[PATCH] S/390 port modernization 1/4
Thu Dec 4 22:11:00 GMT 2003
Jim Blandy wrote:
> The regset_from_core_section stuff is done exactly the way the i386
> and x86-64 do it, so I won't criticize it. But there are two things I
> wonder about:
> - Since core file formats are really OS-specific (ABI's often don't
> specify them), wouldn't it make more sense for the *-nat.c file to
> register an OS/ABI handler, and have that handler register the
> regset_from_core_section method?
Defining the core file format in *-nat.c would mean that a cross-gdb
wouldn't understand core files, right?
> - Why does everyone allocate the 'struct regset' objects dynamically?
> The regset layouts are structures in header files, so there's only a
> static set of them; one could simply define initialized instances of
> 'struct regset', and the regset_from_core_section method could
> return them. That seems simpler than allocating them dynamically,
> and caching them in the tdep structure.
OK, I'll change that.
Dr. Ulrich Weigand
More information about the Gdb-patches