On Tue, Nov 8, 2011 at 8:35 PM, Jeff Law <l...@redhat.com> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On 11/07/11 15:53, Richard Guenther wrote: >> On Mon, Nov 7, 2011 at 10:25 PM, Jakub Jelinek <ja...@redhat.com> >> wrote: >>> Hi! >>> >>> This patch attempts to optimize VEC_BASE if we know that offsetof >>> of base is 0 (unless the compiler is doing something strange, it >>> is true). It doesn't have a clear code size effect, some .text >>> sections grew, supposedly because of more inlining, some .text >>> sections shrunk. >>> >>> Bootstrapped/regtested on x86_64-linux and i686-linux. >> >> I wonder why the compiler doesn't optimize this ... certainly it >> looks backward to, in >> >> <bb 2>: if (c_2(D) != 0B) goto <bb 3>; else goto <bb 4>; >> >> <bb 3>: D.2948_3 = &c_2(D)->fld; goto <bb 5>; >> >> <bb 4>: D.2948_4 = 0B; >> >> <bb 5>: # D.2948_1 = PHI <D.2948_3(3), 0B(4)> return D.2948_1; >> >> see that D.2948_4 is equal to D.2948_3 for c_2 == 0, so I'm not >> sure which pass would be able to detect this (but the optimziation >> opportunity would be on the PHI node, so maybe it should be done in >> phiopt). > Right, I see that now. > > But D2948_3 doesn't flow through block #4, thus we can't use its value > for the PHI arg associated with the edge 4->5. So even with the > equivalency I think to optimize this we'd have to detect the > construct as a whole and collapse it into &c_2->fld. > > Presumably that's what your patch in a later message does.
Yes. It collapses it into c_2. As pointer type differences are no longer important in the GIMPLE IL we can represent &c_2->fld simply as c_2 and thus with copy propagation we'd present phiopt with something it already handles. Unfortunately rewriting all &c_2->fld style expressions simply as c_2 plays foul with _FORTIFY_SOURCE and object size computation (where it makes a difference which field at offset zero you access ...). So instead of canonicalizing address computations using ADDR_EXPR to POINTER_PLUS_EXPR in general for 4.7 the only thing we can do is special case the above case in phiopt (which is probably worthwhile - that pattern ought to be quite common). Richard. > jeff > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.11 (GNU/Linux) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ > > iQEcBAEBAgAGBQJOuYRuAAoJEBRtltQi2kC7E2YH/1ayeLWX3GoOnq5u+JxpyWkx > Fh/Dnar+Ad2jJJMrbx7vUUhAH54OEUDTWKYMQOn8ThUTnMmQtgb4uNqdyO78RfyM > BIVzzpTjD3Ud1+xxlzg8YCjvBw/NekRp9l4HiZpwwRIfTQRpKhWZ+oWe3bBgs+4B > LCVQWA9I5cGBhv09u7OOsYhicuAUa1Tj/XNF4NWE1GVsjilj+ESvZUj5Zd7dvxft > QzGh0fNn+RB9LVu6kktvU5CCX/sVBjVDnTaOP14zYGAEMFrWYjanE0n9+b6cVf9W > degmRowfMzOLvphsWEPFVZCTJPuDfkLlc5k6X3MkgpEFwQu6uo2S/k8vd98PAr8= > =vcgh > -----END PGP SIGNATURE----- >
