Re: setupapi:parser - suspicious test results
Rico Schüller wrote: Hi, concerning to http://test.winehq.org/data/tests/setupapi:parser.html there are some suspicious results for the test from 6-10. January. On my system (F10-x86_64) the setupapi:parser test (setupapi_test.exe) from the test site has 15 additional failures (17 failures) on wine. I checked it with my own build (setupapi_crosstest.exe) which has only 2 failures). Both tests executables succeeded on my windows machine without any failures. Attached is the log from both runs on wine. I have no glue where the difference come from. I guess it has something todo with this commit http://source.winehq.org/git/wine.git/?a=blobdiff;f=dlls/setupapi/tests/parser.c;h=ddfd0f29a813643ec578d7ca50ae49a73ff92703;hp=8869ca11d4870f2beed159afef7c629f9b327184;hb=2e20bd496095ae8b146b4965d4b73081a7ebc818;hpb=b45cfc123eb6c341376edb9ef1d8f63be78255a6 , which introduced the failing tests. But I couldn't run a regression test because my own compiled version didn't trigger the failures. Any suggestions? Cheers Rico Hi Rico, I guess the results on the website came from running the Windows winetest.exe? Could you try the following: - download winetest.exe - remove .wine (or move it away, use a different prefix, whatever) - run 'wine winetest.exe setupapi:parser' - check the res.tmp file in .wine/drive_c/windows/temp On my F10 box this also generates 15 failures, where 'make parser.ok' only shows 2 errors. -- Cheers, Paul.
Re: setupapi:parser - suspicious test results
Paul Vriens schrieb: Hi Rico, I guess the results on the website came from running the Windows winetest.exe? Could you try the following: - download winetest.exe - remove .wine (or move it away, use a different prefix, whatever) - run 'wine winetest.exe setupapi:parser' - check the res.tmp file in .wine/drive_c/windows/temp On my F10 box this also generates 15 failures, where 'make parser.ok' only shows 2 errors. Hi Paul, Yes, this is exactly what I've done. I've also tried 'make parser.ok' and build my own cross test (mingw) which I've run in wine and which both show only 2 errors. The setupapi_test.exe is the extracted version from winetest.exe. Cheers Rico
Re: setupapi:parser - suspicious test results
Rico Schüller schrieb: Paul Vriens schrieb: Hi Rico, I guess the results on the website came from running the Windows winetest.exe? Could you try the following: - download winetest.exe - remove .wine (or move it away, use a different prefix, whatever) - run 'wine winetest.exe setupapi:parser' - check the res.tmp file in .wine/drive_c/windows/temp On my F10 box this also generates 15 failures, where 'make parser.ok' only shows 2 errors. Hi Paul, Yes, this is exactly what I've done. I've also tried 'make parser.ok' and build my own cross test (mingw) which I've run in wine and which both show only 2 errors. The setupapi_test.exe is the extracted version from winetest.exe. Cheers Rico Hi, just want to say problem found, the if (*need) *need = len; in SetupEnumInfSectionsA/W is wrong. I'll send a patch. Cheers Rico
Re: setupapi:parser - suspicious test results
Am Sonntag, den 11.01.2009, 16:02 +0100 schrieb Rico Schüller: just want to say problem found, the if (*need) *need = len; in SetupEnumInfSectionsA/W is wrong. I'll send a patch. This looks like a typical problem valgrind is designed for. It should report an uninitialized memory read at that point. Just as a hint how to attack such beasts, if you encounter things like that again. Regards, Michael Karcher