[PATCH, middle-end]: Backport "[PATCH] Avoid useless bitfield extractions" to 4.3 branch

Uros Bizjak ubizjak@gmail.com
Tue Jan 6 16:31:00 GMT 2009


Eric Botcazou wrote:
>> Eric's patch [1] fixes:
>>
>> FAIL: g++.dg/torture/pr35526.C  -O1  (internal compiler error)
>> FAIL: g++.dg/torture/pr35526.C  -O1  (test for excess errors)
>> FAIL: g++.dg/torture/pr35526.C  -O2  (internal compiler error)
>> FAIL: g++.dg/torture/pr35526.C  -O2  (test for excess errors)
>> FAIL: g++.dg/torture/pr35526.C  -O3 -fomit-frame-pointer  (internal
>> compiler error) FAIL: g++.dg/torture/pr35526.C  -O3 -fomit-frame-pointer 
>> (test for excess errors) FAIL: g++.dg/torture/pr35526.C  -O3 -g  (internal
>> compiler error)
>> FAIL: g++.dg/torture/pr35526.C  -O3 -g  (test for excess errors)
>> FAIL: g++.dg/torture/pr35526.C  -Os  (internal compiler error)
>> FAIL: g++.dg/torture/pr35526.C  -Os  (test for excess errors)
>>
>> for alpha-unknown-linux-gnu on 4.3 branch.
>>
>> 2008-12-01  Uros Bizjak  <ubizjak@gmail.com>
>>
>>     Backport from mainline:
>>     2008-06-12  Eric Botcazou  <ebotcazou@adacore.com>
>>
>>     * expr.c (store_field): Do a block copy from BLKmode to BLKmode-like.
>>     (get_inner_reference): Use BLKmode for byte-aligned BLKmode bitfields.
>>     
>
> FWIW we have been using it for one year in 4.1- and 4.3-based compilers.  You 
> probably just need the first hunk though.
>   

I have committed the complete patch under the assumption that having 
also the second hunk doesn't hurt.

Thanks,
Uros.




More information about the Gcc-patches mailing list