[Bug 466381] New: CIFS client requests write with old FID after connection reset
https://bugzilla.novell.com/show_bug.cgi?id=466381 Summary: CIFS client requests write with old FID after connection reset Classification: openSUSE Product: openSUSE 11.0 Version: Final Platform: Other OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Samba AssignedTo: samba-maintainers@SuSE.de ReportedBy: agraf@novell.com QAContact: samba-maintainers@SuSE.de Found By: --- I am running a cifs client on 2.6.25, connecting to an openSUSE 11.0 samba server. The client runs in a VM on the same machine using KVM. During massive data transfer, I experienced network hangs in KVM's virtio networking, that apparently triggered a bug in the cifs client of the guest: 96549 17.142084 172.16.2.2 -> 172.16.2.1 SMB Write AndX Request, FID: 0x1590, 57344 bytes at offset 2178293760 [ after that a new split capture file starts, that's why the times and ids look off ] [...] 10371 6.460212 172.16.2.1 -> 172.16.2.2 SMB Write AndX Response, 57344 bytes [...] 10418 6.464444 172.16.2.1 -> 172.16.2.2 SMB Write AndX Response, 57344 bytes [...] 10488 10.883599 172.16.2.1 -> 172.16.2.2 TCP [TCP ACKed lost segment] microsoft-ds > 59590 [ACK] Seq=1600385 Ack=9012623 Win=8062 Len=0 TSV=452206937 TSER=4294955533 10489 27.499716 172.16.2.2 -> 172.16.2.1 TCP [TCP Previous segment lost] 59590 > microsoft-ds [FIN, ACK] Seq=9028083 Ack=1600385 Win=5911 Len=0 TSV=4294957195 TSER=452206947 So between packet 10488 and 10489 the connection was lost for 17 seconds. After that the cifs client correctly tried to reestablish the connection. 10490 27.499765 172.16.2.2 -> 172.16.2.1 TCP 59591 > microsoft-ds [SYN] Seq=0 Win=5840 Len=0 MSS=1460 TSV=4294957195 TSER=0 WS=7 10491 27.499799 172.16.2.1 -> 172.16.2.2 TCP microsoft-ds > 59591 [SYN, ACK] Seq=0 Ack=1 Win=5792 Len=0 MSS=1460 TSV=452211091 TSER=4294957195 WS=7 10492 27.500246 172.16.2.2 -> 172.16.2.1 TCP 59591 > microsoft-ds [ACK] Seq=1 Ack=1 Win=5888 Len=0 TSV=4294957195 TSER=452211091 10493 27.500283 172.16.2.2 -> 172.16.2.1 SMB Negotiate Protocol Request 10494 27.500300 172.16.2.1 -> 172.16.2.2 TCP microsoft-ds > 59591 [ACK] Seq=1 Ack=83 Win=5888 Len=0 TSV=452211091 TSER=4294957195 10495 27.513985 172.16.2.1 -> 172.16.2.2 SMB Negotiate Protocol Response 10496 27.514504 172.16.2.2 -> 172.16.2.1 TCP 59591 > microsoft-ds [ACK] Seq=83 Ack=102 Win=5888 Len=0 TSV=4294957196 TSER=452211095 10497 27.514551 172.16.2.2 -> 172.16.2.1 SMB Session Setup AndX Request, User: \contain2 10498 27.517014 172.16.2.1 -> 172.16.2.2 SMB Session Setup AndX Response But just after that tries to send a write request with the old FID: 10559 27.520280 172.16.2.2 -> 172.16.2.1 SMB Write AndX Request, FID: 0x1590, 57344 bytes at offset 2181328896 10566 27.520482 172.16.2.1 -> 172.16.2.2 SMB Write AndX Response, Error: STATUS_INVALID_HANDLE Unfortunately that old FID is unknown to the new connection, because the file hasn't been opened again. This is expressed in the log of the client: CIFS VFS: server not responding CIFS VFS: No response for cmd 50 mid 52571 CIFS VFS: No response to cmd 47 mid 52572 CIFS VFS: Write2 ret -11, wrote 0 CIFS VFS: No response to cmd 47 mid 52570 CIFS VFS: Write2 ret -11, wrote 0 CIFS VFS: Write2 ret -9, wrote 0 CIFS VFS: Write2 ret -9, wrote 0 I would've expected the client to reopen all files again. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 Alexander Graf <agraf@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|samba-maintainers@SuSE.de |sjayaraman@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 Flavio Castelli <fcastelli@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Blocks| |465645 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 Flavio Castelli <fcastelli@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Blocks| |461983 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 Cornelius Schumacher <cschum@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Blocks|461983 | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 Suresh Jayaraman <sjayaraman@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 User sjayaraman@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=466381#c1 Suresh Jayaraman <sjayaraman@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |agraf@novell.com --- Comment #1 from Suresh Jayaraman <sjayaraman@novell.com> 2009-01-20 05:29:36 MST --- We seem to be closing the files in the upstream code, atleast. I'm yet to check 11.0 code, though. Is it possible for you to verify the behavior with 11.1 cifs client? Please enable cifs debugging info as mentioned in point #4: http://en.opensuse.org/Bugs/Samba#Debugging_cifs_vfs_client and attach dmesg o/p Also, a complete packet capture would reveal further information on why the client still uses the old FID or it considers the fid as valid. Do you think it is possible to reproduce this issue without KVM? If so, I could try reproducing this locally.. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=466381 User agraf@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=466381#c2 Alexander Graf <agraf@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|agraf@novell.com | --- Comment #2 from Alexander Graf <agraf@novell.com> 2009-01-20 06:30:14 MST --- I'm having a hard time reproducing this myself, but so far I haven't been able to reproduce it with 11.1. This bug only occurs sparsely, as the root problem is that the virtual machine (or only its network connection) stalls for some time. I can send you a packet capture if you like. How should I send that over? It's pretty big. Reproducing it without KVM should be fairly easy. Use a network card that doesn't do link detection, do a lot of traffic on the network and unplug network cable for ~15 seconds. I haven't tried if that works, but that's roughly what my virtual machine is experiencing. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com