This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [rfa] make bfd_set_section_contents buffer param constant


On Fri, Oct 31, 2003 at 12:22:10PM -0500, Andrew Cagney wrote:

	* section.c (bfd_set_section_contents): Make the "location" buffer
	constant.
	* bfd-in2.h: Re-generate.


OK.

Just commiting. Thanks.


Andrew, if you're in cleanup mode can I suggest another worthwhile
cleanup?  A bit more work than this one.

In bfd we have a number of places that use a bfd_vma to store/pass data
known to require 64 bit integer types.  We ought to use an explicit 64
bit type, as bfd_vma is only 32 bits when configuring for a 32 bit
target (unless you use --enable-64-bit-bfd, which bloats libbfd).  One
example of a place that would benefit is dwarf2.c, where we run into an
assert if processing DW_FORM_data8.  I can't see anything in the DWARF3
spec that precludes use of data8 and ref8 in a 32 bit ELF file.

The cleanups are driven by -Werror problems so don't get your hopes up :-)


But anyway, does BFD export explict but vaguely sensible signed/unsigned 8/16/32/64 bit types(1)? I don't call BFD_HOST_U_64_BIT sane but I might consider bfd_u64. The sim's have unsigned64 but that is also considered too long.

Andrew



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]