Re: [PATCH] Generate an if instead of a switch with one case in genmatch.

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

On Wed, Mar 16, 2022 at 1:28 PM Roger Sayle <roger@nextmovesoftware.com> wrote:
>
>
> This patch is the first of two changes to genmatch that don't affect
> the executable code, but reduce the amount of debugging information
> generated in stage3 of a build, but adhering more closely to GNU style
> guidelines.
>
> This patch avoids generating a switch with a single case statement,
> instead preferring to use an "if (TREE_CODE (...) == SSA_NAME)" idiom.
> These should compile to the same instructions, but the switch requires
> more lines, especially when a debugger may set a break point on the
> switch, the case, or the (obligatory) final "default:;".  This reduces
> the size of gimple-match.o by 53K on x86_64-pc-linux-gnu.
>
> This patch has been tested on x86_64-pc-linux-gnu with make bootstrap
> and make -k check with no new failures.  Ok for mainline?

Hmm, this makes the complicated code emission in gen_kids_1 even more
complicated for a questionable gain which I'd rather not do, debuginfo
savings or not ...

Richard.

>
> 2022-03-16  Roger Sayle  <roger@nextmovesoftware.com>
>
> gcc/ChangeLog
>         * gcc/genmatch.cc (dt_node::gen_kids_1): Introduce use_switch
>         logic that prefers to generate an if statement rather than a
>         switch containing a single case (and a default).
>
>
> Thanks in advance,
> Roger
> --
>