X-Authentication-Warning: delorie.com: mail set sender to djgpp-bounces using -f X-Recipient: djgpp AT delorie DOT com Message-ID: <4F71468E.7010601@iki.fi> Date: Tue, 27 Mar 2012 06:48:14 +0200 From: Andris Pavenis User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.1) Gecko/20120216 Thunderbird/10.0.1 MIME-Version: 1.0 To: djgpp AT delorie DOT com Subject: Re: CC1.EXE/GCC.EXE crashes with SIGILL, hard. References: <16777375 DOT 1302 DOT 1331688966231 DOT JavaMail DOT geo-discussion-forums AT vbat19> <4F60E096 DOT 2060106 AT iki DOT fi> <496fa848-de2c-4020-aced-ef298a0c5ec9 AT r21g2000yqa DOT googlegroups DOT com> <75057a79-95a3-41ec-93c6-8b459c4d9f9a AT h20g2000yqd DOT googlegroups DOT com> <963ed3e2-354f-48ce-90fd-43fb7263fe2d AT v2g2000vbx DOT googlegroups DOT com> In-Reply-To: <963ed3e2-354f-48ce-90fd-43fb7263fe2d@v2g2000vbx.googlegroups.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Reply-To: djgpp AT delorie DOT com On 03/27/2012 02:02 AM, Rugxulo wrote: > Hi, > > On Mar 20, 9:18 am, RayeR wrote: >> >> Does the testcase generate cmove even without >> __attribute__((__target__("sse"))) ? > > No, I don't think it would help, though I haven't tried again since my > previous attempt. The idea is to use detect CPU on which GCC is running and to use best suited code for actual CPU. Who would care that there is some CMOVE instructions in procedure which whould not even be executed. Therefore my inititial approach with objdump --disassemble was not correct The same approach is also used in DJGPP port of GMP (configure option --enable-fat). GMP is also used by GCC, so it would be not too surprising to also see instructions not available to i586 there. Unfortunate the real problem seems to be that GCC perhaps misdetects available features and tries to use code not suitable for i586 One could try to extract related code separately and run tests on different CPUs. I only have Core2 so I do not have systems on which to do the tests Andris