diff options
| author | Rob Clark <[email protected]> | 2019-08-09 09:08:20 -0700 |
|---|---|---|
| committer | Rob Clark <[email protected]> | 2019-09-06 00:13:44 +0000 |
| commit | 9baa72b7fc9989e35f474310ef333c50b2af934a (patch) | |
| tree | 95b63389a7ac926493d95848e4dac379685de411 | |
| parent | d9ad6f54dcf6c4deb43055e45f1636e74c176ce3 (diff) | |
freedreno/ir3: allow copy propagation for relative
This appears to work fine (with the additional constraint of keeping the
indirect load in the same block that a0.x was loaded).
We can probably lift this restriction on earlier gens after testing.
Signed-off-by: Rob Clark <[email protected]>
Reviewed-by: Eric Anholt <[email protected]>
| -rw-r--r-- | src/freedreno/ir3/ir3_cp.c | 28 |
1 files changed, 19 insertions, 9 deletions
diff --git a/src/freedreno/ir3/ir3_cp.c b/src/freedreno/ir3/ir3_cp.c index f0fe8362be0..1baf11971a8 100644 --- a/src/freedreno/ir3/ir3_cp.c +++ b/src/freedreno/ir3/ir3_cp.c @@ -92,11 +92,12 @@ static unsigned cp_flags(unsigned flags) static bool valid_flags(struct ir3_instruction *instr, unsigned n, unsigned flags) { + struct ir3_compiler *compiler = instr->block->shader->compiler; unsigned valid_flags; if ((flags & IR3_REG_HIGH) && (opc_cat(instr->opc) > 1) && - (instr->block->shader->compiler->gpu_id >= 600)) + (compiler->gpu_id >= 600)) return false; flags = cp_flags(flags); @@ -108,14 +109,23 @@ static bool valid_flags(struct ir3_instruction *instr, unsigned n, (flags & IR3_REG_RELATIV)) return false; - /* TODO it seems to *mostly* work to cp RELATIV, except we get some - * intermittent piglit variable-indexing fails. Newer blob driver - * doesn't seem to cp these. Possibly this is hw workaround? Not - * sure, but until that is understood better, lets just switch off - * cp for indirect src's: - */ - if (flags & IR3_REG_RELATIV) - return false; + if (flags & IR3_REG_RELATIV) { + /* TODO need to test on earlier gens.. pretty sure the earlier + * problem was just that we didn't check that the src was from + * same block (since we can't propagate address register values + * across blocks currently) + */ + if (compiler->gpu_id < 600) + return false; + + /* NOTE in the special try_swap_mad_two_srcs() case we can be + * called on a src that has already had an indirect load folded + * in, in which case ssa() returns NULL + */ + struct ir3_instruction *src = ssa(instr->regs[n+1]); + if (src && src->address->block != instr->block) + return false; + } switch (opc_cat(instr->opc)) { case 1: |