Okay, it is the pathing problem, I thought I had backported that fix to the 6.6a branch though.
I'll verify that fix, if it isn't there I'll backport the fix and push out a new Linux client. ----- Rom From: Francisco Sanz García [mailto:[email protected]] Sent: Tuesday, October 20, 2009 11:50 AM To: Rom Walton Cc: Bernd Machenschalk; [email protected] Subject: Re: [boinc_dev] Simple cuda aplication. In the working case: <stderr_out> <![CDATA[ <stderr_txt> argv[0]='../../projects/test.test/hello_cuda2' Adios 17:44:12 (10575): called boinc_finish </stderr_txt> ]]> And in the not working case: surprise: (now it works with 6.10.13) <stderr_out> <![CDATA[ <stderr_txt> argv[0]='../../projects/test.test/hello_cuda'Adios17:49:18 (10928): called boinc_finish </stderr_txt> ]]></stderr_out> But with 6.6.40: (not working case) <stderr_out> <![CDATA[ <message>process exited with code 193 (0xc1, -63)</message><stderr_txt> argv[0]='hello_cuda' SIGSEGV: segmentation violationStack trace (15 frames):hello_cuda(boinc_catch_signal+0x64)[0x805ccb4][0xb806d400]/usr/lib/libcuda.so.1[0xb7be699b]/usr/lib/libcuda.so.1[0xb7bed791]/usr/lib/libcuda.so.1[0xb7bb77ae]/usr/lib/libcuda.so.1[0xb7b5b1f3]/usr/lib/libcuda.so.1[0xb7b6f794]/usr/lib/libcuda.so.1[0xb7b51675]/usr/lib/libcuda.so.1[0xb7b4a992]/usr/lib/libcuda.so.1(cuCtxCreate+0x4f)[0xb7bad5bf]hello_cuda[0x80588ca] hello_cuda[0x80588f5]hello_cuda[0x804acdb]/lib/i686/cmov/libc.so.6(__libc_start_main+0xe5)[0xb7953775]hello_cuda[0x804aa11]Exiting... </stderr_txt>]]> ¿Solved?. One beer for you. 2009/10/20 Rom Walton <[email protected]> Okay, I'm really confused... Could you add: fprintf(stderr, "argv[0] = '%s'\n", argv[0]); To the main function in hello_cuda and run both the working and non-working case again? ----- Rom _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
