othermark wrote: > > I have two kernels, one build from 3/16 and one from this morning 3/18 > built with the exact same kernel config. > > If I boot the 3/16 kernel, I can open /dev/cuad0, if I boot > the 3/18 kernel... > > $ ls -l /dev/cuad0 > crw------- 1 root wheel 13, 64 Mar 18 09:19 /dev/cuad0 > > $ whoami > root > > $ ktrace minicom > minicom: cannot open /dev/cuad0: Permission denied > > $ kdump -f ktrace.out > > [...] > > 821 minicom CALL sigprocmask(0x1,0,0x807097c) > 821 minicom RET sigprocmask 0 > 821 minicom CALL sigaction(0xe,0xbfbfe6c0,0xbfbfe6a0) > 821 minicom RET sigaction 0 > 821 minicom CALL setitimer(0,0xbfbfe6d0,0xbfbfe6c0) > 821 minicom RET setitimer 0 > 821 minicom CALL open(0x806f0c0,0x8006,0x18) > 821 minicom NAMI "/dev/cuad0" > 821 minicom RET open -1 errno 13 Permission denied > 821 minicom CALL setitimer(0,0xbfbfe6d0,0xbfbfe6c0) > 821 minicom RET setitimer 0 > 821 minicom CALL sigaction(0xe,0xbfbfe6c0,0xbfbfe6a0) > 821 minicom RET sigaction 0 > 821 minicom CALL getuid > 821 minicom RET getuid 0 > 821 minicom CALL geteuid > 821 minicom RET geteuid 66/0x42 > 821 minicom CALL write(0x2,0xbfbfdfa0,0x33) > 821 minicom GIO fd 2 wrote 51 bytes > "minicom: cannot open /dev/cuad0: Permission denied > > [...] > > > 0x18 appears to be O_SHLOCK | O_APPEND. > > A simple .c program with O_SHLOCK | O_APPEND on cuad0 returns > "operation not permitted" Apparently there was some ABI change since the 16th in this area, since recompiling minicom fixes the problem. -- othermark atkin901 at nospam dot yahoo dot com (!wired)?(coffee++):(wired);Received on Mon Mar 21 2005 - 15:09:31 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:30 UTC