This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: 2.95: sjlj eh regression on hpux
- To: Bernd Schmidt <bernds at redhat dot com>
- Subject: Re: 2.95: sjlj eh regression on hpux
- From: Richard Henderson <rth at redhat dot com>
- Date: Fri, 12 Jan 2001 09:57:27 -0800
- Cc: gcc-patches at gcc dot gnu dot org, "'David Scott Urban" <urban at ast dot lmco dot com>
- References: <Pine.LNX.4.30.0101121621080.1327-100000@host140.cambridge.redhat.com>
On Fri, Jan 12, 2001 at 04:32:13PM +0000, Bernd Schmidt wrote:
> The patch below fixes it. I suppose something like this is also needed
> for the mainline, but I have installed it only on the branch so far -
> does anyone have a better idea how to fix this cleanly?
Of our 38 ports, 22 already define a barrier insn. Of the 16 that
do not, only 7 of them set INSN_SCHEDULING. These are
a29k
arc
c4x
i960
mcore
pdp11
v850
I think we should require a barrier if a port wants scheduling. I.e.
#ifdef INSN_SCHEDULING
emit_insn (gen_barrier ());
#endif
r~