Merge with /pub/scm/linux/kernel/git/torvalds/linux-2.6.git
[sfrench/cifs-2.6.git] / fs / cifs / cifsfs.c
index 1ebf7dafc1d757128e4084d6a1794485399b7ecc..79eeccd0437f08c2a3ac56a51af2de0cadc36f19 100644 (file)
@@ -32,6 +32,7 @@
 #include <linux/seq_file.h>
 #include <linux/vfs.h>
 #include <linux/mempool.h>
+#include <linux/delay.h>
 #include "cifsfs.h"
 #include "cifspdu.h"
 #define DECLARE_GLOBALS_HERE
@@ -59,6 +60,8 @@ unsigned int ntlmv2_support = 0;
 unsigned int sign_CIFS_PDUs = 1;
 extern struct task_struct * oplockThread; /* remove sparse warning */
 struct task_struct * oplockThread = NULL;
+extern struct task_struct * dnotifyThread; /* remove sparse warning */
+struct task_struct * dnotifyThread = NULL;
 unsigned int CIFSMaxBufSize = CIFS_MAX_MSGSIZE;
 module_param(CIFSMaxBufSize, int, 0);
 MODULE_PARM_DESC(CIFSMaxBufSize,"Network buffer size (not including header). Default: 16384 Range: 8192 to 130048");
@@ -73,6 +76,7 @@ module_param(cifs_max_pending, int, 0);
 MODULE_PARM_DESC(cifs_max_pending,"Simultaneous requests to server. Default: 50 Range: 2 to 256");
 
 static DECLARE_COMPLETION(cifs_oplock_exited);
+static DECLARE_COMPLETION(cifs_dnotify_exited);
 
 extern mempool_t *cifs_sm_req_poolp;
 extern mempool_t *cifs_req_poolp;
@@ -202,6 +206,10 @@ cifs_statfs(struct super_block *sb, struct kstatfs *buf)
 #endif /* CIFS_EXPERIMENTAL */
        rc = CIFSSMBQFSInfo(xid, pTcon, buf);
 
+       /* Old Windows servers do not support level 103, retry with level 
+          one if old server failed the previous call */ 
+       if(rc)
+               rc = SMBOldQFSInfo(xid, pTcon, buf);
        /*     
           int f_type;
           __fsid_t f_fsid;
@@ -253,7 +261,7 @@ cifs_alloc_inode(struct super_block *sb)
        cifs_inode->clientCanCacheAll = FALSE;
        cifs_inode->vfs_inode.i_blksize = CIFS_MAX_MSGSIZE;
        cifs_inode->vfs_inode.i_blkbits = 14;  /* 2**14 = CIFS_MAX_MSGSIZE */
-
+       cifs_inode->vfs_inode.i_flags = S_NOATIME | S_NOCMTIME;
        INIT_LIST_HEAD(&cifs_inode->openFileList);
        return &cifs_inode->vfs_inode;
 }
@@ -398,6 +406,42 @@ static struct quotactl_ops cifs_quotactl_ops = {
 };
 #endif
 
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+static void cifs_umount_begin(struct super_block * sblock)
+{
+       struct cifs_sb_info *cifs_sb;
+       struct cifsTconInfo * tcon;
+
+       cifs_sb = CIFS_SB(sblock);
+       if(cifs_sb == NULL)
+               return;
+
+       tcon = cifs_sb->tcon;
+       if(tcon == NULL)
+               return;
+       down(&tcon->tconSem);
+       if (atomic_read(&tcon->useCount) == 1)
+               tcon->tidStatus = CifsExiting;
+       up(&tcon->tconSem);
+
+       /* cancel_brl_requests(tcon); */
+       /* cancel_notify_requests(tcon); */
+       if(tcon->ses && tcon->ses->server)
+       {
+               cFYI(1,("wake up tasks now - umount begin not complete"));
+               wake_up_all(&tcon->ses->server->request_q);
+               wake_up_all(&tcon->ses->server->response_q);
+               msleep(1); /* yield */
+               /* we have to kick the requests once more */
+               wake_up_all(&tcon->ses->server->response_q);
+               msleep(1);
+       }
+/* BB FIXME - finish add checks for tidStatus BB */
+
+       return;
+}
+#endif 
+
 static int cifs_remount(struct super_block *sb, int *flags, char *data)
 {
        *flags |= MS_NODIRATIME;
@@ -415,7 +459,9 @@ struct super_operations cifs_super_ops = {
    unless later we add lazy close of inodes or unless the kernel forgets to call
    us with the same number of releases (closes) as opens */
        .show_options = cifs_show_options,
-/*    .umount_begin   = cifs_umount_begin, *//* consider adding in the future */
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+       .umount_begin   = cifs_umount_begin,
+#endif
        .remount_fs = cifs_remount,
 };
 
