[CIFS] Avoid extra large buffer allocation (and memcpy) in cifs_readpages

Signed-off-by: Steve French <sfrench@us.ibm.com>

Author: Steve French

fs/cifs/cifs_debug.c

@@ -401,8 +401,8 @@ static read_proc_t ntlmv2_enabled_read;
 static write_proc_t ntlmv2_enabled_write;
 static read_proc_t packet_signing_enabled_read;
 static write_proc_t packet_signing_enabled_write;
-static read_proc_t quotaEnabled_read;
-static write_proc_t quotaEnabled_write;
+static read_proc_t experimEnabled_read;
+static write_proc_t experimEnabled_write;
 static read_proc_t linuxExtensionsEnabled_read;
 static write_proc_t linuxExtensionsEnabled_write;
 
@@ -442,9 +442,9 @@ cifs_proc_init(void)
 		pde->write_proc = oplockEnabled_write;
 
 	pde = create_proc_read_entry("Experimental", 0, proc_fs_cifs,
-				quotaEnabled_read, NULL);
+				experimEnabled_read, NULL);
 	if (pde)
-		pde->write_proc = quotaEnabled_write;
+		pde->write_proc = experimEnabled_write;
 
 	pde = create_proc_read_entry("LinuxExtensionsEnabled", 0, proc_fs_cifs,
 				linuxExtensionsEnabled_read, NULL);
@@ -586,14 +586,13 @@ oplockEnabled_write(struct file *file, const char __user *buffer,
 }
 
 static int
-quotaEnabled_read(char *page, char **start, off_t off,
+experimEnabled_read(char *page, char **start, off_t off,
                    int count, int *eof, void *data)
 {
         int len;
 
         len = sprintf(page, "%d\n", experimEnabled);
-/* could also check if quotas are enabled in kernel
-	as a whole first */
+
         len -= off;
         *start = page + off;
 
@@ -608,21 +607,23 @@ quotaEnabled_read(char *page, char **start, off_t off,
         return len;
 }
 static int
-quotaEnabled_write(struct file *file, const char __user *buffer,
+experimEnabled_write(struct file *file, const char __user *buffer,
                     unsigned long count, void *data)
 {
-        char c;
-        int rc;
+	char c;
+	int rc;
 
-        rc = get_user(c, buffer);
-        if (rc)
-                return rc;
-        if (c == '0' || c == 'n' || c == 'N')
-                experimEnabled = 0;
-        else if (c == '1' || c == 'y' || c == 'Y')
-                experimEnabled = 1;
+	rc = get_user(c, buffer);
+	if (rc)
+		return rc;
+	if (c == '0' || c == 'n' || c == 'N')
+		experimEnabled = 0;
+	else if (c == '1' || c == 'y' || c == 'Y')
+		experimEnabled = 1;
+	else if (c == '2')
+		experimEnabled = 2;
 
-        return count;
+	return count;
 }
 
 static int
@@ -632,8 +633,6 @@ linuxExtensionsEnabled_read(char *page, char **start, off_t off,
         int len;
 
         len = sprintf(page, "%d\n", linuxExtEnabled);
-/* could also check if quotas are enabled in kernel
-	as a whole first */
         len -= off;
         *start = page + off;
 

fs/cifs/cifsfs.c

@@ -733,7 +733,7 @@ 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) 
@@ -742,7 +742,8 @@ 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);

fs/cifs/cifsglob.h

@@ -285,6 +285,7 @@ struct cifs_search_info {
 	unsigned endOfSearch:1;
 	unsigned emptyDir:1;
 	unsigned unicode:1;
+	unsigned smallBuf:1; /* so we know which buf_release function to call */
 };
 
 struct cifsFileInfo {
@@ -420,7 +421,12 @@ struct dir_notify_req {
 #define   MID_RESPONSE_RECEIVED 4
 #define   MID_RETRY_NEEDED      8 /* session closed while this request out */
 #define   MID_NO_RESP_NEEDED 0x10
-#define   MID_SMALL_BUFFER   0x20 /* 112 byte response buffer instead of 4K */
+
+/* Types of response buffer returned from SendReceive2 */
+#define   CIFS_NO_BUFFER        0    /* Response buffer not returned */
+#define   CIFS_SMALL_BUFFER     1
+#define   CIFS_LARGE_BUFFER     2
+#define   CIFS_IOVEC            4    /* array of response buffers */
 
