On Thu, Aug 2, 2012 at 11:30 AM, Al Viro [off-list ref] wrote:
FWIW, I've just pushed (completely untested) arm and alpha
parts of what I described into signal.git#execve2; x86 is
next. Note that after that sys_execve() is identical on
converted architectures and can be merged; ditto for
kernel_execve(). After I do x86 counterpart, I'll
take those guys to fs/exec.c under ifdef for new __ARCH_HAS_...
(and define it on already converted ones, obviously).
Then your patch goes there, except that implementation
gets put into fs/exec.c, under the same ifdef. And with
current_pt_regs() used instead of the extra argument,
of course. From that point on it can be used on any converted
architecture.
OK, that makes sense to me.
What would you need from me, and when? Should I just wait for your
#ifdef-ed sys_execve() in fs/exec.c, and re-spin the patch based on
that?
Is there anything else I can/should do in the meantime to make this
patch acceptable?
Meredydd