1 files changed, 176 insertions, 95 deletions

diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index 159fcc56dc2d..460d87b7cda0 100644
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -1399,7 +1399,10 @@ int cifs_lock(struct file *file, int cmd, struct file_lock *flock)
         return rc;
 }
 
-/* update the file size (if needed) after a write */
+/*
+ * update the file size (if needed) after a write. Should be called with
+ * the inode->i_lock held
+ */
 void
 cifs_update_eof(struct cifsInodeInfo *cifsi, loff_t offset,
                       unsigned int bytes_written)
@@ -1471,7 +1474,9 @@ static ssize_t cifs_write(struct cifsFileInfo *open_file, __u32 pid,
                                 return rc;
                         }
                 } else {
+                        spin_lock(&dentry->d_inode->i_lock);
                         cifs_update_eof(cifsi, *poffset, bytes_written);
+                        spin_unlock(&dentry->d_inode->i_lock);
                         *poffset += bytes_written;
                 }
         }
@@ -1648,6 +1653,27 @@ static int cifs_partialpagewrite(struct page *page, unsigned from, unsigned to)
         return rc;
 }
 
+/*
+ * Marshal up the iov array, reserving the first one for the header. Also,
+ * set wdata->bytes.
+ */
+static void
+cifs_writepages_marshal_iov(struct kvec *iov, struct cifs_writedata *wdata)
+{
+        int i;
+        struct inode *inode = wdata->cfile->dentry->d_inode;
+        loff_t size = i_size_read(inode);
+
+        /* marshal up the pages into iov array */
+        wdata->bytes = 0;
+        for (i = 0; i < wdata->nr_pages; i++) {
+                iov[i + 1].iov_len = min(size - page_offset(wdata->pages[i]),
+                                        (loff_t)PAGE_CACHE_SIZE);
+                iov[i + 1].iov_base = kmap(wdata->pages[i]);
+                wdata->bytes += iov[i + 1].iov_len;
+        }
+}
+
 static int cifs_writepages(struct address_space *mapping,
                            struct writeback_control *wbc)
 {
@@ -1684,7 +1710,8 @@ retry:
                 tofind = min((cifs_sb->wsize / PAGE_CACHE_SIZE) - 1,
                                 end - index) + 1;
 
-                wdata = cifs_writedata_alloc((unsigned int)tofind);
+                wdata = cifs_writedata_alloc((unsigned int)tofind,
+                                             cifs_writev_complete);
                 if (!wdata) {
                         rc = -ENOMEM;
                         break;
@@ -1791,6 +1818,7 @@ retry:
                 wdata->sync_mode = wbc->sync_mode;
                 wdata->nr_pages = nr_pages;
                 wdata->offset = page_offset(wdata->pages[0]);
+                wdata->marshal_iov = cifs_writepages_marshal_iov;
 
                 do {
                         if (wdata->cfile != NULL)
@@ -1802,6 +1830,7 @@ retry:
                                 rc = -EBADF;
                                 break;
                         }
+                        wdata->pid = wdata->cfile->pid;
                         rc = cifs_async_writev(wdata);
                 } while (wbc->sync_mode == WB_SYNC_ALL && rc == -EAGAIN);
 
@@ -2043,7 +2072,7 @@ cifs_write_allocate_pages(struct page **pages, unsigned long num_pages)
         unsigned long i;
 
         for (i = 0; i < num_pages; i++) {
-                pages[i] = alloc_page(__GFP_HIGHMEM);
+                pages[i] = alloc_page(GFP_KERNEL|__GFP_HIGHMEM);
                 if (!pages[i]) {
                         /*
                          * save number of pages we have already allocated and
@@ -2051,15 +2080,14 @@ cifs_write_allocate_pages(struct page **pages, unsigned long num_pages)
                          */
                         num_pages = i;
                         rc = -ENOMEM;
-                        goto error;
+                        break;
                 }
         }
 
-        return rc;
-
-error:
-        for (i = 0; i < num_pages; i++)
-                put_page(pages[i]);
+        if (rc) {
+                for (i = 0; i < num_pages; i++)
+                        put_page(pages[i]);
+        }
         return rc;
 }
 
@@ -2070,9 +2098,7 @@ size_t get_numpages(const size_t wsize, const size_t len, size_t *cur_len)
         size_t clen;
 
         clen = min_t(const size_t, len, wsize);
-        num_pages = clen / PAGE_CACHE_SIZE;
-        if (clen % PAGE_CACHE_SIZE)
-                num_pages++;
+        num_pages = DIV_ROUND_UP(clen, PAGE_SIZE);
 
         if (cur_len)
                 *cur_len = clen;
@@ -2080,24 +2106,79 @@ size_t get_numpages(const size_t wsize, const size_t len, size_t *cur_len)
         return num_pages;
 }
 
