Bug 2257 - Kernel bug and Send error in Close = -9 errors writing files from Java
Kernel bug and Send error in Close = -9 errors writing files from Java
Status: RESOLVED FIXED
Product: CifsVFS
Classification: Unclassified
Component: kernel fs
2.6
x86 Linux
: P3 critical
: ---
Assigned To: Steve French
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2005-01-19 09:02 UTC by Mark
Modified: 2005-03-07 21:19 UTC (History)
0 users

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Mark 2005-01-19 09:02:35 UTC
'No response buffer' messages followed by kernel oops most times a file write is
attempted by Java.

I'm trying to save my eclipse workspace to a cifs network share (server=samba
3.0.6). There is a large delay (20+ seconds) half way through saving as errors
are generated in syslog, followed by kernel bug, then the workspace becomes
corrupted and unusable (perhaps simply due to failure to write all files).
Single file writes from text editors seem to work so far, and simple directory
copies with 'cp', but not saving a workspace from Java.

I also encountered an xml file on the server, part of the workspace that would
have been saved, that had been truncated.

Firewalls at both ends allow all traffic between IPs.

Environment: 64-bit kernel 2.6.8.1-19mdk on Mandrake 10.1 (x86_64)
Java version: Sun j2sdk 1_4_2_06-linux-i586 

Syslog:

<attempt to quit eclipse / save workspace starts here>
Jan 19 15:21:28 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:22:10 localhost last message repeated 2 times
Jan 19 15:37:14 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:37:20 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:37:20 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:40:10 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:40:16 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:40:16 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:40:37 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:40:43 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:41:07 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:41:13 localhost kernel:  CIFS VFS: Send error in Close = -9
Jan 19 15:41:28 localhost kernel:  CIFS VFS: No response buffer
Jan 19 15:41:34 localhost kernel:  CIFS VFS: Send error in Close = -9

Message from syslogd@localhost at Wed Jan 19 15:41:39 2005 ...
localhost kernel: invalid operand: 0000 [3]
Jan 19 15:41:39 localhost kernel: ----------- [cut here ] --------- [please bite
here ] ---------
Jan 19 15:41:39 localhost kernel: Kernel BUG at locks:1751
Jan 19 15:41:39 localhost kernel: invalid operand: 0000 [3]
Jan 19 15:41:39 localhost kernel: CPU 0
Jan 19 15:41:39 localhost kernel: Modules linked in: nls_iso8859-15 nls_cp850
vfat fat usb-storage md5 ipv6 nls_iso8859-1 cifs rfcomm l2                   cap
bluetooth snd-seq-oss snd-seq-midi-event snd-seq snd-pcm-oss snd-mixer-oss
snd-intel8x0 snd-ac97-codec snd-pcm snd-timer snd-page-al                   loc
gameport snd-mpu401-uart snd-rawmidi snd-seq-device snd soundcore ipt_TOS
ipt_REJECT ipt_pkttype ipt_LOG ipt_state ipt_multiport ipt                  
_conntrack iptable_mangle ip_nat_irc ip_nat_tftp ip_nat_ftp iptable_nat
ip_conntrack_irc ip_conntrack_tftp ip_conntrack_ftp ip_conntrack               
    iptable_filter ip_tables af_packet floppy thermal tc1100-wmi processor fan
button battery ac r8169 forcedeth ide-cd cdrom loop supermou                  
nt ehci-hcd ohci-hcd usbcore ext3 jbd raid0 sd_mod sata_nv libata scsi_mod
Jan 19 15:41:39 localhost kernel: Pid: 16556, comm: java Not tainted 2.6.8.1-19mdk
Jan 19 15:41:39 localhost kernel: RIP: 0010:[locks_remove_flock+87/128]
<ffffffff80180ba7>{locks_remove_flock+87}
Jan 19 15:41:39 localhost kernel: RIP: 0010:[<ffffffff80180ba7>]
<ffffffff80180ba7>{locks_remove_flock+87}
Jan 19 15:41:39 localhost kernel: RSP: 0018:000001002103de78  EFLAGS: 00210246
Jan 19 15:41:39 localhost kernel: RAX: 000001003f24a948 RBX: 000001000dfb73f8
RCX: 0000000000000000
Jan 19 15:41:39 localhost kernel: RDX: 0000000000000000 RSI: 00000100247fafc0
RDI: 000001003b34c1c0
Jan 19 15:41:39 localhost kernel: RBP: 000001003b34c1c0 R08: 0000000000000000
R09: 00000000c0000001
Jan 19 15:41:39 localhost kernel: R10: 000001000edd19c8 R11: ffffffff801524d0
R12: 000001000dfb7300
Jan 19 15:41:39 localhost kernel: R13: 000001003b661e58 R14: 0000000000000000
R15: 0000000000000000
Jan 19 15:41:39 localhost kernel: FS:  0000002a958964c0(0000)
GS:ffffffff803fe180(0000) knlGS:0000000000000000
Jan 19 15:41:39 localhost kernel: CS:  0010 DS: 002b ES: 002b CR0: 000000008005003b
Jan 19 15:41:39 localhost kernel: CR2: 00000000ff1ff594 CR3: 0000000000101000
CR4: 00000000000006e0
Jan 19 15:41:39 localhost kernel: Process java (pid: 16556, threadinfo
000001002103c000, task 0000010006acac40)
Jan 19 15:41:39 localhost kernel: Stack: 0000000000000000 000001003b34c1c0
000001003f23f300 ffffffff8016c2f8
Jan 19 15:41:39 localhost kernel:        0000000000200212 000001003b34c1c0
0000000000000000 00000100085d23c0
Jan 19 15:41:39 localhost kernel:        0000000000000001 ffffffff8016acbe
Jan 19 15:41:39 localhost kernel: Call Trace:<ffffffff8016c2f8>{__fput+72}
<ffffffff8016acbe>{filp_close+126}
Jan 19 15:41:39 localhost kernel:       
<ffffffff801345b3>{put_files_struct+115} <ffffffff80134c79>{do_exit+505}
Jan 19 15:41:39 localhost kernel:       
<ffffffff80110010>{sys_rt_sigsuspend+176} <ffffffff8013563b>{do_group_exit+171}
Jan 19 15:41:39 localhost kernel:        <ffffffff801217a1>{ia32_sysret+0}
Jan 19 15:41:39 localhost kernel:
Jan 19 15:41:39 localhost kernel: Code: 0f 0b 69 da 2e 80 ff ff ff ff d7 06 48
89 c3 48 8b 03 48 85
Jan 19 15:41:39 localhost kernel: RIP <ffffffff80180ba7>{locks_remove_flock+87}
RSP <000001002103de78>
Comment 1 Steve French 2005-01-19 09:37:22 UTC
This looks like the problem with socket peek handling in fs/cifs/connect.c that 
was fixed in 2.6.10

I have a patch for 2.6.9, not sure it will build for 2.6.8 but should only 
require minor changes (if any) if you are comfortable building the kernel.
Comment 2 Steve French 2005-03-07 21:19:39 UTC
fixed in 2.6.10 or later kernels (patches available for some earlier kernel 
versions on the project page).