@@ -443,56 +489,40 @@ cifs_get_sb(struct file_system_type *fs_type,
        return sb;
 }
 
-static ssize_t
-cifs_read_wrapper(struct file * file, char __user *read_data, size_t read_size,
-          loff_t * poffset)
+static ssize_t cifs_file_writev(struct file *file, const struct iovec *iov,
+                               unsigned long nr_segs, loff_t *ppos)
 {
-       if(file->f_dentry == NULL)
-               return -EIO;
-       else if(file->f_dentry->d_inode == NULL)
-               return -EIO;
-
-       cFYI(1,("In read_wrapper size %zd at %lld",read_size,*poffset));
+       struct inode *inode = file->f_dentry->d_inode;
+       ssize_t written;
 
-       if(CIFS_I(file->f_dentry->d_inode)->clientCanCacheRead) {
-               return generic_file_read(file,read_data,read_size,poffset);
-       } else {
-               /* BB do we need to lock inode from here until after invalidate? */
-/*             if(file->f_dentry->d_inode->i_mapping) {
-                       filemap_fdatawrite(file->f_dentry->d_inode->i_mapping);
-                       filemap_fdatawait(file->f_dentry->d_inode->i_mapping);
-               }*/
-/*             cifs_revalidate(file->f_dentry);*/ /* BB fixme */
-
-               /* BB we should make timer configurable - perhaps 
-                  by simply calling cifs_revalidate here */
-               /* invalidate_remote_inode(file->f_dentry->d_inode);*/
-               return generic_file_read(file,read_data,read_size,poffset);
-       }
+       written = generic_file_writev(file, iov, nr_segs, ppos);
+       if (!CIFS_I(inode)->clientCanCacheAll)
+               filemap_fdatawrite(inode->i_mapping);
+       return written;
 }
 
-static ssize_t
-cifs_write_wrapper(struct file * file, const char __user *write_data,
-           size_t write_size, loff_t * poffset) 
+static ssize_t cifs_file_aio_write(struct kiocb *iocb, const char __user *buf,
+                                  size_t count, loff_t pos)
 {
+       struct inode *inode = iocb->ki_filp->f_dentry->d_inode;
        ssize_t written;
 
-       if(file->f_dentry == NULL)
-               return -EIO;
-       else if(file->f_dentry->d_inode == NULL)
-               return -EIO;
-
-       cFYI(1,("In write_wrapper size %zd at %lld",write_size,*poffset));
-
-       written = generic_file_write(file,write_data,write_size,poffset);
-       if(!CIFS_I(file->f_dentry->d_inode)->clientCanCacheAll)  {
-               if(file->f_dentry->d_inode->i_mapping) {
-                       filemap_fdatawrite(file->f_dentry->d_inode->i_mapping);
-               }
-       }
+       written = generic_file_aio_write(iocb, buf, count, pos);
+       if (!CIFS_I(inode)->clientCanCacheAll)
+               filemap_fdatawrite(inode->i_mapping);
        return written;
 }
 
+static loff_t cifs_llseek(struct file *file, loff_t offset, int origin)
+{
+       /* origin == SEEK_END => we must revalidate the cached file length */
+       if (origin == 2) {
+               int retval = cifs_revalidate(file->f_dentry);
+               if (retval < 0)
+                       return (loff_t)retval;
+       }
+       return remote_llseek(file, offset, origin);
+}
 
 static struct file_system_type cifs_fs_type = {
        .owner = THIS_MODULE,
@@ -554,8 +584,12 @@ struct inode_operations cifs_symlink_inode_ops = {
 };
 
 struct file_operations cifs_file_ops = {
-       .read = cifs_read_wrapper,
-       .write = cifs_write_wrapper, 
+       .read = do_sync_read,
+       .write = do_sync_write,
+       .readv = generic_file_readv,
+       .writev = cifs_file_writev,
+       .aio_read = generic_file_aio_read,
+       .aio_write = cifs_file_aio_write,
        .open = cifs_open,
        .release = cifs_close,
        .lock = cifs_lock,
@@ -563,15 +597,12 @@ struct file_operations cifs_file_ops = {
        .flush = cifs_flush,
        .mmap  = cifs_file_mmap,
        .sendfile = generic_file_sendfile,
+       .llseek = cifs_llseek,
 #ifdef CONFIG_CIFS_POSIX
        .ioctl  = cifs_ioctl,
 #endif /* CONFIG_CIFS_POSIX */
 
