https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #7 from Richard Biener ---
Yeah, guess we still have target code doing ABI decisions based on mode ...
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #6 from richard.sandiford at arm dot com ---
Here's a proof of concept patch that fixes the testcase for
-mstrict-align. The VECTOR_MODE_P part would need to be behind
a new target hook, to avoid accidentally breaking someone's ABI.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #5 from Richard Sandiford ---
I think the loop in compute_mode_layout needs to be smarter
for unions. At the moment it's sensitive to field order,
which doesn't make much conceptual sense.
E.g. for the admittedly contrived
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #4 from Richard Biener ---
I think for SRA the issue is that there's no access to any of the unions
components and as Martin says it disqualifies total scalarization.
I think total scalarization would be possible if we can use the
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #3 from avieira at gcc dot gnu.org ---
Err that should be 'double d[4];' so:
typedef struct
{
float __attribute__ ((vector_size(16))) v[2];
} STRUCT;
#ifdef GOOD
typedef STRUCT TYPE;
#else
typedef union
{
STRUCT s;
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
--- Comment #2 from avieira at gcc dot gnu.org ---
Sorry for the delay. Here's the typedefs with GNU vectors.
typedef struct
{
float __attribute__ ((vector_size(16))) v[2];
} STRUCT;
#ifdef GOOD
typedef STRUCT TYPE;
#else
typedef union
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109543
Richard Biener changed:
What|Removed |Added
Keywords||missed-optimization
