Re: [Patch] [regression] libvgl and r197330 (kbd)

From: John Baldwin <jhb_at_freebsd.org>
Date: Sat, 16 Jul 2011 21:22:00 -0400
On Friday, June 17, 2011 08:18:03 pm Ben Kaduk wrote:
> On Fri, Sep 25, 2009 at 8:39 AM, Ed Schouten <ed_at_80386.nl> wrote:
> > Hi all,
> > 
> > * Kostik Belousov <kostikbel_at_gmail.com> wrote:
> >> > Ah, it seems SDL also calls GIO_KEYMAP. Just rebuilding SDL should fix
> >> > this. I promised to add a message to UPDATING as well, so I'll also
> >> > mention SDL should be rebuilt as well.
> >> 
> >> I consider this as a very strong argument to keep the existing ioctl
> >> as is, and provide new ioctl that takes new table.
> > 
> > I've attached a patch that should restore binary compatibility. I first
> > thought this wasn't really needed, because most applications would use
> > K_RAW instead of K_XLATE anyway.
> > 
> > Just breaking binary compatibility with kbdcontrol(1) wouldn't have been
> > too bad, but it turns out things like SDL use this as well. I've
> > attached a patch that should restore binary compatibility. Anyone
> > interested in testing this before I commit it to SVN?
> 
> Replying to ancient history, it looks like this patch never got
> committed?  The Debian kFreeBSD folks have run into a similar issue:
> http://lists.debian.org/debian-bsd/2011/06/msg00238.html
> proposing
> ===============
> Upstream could do it properly, without ABI breaking, i.e. by
> 
> #define GIO_KEYMAP_OLD   _IOR('k', 6, keymap_t)
> #define PIO_KEYMAP_OLD   _IOW('k', 7, keymap_t)
> ...
> #define GIO_KEYMAP       _IO('k', 16)
> #define PIO_KEYMAP       _IO('k', 17)
> ===============
> 
> Something to keep the ABI between 8 and 9 is probably still useful,
> even at this juncture.

I just asked the kFreeBSD folks to test Ed's patch and it does restore the 
ABI.  Ed, can you commit this patch as it's been successfully tested now?

-- 
John Baldwin
Received on Sat Jul 16 2011 - 23:22:03 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:15 UTC