[PATCH] TTY layer buffering revamp

The API and code have been through various bits of initial review by
serial driver people but they definitely need to live somewhere for a
while so the unconverted drivers can get knocked into shape, existing
drivers that have been updated can be better tuned and bugs whacked out.

This replaces the tty flip buffers with kmalloc objects in rings. In the
normal situation for an IRQ driven serial port at typical speeds the
behaviour is pretty much the same, two buffers end up allocated and the
kernel cycles between them as before.

When there are delays or at high speed we now behave far better as the
buffer pool can grow a bit rather than lose characters. This also means
that we can operate at higher speeds reliably.

For drivers that receive characters in blocks (DMA based, USB and
especially virtualisation) the layer allows a lot of driver specific
code that works around the tty layer with private secondary queues to be
removed. The IBM folks need this sort of layer, the smart serial port
people do, the virtualisers do (because a virtualised tty typically
operates at infinite speed rather than emulating 9600 baud).

Finally many drivers had invalid and unsafe attempts to avoid buffer
overflows by directly invoking tty methods extracted out of the innards
of work queue structs. These are no longer needed and all go away. That
fixes various random hangs with serial ports on overflow.

The other change in here is to optimise the receive_room path that is
used by some callers. It turns out that only one ldisc uses receive room
except asa constant and it updates it far far less than the value is
read. We thus make it a variable not a function call.

I expect the code to contain bugs due to the size alone but I'll be
watching and squashing them and feeding out new patches as it goes.

Because the buffers now dynamically expand you should only run out of
buffering when the kernel runs out of memory for real.  That means a lot of
the horrible hacks high performance drivers used to do just aren't needed any
more.

Description:

tty_insert_flip_char is an old API and continues to work as before, as does
tty_flip_buffer_push() [this is why many drivers dont need modification].  It
does now also return the number of chars inserted

There are also

tty_buffer_request_room(tty, len)

which asks for a buffer block of the length requested and returns the space
found.  This improves efficiency with hardware that knows how much to
transfer.

and tty_insert_flip_string_flags(tty, str, flags, len)

to insert a string of characters and flags

For a smart interface the usual code is

    len = tty_request_buffer_room(tty, amount_hardware_says);
    tty_insert_flip_string(tty, buffer_from_card, len);

More description!

At the moment tty buffers are attached directly to the tty.  This is causing a
lot of the problems related to tty layer locking, also problems at high speed
and also with bursty data (such as occurs in virtualised environments)

I'm working on ripping out the flip buffers and replacing them with a pool of
dynamically allocated buffers.  This allows both for old style "byte I/O"
devices and also helps virtualisation and smart devices where large blocks of
data suddenely materialise and need storing.

So far so good.  Lots of drivers reference tty->flip.*.  Several of them also
call directly and unsafely into function pointers it provides.  This will all
break.  Most drivers can use tty_insert_flip_char which can be kept as an API
but others need more.

At the moment I've added the following interfaces, if people think more will
be needed now is a good time to say

 int tty_buffer_request_room(tty, size)

Try and ensure at least size bytes are available, returns actual room (may be
zero).  At the moment it just uses the flipbuf space but that will change.
Repeated calls without characters being added are not cumulative.  (ie if you
call it with 1, 1, 1, and then 4 you'll have four characters of space.  The
other functions will also try and grow buffers in future but this will be a
more efficient way when you know block sizes.

 int tty_insert_flip_char(tty, ch, flag)

As before insert a character if there is room.  Now returns 1 for success, 0
for failure.

 int tty_insert_flip_string(tty, str, len)

Insert a block of non error characters.  Returns the number inserted.

 int tty_prepare_flip_string(tty, strptr, len)

Adjust the buffer to allow len characters to be added.  Returns a buffer
pointer in strptr and the length available.  This allows for hardware that
needs to use functions like insl or mencpy_fromio.

Signed-off-by: Alan Cox <alan@redhat.com>
Cc: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: John Hawkes <hawkes@sgi.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