 /*
  *****************************************************************

fs/cifs/cifspdu.h

@@ -80,7 +80,11 @@
 #define NT_TRANSACT_GET_USER_QUOTA    0x07
 #define NT_TRANSACT_SET_USER_QUOTA    0x08
 
-#define MAX_CIFS_HDR_SIZE 256	/* is future chained NTCreateXReadX bigger? */
+#define MAX_CIFS_SMALL_BUFFER_SIZE 448 /* big enough for most */
+/* future chained NTCreateXReadX bigger, but for time being NTCreateX biggest */
+/* among the requests (NTCreateX response is bigger with wct of 34) */
+#define MAX_CIFS_HDR_SIZE 0x58 /* 4 len + 32 hdr + (2*24 wct) + 2 bct + 2 pad */
+#define CIFS_SMALL_PATH 120 /* allows for (448-88)/3 */
 
 /* internal cifs vfs structures */
 /*****************************************************************

fs/cifs/cifsproto.h

@@ -49,7 +49,7 @@ extern int SendReceive(const unsigned int /* xid */ , struct cifsSesInfo *,
 			int * /* bytes returned */ , const int long_op);
 extern int SendReceive2(const unsigned int /* xid */ , struct cifsSesInfo *,
 			struct kvec *, int /* nvec to send */, 
-			int * /* bytes returned */ , const int long_op);
+			int * /* type of buf returned */ , const int long_op);
 extern int checkSMBhdr(struct smb_hdr *smb, __u16 mid);
 extern int checkSMB(struct smb_hdr *smb, __u16 mid, int length);
 extern int is_valid_oplock_break(struct smb_hdr *smb);
@@ -93,11 +93,12 @@ extern int CIFSTCon(unsigned int xid, struct cifsSesInfo *ses,
 			const struct nls_table *);
 
 extern int CIFSFindFirst(const int xid, struct cifsTconInfo *tcon,
-            const char *searchName, const struct nls_table *nls_codepage,
-            __u16 *searchHandle, struct cifs_search_info * psrch_inf, int map, const char dirsep);
+		const char *searchName, const struct nls_table *nls_codepage,
+		__u16 *searchHandle, struct cifs_search_info * psrch_inf, 
+		int map, const char dirsep);
 
 extern int CIFSFindNext(const int xid, struct cifsTconInfo *tcon,
-            __u16 searchHandle, struct cifs_search_info * psrch_inf);
+		__u16 searchHandle, struct cifs_search_info * psrch_inf);
 
 extern int CIFSFindClose(const int, struct cifsTconInfo *tcon,
 			const __u16 search_handle);
@@ -230,8 +231,9 @@ extern int CIFSSMBClose(const int xid, struct cifsTconInfo *tcon,
 			const int smb_file_id);
 
 extern int CIFSSMBRead(const int xid, struct cifsTconInfo *tcon,
-			const int netfid, unsigned int count,
-			const __u64 lseek, unsigned int *nbytes, char **buf);
+                        const int netfid, unsigned int count,
+                        const __u64 lseek, unsigned int *nbytes, char **buf,
+			int * return_buf_type);
 extern int CIFSSMBWrite(const int xid, struct cifsTconInfo *tcon,
 			const int netfid, const unsigned int count,
 			const __u64 lseek, unsigned int *nbytes,

fs/cifs/cifssmb.c

@@ -958,21 +958,19 @@ CIFSSMBOpen(const int xid, struct cifsTconInfo *tcon,
 	return rc;
 }
 
-/* If no buffer passed in, then caller wants to do the copy
-	as in the case of readpages so the SMB buffer must be
-	freed by the caller */
-
 int
 CIFSSMBRead(const int xid, struct cifsTconInfo *tcon,
-	    const int netfid, const unsigned int count,
-	    const __u64 lseek, unsigned int *nbytes, char **buf)
+            const int netfid, const unsigned int count,
+            const __u64 lseek, unsigned int *nbytes, char **buf,
+	    int * pbuf_type)
 {
 	int rc = -EACCES;
 	READ_REQ *pSMB = NULL;
 	READ_RSP *pSMBr = NULL;
 	char *pReadData = NULL;
-	int bytes_returned;
 	int wct;
+	int resp_buf_type = 0;
+	struct kvec iov[1];
 
 	cFYI(1,("Reading %d bytes on fid %d",count,netfid));
 	if(tcon->ses->capabilities & CAP_LARGE_FILES)
@@ -981,22 +979,21 @@ CIFSSMBRead(const int xid, struct cifsTconInfo *tcon,
 		wct = 10; /* old style read */
 
