This is the mail archive of the
mailing list for the binutils project.
Re: [PING] Contributing new gcc targets: i386-*-dragonfly and x86-64-*-dragonfly
- From: John Marino <gnugcc at marino dot st>
- To: Ian Lance Taylor <iant at google dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>, libstdc++ <libstdc++ at gcc dot gnu dot org>, Binutils <binutils at sourceware dot org>, gdb-patches at sourceware dot org
- Date: Tue, 29 Apr 2014 20:37:39 +0200
- Subject: Re: [PING] Contributing new gcc targets: i386-*-dragonfly and x86-64-*-dragonfly
- Authentication-results: sourceware.org; auth=none
- References: <5352D100 dot 9040108 at marino dot st> <CAH6eHdS7JsVKz=c1T_b1B6uSixCnZsqhurOPUivEm=-w3T_j3Q at mail dot gmail dot com> <53541A34 dot 7050908 at marino dot st> <535FC71E dot 8070406 at marino dot st> <CAKOQZ8xtnVwweDW0Nh9Hbf8bOrySTrD_Z114dBF=cRFDPq=H7g at mail dot gmail dot com>
On 4/29/2014 19:23, Ian Lance Taylor wrote:
> On Tue, Apr 29, 2014 at 11:37 AM, John Marino <email@example.com> wrote:
>> Does anyone have any issues with this set of patches to add support for
>> the DragonFly targets? It's a blocker for other patches of mine that
>> have a more general benefit, but this (relatively simple) one has to go
>> in first.
> It's inconvenient, but patches are much more likely to be reviewed
> when they cover a separate part of the tree, as different people
> maintain different parts. I expect your libitm and libcilkrts could
> be approved trivially if you send them separately.
I was trying to identify specific people (e.g. an libitm person) and
have them approve specific files since they are trivial as you saw. I
decided to keep the patch set as an atomic unit because it needs to be
committed as a unit, and also because I assumed it provided the
> The change to include/libiberty.h is fine.
> I don't understand the benefit of libgcc/enable-execute-stack-bsd.c.
> The code seems the same as the existing
> libgcc/enable-execute-stack-mprotect.c. All you are changing is
> omitting need_enable_exec_stack. If you just drop the FreeBSD
> constructor, you will get the behaviour you want.
With the caveat that this patch is over 2 years old, I just took a look
at both files. I would have not needed to modify this file at all for
DragonFly. In fact, I seem to recall that I didn't modify it for
DragonFly, but rather for FreeBSD. If I had to guess, it would be that
I found mprotect() was needed regardless of value of kern.stackprot. I
must have traced some test failures back to this.
Which I guess that's what you mean - just delete the block between "#if
defined __FreeBSD__" and the next #elif which should be equivalent. I
can tweak the patch set to do that.
And what about the dl_iterate_phdr changes? Do they look good to you?