diff options
Diffstat (limited to 'doc/bugs/SMB__58___git_annex_clone_works__44___get_fails_on_transfer_lock/comment_5_8b5d03f64127aa6b4073ff10796b9771._comment')
| -rw-r--r-- | doc/bugs/SMB__58___git_annex_clone_works__44___get_fails_on_transfer_lock/comment_5_8b5d03f64127aa6b4073ff10796b9771._comment | 59 |
1 files changed, 59 insertions, 0 deletions
diff --git a/doc/bugs/SMB__58___git_annex_clone_works__44___get_fails_on_transfer_lock/comment_5_8b5d03f64127aa6b4073ff10796b9771._comment b/doc/bugs/SMB__58___git_annex_clone_works__44___get_fails_on_transfer_lock/comment_5_8b5d03f64127aa6b4073ff10796b9771._comment new file mode 100644 index 000000000..64b1feb8b --- /dev/null +++ b/doc/bugs/SMB__58___git_annex_clone_works__44___get_fails_on_transfer_lock/comment_5_8b5d03f64127aa6b4073ff10796b9771._comment @@ -0,0 +1,59 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 5""" + date="2017-06-06T17:54:42Z" + content=""" +Here's the relevant parts of the dtruss: + + 7003/0x16c3126: 121483 3371 127 open("/Volumes/music/ga-test/.git/annex/locktmp16807282475249\0", 0x20A06, 0x180) = 18 0 + ... + 7003/0x16c3126: 121792 10 6 stat64("/Volumes/music/ga-test/.git/annex/locktmp16807282475249\0", 0x200004940, 0x33) = 0 0 + 7003/0x16c3126: 122041 2341 235 link("/Volumes/music/ga-test/.git/annex/locktmp16807282475249\0", "/Volumes/music/ga-test/.git/annex/locktmp16807282475249.lnk\0") = -1 Err#45 + 7003/0x16c3126: 122115 885 49 unlink("/Volumes/music/ga-test/.git/annex/locktmp16807282475249.lnk\0", 0x200004D80, 0x33) = -1 Err#2 + 7003/0x16c3126: 122126 5 1 sigreturn(0x7FFF56C9E390, 0x1E, 0x33) = 0 Err#-2 + 7003/0x16c3126: 122247 1050 90 open("/Volumes/music/ga-test/.git/annex/pidlock\0", 0xA01, 0x124) = 18 0 + 7003/0x16c3126: 122256 4 1 fcntl(0x12, 0x3, 0x0) = 1 0 + 7003/0x16c3126: 122260 5 2 fstat64(0x12, 0x20001A120, 0x0) = 0 0 + 7003/0x16c3126: 122273 4 1 ioctl(0x12, 0x4004667A, 0x7FFF56C9E3FC) = -1 Err#25 + 7003/0x16c3126: 122274 2 0 ioctl(0x12, 0x40487413, 0x7FFF56C9E400) = -1 Err#25 + 7003/0x16c3126: 122314 677 32 open("/Volumes/music/ga-test/.git/annex/locktmp16807282475249\0", 0x20004, 0x1B6) = 19 0 + 7003/0x16c3126: 122319 5 2 fstat64(0x13, 0x20001A310, 0x1B6) = 0 0 + 7003/0x16c3126: 122326 2 0 ioctl(0x13, 0x4004667A, 0x7FFF56C9E3FC) = -1 Err#25 + 7003/0x16c3126: 122327 2 0 ioctl(0x13, 0x40487413, 0x7FFF56C9E400) = -1 Err#25 + 7003/0x16c3126: 122381 1455 46 read(0x13, "PidLock {lockingPid = 7003, lockingHost = \"ashram\"}\0", 0x1FA0) = 51 0 + 7003/0x16c3126: 122392 5 2 read(0x13, "\0", 0x1FA0) = 0 0 + 7003/0x16c3126: 122427 1339 27 close(0x13) = 0 0 + 7003/0x16c3126: 122441 10 4 select(0x13, 0x7FFF56C9E340, 0x7FFF56C9E3C0, 0x0, 0x7FFF56C9E330) = 1 0 + 7003/0x16c3126: 122469 661 23 write(0x12, "PidLock {lockingPid = 7003, lockingHost = \"ashram\"}\0", 0x33) = 51 0 + 7003/0x16c3126: 122577 6028 99 close(0x12) = 0 0 + +Avove shows hard linking failed, and it fell back to using open to create +the pidlock file. That seemed to finish successfully. + + 7003/0x16c3126: 122583 4 0 sigreturn(0x7FFF56C9E3B0, 0x1E, 0x33) = 0 Err#-2 + +Unsure what this is? + + 7003/0x16c3126: 122697 2126 101 unlink("/Volumes/music/ga-test/.git/annex/locktmp16807282475249\0", 0x1E, 0x33) = 0 0 + +Here linkToLock must has succeeded, because it deletes the tmp file that was passed to that function. +So, it looks like tryLock succeeded. + + 7003/0x16c3126: 122746 1423 28 stat64(".git/annex/pidlock\0", 0x20001A510, 0x33) = 0 0 + 7003/0x16c3126: 122775 10 5 select(0x2, 0x7FFF56C9E340, 0x7FFF56C9E3C0, 0x0, 0x7FFF56C9E330) = 1 0 + 7003/0x16c3126: 122786 10 8 write(0x1, "(transfer already in progress, or unable to take transfer lock) \0", 0x40) = 64 0 + +I think this is a call to checkSaneLock, which calls getFileStatus, +and since the pid lock was created earlier, that succeeds. +runTransfer does call checkSaneLock, and if it fails +we get the "transfer already in progress, or unable to take transfer lock" +message. + +So, why did checkSaneLock fail? + +Well, the device and inode of the lock file +are compared with the device and inode of the temp file +that was passed to linkToLock. However, since the hard link didn't +happen, the device and inode will be different! +And that's the bug. +"""]] |