This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: Cleanups for the m68k backend

From: Jim Wilson <wilson at tuliptree dot org>
To: Peter Barada <pbarada at mail dot wm dot sps dot mot dot com>
Cc: gcc at gcc dot gnu dot org, ghazi at caip dot rutgers dot edu, rth at redhat dot com
Date: Tue, 08 Jul 2003 12:04:03 -0700
Subject: Re: Cleanups for the m68k backend
References: <200307051523.01329.bernie@develer.com> <20030708000911.GL14818@redhat.com> <jefzlhwaq0.fsf@sykes.suse.de> <200307081515.18581.bernie@develer.com> <200307081512.h68FCTQ20379@hyper.wm.sps.mot.com>

Peter Barada wrote:

[(set (mem:SI (pre_dec:SI (reg:SI 15)) (match_operand:SI 0 "address_operand" "a")))
 (set (match_dup 0) (reg:SI 15))
 (set (reg:SI 15) (add:SI (reg:SI 15) (match_operand:SI 1 "const_int_operand" "n")))]

You are assuming sequential semantics here, however, the compiler implements this with a PARALLEL which implies that everything happens at once. Thus you can't write it this way. You need to do something more like (set (mem (plus (reg 15) (const_int -4)) operand0) (set operand0 (plus (reg 15) (const_int -4)) (set (reg 15) (plus (reg 15) operand1))

Similarly for unlink.

Jim

References:
- Cleanups for the m68k backend
  - From: Bernardo Innocenti
- Re: Cleanups for the m68k backend
  - From: Richard Henderson
- Re: Cleanups for the m68k backend
  - From: Andreas Schwab
- Re: Cleanups for the m68k backend
  - From: Bernardo Innocenti
- Re: Cleanups for the m68k backend
  - From: Peter Barada

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]