Re: [PATCH] gold: fix some signed-unsigned comparison warnings

[Cary Coutant <ccoutant at google dot com>]

> diff --git a/gold/arm.cc b/gold/arm.cc
> index 5770c8a..d53ef6b 100644
> --- a/gold/arm.cc
> +++ b/gold/arm.cc
> @@ -9688,7 +9688,7 @@ Target_arm<big_endian>::relocate_special_relocatable(
>
>    Arm_address offset = reloc.get_r_offset();
>    Arm_address new_offset;
> -  if (offset_in_output_section != invalid_address)
> +  if (offset_in_output_section != (off_t) invalid_address)
>      new_offset = offset + offset_in_output_section;
>    else
>      {
> @@ -9707,7 +9707,7 @@ Target_arm<big_endian>::relocate_special_relocatable(
>    if (!parameters->options().relocatable())
>      {
>        new_offset += view_address;
> -      if (offset_in_output_section != invalid_address)
> +      if (offset_in_output_section != (off_t) invalid_address)
>         new_offset -= offset_in_output_section;
>      }

Seems to me that offset_in_output_section is misdeclared here. As
passed from relocate_relocs(), the actual parameter is declared as an
Elf_Addr, which is unsigned. I think offset_in_output_section should
be Arm_address here.

> diff --git a/gold/dwarf_reader.cc b/gold/dwarf_reader.cc
> index c80e8cb..50942d6 100644
> --- a/gold/dwarf_reader.cc
> +++ b/gold/dwarf_reader.cc
> @@ -57,7 +57,7 @@ Sized_elf_reloc_mapper<size, big_endian>::symbol_section(
>      unsigned int symndx, Address* value, bool* is_ordinary)
>  {
>    const int symsize = elfcpp::Elf_sizes<size>::sym_size;
> -  gold_assert((symndx + 1) * symsize <= this->symtab_size_);
> +  gold_assert((off_t) ((symndx + 1) * symsize) <= this->symtab_size_);
>    elfcpp::Sym<size, big_endian> elfsym(this->symtab_ + symndx * symsize);
>    *value = elfsym.get_st_value();
>    return this->object_->adjust_sym_shndx(symndx, elfsym.get_st_shndx(),

And here, I think the right thing to do is declare symsize as an
off_t, since that's what Sized_elf_reloc_mapper::symtab_size_ is.

> diff --git a/gold/incremental.cc b/gold/incremental.cc
> index acabaea..9898434 100644
> --- a/gold/incremental.cc
> +++ b/gold/incremental.cc
> @@ -1432,7 +1432,7 @@ Output_section_incremental_inputs<size,
> big_endian>::do_write(Output_file* of)
>    gold_assert(pov - oview == oview_size);
>
>    // Write the .gnu_incremental_symtab section.
> -  gold_assert(global_sym_count * 4 == symtab_size);
> +  gold_assert((off_t) global_sym_count * 4 == symtab_size);

This is the right thing here, but it's preferable to use C++
static_cast<off_t>(...).

> --- a/gold/output.cc
> +++ b/gold/output.cc
> @@ -1935,7 +1935,7 @@ Output_symtab_xindex::endian_do_write(unsigned
> char* const oview)
>         ++p)
>      {
>        unsigned int symndx = p->first;
> -      gold_assert(symndx * 4 < this->data_size());
> +      gold_assert((off_t) symndx * 4 < this->data_size());

Likewise.

-cary