Avoid using abs() on unsigned quantity in powerpc's reloc_jmpslot()

[dim]

Summary

During powerpc tests in the clang350-import branch, I noticed this -Werror warning:

libexec/rtld-elf/powerpc/reloc.c:486:6: error: taking the absolute value of unsigned type 'Elf_Addr' (aka 'unsigned int') has no effect [-Werror,-Wabsolute-value]
        if (abs(offset) < 32*1024*1024) {     /* inside 32MB? */
            ^
libexec/rtld-elf/powerpc/reloc.c:486:6: note: remove the call to 'abs' since unsigned values cannot be negative
        if (abs(offset) < 32*1024*1024) {     /* inside 32MB? */
            ^~~
1 error generated.

Just before this line, offset is calculated as the difference between two other Elf_Addr quantities, so the result may well be "negative" in the sense that it has wrapped around UINT_MAX.

So I propose to avoid using abs() altogether, and just test explicitly for the correct ranges. I replaced the decimal constants by hexadecimal ones, since that seemed a little less magical to me. Otherwise the 0xfe000000 constant can be changed to 4064*1024*1024.

Test Plan

Compile, and check if it does not blow up on actual powerpc32 hardware. (Justin, maybe you can try this?)

[kib]

I think the patch is obfuscation.

FWIW, why clang issues the warning ? Rules for C types cast definitely require uint to be converted to int there, with preservation of the bit-value. This, combining with the fact that ppc32 uses two-compliment representation of the integers, makes the code well-defined and behaving as the author intended. The 'arguments' about portability do not make any sense in this context, since the code implements C runtime for ppc.

Better approach to satisfy clang non-suitability for system programming, is to explicitely cast abs() argument to int.
Does it fix the warning ?

[dim]

To avoid the warning about abs(), explicitly cast the input argument to int instead.

[dim]

If casting the input argument explicitly to int makes you happier, then it is fine with me. It will obviously also fix the warning.

[kib]

I am unhappy that perfectly valid code has to be mangled to satisfy the compiler.

[dim]

Closed by commit rS276342 (authored by @dim).