RFA: Fix PR21767 (Was: Re: RFD: what to do about stale REG_EQUAL notes in dead_or_predictable)

Joern RENNECKE joern.rennecke@st.com
Wed Jun 1 11:31:00 GMT 2005


Bernd Schmidt wrote:

>> 2005-05-27  J"orn Rennecke <joern.rennecke@st.com>
>>
>>     * rtl.h (function_invariant_p): Re-add declaration.
>>     * reload1.c (function_invariant_p): No longer static.
>>     * ifcvt.c (dead_or_predicable): Remove REG_EQUAL notes that
>>     might have become invalid.
>
>
> This is OK.  I think we have a convention for mentioning the PR number 
> in ChangeLogs though.

Thanks.  Yes, I'll add the PR number.  In the meantime, I've 
bootstrapped & regtested (*) this on
i686-pc-linux-gnu, so I'm checking this in now.

(*) Since CVS won't bootstrap out of the box, I have used as baseline 
sources from yesterday
16:00 UTC with this patch:

005-05-31  J"orn Rennecke <joern.rennecke@st.com>

	* calls.c (emit_library_call_value_1): For const functions, add
	USEs of the stack slots to CALL_INSN_FUNCTION_USAGE.
http://gcc.gnu.org/ml/gcc-patches/2005-05/msg02946.html

Yes, and this also should say PR rtl-optimization/21848 .



More information about the Gcc-patches mailing list