Re: [PATCH v4 1/1] x86, relocs: add printf attribute to die()

[Nicolas Iooss <nicolas.iooss_linux@m4x.org>]

On Tue, Mar 7, 2017 at 10:16 AM, Ingo Molnar <mingo@kernel.org> wrote:
>
> * Nicolas Iooss <nicolas.iooss_linux@m4x.org> wrote:
>
>> Adding such an attribute helps to detect errors in the format string at
>> build time. After doing this, the compiler complains about some issues:
>>
>>     arch/x86/tools/relocs.c:460:5: error: format specifies type 'int'
>>     but the argument has type 'Elf64_Xword' (aka 'unsigned long')
>>     [-Werror,-Wformat]
>>                                     sec->shdr.sh_size);
>>                                     ^~~~~~~~~~~~~~~~~
>>     arch/x86/tools/relocs.c:464:5: error: format specifies type 'int'
>>     but the argument has type 'Elf64_Off' (aka 'unsigned long')
>>     [-Werror,-Wformat]
>>                                     sec->shdr.sh_offset, strerror(errno));
>>                                     ^~~~~~~~~~~~~~~~~~~
>>
>> When relocs.c is included by relocs_32.c, sec->shdr.sh_size and
>> sec->shdr.sh_offset are 32-bit unsigned integers. When the file is
>> included by relocs_64.c, these expressions are 64-bit unsigned integers.
>>
>> Add casts to unsigned long long, which length is always 64-bit, and use
>> %llu to format sec->shdr.sh_size and sec->shdr.sh_offset in relocs.c.
>>
>> While at it, constify the format attribute of die().
>>
>> Signed-off-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
>> ---
>>  arch/x86/tools/relocs.c        | 31 +++++++++++++++++--------------
>>  arch/x86/tools/relocs.h        |  3 ++-
>>  arch/x86/tools/relocs_common.c |  2 +-
>>  3 files changed, 20 insertions(+), 16 deletions(-)
>>
>> diff --git a/arch/x86/tools/relocs.c b/arch/x86/tools/relocs.c
>> index 73eb7fd4aec4..3cc02065c677 100644
>> --- a/arch/x86/tools/relocs.c
>> +++ b/arch/x86/tools/relocs.c
>> @@ -397,8 +397,8 @@ static void read_shdrs(FILE *fp)
>>                   ehdr.e_shnum);
>>       }
>>       if (fseek(fp, ehdr.e_shoff, SEEK_SET) < 0) {
>> -             die("Seek to %d failed: %s\n",
>> -                     ehdr.e_shoff, strerror(errno));
>> +             die("Seek to %llu failed: %s\n",
>> +                     (unsigned long long)ehdr.e_shoff, strerror(errno));
>
> Isn't "(u64)" shorter to write?

u64 does not seem to be defined in this file right now. Adding
"#include <linux/types.h>" defines u64 and __u64 in the following way:
- "typedef uint64_t u64;" from tools/include/linux/types.h
- "typedef unsigned long long __u64;" from /usr/include/asm-generic/int-ll64.h

uint64_t is unsigned long on x86-64 and gcc complains when using %llu
on such a type, so using a cast to u64 forces using PRIu64 too.

Nevertheless "(__u64)" is shorter than "(unsigned long long)" and
seems to work fine in my quick tests because it is always unsigned
long long (on both x86-32 and x86-64). Would you prefer to use this
cast?

Thanks,
Nicolas