This is the mail archive of the
mailing list for the GCC project.
Re: [patch]: builtin_memset support (new standard pattern setmem)
- From: Richard Henderson <rth at redhat dot com>
- To: Adrian Straetling <straetling at de dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Sun, 26 Jun 2005 23:19:04 -0700
- Subject: Re: [patch]: builtin_memset support (new standard pattern setmem)
- References: <20050623163650.GA21722@de.ibm.com>
On Thu, Jun 23, 2005 at 06:36:51PM +0200, Adrian Straetling wrote:
> Actually, I already have a patch on my disk that does so by
> simply removing "clear_storage_via_clrmem" and the 'clrmem' pattern
> itself. On the back-end side it should IMHO suffice to rename the
> relevant expander patterns, add a new operand x which holds the byte
> value and add:
> if (operands[x] != const0_rtx) FAIL;
> as their first line. They should then behave as 'clrmem' pattern do now.
> I tested that on s390x and it behaved as desired.
I would prefer this approach. It does sound very much like your
patch does the right thing; I'd like to see it before I approve