Build of gcc 4.4.0 on Solaris 10 Sparc ok, most tests failed.
Hi GCC team, I built GCC 4.4.0 on a SPARC system running Solaris 10 The build went through without any error, but most of the tests failed in "make check". unexpected failures = 6472 and passed = 52. Output from gcc -v and fragment from make check are below. What am I doing wrong? Is this build usable? Thanks for your help. Best regards Amitava Dutta SRCDIR=/big3/src/gcc/src/gcc-4.4.0 $ $SRCDIR/config.guess sparc-sun-solaris2.10 $ $SRCDIR/configure \ --srcdir=$SRCDIR \ --prefix=/usr/local/gcc-4.4.0 \ --without-gnu-as --with-as=/usr/ccs/bin/as \ --without-gnu-ld --with-ld=/usr/ccs/bin/ld \ --enable-languages=c,c++,fortran $ /usr/local/gcc-4.4.0/bin/gcc -v Using built-in specs. Target: sparc-sun-solaris2.10 Configured with: /big3/src/gcc/src/gcc-4.4.0/configure --srcdir=/big3/src/gcc/src/gcc-4.4.0 --prefix=/usr/local/gcc-4.4.0 --without-gnu-as --with-as=/usr/ccs/bin/as --without-gnu-ld --with-ld=/usr/ccs/bin/ld --enable-languages=c,c++,fortran Thread model: posix gcc version 4.4.0 (GCC) // BEGIN last fragment from make check / FAIL: .. FAIL: gcc.c-torture/compile/zero-strct-5.c -O3 -g (test for excess errors) FAIL: gcc.c-torture/compile/zero-strct-5.c -Os (test for excess errors) === gcc Summary === # of expected passes52 # of unexpected failures6472 # of expected failures 16 # of unresolved testcases 24 # of unsupported tests 106 /big3/src/gcc/build/gcc440-s10-sparc/gcc/xgcc version 4.4.0 (GCC) make[3]: [check-parallel-gcc_3] Error 1 (ignored) make[3]: Leaving directory `/big3/src/gcc/build/gcc440-s10-sparc/gcc' nawk: /tmp/dg-combine-results-25920-2MaGOY/list38 makes too many open files input record number 15646 source line number 40 usage: tail [+/-[n][lbc][f]] [file] tail [+/-[n][l][r|f]] [file] grep: illegal option -- q Usage: grep -hblcnsviw pattern file . . . usage: tail [+/-[n][lbc][f]] [file] tail [+/-[n][l][r|f]] [file] nawk: /tmp/dg-combine-results-25966-DMaqUY/list26 makes too many open files input record number 54342 source line number 44 usage: tail [+/-[n][lbc][f]] [file] tail [+/-[n][l][r|f]] [file] grep: illegal option -- q Usage: grep -hblcnsviw pattern file . . . usage: tail [+/-[n][lbc][f]] [file] tail [+/-[n][l][r|f]] [file] make[2]: Leaving directory `/big3/src/gcc/build/gcc440-s10-sparc/gcc' make[1]: Target `check-host' not remade because of errors. make[1]: Leaving directory `/big3/src/gcc/build/gcc440-s10-sparc' make: *** [do-check] Error 2 make: Target `check' not remade because of errors. // END last fragment from make check /
Re: Build of gcc 4.4.0 on Solaris 10 Sparc ok, most tests failed.
Here's from the output of running "make -k check" (without "-j"), notice that there's no # for libstdc++. === gcc Summary === # of expected passes18369 # of unexpected failures22530 # of expected failures 226 # of unresolved testcases 7716 # of unsupported tests 1016 === g++ Summary === # of expected passes9839 # of unexpected failures8301 # of unexpected successes 1 # of expected failures 143 # of unresolved testcases 122 # of unsupported tests 291 === gfortran Summary === # of expected passes15952 # of unexpected failures9498 # of expected failures 9 # of unsupported tests 1503 === libstdc++ Summary === === libgomp Summary === # of expected passes1057 # of unexpected failures1061 # of unsupported tests 137 What am I doing wrong? Best regards Amitava Dutta --- On Thu, 5/14/09, Eric Botcazou wrote: > From: Eric Botcazou > Subject: Re: Build of gcc 4.4.0 on Solaris 10 Sparc ok, most tests failed. > To: ad_...@yahoo.com > Cc: gcc@gcc.gnu.org > Date: Thursday, May 14, 2009, 12:14 PM > > The build went through without > any error, > > but most of the tests failed in "make check". > > unexpected failures = 6472 and passed = 52. > > Try with "make -k check" and no -j, parallel testing is > broken on Solaris. > > -- > Eric Botcazou >
gcc 4.4.0 on Solaris 10 Sparc, some tests failed.
