Re: [leaf-devel] compiling openswan for BuC4 - update
On Wed, 2010-11-17 at 20:05 +0200, Andrew wrote: > 17.11.2010 19:51, Erich Titl пишет: > > Hi Folks > > > > Indeed setting > > > > UCLIBC_HAS_GNU_GLOB=y > > > > in the buildenv makes the compile error for openswan go away. I would > > like to suggest to include this in the main trunk. > > > > cheers > > > > Erich > Yes, of course. You have write access for v4 tree? Andrew, Not yet. See below: Forwarded Message From: Mike Noyes Reply-to: Mike Noyes To: leaf-devel@lists.sourceforge.net Subject: Re: [leaf-devel] trying to get to BuC4 Date: Sun, 14 Nov 2010 07:53:28 -0800 On Sun, 2010-11-14 at 13:53 +0100, Erich Titl wrote: > on 14.11.2010 12:34, davidMbrooke wrote: > > The WRAP boards are a key target platform for BuC 4.x. However, I do not > > have one so personally I cannot test compatibility. > > I will port the wd1100 code again. What is the current canonical form of > entering it into the mainline? Still a patch against the kernel? Andrew & KP, I move that Erich receive write access to BuC 4.x in CVS. CVSROOT/avail avail|arneb, dorus, espakman, hejl, hun, lfcorreia, nangel, nitr0man, davidmbrooke|src/bering-uclibc4 -- Mike Noyes http://sourceforge.net/users/mhnoyes/ SF.net Projects: leaf, sourceforge/sitedocs -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] compiling openswan for BuC4 - update
17.11.2010 19:51, Erich Titl пишет: > Hi Folks > > Indeed setting > > UCLIBC_HAS_GNU_GLOB=y > > in the buildenv makes the compile error for openswan go away. I would > like to suggest to include this in the main trunk. > > cheers > > Erich Yes, of course. You have write access for v4 tree? -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] compiling openswan for BuC4 - update
Hi Folks Indeed setting UCLIBC_HAS_GNU_GLOB=y in the buildenv makes the compile error for openswan go away. I would like to suggest to include this in the main trunk. cheers Erich -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
[leaf-devel] compiling openswan for BuC4
Hi Folks I am having a little problem compiling the openswan config parser on BuC4. there is a declaration of a few macros in glob.h which depend on the definition of #. && defined __UCLIBC_HAS_GNU_GLOB__ If I force the definition with -D__UCLIBC_HAS_GNU_GLOB__ , it is not recognized. There appears to be no way to make the C preprocessor accept this definition unless I add it to the parser input file _before_ the inclusion of glob.h. I _believe_ this can be improved by rebuilding buildenv with setting the respective parameters in .config for uClibc UCLIBC_HAS_GLOB=y # UCLIBC_HAS_GNU_GLOB is not set cheers Erich -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
Hi Andrew at 17.11.2010 13:58, Andrew wrote: > if [ "$ARCH" = "i486" ]; then >SLKCFLAGS="-O2 -march=i486 -mtune=i686" > elif [ "$ARCH" = "i686" ]; then >SLKCFLAGS="-O2 -march=i686 -mtune=i686" > fi Right, apparently they do it that way, nevertheless the wording in the manual especially the word _implies_ ... implies a different meaning to me. Whatever, I have still other problems with openswan, maybe you know where these defines come from (they are used in glob.h to trigger the define of a bunch of others) __USE_POSIX2 __USE_BSD __USE_GNU __UCLIBC_HAS_GNU_GLOB__ cheers Erich -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
17.11.2010 14:51, Andrew пишет: > IMHO - this means that -march option also sets by default -mtune for > arch if it isn't specified in command line. At least, performance tests > show this behavior. > I tested different -mrach and -mtune (look at dry2 package - dhrystone > benchmark) and it shows different performance. Strange, but -march=i686 > for me shows slightly worse results that -march=i486 -mtune=i686. > After brief look - I found that slackware used same behavior: http://slackbuilds.org/howto/versions/ if [ "$ARCH" = "i486" ]; then SLKCFLAGS="-O2 -march=i486 -mtune=i686" elif [ "$ARCH" = "i686" ]; then SLKCFLAGS="-O2 -march=i686 -mtune=i686" fi -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
17.11.2010 14:37, Erich Titl пишет: > Andrew > > at 17.11.2010 12:30, Andrew wrote: >> 17.11.2010 12:03, Erich Titl пишет: >>> Hi Folks >>> >>> I am attempting to build openswan for BuC4 and I am worried a bit about >>> the environment preset in MasteriInclude.mk. >>> >>> We define GNU_TUNE as pentiumpro, which according to the gcc doc should >>> produce code for this kind of architecture, whereas GNU_ARCH is set to >>> I486 which definitely is not a pentium pro. >>> >> No, you're miss about this. Key difference between arch& tune - arch >> specifies target CPU, for which code is generated (which instruction set >> will be used in executable), and tune - specifies for which architecture >> code will be optimized; on