Re: Make ifcvt clean up dead comparisons

[Richard Biener <richard.guenther@gmail.com>]

On Fri, Jul 12, 2019 at 10:00 AM Richard Sandiford
<richard.sandiford@arm.com> wrote:
>
> This change is needed to avoid a regression in gcc.dg/ifcvt-3.c
> for a later patch.  Without it, we enter CSE with a dead comparison left
> by if-conversion and then eliminate the second (live) comparison in
> favour of the dead one.  That's functionally correct in itself, but it
> meant that we'd combine the subtraction and comparison into a SUBS
> before we have a chance to fold away the subtraction.
>
> Tested on aarch64-linux-gnu, aarch64_be-elf and x86_64-linux-gnu.
> OK to install?

cfg_cleanup if it does, runs fast-dce after cleaning up the CFG so does
it make sense to do this in the caller instead?  (and after removing the
live problem just in case dce tries to keep that updated?)

Otherwise this is of course ok.

Richard.

> Richard
>
>
> 2019-07-11  Richard Sandiford  <richard.sandiford@arm.com>
>
> gcc/
>         * ifcvt.c: Include dce.h.
>         (if_convert): Call run_fast_dce if the pass changed something.
>
> Index: gcc/ifcvt.c
> ===================================================================
> --- gcc/ifcvt.c 2019-03-08 18:15:26.828777872 +0000
> +++ gcc/ifcvt.c 2019-07-12 08:58:21.787403345 +0100
> @@ -46,6 +46,7 @@
>  #include "rtl-iter.h"
>  #include "ifcvt.h"
>  #include "params.h"
> +#include "dce.h"
>
>  #ifndef MAX_CONDITIONAL_EXECUTE
>  #define MAX_CONDITIONAL_EXECUTE \
> @@ -5443,6 +5444,10 @@ if_convert (bool after_combine)
>                num_true_changes);
>      }
>
> +  if (num_updated_if_blocks)
> +    /* Get rid of any dead CC-related instructions.  */
> +    run_fast_dce ();
> +
>    if (optimize == 1)
>      df_remove_problem (df_live);
>