Hello, Linux execve(2) appears to have a bug: The execve(2) of a FIFO file with execute permission results in the kernel opening the named pipe. Since there are no writers and this is a blocking call, the syscall goes to sleep waiting for writer(s) on the file. Stack trace: #0 [ffff880b978bfb58] schedule at ffffffff814dad09 #1 [ffff880b978bfc20] pipe_wait at ffffffff8117c1cb #2 [ffff880b978bfc70] fifo_open at ffffffff8118885c #3 [ffff880b978bfcb0] __dentry_open at ffffffff8116fb7a #4 [ffff880b978bfd10] nameidata_to_filp at ffffffff8116fee4 #5 [ffff880b978bfd30] do_filp_open at ffffffff81182b40 #6 [ffff880b978bfe80] open_exec at ffffffff8117b033 #7 [ffff880b978bfeb0] do_execve at ffffffff8117b1ef #8 [ffff880b978bff20] sys_execve at ffffffff810095ca #9 [ffff880b978bff50] stub_execve at ffffffff8100b5ca The issue appears to be that in open_exec(), before we've had a chance to verify that it's a regular file, the do_filp_open() ends up in pipe_wait(). Other OSes return EACCES as expected in the same situation. Respectfully, Kuthonuzo -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