+static void
+cifs_uncached_marshal_iov(struct kvec *iov, struct cifs_writedata *wdata)
+{
+        int i;
+        size_t bytes = wdata->bytes;
+
+        /* marshal up the pages into iov array */
+        for (i = 0; i < wdata->nr_pages; i++) {
+                iov[i + 1].iov_len = min_t(size_t, bytes, PAGE_SIZE);
+                iov[i + 1].iov_base = kmap(wdata->pages[i]);
+                bytes -= iov[i + 1].iov_len;
+        }
+}
+
+static void
+cifs_uncached_writev_complete(struct work_struct *work)
+{
+        int i;
+        struct cifs_writedata *wdata = container_of(work,
+                                        struct cifs_writedata, work);
+        struct inode *inode = wdata->cfile->dentry->d_inode;
+        struct cifsInodeInfo *cifsi = CIFS_I(inode);
+
+        spin_lock(&inode->i_lock);
+        cifs_update_eof(cifsi, wdata->offset, wdata->bytes);
+        if (cifsi->server_eof > inode->i_size)
+                i_size_write(inode, cifsi->server_eof);
+        spin_unlock(&inode->i_lock);
+
+        complete(&wdata->done);
+
+        if (wdata->result != -EAGAIN) {
+                for (i = 0; i < wdata->nr_pages; i++)
+                        put_page(wdata->pages[i]);
+        }
+
+        kref_put(&wdata->refcount, cifs_writedata_release);
+}
+
+/* attempt to send write to server, retry on any -EAGAIN errors */
+static int
+cifs_uncached_retry_writev(struct cifs_writedata *wdata)
+{
+        int rc;
+
+        do {
+                if (wdata->cfile->invalidHandle) {
+                        rc = cifs_reopen_file(wdata->cfile, false);
+                        if (rc != 0)
+                                continue;
+                }
+                rc = cifs_async_writev(wdata);
+        } while (rc == -EAGAIN);
+
+        return rc;
+}
+
 static ssize_t
 cifs_iovec_write(struct file *file, const struct iovec *iov,
                  unsigned long nr_segs, loff_t *poffset)
 {
-        unsigned int written;
-        unsigned long num_pages, npages, i;
+        unsigned long nr_pages, i;
         size_t copied, len, cur_len;
         ssize_t total_written = 0;
-        struct kvec *to_send;
-        struct page **pages;
+        loff_t offset = *poffset;
         struct iov_iter it;
-        struct inode *inode;
         struct cifsFileInfo *open_file;
-        struct cifs_tcon *pTcon;
+        struct cifs_tcon *tcon;
         struct cifs_sb_info *cifs_sb;
-        struct cifs_io_parms io_parms;
-        int xid, rc;
-        __u32 pid;
+        struct cifs_writedata *wdata, *tmp;
+        struct list_head wdata_list;
+        int rc;
+        pid_t pid;
 
         len = iov_length(iov, nr_segs);
         if (!len)
@@ -2107,103 +2188,103 @@ cifs_iovec_write(struct file *file, const struct iovec *iov,
         if (rc)
                 return rc;
 
+        INIT_LIST_HEAD(&wdata_list);
         cifs_sb = CIFS_SB(file->f_path.dentry->d_sb);
-        num_pages = get_numpages(cifs_sb->wsize, len, &cur_len);
-
-        pages = kmalloc(sizeof(struct pages *)*num_pages, GFP_KERNEL);
-        if (!pages)
-                return -ENOMEM;
-
-        to_send = kmalloc(sizeof(struct kvec)*(num_pages + 1), GFP_KERNEL);
-        if (!to_send) {
-                kfree(pages);
-                return -ENOMEM;
-        }
-
-        rc = cifs_write_allocate_pages(pages, num_pages);
-        if (rc) {
-                kfree(pages);
-                kfree(to_send);
-                return rc;
-        }
-
-        xid = GetXid();
         open_file = file->private_data;
+        tcon = tlink_tcon(open_file->tlink);
 
         if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_RWPIDFORWARD)
                 pid = open_file->pid;
         else
                 pid = current->tgid;
 
