4e68b4052f7ce71c58540a2b80a90d67674 /gcc/ada/decl.c | |
| parent | 56141a6ad7f965b156427f25ab45bd1093b0b565 (diff) | |
| download | gcc-d935a36e27c5a15e05cabdc6b6e9b6d68538a14f.zip gcc-d935a36e27c5a15e05cabdc6b6e9b6d68538a14f.tar.gz gcc-d935a36e27c5a15e05cabdc6b6e9b6d68538a14f.tar.bz2 | |
[multiple changes]
2004-04-29 Ed Schonberg <schonberg@gnat.com>
* checks.adb (Enable_Range_Check): If the prefix of an index component
is an access to an unconstrained array, perform check unconditionally.
2004-04-29 Richard Kenner <kenner@vlsi1.ultra.nyu.edu>
* decl.c (gnat_to_gnu_field): Also call make_packable_type if
Component_Clause.
2004-04-29 Olivier Hainque <hainque@act-europe.fr>
* init.c (__gnat_install_handler, __gnat_error_handler): Remove
alternate stack setting. There was no support for the tasking cases
and the changes eventually caused a number of side-effect failures in
the non-tasking case too.
2004-04-29 Eric Botcazou <ebotcazou@act-europe.fr>
lang-specs.h: Redirect output to /dev/null if -gnatc or -gnatz or
-gnats is passed.
2004-04-29 Vincent Celier <celier@gnat.com>
* make.adb (Gnatmake): Increase max size of argument array for
gnatbind for the potential addition of -F.
If there are Stand-Alone Library projects, invoke gnatbind with -F to
be sure that elaboration flags will be checked.
* switch-c.adb: Correct call to Scan_Pos for -gnateI
2004-04-29 Thomas Quinot <quinot@act-europe.fr>
* sem_warn.adb (Check_References): Move '<access-variable> may be
null' warning out of under Warn_On_No_Value_Assigned.
2004-04-29 Ed Falis <falis@gnat.com>
* gnat_ugn.texi: Fixed texi error
2004-04-29 Robert Dewar <dewar@gnat.com>
* sem_ch4.adb (Remove_Abstract_Operations): Unconditionally remove
abstract operations if they come from predefined files.
* gnat_rm.texi: Fix bad doc for pragma Elaboration_Checks (should be
Dynamic, not RM).
* s-addope.adb: Correct obvious error in mod function
From-SVN: r81289
Diffstat (limited to 'gcc/ada/decl.c')
| -rw-r--r-- | gcc/ada/decl.c | 13 |
1 files changed, 7 insertions, 6 deletions
diff --git a/gcc/ada/decl.c b/gcc/ada/decl.c index 73c35c5..92e1342 100644 --- a/gcc/ada/decl.c +++ b/gcc/ada/decl.c @@ -4998,11 +4998,11 @@ gnat_to_gnu_field (Entity_Id gnat_field, && TYPE_LEFT_JUSTIFIED_MODULAR_P (gnu_field_type)) gnu_field_type = TREE_TYPE (TYPE_FIELDS (gnu_field_type)); - /* If we are packing this record or we have a specified size that's - smaller than that of the field type and the field type is also a record - that's BLKmode and with a small constant size, see if we can get a - better form of the type that allows more packing. If we can, show - a size was specified for it if there wasn't one so we know to + /* If we are packing this record, have a specified size that's smaller than + that of the field type, or a position is specified, and the field type + is also a record that's BLKmode and with a small constant size, see if + we can get a better form of the type that allows more packing. If we + can, show a size was specified for it if there wasn't one so we know to make this a bitfield and avoid making things wider. */ if (TREE_CODE (gnu_field_type) == RECORD_TYPE && TYPE_MODE (gnu_field_type) == BLKmode @@ -5010,7 +5010,8 @@ gnat_to_gnu_field (Entity_Id gnat_field, && compare_tree_int (TYPE_SIZE (gnu_field_type), BIGGEST_ALIGNMENT) <= 0 && (packed || (gnu_size != 0 && tree_int_cst_lt (gnu_size, - TYPE_SIZE (gnu_field_type))))) + TYPE_SIZE (gnu_field_type))) + || Present (Component_Clause (gnat_field)))) { gnu_field_type = make_packable_type (gnu_field_type); |