On 08/22/2014 12:18 AM, Mike Stump wrote: > On Aug 18, 2014, at 3:06 AM, Chen Gang <gang.chen.5...@gmail.com> wrote: >> - one for original latest gcc source code (master for 20140816). >> >> - one for my modification based on the original latest gcc source code. >> >> - they passed building, but for make check, they reported same issues: > > So, I see no evidence that you ran contrib/compare_tests yet. Did you? If > so, what was the output? > >> For only sending my patch, may I bypass them, > > So, the output of the make check is uninteresting, only the regressions > identified by compare_tests are interesting. If none, then you’re set wrt > the current patch. >
OK, thanks. At present, under Fedora 20 x86_64, it passed normal testsuite: - "../gcc/configure && make && make check" succeed (echo $? == 0). - "contrib/compare_test dir_orig dir_new" succeed (echo $? == 0). - The related output of "contrib/compare_test" are: # Comparing directories ## Dir1=/upstream/build-gcc: 11 sum files ## Dir2=/upstream/build-gcc-new: 11 sum files # Comparing 11 common sum files ## /bin/sh ./compare_tests /tmp/gxx-sum1.7678 /tmp/gxx-sum2.7678 # No differences found in 11 common sum files >> and then I shall try to analyze them one by one in another time, next? > > You can, if you want. For example, there is a PR on the darwin issue that > has state in it. Roughly binds local is slightly wrong and need fixing. > OK, thank. I shall still try the Darwin, next, in another time, although it is not related with my new patches for sending. At last: Thank all of you very much for your much help, and I shall send related patch next. Thanks. -- Chen Gang Open share and attitude like air water and life which God blessed