 #ifdef CONFIG_CIFS_EXPERIMENTAL
-       .readv = generic_file_readv,
-       .writev = generic_file_writev,
-       .aio_read = generic_file_aio_read,
-       .aio_write = generic_file_aio_write,
        .dir_notify = cifs_dir_notify,
 #endif /* CONFIG_CIFS_EXPERIMENTAL */
 };
@@ -590,12 +621,53 @@ struct file_operations cifs_file_direct_ops = {
 #ifdef CONFIG_CIFS_POSIX
        .ioctl  = cifs_ioctl,
 #endif /* CONFIG_CIFS_POSIX */
+       .llseek = cifs_llseek,
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+       .dir_notify = cifs_dir_notify,
+#endif /* CONFIG_CIFS_EXPERIMENTAL */
+};
+struct file_operations cifs_file_nobrl_ops = {
+       .read = do_sync_read,
+       .write = do_sync_write,
+       .readv = generic_file_readv,
+       .writev = cifs_file_writev,
+       .aio_read = generic_file_aio_read,
+       .aio_write = cifs_file_aio_write,
+       .open = cifs_open,
+       .release = cifs_close,
+       .fsync = cifs_fsync,
+       .flush = cifs_flush,
+       .mmap  = cifs_file_mmap,
+       .sendfile = generic_file_sendfile,
+       .llseek = cifs_llseek,
+#ifdef CONFIG_CIFS_POSIX
+       .ioctl  = cifs_ioctl,
+#endif /* CONFIG_CIFS_POSIX */
 
 #ifdef CONFIG_CIFS_EXPERIMENTAL
        .dir_notify = cifs_dir_notify,
 #endif /* CONFIG_CIFS_EXPERIMENTAL */
 };
 
+struct file_operations cifs_file_direct_nobrl_ops = {
+       /* no mmap, no aio, no readv - 
+          BB reevaluate whether they can be done with directio, no cache */
+       .read = cifs_user_read,
+       .write = cifs_user_write,
+       .open = cifs_open,
+       .release = cifs_close,
+       .fsync = cifs_fsync,
+       .flush = cifs_flush,
+       .sendfile = generic_file_sendfile, /* BB removeme BB */
+#ifdef CONFIG_CIFS_POSIX
+       .ioctl  = cifs_ioctl,
+#endif /* CONFIG_CIFS_POSIX */
+       .llseek = cifs_llseek,
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+       .dir_notify = cifs_dir_notify,
+#endif /* CONFIG_CIFS_EXPERIMENTAL */
+};
+
 struct file_operations cifs_dir_ops = {
        .readdir = cifs_readdir,
        .release = cifs_closedir,
@@ -674,7 +746,7 @@ cifs_init_request_bufs(void)
                kmem_cache_destroy(cifs_req_cachep);
                return -ENOMEM;
        }
-       /* 256 (MAX_CIFS_HDR_SIZE bytes is enough for most SMB responses and
+       /* MAX_CIFS_SMALL_BUFFER_SIZE bytes is enough for most SMB responses and
        almost all handle based requests (but not write response, nor is it
        sufficient for path based requests).  A smaller size would have
        been more efficient (compacting multiple slab items on one 4k page) 
@@ -683,7 +755,8 @@ cifs_init_request_bufs(void)
        efficient to alloc 1 per page off the slab compared to 17K (5page) 
        alloc of large cifs buffers even when page debugging is on */
        cifs_sm_req_cachep = kmem_cache_create("cifs_small_rq",
-                       MAX_CIFS_HDR_SIZE, 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
+                       MAX_CIFS_SMALL_BUFFER_SIZE, 0, SLAB_HWCACHE_ALIGN, 
+                       NULL, NULL);
        if (cifs_sm_req_cachep == NULL) {
                mempool_destroy(cifs_req_poolp);
                kmem_cache_destroy(cifs_req_cachep);
@@ -783,9 +856,7 @@ static int cifs_oplock_thread(void * dummyarg)
        do {
                if (try_to_freeze()) 
                        continue;
-               set_current_state(TASK_INTERRUPTIBLE);
                