4 files changed, 150 insertions, 41 deletions

diff --git a/drivers/isdn/capi/capi.c b/drivers/isdn/capi/capi.c
index 11ae0fddea0..623adbb0d13 100644
--- a/drivers/isdn/capi/capi.c
+++ b/drivers/isdn/capi/capi.c
@@ -463,8 +463,7 @@ static int handle_recv_skb(struct capiminor *mp, struct sk_buff *skb)
 #endif
 		goto bad;
 	}
-	if (ld->receive_room &&
-	    ld->receive_room(mp->tty) < datalen) {
+	if (mp->tty->receive_room < datalen) {
 #if defined(_DEBUG_DATAFLOW) || defined(_DEBUG_TTYFUNCS)
 		printk(KERN_DEBUG "capi: no room in tty\n");
 #endif
diff --git a/drivers/isdn/i4l/isdn_common.c b/drivers/isdn/i4l/isdn_common.c
index 4643df097bf..22759c01746 100644
--- a/drivers/isdn/i4l/isdn_common.c
+++ b/drivers/isdn/i4l/isdn_common.c
@@ -857,6 +857,118 @@ isdn_readbchan(int di, int channel, u_char * buf, u_char * fp, int len, wait_que
 	return count;
 }
 
+/*
+ * isdn_readbchan_tty() tries to get data from the read-queue.
+ * It MUST be called with interrupts off.
+ *
+ * Be aware that this is not an atomic operation when sleep != 0, even though
+ * interrupts are turned off! Well, like that we are currently only called
+ * on behalf of a read system call on raw device files (which are documented
+ * to be dangerous and for for debugging purpose only). The inode semaphore
+ * takes care that this is not called for the same minor device number while
+ * we are sleeping, but access is not serialized against simultaneous read()
+ * from the corresponding ttyI device. Can other ugly events, like changes
+ * of the mapping (di,ch)<->minor, happen during the sleep? --he
+ */
+int
+isdn_readbchan_tty(int di, int channel, struct tty_struct *tty, int cisco_hack)
+{
+	int count;
+	int count_pull;
+	int count_put;
+	int dflag;
+	struct sk_buff *skb;
+	char last = 0;
+	int len;
+
+	if (!dev->drv[di])
+		return 0;
+	if (skb_queue_empty(&dev->drv[di]->rpqueue[channel]))
+			return 0;
+
+	len = tty_buffer_request_room(tty, dev->drv[di]->rcvcount[channel]);
+	if(len == 0)
+		return len;
+
+	count = 0;
+	while (len) {
+		if (!(skb = skb_peek(&dev->drv[di]->rpqueue[channel])))
+			break;
+#ifdef CONFIG_ISDN_AUDIO
+		if (ISDN_AUDIO_SKB_LOCK(skb))
+			break;
+		ISDN_AUDIO_SKB_LOCK(skb) = 1;
+		if ((ISDN_AUDIO_SKB_DLECOUNT(skb)) || (dev->drv[di]->DLEflag & (1 << channel))) {
+			char *p = skb->data;
+			unsigned long DLEmask = (1 << channel);
+
+			dflag = 0;
+			count_pull = count_put = 0;
+			while ((count_pull < skb->len) && (len > 0)) {
+				len--;
+				if (dev->drv[di]->DLEflag & DLEmask) {
+					last = DLE;
+					dev->drv[di]->DLEflag &= ~DLEmask;
+				} else {
+					last = *p;
+					if (last == DLE) {
+						dev->drv[di]->DLEflag |= DLEmask;
+						(ISDN_AUDIO_SKB_DLECOUNT(skb))--;
+					}
+					p++;
+					count_pull++;
+				}
+				count_put++;
+			}
+			if (count_pull >= skb->len)
+				dflag = 1;
+		} else {
+#endif
+			/* No DLE's in buff, so simply copy it */
+			dflag = 1;
+			if ((count_pull = skb->len) > len) {
+				count_pull = len;
+				dflag = 0;
+			}
+			count_put = count_pull;
+			if(count_put > 1)
+				tty_insert_flip_string(tty, skb->data, count_put - 1);
+			last = skb->data[count_put] - 1;
+			len -= count_put;
+#ifdef CONFIG_ISDN_AUDIO
+		}
+#endif
+		count += count_put;
+		if (dflag) {
+			/* We got all the data in this buff.
+			 * Now we can dequeue it.
+			 */
+			if(cisco_hack)
+				tty_insert_flip_char(tty, last, 0xFF);
+			else
+				tty_insert_flip_char(tty, last, TTY_NORMAL);
+#ifdef CONFIG_ISDN_AUDIO
+			ISDN_AUDIO_SKB_LOCK(skb) = 0;
+#endif
+			skb = skb_dequeue(&dev->drv[di]->rpqueue[channel]);
+			dev_kfree_skb(skb);
+		} else {
+			tty_insert_flip_char(tty, last, TTY_NORMAL);
+			/* Not yet emptied this buff, so it
+			 * must stay in the queue, for further calls
+			 * but we pull off the data we got until now.
+			 */
+			skb_pull(skb, count_pull);
+#ifdef CONFIG_ISDN_AUDIO
+			ISDN_AUDIO_SKB_LOCK(skb) = 0;
+#endif
+		}
+		dev->drv[di]->rcvcount[channel] -= count_put;
+	}
+	return count;
+}
+
+
 static __inline int
 isdn_minor2drv(int minor)
 {
diff --git a/drivers/isdn/i4l/isdn_common.h b/drivers/isdn/i4l/isdn_common.h
index e27e9c3a81e..082735dbb41 100644
--- a/drivers/isdn/i4l/isdn_common.h
+++ b/drivers/isdn/i4l/isdn_common.h
@@ -37,6 +37,7 @@ extern void isdn_timer_ctrl(int tf, int onoff);
 extern void isdn_unexclusive_channel(int di, int ch);
 extern int  isdn_getnum(char **);
 extern int  isdn_readbchan(int, int, u_char *, u_char *, int, wait_queue_head_t *);
+extern int  isdn_readbchan_tty(int, int, struct tty_struct *, int);
 extern int  isdn_get_free_channel(int, int, int, int, int, char *);
 extern int  isdn_writebuf_skb_stub(int, int, int, struct sk_buff *);
 extern int  register_isdn(isdn_if * i);
diff --git a/drivers/isdn/i4l/isdn_tty.c b/drivers/isdn/i4l/isdn_tty.c
index 8c404b4e248..f190a99604f 100644
--- a/drivers/isdn/i4l/isdn_tty.c
+++ b/drivers/isdn/i4l/isdn_tty.c
@@ -64,37 +64,42 @@ isdn_tty_try_read(modem_info * info, struct sk_buff *skb)
 	int c;
 	int len;
 	struct tty_struct *tty;
+	char last;
 
