1. 04 11月, 2008 1 次提交
  2. 31 10月, 2008 1 次提交
  3. 29 10月, 2008 1 次提交
    • S
      [CIFS] Reduce number of socket retries in large write path · edf1ae40
      Steve French 提交于
      CIFS in some heavy stress conditions cifs could get EAGAIN
      repeatedly in smb_send2 which led to repeated retries and eventually
      failure of large writes which could lead to data corruption.
      
      There are three changes that were suggested by various network
      developers:
      
      1) convert cifs from non-blocking to blocking tcp sendmsg
      (we left in the retry on failure)
      2) change cifs to not set sendbuf and rcvbuf size for the socket
      (let tcp autotune the buffer sizes since that works much better
      in the TCP stack now)
      3) if we have a partial frame sent in smb_send2, mark the tcp
      session as invalid (close the socket and reconnect) so we do
      not corrupt the remaining part of the SMB with the beginning
      of the next SMB.
      
      This does not appear to hurt performance measurably and has
      been run in various scenarios, but it definately removes
      a corruption that we were seeing in some high stress
      test cases.
      Acked-by: NShirish Pargaonkar <shirishp@us.ibm.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      edf1ae40
  4. 23 10月, 2008 5 次提交
    • M
      [PATCH] move executable checking into ->permission() · f696a365
      Miklos Szeredi 提交于
      For execute permission on a regular files we need to check if file has
      any execute bits at all, regardless of capabilites.
      
      This check is normally performed by generic_permission() but was also
      added to the case when the filesystem defines its own ->permission()
      method.  In the latter case the filesystem should be responsible for
      performing this check.
      
      Move the check from inode_permission() inside filesystems which are
      not calling generic_permission().
      
      Create a helper function execute_ok() that returns true if the inode
      is a directory or if any execute bits are present in i_mode.
      
      Also fix up the following code:
      
       - coda control file is never executable
       - sysctl files are never executable
       - hfs_permission seems broken on MAY_EXEC, remove
       - hfsplus_permission is eqivalent to generic_permission(), remove
      Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
      f696a365
    • C
      [PATCH] fix ->llseek for more directories · 3222a3e5
      Christoph Hellwig 提交于
      With this patch all directory fops instances that have a readdir
      that doesn't take the BKL are switched to generic_file_llseek.
      Signed-off-by: NChristoph Hellwig <hch@lst.de>
      3222a3e5
    • J
      cifs: handle the TCP_Server_Info->tsk field more carefully · b1c8d2b4
      Jeff Layton 提交于
      cifs: handle the TCP_Server_Info->tsk field more carefully
      
      We currently handle the TCP_Server_Info->tsk field without any locking,
      but with some half-measures to try and prevent races. These aren't
      really sufficient though. When taking down cifsd, use xchg() to swap
      the contents of the tsk field with NULL so we don't end up trying
      to send it more than one signal. Also, don't allow cifsd to exit until
      the signal is received if we expect one.
      Signed-off-by: NJeff Layton <jlayton@redhat.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      b1c8d2b4
    • J
      cifs: fix unlinking of rename target when server doesn't support open file renames · 8d281efb
      Jeff Layton 提交于
      cifs: fix unlinking of rename target when server doesn't support open file renames
      
      The patch to make cifs_rename undoable broke renaming one file on top of
      another when the server doesn't support busy file renames. Remove the
      code that uses busy file renames to unlink the target file, and just
      have it call cifs_unlink. If the rename of the source file fails, then
      the unlink won't be undoable, but hopefully that's rare enough that it
      won't be a problem.
      Signed-off-by: NJeff Layton <jlayton@redhat.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      8d281efb
    • S
      [CIFS] improve setlease handling · 84210e91
      Steve French 提交于
      fcntl(F_SETLEASE) currently is not exported by cifs (nor by local file
      systems) so cifs grants leases based on how other local processes have
      opened the file not by whether the file is cacheable (oplocked).  This
      adds the check to make sure that the file is cacheable on the client
      before checking whether we can grant the lease locally
      (generic_setlease).  It also adds a mount option for cifs (locallease)
      if the user wants to override this and try to grant leases even
      if the server did not grant oplock.
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      84210e91
  5. 21 10月, 2008 3 次提交
  6. 20 10月, 2008 4 次提交
  7. 17 10月, 2008 3 次提交
    • J
      cifs: don't use CREATE_DELETE_ON_CLOSE in cifs_rename_pending_delete · dd1db2de
      Jeff Layton 提交于
      cifs: don't use CREATE_DELETE_ON_CLOSE in cifs_rename_pending_delete
      
      CREATE_DELETE_ON_CLOSE apparently has different semantics than when you
      set the DELETE_ON_CLOSE bit after opening the file. Setting it in the
      open says "delete this file as soon as this filehandle is closed". That's
      not what we want for cifs_rename_pending_delete.
      
