[Bug target/53101] Recognize casts to sub-vectors

[marc.glisse at normalesup dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53101

--- Comment #3 from Marc Glisse <marc.glisse at normalesup dot org> 2012-05-01 17:17:42 UTC ---
(In reply to comment #2)
> but operands[2] and operands[3] don't compare equal with rtx_equal_p, and
> trying a match_dup refuses to compile because of the mode mismatch, so I don't
> know how to constrain 2 and 3 to be "the same".

rtx_equal_p (XEXP (operands[2], 0), XEXP (operands[3], 0))

seems to give the right answer in the 3 manual tests I did. Currently checking
if the testsuite finds something. It is very likely not the right way to do it,
but I didn't find any inspiring pattern in the .md files.

Then I'll see if I understand how the fancy macros make it possible to have a
single piece of code for all modes, and if instead of calling
gen_vec_extract_lo_v8sf I shouldn't give a replacement pattern like (set
(match_dup 0) (vec_select (match_dup 1) (const_int 0))).