Ran the "make -k check" without the -j option, after creating a symlink to /usr/local/bin/stty (noticed many errors about that) Is this as good a build as I can expect? Here are the results (output from config.guess and "gcc -v" are below) : === gcc Summary === # of expected passes50284 # of unexpected failures19 # of expected failures 234 # of unsupported tests 656 === g++ Summary === # of expected passes19007 # of unexpected failures1 # of unexpected successes 1 # of expected failures 144 # of unsupported tests 185 === gfortran Summary === # of expected passes29193 # of expected failures 12 # of unsupported tests 135 === libstdc++ Summary === # of expected passes5733 # of unexpected failures1 # of unexpected successes 2 # of expected failures 80 # of unsupported tests 393 === libgomp Summary === # of expected passes2228 # of unexpected failures80 # of unsupported tests 9 / gcc -v for the new compiler / $ /usr/local/gcc-4.4.0/bin/gcc -v Using built-in specs. Target: sparc-sun-solaris2.10 Configured with: /big3/src/gcc/src/gcc-4.4.0/configure --srcdir=/big3/src/gcc/src/gcc-4.4.0 --prefix=/usr/local/gcc-4.4.0 --without-gnu-as --with-as=/usr/ccs/bin/as --without-gnu-ld --with-ld=/usr/ccs/bin/ld --enable-languages=c,c++,fortran Thread model: posix gcc version 4.4.0 (GCC) // config.guess from the source directory $ /usr/local/src/gcc/src/gcc-4.4.0/config.guess sparc-sun-solaris2.10 Best regards Amitava --- On Fri, 5/15/09, Eric Botcazou wrote: > From: Eric Botcazou > Subject: Re: Build of gcc 4.4.0 on Solaris 10 Sparc ok, most tests failed. > To: "Kaveh R. GHAZI" > Cc: ad_...@yahoo.com, gcc@gcc.gnu.org > Date: Friday, May 15, 2009, 5:20 PM > > To clarify, is it "make -l #" > that fails or "make -j #" on Solaris? > > "make -j" > > > Parallel testing with -j# used to work fine, but > admitedly its been a long > > while since I lost my solaris box... (I don't > know if the load avg based > > mechanism for -l ever worked.) Is there a PR > number? > > That's already fixed. > > -- > Eric Botcazou >
Re: gcc 4.4.0 on Solaris 10 Sparc, some tests failed.
Thanks. Now I have a few questions. You mentioned that the "make -k # check" is broken on Solaris. I built the compiler using the -k option, is that a concern? or is it just the "check" target is broken? IOW, should I build the compiler without the "-k" option? At this point, should I do the next build using the recently built gcc 4.4.0? btw, is there a quick tutorial on running the tests against an arbitrary compiler, that's installed on a given machine ? Thanks for your help. Best regards Amitava Dutta --- On Mon, 5/18/09, Eric Botcazou wrote: From: Eric Botcazou Subject: Re: gcc 4.4.0 on Solaris 10 Sparc, some tests failed. To: ad_...@yahoo.com Cc: gcc@gcc.gnu.org, "Kaveh R. GHAZI" Date: Monday, May 18, 2009, 1:37 AM > Ran the "make -k check" without the -j option, > after creating a symlink to /usr/local/bin/stty > (noticed many errors about that) > > Is this as good a build as I can expect? Probably, although the number of libgomp failures is high. You might want to try with the GNU assembler instead of the Sun assembler, the latter is barely maintained by Sun. -- Eric Botcazou
Re: gcc 4.4.0 on Solaris 10 Sparc, some tests failed.
Rebuilt using GNU as and ld and the number of libgomp failure came down to 18 from 80: === gcc Summary === # of expected passes50313 # of unexpected failures14 # of unexpected successes 3 # of expected failures 231 # of unsupported tests 651 === g++ Summary === # of expected passes19043 # of unexpected failures1 # of unexpected successes 2 # of expected failures 143 # of unsupported tests 173 === gfortran Summary === # of expected passes29193 # of expected failures 12 # of unsupported tests 135 === libstdc++ Summary === # of expected passes5750 # of unexpected successes 2 # of expected failures 80 # of unsupported tests 385 === libgomp Summary === # of expected passes2370 # of unexpected failures18 # of unsupported tests 9 I'll submit the full report soon. Thanks for your help. Best regards Amitava Dutta --- On Mon, 5/18/09, Eric Botcazou wrote: > From: Eric Botcazou > Subject: Re: gcc 4.4.0 on Solaris 10 Sparc, some tests failed. > To: ad_...@yahoo.com > Cc: gcc@gcc.gnu.org, "Kaveh R. GHAZI" > Date: Monday, May 18, 2009, 1:37 AM > > Ran the "make -k check" without > the -j option, > > after creating a symlink to /usr/local/bin/stty > > (noticed many errors about that) > > > > Is this as good a build as I can expect? > > Probably, although the number of libgomp failures is > high. You might want to > try with the GNU assembler instead of the Sun assembler, > the latter is barely > maintained by Sun. > > -- > Eric Botcazou >
