From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 12004 invoked by alias); 12 May 2007 20:27:48 -0000 Received: (qmail 11967 invoked by alias); 12 May 2007 20:27:36 -0000 Date: Sat, 12 May 2007 20:27:00 -0000 Message-ID: <20070512202736.11966.qmail@sourceware.org> From: "kreckel at ginac dot de" To: glibc-bugs@sources.redhat.com In-Reply-To: <20061026110459.3427.kreckel@ginac.de> References: <20061026110459.3427.kreckel@ginac.de> Reply-To: sourceware-bugzilla@sourceware.org Subject: [Bug math/3427] feholdexcept does not clear status flags X-Bugzilla-Reason: CC Mailing-List: contact glibc-bugs-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Post: List-Help: , Sender: glibc-bugs-owner@sourceware.org X-SW-Source: 2007-05/txt/msg00054.txt.bz2 ------- Additional Comments From kreckel at ginac dot de 2007-05-12 21:27 ------- Subject: Re: feholdexcept does not clear status flags Jakub, jakub at redhat dot com wrote: > ------- Additional Comments From jakub at redhat dot com 2007-05-10 12:50 ------- > This is fixed in CVS for some time already. I just read your patch at and re-read the Intel instruction set reference. And now the patch appears to be overly complex to me. Regarding the fnstenv insn used, the Intel instruction set reference says: "Store FPU environment to [...] m28byte without checking for pending unmasked floating-point exceptions. Then mask all floating-point exceptions." Note the last sentence. The part after the comment "Now set all exceptions to non-stop" is unnecessary. The part "And clear all exceptions" is more easily done with the fnclex instruction. This way, the temp fenv_t is not needed at all. Cheers -richy. -- http://sourceware.org/bugzilla/show_bug.cgi?id=3427 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.