This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch fixing a typo.
On Fri, Apr 24, 2009 at 5:36 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Wed, Apr 22, 2009 at 1:22 PM, Vladimir Makarov <vmakarov@redhat.com> wrote:
>> I've made a mistake when prepared a big patch in Dec (this patch without
>> #ifdef was successfully tested on m32c. ?#ifdef and #endif were added a bit
>> lately and the final patch was bootstrapped and tested on x86/x86_64). ?Ian
>> recently found this typo
>>
>> http://gcc.gnu.org/ml/gcc/2009-04/msg00544.html
>>
>> The typo was not a serious one. ?It could have resulted in worse code for
>> most targets when -fira-algorithm=priority is used (it is not a default for
>> all architectures except for m32c). ?Although it might have created problems
>> for m32c (a weird architecture which uses priority coloring by default).
>>
>> Here is the fix
>>
>> 2009-04-22 ?Vladimir Makarov ?<vmakarov@redhat.com>
>>
>> ? * genpreds.c (write_enum_constraint_num): Output definition of
>> ? CONSTRAINT_NUM_DEFINED_P macro.
>> ? * ira.c (setup_cover_and_important_classes): Use
>> ? CONSTRAINT_NUM_DEFINED_P instead of CONSTRAINT__LIMIT in #ifdef.
>>
>> The patch was successfully bootstrapped on x86_64. ?I hope it will also fix
>> a m32c failure reported by DJ recently.
>>
>> Is it ok to commit?
>>
>>
>> Index: genpreds.c
>> ===================================================================
>> --- genpreds.c ?(revision 146528)
>> +++ genpreds.c ?(working copy)
>> @@ -954,6 +954,7 @@ write_enum_constraint_num (void)
>> ?{
>> ? struct constraint_data *c;
>>
>> + ?fputs ("#define CONSTRAINT_NUM_DEFINED_P 1\n", stdout);
>> ? fputs ("enum constraint_num\n"
>> ? ? ? ? "{\n"
>> ? ? ? ? " ?CONSTRAINT__UNKNOWN = 0", stdout);
>> Index: ira.c
>> ===================================================================
>> --- ira.c ? ? ? (revision 146528)
>> +++ ira.c ? ? ? (working copy)
>> @@ -754,9 +754,9 @@ setup_cover_and_important_classes (void)
>> ? ? ? ?{
>> ? ? ? ? ?if (i == NO_REGS)
>> ? ? ? ? ? ?continue;
>> -#ifdef CONSTRAINT__LIMIT
>> +#ifdef CONSTRAINT_NUM_DEFINED_P
>> ? ? ? ? ?for (j = 0; j < CONSTRAINT__LIMIT; j++)
>> - ? ? ? ? ? if ((int) regclass_for_constraint (j) == i)
>> + ? ? ? ? ? if ((int) regclass_for_constraint ((enum constraint_num) j) ==
>> i)
>> ? ? ? ? ? ? ?break;
>> ? ? ? ? ?if (j < CONSTRAINT__LIMIT)
>> ? ? ? ? ? ?{
>>
>
> It breaks gcc:
>
> cc1: warnings being treated as errors
> ../../src-trunk/gcc/ira.c: In function 'setup_cover_and_important_classes':
> ../../src-trunk/gcc/ira.c:763: error: enum conversion in assignment is
> invalid in C++
> make[6]: *** [ira.o] Error 1
>
I am checking it in as an obvious fix.
H.J.
Index: ira.c
===================================================================
--- ira.c (revision 146749)
+++ ira.c (working copy)
@@ -760,7 +760,7 @@ setup_cover_and_important_classes (void)
break;
if (j < CONSTRAINT__LIMIT)
{
- classes[n++] = i;
+ classes[n++] = (enum reg_class) i;
continue;
}
#endif