 	*nbytes = 0;
-	rc = smb_init(SMB_COM_READ_ANDX, wct, tcon, (void **) &pSMB,
-		      (void **) &pSMBr);
+	rc = small_smb_init(SMB_COM_READ_ANDX, wct, tcon, (void **) &pSMB);
 	if (rc)
 		return rc;
 
 	/* tcon and ses pointer are checked in smb_init */
 	if (tcon->ses->server == NULL)
 		return -ECONNABORTED;
 
-	pSMB->AndXCommand = 0xFF;	/* none */
+	pSMB->AndXCommand = 0xFF;       /* none */
 	pSMB->Fid = netfid;
 	pSMB->OffsetLow = cpu_to_le32(lseek & 0xFFFFFFFF);
 	if(wct == 12)
 		pSMB->OffsetHigh = cpu_to_le32(lseek >> 32);
-        else if((lseek >> 32) > 0) /* can not handle this big offset for old */
-                return -EIO;
+	else if((lseek >> 32) > 0) /* can not handle this big offset for old */
+		return -EIO;
 
 	pSMB->Remaining = 0;
 	pSMB->MaxCount = cpu_to_le16(count & 0xFFFF);
@@ -1005,14 +1002,18 @@ CIFSSMBRead(const int xid, struct cifsTconInfo *tcon,
 		pSMB->ByteCount = 0;  /* no need to do le conversion since 0 */
 	else {
 		/* old style read */
-		struct smb_com_readx_req * pSMBW = 
+		struct smb_com_readx_req * pSMBW =
 			(struct smb_com_readx_req *)pSMB;
-		pSMBW->ByteCount = 0;	
+		pSMBW->ByteCount = 0;
 	}
-	
-	rc = SendReceive(xid, tcon->ses, (struct smb_hdr *) pSMB,
-			 (struct smb_hdr *) pSMBr, &bytes_returned, 0);
+
+	iov[0].iov_base = (char *)pSMB;
+	iov[0].iov_len = pSMB->hdr.smb_buf_length + 4;
+	rc = SendReceive2(xid, tcon->ses, iov, 
+			  1 /* num iovecs */,
+			  &resp_buf_type, 0); 
 	cifs_stats_inc(&tcon->num_reads);
+	pSMBr = (READ_RSP *)iov[0].iov_base;
 	if (rc) {
 		cERROR(1, ("Send error in read = %d", rc));
 	} else {
@@ -1022,33 +1023,43 @@ CIFSSMBRead(const int xid, struct cifsTconInfo *tcon,
 		*nbytes = data_length;
 
 		/*check that DataLength would not go beyond end of SMB */
-		if ((data_length > CIFSMaxBufSize) 
+		if ((data_length > CIFSMaxBufSize)
 				|| (data_length > count)) {
 			cFYI(1,("bad length %d for count %d",data_length,count));
 			rc = -EIO;
 			*nbytes = 0;
 		} else {
-			pReadData =
-			    (char *) (&pSMBr->hdr.Protocol) +
+			pReadData = (char *) (&pSMBr->hdr.Protocol) +
 			    le16_to_cpu(pSMBr->DataOffset);
-/*			if(rc = copy_to_user(buf, pReadData, data_length)) {
-				cERROR(1,("Faulting on read rc = %d",rc));
-				rc = -EFAULT;
-			}*/ /* can not use copy_to_user when using page cache*/
+/*                      if(rc = copy_to_user(buf, pReadData, data_length)) {
+                                cERROR(1,("Faulting on read rc = %d",rc));
+                                rc = -EFAULT;
+                        }*/ /* can not use copy_to_user when using page cache*/
 			if(*buf)
-			    memcpy(*buf,pReadData,data_length);
+				memcpy(*buf,pReadData,data_length);
 		}
 	}
-	if(*buf)
-		cifs_buf_release(pSMB);
-	else
-		*buf = (char *)pSMB;
 