 	if (info->online) {
 		if ((tty = info->tty)) {
 			if (info->mcr & UART_MCR_RTS) {
-				c = TTY_FLIPBUF_SIZE - tty->flip.count;
 				len = skb->len
 #ifdef CONFIG_ISDN_AUDIO
 					+ ISDN_AUDIO_SKB_DLECOUNT(skb)
 #endif
 					;
+
+				c = tty_buffer_request_room(tty, len);
 				if (c >= len) {
 #ifdef CONFIG_ISDN_AUDIO
-					if (ISDN_AUDIO_SKB_DLECOUNT(skb))
-						while (skb->len--) {
+					if (ISDN_AUDIO_SKB_DLECOUNT(skb)) {
+						int l = skb->len;
+						unsigned char *dp = skb->data;
+						while (--l) {
 							if (*skb->data == DLE)
 								tty_insert_flip_char(tty, DLE, 0);
-							tty_insert_flip_char(tty, *skb->data++, 0);
+							tty_insert_flip_char(tty, *dp++, 0);
+						}
+						last = *dp;
 					} else {
 #endif
-						memcpy(tty->flip.char_buf_ptr,
-						       skb->data, len);
-						tty->flip.count += len;
-						tty->flip.char_buf_ptr += len;
-						memset(tty->flip.flag_buf_ptr, 0, len);
-						tty->flip.flag_buf_ptr += len;
+						if(len > 1)
+							tty_insert_flip_string(tty, skb->data, len - 1);
+						last = skb->data[len - 1];
 #ifdef CONFIG_ISDN_AUDIO
 					}
 #endif
 					if (info->emu.mdmreg[REG_CPPP] & BIT_CPPP)
-						tty->flip.flag_buf_ptr[len - 1] = 0xff;
-					schedule_delayed_work(&tty->flip.work, 1);
+						tty_insert_flip_char(tty, last, 0xFF);
+					else
+						tty_insert_flip_char(tty, last, TTY_NORMAL);
+					tty_flip_buffer_push(tty);
 					kfree_skb(skb);
 					return 1;
 				}
@@ -114,7 +119,6 @@ isdn_tty_readmodem(void)
 	int resched = 0;
 	int midx;
 	int i;
-	int c;
 	int r;
 	struct tty_struct *tty;
 	modem_info *info;
@@ -131,20 +135,13 @@ isdn_tty_readmodem(void)
 #endif
 				if ((tty = info->tty)) {
 					if (info->mcr & UART_MCR_RTS) {
-						c = TTY_FLIPBUF_SIZE - tty->flip.count;
-						if (c > 0) {
-							r = isdn_readbchan(info->isdn_driver, info->isdn_channel,
-									   tty->flip.char_buf_ptr,
-									   tty->flip.flag_buf_ptr, c, NULL);
-							/* CISCO AsyncPPP Hack */
-							if (!(info->emu.mdmreg[REG_CPPP] & BIT_CPPP))
-								memset(tty->flip.flag_buf_ptr, 0, r);
-							tty->flip.count += r;
-							tty->flip.flag_buf_ptr += r;
-							tty->flip.char_buf_ptr += r;
-							if (r)
-								schedule_delayed_work(&tty->flip.work, 1);
-						}
+						/* CISCO AsyncPPP Hack */
+						if (!(info->emu.mdmreg[REG_CPPP] & BIT_CPPP))
+							r = isdn_readbchan_tty(info->isdn_driver, info->isdn_channel, tty, 0);
+						else
+							r = isdn_readbchan_tty(info->isdn_driver, info->isdn_channel, tty, 1);
+						if (r)
+							tty_flip_buffer_push(tty);
 					} else
 						r = 1;
 				} else
@@ -249,7 +246,7 @@ isdn_tty_rcv_skb(int i, int di, int channel, struct sk_buff *skb)
 	}
 #endif
 #endif
-	/* Try to deliver directly via tty-flip-buf if queue is empty */
+	/* Try to deliver directly via tty-buf if queue is empty */
 	spin_lock_irqsave(&info->readlock, flags);
 	if (skb_queue_empty(&dev->drv[di]->rpqueue[channel]))
 		if (isdn_tty_try_read(info, skb)) {
@@ -534,7 +531,7 @@ isdn_tty_senddown(modem_info * info)
 /* The next routine is called once from within timer-interrupt
  * triggered within isdn_tty_modem_ncarrier(). It calls
  * isdn_tty_modem_result() to stuff a "NO CARRIER" Message
- * into the tty's flip-buffer.
+ * into the tty's buffer.
  */
 static void
 isdn_tty_modem_do_ncarrier(unsigned long data)
@@ -2347,6 +2344,7 @@ isdn_tty_at_cout(char *msg, modem_info * info)
 	u_long flags;
 	struct sk_buff *skb = NULL;
 	char *sp = NULL;
+	int l = strlen(msg);
 
