From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Hubicka To: Richard Kenner Cc: jh@suse.cz, gcc-patches@gcc.gnu.org Subject: Re: spec2000 regression Date: Wed, 15 Aug 2001 07:10:00 -0000 Message-id: <20010815161053.L19872@atrey.karlin.mff.cuni.cz> References: <10108151413.AA18046@vlsi1.ultra.nyu.edu> X-SW-Source: 2001-08/msg00914.html > If I understand the purpose of original patch properly, it was hooting > for case variable gets unique class, but this class is recomputed > several times so it get several unique classes. For class 0 this does > not matter. > > No, that's backwards. This is a case where the RTL was class 0 and so you > want to make sure the new class is 0. Your change will cause it to use > the alias set of the type, which might be nonzero and hence is wrong. Uhh, I see, then we probably need to go with Jason's patch, as I don't see any way how to check case someone call set_memory_attributes after doing DECL_RTL... Honza