Message boards :
Number crunching :
sixtracktest v450.09 (sse3) windows x86 : CreateProcess() failed
Message board moderation
Author | Message |
---|---|
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
here and on all others I received on this box When I start the program outside of BOINC, it tells me that it is not a "zulässige Win32-Anwendung" (valid application for Win32) The binary has an unusual large number of 0x00 bytes at the end. I downloaded a sixtrack_win32_4463_sse3.exe for comparison, that one does start properly and creates stderr with "15:49:08 (7132): Can't open init data file - running in standalone mode" ... edit : this is a basic test you should do on a machine without the development environment before you upload the new program An x64 wingman has the same problem : couldn't start CreateProcess() failed - : -148 |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
here and on all others I received on this box Right, this is really the problem. Many thanks for your help. More tests coming. Eric. [list=][/list] |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
The "file" command checks for some typical details for a file type, e.g. "MZ" as the first two letters, it cannot check for missing shared libraries, illegal use of privileged instructions and stuff like that. Maybe an interesting detail that I did not notice in my first offline test : After confirming the message box ("not a valid application"), it sends the text "Zugriff verweigert" / "access denied" to stderr (console, not disk file). The reason why an offline test should run on a machine without IDE is, that shell environment, paths, debug environment, special libraries ... on a development machine might be untypical and often cannot be compared to a "plain" windows installation. |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
Suspicious (unresolved) dependencies : IESHIMS.DLL and WER.DLL Try static linking. After downloading those two DLLs, it still isn't happy. Missing now : API-MS-WIN-DOWNLEVEL-ADVAPI32-L1-1-0.DLL (referenced from IESHIMS.DLL) API-MS-WIN-DOWNLEVEL-OLE32-L1-1-0.DLL (referenced from IESHIMS.DLL) API-MS-WIN-DOWNLEVEL-SHLWAPI-L1-1-0.DLL (referenced from IESHIMS.DLL) NCRYPT.DLL (referenced from WER.DLL) Actually it seems that you have (dyn-) linked half the IE8 (or higher) API to the program, I wonder if that could not be excluded (project properties or so). Maybe some specific module of the .net framework requires that stuff. p.s.: Somewhere in a newsgroup I saw someone mention the same problem with Visual Fortran and he solved it by recompiling his stuff in "release mode" - sounds good as this usually produces files with much less overhead and without referring to non-distributable files that are part of the IDE and/or debug environment |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Agh...all this is very helpful indeed. I will be looking at the "build" procedure (new on a virtual machine) this afternoon with my colleagues. In the past I always built static and I thought we still were. many many thanks for the detailed diagnostics. Eric. |
Send message Joined: 27 Oct 07 Posts: 186 Credit: 3,297,640 RAC: 0 |
A similar set of missing dependencies turned up in an Einstein application recently. Bernd's response also implicated static linking: The problem seems to be that the libstdc++-6.dll is not linked statically into the App. The version of the MinGW compiler that I used for the first time apparently requires an addional option for this (-static-libstdc++). |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Right; it looks like a problem on Windows XP. We are now forced to build on Windows 7 and the executables do NOT work on XP and on "many/most" server versions of Windows. Our new colleague Xavier is looking at building with backwards compatibility. This is tricky because when my Windows PC broke I was forced to Windows 7 and we have no XP system at CERN for testing. Ah well.....I have no intention of forcing upgrades to Windows 7 0r Vista which seem to be OK as well. In the worst case I will try our next executables on boinctest again and wait for feedback. This may seem an obvious issue but not so easy to check in the database (at least not for me as I am not a MYSQL expert). After that we shall see if there are other issues with the API etc. (Maybe I shall post one day to the cafe on how we are supposed to upgrade to Windows 7, perhaps being forced to buy new hardware, and PAY more money for software and hardware....) Xavier is also counting hosts by OS Windows version. Hope to have some good news very soon and get back to the physics. Thanks for your patience. Eric. |
Send message Joined: 25 Jan 08 Posts: 1 Credit: 85,738 RAC: 0 |
Right; it looks like a problem on Windows XP. We are now forced to win7 has "xp-mode". you simply need to install it. |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
The page title sounds somehow promising : https://software.intel.com/en-us/articles/linking-applications-using-visual-studio-2012-to-run-on-windows-xp Here's some more on setting the targeted environment with screenshots : http://blogs.msdn.com/b/vcblog/archive/2012/10/08/10357555.aspx p.s.: If you need someone to run the basic standalone (offline) test, just upload the binary somewhere and post the link here, I'm 100% sure a lot of people will jump in and see if it runs on their systems (including me) |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
sixtracktest_win32_4509_pni.exe with size 3,899,392 (April 25, 08:30) does not work on my XP x86, the dependency walker reports the same unresolved DLLs |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Thanks; I think we have a fix but failed to install it properly. It also had same ID :-( NOT good. Xavier has built 4511 and we shall install and test again today. Eric. |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Thanks; I think we have a fix but failed to install it properly. It also had same ID :-( NOT good. Xavier has built 4511 and we shall install and test again today. Eric. |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
Standalone test : failed on XP :-/ All three sse3/pni binaries are identical btw., win64 and both win32 are the same file. |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Thanks; that is OK. We have three 32-bit binaries, ia32 (generic), sse2 and sse3. pni is a copy of sse3. The 64-bit versions are just copies. (Hope to have an axp when current issues resolved.) Can run now on some Window 2008 but still fighting XP. More tests coming. Eric. |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
Thumbs up for sixtracktest_win32_4512_sse3.exe (basic standalone test) : 14:37:09 (7720): Can't open init data file - running in standalone mode 14:37:09 (7720): called boinc_finish It created all dummy result files (including the final ZIP) and shows no weird message boxes anymore :-) The dependency walker still finds the same missing DLLs but this binary runs anyway. |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
Same for sixtrack_win32_4513_sse3.exe, works fine on XP |
Send message Joined: 17 Jul 05 Posts: 102 Credit: 542,016 RAC: 0 |
Same for sixtrack_win32_4513_sse3.exe, works fine on XP Btw., 451.03 is listed as "sixtracktest" but filename is "sixtrack". This might be purpose of course but Justin Case told me to mention it |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
Apologies for Version 4515 451.5. Totally wrong build. Mea Culpa. Eric. (Version 4513 451.3 looks best so far and runs on XP...I'll keep trying) |
Send message Joined: 22 Dec 08 Posts: 6 Credit: 1,476,859 RAC: 3 |
the boinc 7.7.42 x64 produce a lot of errors for the ne wus on my win7 system |
Send message Joined: 12 Jul 11 Posts: 857 Credit: 1,619,050 RAC: 0 |
thanks; I'll look at this which seems to be an issue now we have apparently fixed WXP. Eric. |
©2024 CERN