This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] make data dependence analyzer more strict
- From: "Giovanni Bajo" <rasky at develer dot com>
- To: "Sebastian Pop" <sebastian dot pop at cri dot ensmp dot fr>
- Cc: <gcc-patches at gcc dot gnu dot org>
- Date: Mon, 9 May 2005 15:37:31 +0200
- Subject: Re: [patch] make data dependence analyzer more strict
- References: <20050509093155.GA7153@napoca.cri.ensmp.fr>
Sebastian Pop <sebastian.pop@cri.ensmp.fr> wrote:
> typedef struct rtx_
> {
> } *rtx;
> static rtx regno_save_mem[53][16 / 4 + 1];
> extern set_mem_alias_set (rtx, rtx);
> int main(void)
> {
> int i, j;
> for (i = 0; i < 53; i++)
> for (j = (16 / (0 ? 8 : 4)); j > 0; j--)
> if (regno_save_mem[i][j] != 0)
> set_mem_alias_set (regno_save_mem[i][j], 0);
> }
>
> In order to transform this loop, we have to prove that set_mem_alias_set
> does not contain side effects that may change the contents of the array.
> I'm proposing to remove this testcase for the moment.
Is it enough to mark set_mem_alias_set with attribute pure?
--
Giovanni Bajo