diff options
Diffstat (limited to 'gdb/i386-tdep.c')
| -rw-r--r-- | gdb/i386-tdep.c | 26 |
1 files changed, 26 insertions, 0 deletions
diff --git a/gdb/i386-tdep.c b/gdb/i386-tdep.c index bf4ca54..fec74ee 100644 --- a/gdb/i386-tdep.c +++ b/gdb/i386-tdep.c @@ -8347,6 +8347,31 @@ i386_validate_tdesc_p (struct gdbarch_tdep *tdep, } + +/* Implement the type_align gdbarch function. */ + +static ULONGEST +i386_type_align (struct gdbarch *gdbarch, struct type *type) +{ + type = check_typedef (type); + + if (gdbarch_ptr_bit (gdbarch) == 32) + { + if ((TYPE_CODE (type) == TYPE_CODE_INT + || TYPE_CODE (type) == TYPE_CODE_FLT) + && TYPE_LENGTH (type) > 4) + return 4; + + /* Handle x86's funny long double. */ + if (TYPE_CODE (type) == TYPE_CODE_FLT + && gdbarch_long_double_bit (gdbarch) == TYPE_LENGTH (type) * 8) + return 4; + } + + return TYPE_LENGTH (type); +} + + /* Note: This is called for both i386 and amd64. */ static struct gdbarch * @@ -8405,6 +8430,7 @@ i386_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches) tdep->record_regmap = i386_record_regmap; set_gdbarch_long_long_align_bit (gdbarch, 32); + set_gdbarch_type_align (gdbarch, i386_type_align); /* The format used for `long double' on almost all i386 targets is the i387 extended floating-point format. In fact, of all targets |