This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch 3/4] Nios II gdbserver support
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Sandra Loosemore <sandra at codesourcery dot com>
- Cc: <gdb-patches at sourceware dot org>
- Date: Tue, 23 Apr 2013 13:54:54 +0000
- Subject: Re: [patch 3/4] Nios II gdbserver support
- References: <5174820E dot 6090500 at codesourcery dot com> <Pine dot LNX dot 4 dot 64 dot 1304222101460 dot 8464 at digraph dot polyomino dot org dot uk> <5175CC11 dot 7060905 at codesourcery dot com>
On Mon, 22 Apr 2013, Sandra Loosemore wrote:
> > Copyright (C) 2008-2013 Free Software Foundation, Inc.
> >
> > (all on one line) and likewise for any other new files in this patch
> > series.
>
> Hmmm, I thought the year range was only acceptable when there were actually
> modifications to the file in all of those years?
No, modifications to the package as a whole (presuming it had public
version control, or a public release, for each year in question), not to
the individual file.
> > > +#ifndef PTRACE_GET_THREAD_AREA
> > > +#define PTRACE_GET_THREAD_AREA 25
> > > +#endif
> >
> > The same comment I made on the AArch64 port applies here: for native code
> > in new ports, as opposed to use of new definitions in existing ports, you
> > should just be able to assume recent-enough headers (at least until the
> > Linux kernel port goes upstream, at which point the first upstream release
> > with the code can be the minimum version of kernel headers supported)
> > rather than having such #if conditionals.
>
> I'm not sure what change you want here. If I remove that block of code, the
> file fails to compile with the kernel headers currently being provided by
> Altera, which do not define PTRACE_GET_THREAD_AREA.
That sounds like the kernel headers need to be fixed before the kernel
port goes upstream, at which point GDB can depend on the first kernel
version to go upstream rather than allowing for old kernel header versions
(which will be ABI-incompatible with the upstream version anyway because
of not using the generic syscall ABI).
--
Joseph S. Myers
joseph@codesourcery.com