 	if (!msg) {
 		printk(KERN_WARNING "isdn_tty: Null-Message in isdn_tty_at_cout\n");
@@ -2359,16 +2357,16 @@ isdn_tty_at_cout(char *msg, modem_info * info)
 		return;
 	}
 
-	/* use queue instead of direct flip, if online and */
-	/* data is in queue or flip buffer is full */
-	if ((info->online) && (((tty->flip.count + strlen(msg)) >= TTY_FLIPBUF_SIZE) ||
-	    (!skb_queue_empty(&dev->drv[info->isdn_driver]->rpqueue[info->isdn_channel])))) {
-		skb = alloc_skb(strlen(msg), GFP_ATOMIC);
+	/* use queue instead of direct, if online and */
+	/* data is in queue or buffer is full */
+	if ((info->online && tty_buffer_request_room(tty, l) < l) ||
+	    (!skb_queue_empty(&dev->drv[info->isdn_driver]->rpqueue[info->isdn_channel]))) {
+		skb = alloc_skb(l, GFP_ATOMIC);
 		if (!skb) {
 			spin_unlock_irqrestore(&info->readlock, flags);
 			return;
 		}
-		sp = skb_put(skb, strlen(msg));
+		sp = skb_put(skb, l);
 #ifdef CONFIG_ISDN_AUDIO
 		ISDN_AUDIO_SKB_DLECOUNT(skb) = 0;
 		ISDN_AUDIO_SKB_LOCK(skb) = 0;
@@ -2392,9 +2390,8 @@ isdn_tty_at_cout(char *msg, modem_info * info)
 		if (skb) {
 			*sp++ = c;
 		} else {
-			if (tty->flip.count >= TTY_FLIPBUF_SIZE)
+			if(tty_insert_flip_char(tty, c, TTY_NORMAL) == 0)
 				break;
-			tty_insert_flip_char(tty, c, 0);
 		}
 	}
 	if (skb) {
@@ -2402,12 +2399,12 @@ isdn_tty_at_cout(char *msg, modem_info * info)
 		dev->drv[info->isdn_driver]->rcvcount[info->isdn_channel] += skb->len;
 		spin_unlock_irqrestore(&info->readlock, flags);
 		/* Schedule dequeuing */
-		if ((dev->modempoll) && (info->rcvsched))
+		if (dev->modempoll && info->rcvsched)
 			isdn_timer_ctrl(ISDN_TIMER_MODEMREAD, 1);
 
 	} else {
 		spin_unlock_irqrestore(&info->readlock, flags);
-		schedule_delayed_work(&tty->flip.work, 1);
+		tty_flip_buffer_push(tty);
 	}
 }