-        pTcon = tlink_tcon(open_file->tlink);
-        inode = file->f_path.dentry->d_inode;
-
         iov_iter_init(&it, iov, nr_segs, len, 0);
-        npages = num_pages;
-
         do {
-                size_t save_len = cur_len;
-                for (i = 0; i < npages; i++) {
-                        copied = min_t(const size_t, cur_len, PAGE_CACHE_SIZE);
-                        copied = iov_iter_copy_from_user(pages[i], &it, 0,
-                                                         copied);
+                size_t save_len;
+
+                nr_pages = get_numpages(cifs_sb->wsize, len, &cur_len);
+                wdata = cifs_writedata_alloc(nr_pages,
+                                             cifs_uncached_writev_complete);
+                if (!wdata) {
+                        rc = -ENOMEM;
+                        break;
+                }
+
+                rc = cifs_write_allocate_pages(wdata->pages, nr_pages);
+                if (rc) {
+                        kfree(wdata);
+                        break;
+                }
+
+                save_len = cur_len;
+                for (i = 0; i < nr_pages; i++) {
+                        copied = min_t(const size_t, cur_len, PAGE_SIZE);
+                        copied = iov_iter_copy_from_user(wdata->pages[i], &it,
+                                                         0, copied);
                         cur_len -= copied;
                         iov_iter_advance(&it, copied);
-                        to_send[i+1].iov_base = kmap(pages[i]);
-                        to_send[i+1].iov_len = copied;
                 }
-
                 cur_len = save_len - cur_len;
 
-                do {
-                        if (open_file->invalidHandle) {
-                                rc = cifs_reopen_file(open_file, false);
-                                if (rc != 0)
-                                        break;
-                        }
-                        io_parms.netfid = open_file->netfid;
-                        io_parms.pid = pid;
-                        io_parms.tcon = pTcon;
-                        io_parms.offset = *poffset;
-                        io_parms.length = cur_len;
-                        rc = CIFSSMBWrite2(xid, &io_parms, &written, to_send,
-                                           npages, 0);
-                } while (rc == -EAGAIN);
-
-                for (i = 0; i < npages; i++)
-                        kunmap(pages[i]);
-
-                if (written) {
-                        len -= written;
-                        total_written += written;
-                        cifs_update_eof(CIFS_I(inode), *poffset, written);
-                        *poffset += written;
-                } else if (rc < 0) {
-                        if (!total_written)
-                                total_written = rc;
+                wdata->sync_mode = WB_SYNC_ALL;
+                wdata->nr_pages = nr_pages;
+                wdata->offset = (__u64)offset;
+                wdata->cfile = cifsFileInfo_get(open_file);
+                wdata->pid = pid;
+                wdata->bytes = cur_len;
+                wdata->marshal_iov = cifs_uncached_marshal_iov;
+                rc = cifs_uncached_retry_writev(wdata);
+                if (rc) {
+                        kref_put(&wdata->refcount, cifs_writedata_release);
                         break;
                 }
 
-                /* get length and number of kvecs of the next write */
-                npages = get_numpages(cifs_sb->wsize, len, &cur_len);
+                list_add_tail(&wdata->list, &wdata_list);
+                offset += cur_len;
+                len -= cur_len;
         } while (len > 0);
 
-        if (total_written > 0) {
-                spin_lock(&inode->i_lock);
-                if (*poffset > inode->i_size)
-                        i_size_write(inode, *poffset);
-                spin_unlock(&inode->i_lock);
+        /*
+         * If at least one write was successfully sent, then discard any rc
+         * value from the later writes. If the other write succeeds, then
+         * we'll end up returning whatever was written. If it fails, then
+         * we'll get a new rc value from that.
+         */
+        if (!list_empty(&wdata_list))
+                rc = 0;
+
+        /*
+         * Wait for and collect replies for any successful sends in order of
+         * increasing offset. Once an error is hit or we get a fatal signal
+         * while waiting, then return without waiting for any more replies.
+         */
+restart_loop:
+        list_for_each_entry_safe(wdata, tmp, &wdata_list, list) {
+                if (!rc) {
+                        /* FIXME: freezable too? */
+                        rc = wait_for_completion_killable(&wdata->done);
+                        if (rc)
+                                rc = -EINTR;
+                        else if (wdata->result)
+                                rc = wdata->result;
+                        else
+                                total_written += wdata->bytes;
+
+                        /* resend call if it's a retryable error */
+                        if (rc == -EAGAIN) {
+                                rc = cifs_uncached_retry_writev(wdata);
+                                goto restart_loop;
+                        }
+                }
+                list_del_init(&wdata->list);
+                kref_put(&wdata->refcount, cifs_writedata_release);
         }
 
-        cifs_stats_bytes_written(pTcon, total_written);
-        mark_inode_dirty_sync(inode);
+        if (total_written > 0)
+                *poffset += total_written;
 
-        for (i = 0; i < num_pages; i++)
-                put_page(pages[i]);
-        kfree(to_send);
-        kfree(pages);
-        FreeXid(xid);
-        return total_written;
+        cifs_stats_bytes_written(tcon, total_written);
+        return total_written ? total_written : (ssize_t)rc;
 }
 
 ssize_t cifs_user_writev(struct kiocb *iocb, const struct iovec *iov,