other architectures code will work but >> possible some less effective. > Right, that is what I read from the gcc manual: > > -mtune=cpu-type > Tune to cpu-type everything applicable about the generated code, > except for the ABI and the set of available instructions. The choices > for cpu-type are: > > > > And here is what the manual sais about -march > > -march=cpu-type > Generate instructions for the machine type cpu-type. The choices for > cpu-type are the same as for -mtune. Moreover, specifying > -march=cpu-type implies -mtune=cpu-type. > > And this reads to me that, whatever you do in -mtune, if you specify > -march then it is automatically overwritten. > > And then it does not make sense. > > cheers > > Erich IMHO - this means that -march option also sets by default -mtune for arch if it isn't specified in command line. At least, performance tests show this behavior. I tested different -mrach and -mtune (look at dry2 package - dhrystone benchmark) and it shows different performance. Strange, but -march=i686 for me shows slightly worse results that -march=i486 -mtune=i686. -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
Andrew at 17.11.2010 12:30, Andrew wrote: > 17.11.2010 12:03, Erich Titl пишет: >> Hi Folks >> >> I am attempting to build openswan for BuC4 and I am worried a bit about >> the environment preset in MasteriInclude.mk. >> >> We define GNU_TUNE as pentiumpro, which according to the gcc doc should >> produce code for this kind of architecture, whereas GNU_ARCH is set to >> I486 which definitely is not a pentium pro. >> > No, you're miss about this. Key difference between arch & tune - arch > specifies target CPU, for which code is generated (which instruction set > will be used in executable), and tune - specifies for which architecture > code will be optimized; on other architectures code will work but > possible some less effective. Right, that is what I read from the gcc manual: -mtune=cpu-type Tune to cpu-type everything applicable about the generated code, except for the ABI and the set of available instructions. The choices for cpu-type are: And here is what the manual sais about -march -march=cpu-type Generate instructions for the machine type cpu-type. The choices for cpu-type are the same as for -mtune. Moreover, specifying -march=cpu-type implies -mtune=cpu-type. And this reads to me that, whatever you do in -mtune, if you specify -march then it is automatically overwritten. And then it does not make sense. cheers Erich -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
17.11.2010 12:03, Erich Titl пишет: > Hi Folks > > I am attempting to build openswan for BuC4 and I am worried a bit about > the environment preset in MasteriInclude.mk. > > We define GNU_TUNE as pentiumpro, which according to the gcc doc should > produce code for this kind of architecture, whereas GNU_ARCH is set to > I486 which definitely is not a pentium pro. > No, you're miss about this. Key difference between arch & tune - arch specifies target CPU, for which code is generated (which instruction set will be used in executable), and tune - specifies for which architecture code will be optimized; on other architectures code will work but possible some less effective. If we specified -march=i486 -mtune=pentiumpro for gcc - code will be generated using i486 instruction set (compatible with i486 and above), but optimized for execution on i686 & higher. This makes good performance boost (10-20%) for both geode & powerful modern CPUs, and doesn't hurt compatibility with older hardware. -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] trying to get to BuC4
Hi Folks I am attempting to build openswan for BuC4 and I am worried a bit about the environment preset in MasteriInclude.mk. We define GNU_TUNE as pentiumpro, which according to the gcc doc should produce code for this kind of architecture, whereas GNU_ARCH is set to I486 which definitely is not a pentium pro. The BT_COPT_FLAGS use these definitions and pass it to the compiler in some locations. Now for the kernel and modules this probably gets overwritten, but for other use of BT_COPT_FLAGS this appears not to be correct. export GNU_ARCH:=i486 # set target optimization here export GNU_TUNE:=pentiumpro export BT_COPT_FLAGS=-Os -march=$(GNU_ARCH) -mtune=$(GNU_TUNE) This leads to USERCOMPILE="-g -Os -march=i486 -mtune=pentiumpro" It might make no difference, but does not look right to me. cheers Erich -- Beautiful is writing same markup. Internet Explorer 9 supports standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3. Spend less time writing and rewriting code and more time creating great experiences on the web. Be a part of the beta today http://p.sf.net/sfu/msIE9-sfdev2dev ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