-               schedule_timeout(1*HZ);  
                spin_lock(&GlobalMid_Lock);
                if(list_empty(&GlobalOplock_Q)) {
                        spin_unlock(&GlobalMid_Lock);
@@ -803,9 +874,9 @@ static int cifs_oplock_thread(void * dummyarg)
                                DeleteOplockQEntry(oplock_item);
                                /* can not grab inode sem here since it would
                                deadlock when oplock received on delete 
-                               since vfs_unlink holds the i_sem across
+                               since vfs_unlink holds the i_mutex across
                                the call */
-                               /* down(&inode->i_sem);*/
+                               /* mutex_lock(&inode->i_mutex);*/
                                if (S_ISREG(inode->i_mode)) {
                                        rc = filemap_fdatawrite(inode->i_mapping);
                                        if(CIFS_I(inode)->clientCanCacheRead == 0) {
@@ -814,7 +885,7 @@ static int cifs_oplock_thread(void * dummyarg)
                                        }
                                } else
                                        rc = 0;
-                               /* up(&inode->i_sem);*/
+                               /* mutex_unlock(&inode->i_mutex);*/
                                if (rc)
                                        CIFS_I(inode)->write_behind_rc = rc;
                                cFYI(1,("Oplock flush inode %p rc %d",inode,rc));
@@ -834,10 +905,42 @@ static int cifs_oplock_thread(void * dummyarg)
                                }
                        } else
                                spin_unlock(&GlobalMid_Lock);
+                       set_current_state(TASK_INTERRUPTIBLE);
+                       schedule_timeout(1);  /* yield in case q were corrupt */
                }
        } while(!signal_pending(current));
-       complete_and_exit (&cifs_oplock_exited, 0);
        oplockThread = NULL;
+       complete_and_exit (&cifs_oplock_exited, 0);
+}
+
+static int cifs_dnotify_thread(void * dummyarg)
+{
+       struct list_head *tmp;
+       struct cifsSesInfo *ses;
+
+       daemonize("cifsdnotifyd");
+       allow_signal(SIGTERM);
+
+       dnotifyThread = current;
+       do {
+               if(try_to_freeze())
+                       continue;
+               set_current_state(TASK_INTERRUPTIBLE);
+               schedule_timeout(15*HZ);
+               read_lock(&GlobalSMBSeslock);
+               /* check if any stuck requests that need
+                  to be woken up and wakeq so the
+                  thread can wake up and error out */
+               list_for_each(tmp, &GlobalSMBSessionList) {
+                       ses = list_entry(tmp, struct cifsSesInfo, 
+                               cifsSessionList);
+                       if(ses && ses->server && 
+                            atomic_read(&ses->server->inFlight))
+                               wake_up_all(&ses->server->response_q);
+               }
+               read_unlock(&GlobalSMBSeslock);
+       } while(!signal_pending(current));
+       complete_and_exit (&cifs_dnotify_exited, 0);
 }
 
 static int __init
@@ -851,6 +954,10 @@ init_cifs(void)
        INIT_LIST_HEAD(&GlobalSMBSessionList);
        INIT_LIST_HEAD(&GlobalTreeConnectionList);
        INIT_LIST_HEAD(&GlobalOplock_Q);
+#ifdef CONFIG_CIFS_EXPERIMENTAL
+       INIT_LIST_HEAD(&GlobalDnotifyReqList);
+       INIT_LIST_HEAD(&GlobalDnotifyRsp_Q);
+#endif 
 /*
  *  Initialize Global counters
  */
@@ -861,6 +968,12 @@ init_cifs(void)
        atomic_set(&tconInfoReconnectCount, 0);
 
        atomic_set(&bufAllocCount, 0);
+       atomic_set(&smBufAllocCount, 0);
+#ifdef CONFIG_CIFS_STATS2
+       atomic_set(&totBufAllocCount, 0);
+       atomic_set(&totSmBufAllocCount, 0);
+#endif /* CONFIG_CIFS_STATS2 */
+
        atomic_set(&midCount, 0);
        GlobalCurrentXid = 0;
        GlobalTotalActiveXid = 0;
@@ -886,10 +999,16 @@ init_cifs(void)
                                if (!rc) {                
                                        rc = (int)kernel_thread(cifs_oplock_thread, NULL, 
                                                CLONE_FS | CLONE_FILES | CLONE_VM);
-                                       if(rc > 0)
-                                               return 0;
-                                       else 
+                                       if(rc > 0) {
+                                               rc = (int)kernel_thread(cifs_dnotify_thread, NULL,
+                                                       CLONE_FS | CLONE_FILES | CLONE_VM);
+                                               if(rc > 0)
+                                                       return 0;
+                                               else
+                                                       cERROR(1,("error %d create dnotify thread", rc));
+                                       } else {
                                                cERROR(1,("error %d create oplock thread",rc));
+                                       }
                                }
                                cifs_destroy_request_bufs();
                        }
@@ -918,6 +1037,10 @@ exit_cifs(void)
                send_sig(SIGTERM, oplockThread, 1);
                wait_for_completion(&cifs_oplock_exited);
        }
+       if(dnotifyThread) {
+               send_sig(SIGTERM, dnotifyThread, 1);
+               wait_for_completion(&cifs_dnotify_exited);
+       }
 }
 
 MODULE_AUTHOR("Steve French <sfrench@us.ibm.com>");