-	/* Note: On -EAGAIN error only caller can retry on handle based calls 
+	cifs_small_buf_release(pSMB);
+	if(*buf) {
+		if(resp_buf_type == CIFS_SMALL_BUFFER)
+			cifs_small_buf_release(iov[0].iov_base);
+		else if(resp_buf_type == CIFS_LARGE_BUFFER)
+			cifs_buf_release(iov[0].iov_base);
+	} else /* return buffer to caller to free */ /* BB FIXME how do we tell caller if it is not a large buffer */ {
+		*buf = iov[0].iov_base;
+		if(resp_buf_type == CIFS_SMALL_BUFFER)
+			*pbuf_type = CIFS_SMALL_BUFFER;
+		else if(resp_buf_type == CIFS_LARGE_BUFFER)
+			*pbuf_type = CIFS_LARGE_BUFFER;
+	}
+
+	/* Note: On -EAGAIN error only caller can retry on handle based calls
 		since file handle passed in no longer valid */
 	return rc;
 }
 
+
 int
 CIFSSMBWrite(const int xid, struct cifsTconInfo *tcon,
 	     const int netfid, const unsigned int count,
@@ -1163,10 +1174,10 @@ CIFSSMBWrite2(const int xid, struct cifsTconInfo *tcon,
 {
 	int rc = -EACCES;
 	WRITE_REQ *pSMB = NULL;
-	int bytes_returned, wct;
+	int wct;
 	int smb_hdr_len;
+	int resp_buf_type = 0;
 
-	/* BB removeme BB */
 	cFYI(1,("write2 at %lld %d bytes", (long long)offset, count));
 
 	if(tcon->ses->capabilities & CAP_LARGE_FILES)
@@ -1209,22 +1220,34 @@ CIFSSMBWrite2(const int xid, struct cifsTconInfo *tcon,
 		pSMBW->ByteCount = cpu_to_le16(count + 5);
 	}
 	iov[0].iov_base = pSMB;
-	iov[0].iov_len = smb_hdr_len + 4;
+	if(wct == 14)
+		iov[0].iov_len = smb_hdr_len + 4;
+	else /* wct == 12 pad bigger by four bytes */
+		iov[0].iov_len = smb_hdr_len + 8;
+	
 
-	rc = SendReceive2(xid, tcon->ses, iov, n_vec + 1, &bytes_returned,
+	rc = SendReceive2(xid, tcon->ses, iov, n_vec + 1, &resp_buf_type,
 			  long_op);
 	cifs_stats_inc(&tcon->num_writes);
 	if (rc) {
 		cFYI(1, ("Send error Write2 = %d", rc));
 		*nbytes = 0;
+	} else if(resp_buf_type == 0) {
+		/* presumably this can not happen, but best to be safe */
+		rc = -EIO;
+		*nbytes = 0;
 	} else {
-		WRITE_RSP * pSMBr = (WRITE_RSP *)pSMB;
+		WRITE_RSP * pSMBr = (WRITE_RSP *)iov[0].iov_base;
 		*nbytes = le16_to_cpu(pSMBr->CountHigh);
 		*nbytes = (*nbytes) << 16;
 		*nbytes += le16_to_cpu(pSMBr->Count);
 	} 
 
 	cifs_small_buf_release(pSMB);
+	if(resp_buf_type == CIFS_SMALL_BUFFER)
+		cifs_small_buf_release(iov[0].iov_base);
+	else if(resp_buf_type == CIFS_LARGE_BUFFER)
+		cifs_buf_release(iov[0].iov_base);
 
 	/* Note: On -EAGAIN error only caller can retry on handle based calls 
 		since file handle passed in no longer valid */

fs/cifs/connect.c

@@ -514,7 +514,7 @@ cifs_demultiplex_thread(struct TCP_Server_Info *server)
 		/* else length ok */
 		reconnect = 0;
 
-		if(pdu_length > MAX_CIFS_HDR_SIZE - 4) {
+		if(pdu_length > MAX_CIFS_SMALL_BUFFER_SIZE - 4) {
 			isLargeBuf = TRUE;
 			memcpy(bigbuf, smallbuf, 4);
 			smb_buffer = bigbuf;