Re: [PATCH] microblaze: fix build error on 32-bit hosts

[Michael Eager <eager@eagercon.com>]

On 10/7/23 15:53, Michael Eager wrote:
> On 10/7/23 15:01, Mark Wielaard wrote:
>> commit 6bbf24955 opcodes: microblaze: Add new bit-field instructions
>> introduced a build error on 32-bit hosts:
>>
>> elf32-microblaze.c: In function ‘microblaze_elf_relax_section’:
>> elf32-microblaze.c:1989:53: error: format ‘%lx’ expects argument of 
>> type ‘long unsigned int’, but argument 5 has type ‘bfd_vma’ {aka 
>> ‘unsigned int’} [-Werror=format=]
>>   1989 |       fprintf(stderr, "%d: CORRUPT relax reloc %x %lx\n", 
>> __LINE__, val, irel->r_addend);
>>        |                                                   
>> ~~^                    ~~~~~~~~~~~~~~
>>        |                                                     
>> |                        |
>>        |                                                     long 
>> unsigned int        bfd_vma {aka unsigned int}
>>        |                                                   %x
>> elf32-microblaze.c:2074:51: error: format ‘%lx’ expects argument of 
>> type ‘long unsigned int’, but argument 5 has type ‘bfd_vma’ {aka 
>> ‘unsigned int’} [-Werror=format=]
>>   2074 |    fprintf(stderr, "%d: CORRUPT relax reloc! %x %lx\n", 
>> __LINE__, val, irelscan->r_addend);
>>        |                                                 
>> ~~^                    ~~~~~~~~~~~~~~~~~~
>>        |                                                   
>> |                            |
>>        |                                                   long 
>> unsigned int            bfd_vma {aka unsigned int}
>>        |                                                 %x
>>
>> Fix by explicitly casting the r_addend to long.
>> ---
>>   bfd/elf32-microblaze.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/bfd/elf32-microblaze.c b/bfd/elf32-microblaze.c
>> index a8ced43c08a..2c584f91a4e 100644
>> --- a/bfd/elf32-microblaze.c
>> +++ b/bfd/elf32-microblaze.c
>> @@ -1986,7 +1986,7 @@ microblaze_elf_relax_section (bfd *abfd,
>>           /* Validate the in-band val.  */
>>           val = bfd_get_32 (abfd, contents + irel->r_offset);
>>           if (val != irel->r_addend && ELF32_R_TYPE (irel->r_info) == 
>> R_MICROBLAZE_32_NONE) {
>> -            fprintf(stderr, "%d: CORRUPT relax reloc %x %lx\n", 
>> __LINE__, val, irel->r_addend);
>> +            fprintf(stderr, "%d: CORRUPT relax reloc %x %lx\n", 
>> __LINE__, val, (long) irel->r_addend);
>>           }
>>           irel->r_addend -= (efix - sfix);
>>           /* Should use HOWTO.  */
>> @@ -2071,7 +2071,7 @@ microblaze_elf_relax_section (bfd *abfd,
>>             val = bfd_get_32 (abfd, ocontents + irelscan->r_offset);
>>             if (val != irelscan->r_addend) {
>> -            fprintf(stderr, "%d: CORRUPT relax reloc! %x %lx\n", 
>> __LINE__, val, irelscan->r_addend);
>> +            fprintf(stderr, "%d: CORRUPT relax reloc! %x %lx\n", 
>> __LINE__, val, (long) irelscan->r_addend);
>>             }
>>             irelscan->r_addend -= calc_fixup (irelscan->r_addend, 0, 
>> sec);
> 
> I didn't see any build errors, building with GCC-12.3.1.

Building with GCC-13.2.1 on Fedora 38 -- no error messages.

> Which version of GCC are you using?
> 
> (Patch was reverted.)
> 

-- 
Michael Eager