From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 24979 invoked by alias); 27 Mar 2004 07:07:55 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 24965 invoked from network); 27 Mar 2004 07:07:53 -0000 Received: from unknown (HELO yosemite.airs.com) (209.128.65.135) by sources.redhat.com with SMTP; 27 Mar 2004 07:07:53 -0000 Received: (qmail 5049 invoked by uid 10); 27 Mar 2004 07:07:53 -0000 Received: (qmail 32415 invoked by uid 500); 27 Mar 2004 07:07:44 -0000 Message-ID: <20040327070744.32414.qmail@gossamer.airs.com> Date: Sat, 27 Mar 2004 19:38:00 -0000 From: Ian Lance Taylor To: geoffk@apple.com CC: gcc@gcc.gnu.org Subject: PR 14400--PCH problem with large files X-SW-Source: 2004-03/txt/msg01617.txt.bz2 Geoff, could you take a look at PR 14400? It is a PCH problem which arises with large files. Basically, gcc currently reads the entire source file into memory before it loads any PCH (in read_file_guts() called eventually by cpp_read_main_file()). I think this means that on a system for which PCH requires that mmap generally return the same address--i.e., any non-Darwin system--PCH will fail for a sufficiently large source file. I'm a bit tired, so I may be missing something. If not, I think we should add a test to the PCH testsuite which tries to compile a large file. I'm not sure what a good fix would be. Ian