Date: Tue, 20 Jun 2000 18:44:29 +0200 (MET DST)
From: "Maciej W. Rozycki" <[EMAIL PROTECTED]>
On Tue, 20 Jun 2000, Mark Kettenis wrote:
> That's basically right. What flavor of Linux are you using?
> Linux/x86 doesn't use core-regset.c, so it doesn't suffer from this
> braindamage. There are some Linux variants where the gregset_t and
> elf_gregset_t are one and the same type. Other Linuxen have
> maintainers who are still confused by this stuff :-(.
I am using mipsel-linux: linux 2.3.99-pre6, glibc 2.1.90. While support
for it is not yet included in gdb, I have a few patches that make it work
very well for me. I sent a few of them here yesterday. The two remaining
ones are bfd/opcodes updates by Ulf Carlsson, which I believe are already
applied to the main binutils tree, and a port of MIPS/Linux specific
changes that are currently available from oss.sgi.com (and reflect H.J.
Lu's gdb 4.17.0.11). I'm contacting authors of these changes; if no one
responds, I'll submit this patch myself.
OK, that explains things. Please keep in mind that we need copyright
assignments for any substantial patches, in this case from the
*origional* authors.
By the way, did you post your patches to [EMAIL PROTECTED]? If so,
please post them to [EMAIL PROTECTED] too.
[EMAIL PROTECTED] is for reporting bugs in GDB, and doesn't really get
that much attention from GDB developers.
I've seen it may also affect arm and m68k, at least.
AFAIK it doesn't. But that isn't a reason to stop us from fixing the
(potential) problems of course.
> Some work has been done on rationalizing the regset mess already. GDB
> now defines a type gdb_gregset_t (and gdb_fpregset_t), which should be
> a typedef for elf_gregset_t on Linux. I think core-regset.c should use
> these types instead.
OK, I'll see how I can update the patch.
Great!
Mark
