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: df.c bug fix for gcc_3_4_0_release

From: Jim Wilson <wilson at specifixinc dot com>
To: Leehod Baruch <LEEHOD at il dot ibm dot com>
Cc: Mircea Namolaru <NAMOLARU at il dot ibm dot com>, gcc at gcc dot gnu dot org
Date: Tue, 06 Jul 2004 11:38:41 -0700
Subject: Re: df.c bug fix for gcc_3_4_0_release
References: <OF9FB3B2E7.DDB519B1-ONC2256EBE.004241C3-C2256EBE.0045766A@il.ibm.com>

Leehod Baruch wrote:

From the code it seems that the problem exists even in gcc 3.4.0 release.
Look at the above patch for my suggested changes to fix this problem.

Your suggested patch looks correct to me. However, in order to get this into gcc-3.4.x, we need a bug report filed against 3.4.x that can be used to reproduce it, and it needs to be a regression from an earlier release. If the only testcase you have is from your own development work, then that probably won't be good enough.

Meanwhile, as you mentioned, it has already been fixed on mainline.

This doesn't seem to be a serious problem for the gcc-3.4.x release. The df.c module is only used by a few new optimization passes that aren't on by default (-fweb, -fnew-ra), and it isn't clear if they use the backwards propagation code that is buggy.

This probably got fixed on mainline because there are more uses of this code in current sources. -- Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com

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