RFA:fix PR middle-end/23467

Joern RENNECKE joern.rennecke@st.com
Fri Aug 26 15:12:00 GMT 2005


Ranjit Mathew wrote:

>Joern RENNECKE wrote:
>  
>
>>2005-08-18  J"orn Rennecke <joern.rennecke@st.com>
>>
>>gcc:
>>	PR middle-end/23467
>>	* stor-layout.c (finalize_type_size): Dont override
>>	existing alignment with a smaller alignment from the mode.
>>    
>>
>
>The testcase passes for me (i686-pc-linux-gnu) with or without this
>patch. 
>
Do you mean a java testcase?
Or the pr23467.c testcase?  If the latter, that is expected, because the 
problem
that the testcase is written for pertains to STRICT_ALIGNMENT targets only.

>Note that the part you have patched is executed in some
>cases even for non-STRICT_ALIGNMENT cases.
>  
>
Yes, I see that, and it would not conflict with fixing pr23467 to 
restore the
old behaviour there.  However, I'd like to know why we ever would start out
with a larger alignment, and then want to reduce it according to the mode.
I've modified my local copy of the code to abort under the circumstances in
question, and re-build java with this patch, but if anyone has some insight
what is going on, it would be appreciated.



More information about the Gcc-patches mailing list