      Don't set this bit in the CreateFlags. Experimentation shows that
      setting this flag in the SET_FILE_INFO call has no effect.
      Signed-off-by: NJeff Layton <jlayton@redhat.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      dd1db2de
    • J
      [CIFS] eliminate usage of kthread_stop for cifsd · 469ee614
      Jeff Layton 提交于
      When cifs_demultiplex_thread was converted to a kthread based kernel
      thread, great pains were taken to make it so that kthread_stop would be
      used to bring it down. This just added unnecessary complexity since we
      needed to use a signal anyway to break out of kernel_recvmsg.
      
      Also, cifs_demultiplex_thread does a bit of cleanup as it's exiting, and
      we need to be certain that this gets done. It's possible for a kthread
      to exit before its main function is ever run if kthread_stop is called
      soon after its creation. While I'm not sure that this is a real problem
      with cifsd now, it could be at some point in the future if cifs_mount is
      ever changed to bring down the thread quickly.
      
      The upshot here is that using kthread_stop to bring down the thread just
      adds extra complexity with no real benefit. This patch changes the code
      to use the original method to bring down the thread, but still leaves it
      so that the thread is actually started with kthread_run.
      
      This seems to fix the deadlock caused by the reproducer in this bug
      report:
      
      https://bugzilla.samba.org/show_bug.cgi?id=5720Signed-off-by: NJeff Layton <jlayton@redhat.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      469ee614
    • S
      [CIFS] Add nodfs mount option · 2c1b8615
      Steve French 提交于
      Older samba server (eg. 3.0.24 from Debian etch) don't work correctly,
      if DFS paths are used. Such server claim that they support DFS, but fail
      to process some requests with DFS paths. Starting with Linux 2.6.26,
      the cifs clients starts sending DFS paths in such situations, rendering
      it unuseable with older samba servers.
      
      The nodfs mount options forces a share to be used with non DFS paths,
      even if the server claims, that it supports it.
      Signed-off-by: NMartin Koegler <mkoegler@auto.tuwien.ac.at>
      Acked-by: NJeff Layton <jlayton@redhat.com>
      Acked-by: NIgor Mammedov <niallain@gmail.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      2c1b8615
  8. 12 10月, 2008 1 次提交
  9. 09 10月, 2008 1 次提交
  10. 08 10月, 2008 2 次提交
    • S
      [CIFS] make sure we have the right resume info before calling CIFSFindNext · 0752f152
      Steve French 提交于
      When we do a seekdir() or equivalent, we usually end up doing a
      FindFirst call and then call FindNext until we get to the offset that we
      want. The problem is that when we call FindNext, the code usually
      doesn't have the proper info (mostly, the filename of the entry from the
      last search) to resume the search.
      
      Add a "last_entry" field to the cifs_search_info that points to the last
      entry in the search. We calculate this pointer by using the
      LastNameOffset field from the search parms that are returned. We then
      use that info to do a cifs_save_resume_key before we call CIFSFindNext.
      
      This patch allows CIFS to reliably pass the "telldir" connectathon test.
      Signed-off-by: NJeff Layton <jlayton@redhat.com>
      CC: Stable <stable@kernel.org>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      0752f152
    • S
      [CIFS] clean up error handling in cifs_unlink · 6050247d
      Steve French 提交于
      Currently, if a standard delete fails and we end up getting -EACCES
      we try to clear ATTR_READONLY and try the delete again. If that
      then fails with -ETXTBSY then we try a rename_pending_delete. We
      aren't handling other errors appropriately though.
      
      Another client could have deleted the file in the meantime and
      we get back -ENOENT, for instance. In that case we wouldn't do a
      d_drop. Instead of retrying in a separate call, simply goto the
      original call and use the error handling from that.
      
      Also, we weren't properly undoing any attribute changes that
      were done before returning an error back to the caller.
      
      CC: Jeff Layton <jlayton@redhat.com>
      Signed-off-by: NSteve French <sfrench@us.ibm.com>
      6050247d
  11. 07 10月, 2008 1 次提交
  12. 25 9月, 2008 6 次提交
  13. 24 9月, 2008 5 次提交
  14. 23 9月, 2008 2 次提交
  15. 17 9月, 2008 2 次提交
  16. 28 8月, 2008 2 次提交