revert broken select optimizations

Cset exclude: torvalds@penguin.transmeta.com|ChangeSet|20020619003306|07760
Cset exclude: ak@muc.de|ChangeSet|20020618172743|19150

fs/select.c

@@ -12,9 +12,6 @@
  *  24 January 2000
  *     Changed sys_poll()/do_poll() to use PAGE_SIZE chunk-based allocation 
  *     of fds to overcome nfds < 16390 descriptors limit (Tigran Aivazian).
- * 
- *  Dec 2001
- *     Stack allocation and fast path (Andi Kleen) 
  */
 
 #include <linux/slab.h>
@@ -29,6 +26,21 @@
 #define ROUND_UP(x,y) (((x)+(y)-1)/(y))
 #define DEFAULT_POLLMASK (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM)
 
+struct poll_table_entry {
+	struct file * filp;
+	wait_queue_t wait;
+	wait_queue_head_t * wait_address;
+};
+
+struct poll_table_page {
+	struct poll_table_page * next;
+	struct poll_table_entry * entry;
+	struct poll_table_entry entries[0];
+};
+
+#define POLL_TABLE_FULL(table) \
+	((unsigned long)((table)->entry+1) > PAGE_SIZE + (unsigned long)(table))
+
 /*
  * Ok, Peter made a complicated, but straightforward multiple_wait() function.
  * I have rewritten this, taking some shortcuts: This code may not be easy to
@@ -50,14 +62,13 @@ void poll_freewait(poll_table* pt)
 		struct poll_table_page *old;
 
 		entry = p->entry;
-		while (entry > p->entries) {
+		do {
 			entry--;
 			remove_wait_queue(entry->wait_address,&entry->wait);
 			fput(entry->filp);
-		}
+		} while (entry > p->entries);
 		old = p;
 		p = p->next;
-		if (old != &pt->inline_page) 
 		free_page((unsigned long) old);
 	}
 }
@@ -65,24 +76,16 @@ void poll_freewait(poll_table* pt)
 void __pollwait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p)
 {
 	struct poll_table_page *table = p->table;
-	struct poll_table_page *new_table = NULL;
-	int sz;
-
-	if (!table) { 
-		new_table = &p->inline_page; 
-	} else { 
-		sz = (table == &p->inline_page) ? POLL_INLINE_TABLE_LEN : PAGE_SIZE; 
-		if ((char*)table->entry >= (char*)table + sz) {
-			new_table = (struct poll_table_page *)__get_free_page(GFP_KERNEL);
+
+	if (!table || POLL_TABLE_FULL(table)) {
+		struct poll_table_page *new_table;
+
+		new_table = (struct poll_table_page *) __get_free_page(GFP_KERNEL);
 		if (!new_table) {
 			p->error = -ENOMEM;
 			__set_current_state(TASK_RUNNING);
 			return;
 		}
-		}
-	} 
-
-	if (new_table) { 
 		new_table->entry = new_table->entries;
 		new_table->next = table;
 		p->table = new_table;
@@ -110,6 +113,48 @@ void __pollwait(struct file * filp, wait_queue_head_t * wait_address, poll_table
 
 #define BITS(fds, n)		(*__IN(fds, n)|*__OUT(fds, n)|*__EX(fds, n))
 
+static int max_select_fd(unsigned long n, fd_set_bits *fds)
+{
+	unsigned long *open_fds;
+	unsigned long set;
+	int max;
+
+	/* handle last in-complete long-word first */
+	set = ~(~0UL << (n & (__NFDBITS-1)));
+	n /= __NFDBITS;
+	open_fds = current->files->open_fds->fds_bits+n;
+	max = 0;
+	if (set) {
+		set &= BITS(fds, n);
+		if (set) {
+			if (!(set & ~*open_fds))
+				goto get_max;
+			return -EBADF;
+		}
+	}
+	while (n) {
+		open_fds--;
+		n--;
+		set = BITS(fds, n);
+		if (!set)
+			continue;
+		if (set & ~*open_fds)
+			return -EBADF;
+		if (max)
+			continue;
+get_max:
+		do {
+			max++;
+			set >>= 1;
+		} while (set);
+		max += n * __NFDBITS;
+	}
+
+	return max;
+}
+
+#define BIT(i)		(1UL << ((i)&(__NFDBITS-1)))
+#define MEM(i,m)	((m)+(unsigned)(i)/__NFDBITS)
 #define ISSET(i,m)	(((i)&*(m)) != 0)
 #define SET(i,m)	(*(m) |= (i))
 
@@ -120,52 +165,46 @@ void __pollwait(struct file * filp, wait_queue_head_t * wait_address, poll_table
 int do_select(int n, fd_set_bits *fds, long *timeout)
 {
 	poll_table table, *wait;
-	int retval, off, maxoff;
+	int retval, i, off;
 	long __timeout = *timeout;
 
+ 	read_lock(&current->files->file_lock);
+	retval = max_select_fd(n, fds);
+	read_unlock(&current->files->file_lock);
+
+	if (retval < 0)
+		return retval;
+	n = retval;
+
 	poll_initwait(&table);
 	wait = &table;
 	if (!__timeout)
 		wait = NULL;
-	
 	retval = 0;
-	maxoff = FDS_LONGS(n);
 	for (;;) {
 		set_current_state(TASK_INTERRUPTIBLE);
-		for (off = 0; off <= maxoff; off++) { 
-			unsigned long val = BITS(fds, off); 
-
-			while (val) { 
-				int k = ffz(~val), index; 
-				unsigned long mask, bit;
+		for (i = 0 ; i < n; i++) {
+			unsigned long bit = BIT(i);
+			unsigned long mask;
 			struct file *file;
 
-				bit = (1UL << k); 
-				val &= ~bit; 
-
-				index = off*BITS_PER_LONG + k;
-				if (index >= n) 
-					break;
-
-				file = fget(index);
+			off = i / __NFDBITS;
+			if (!(bit & BITS(fds, off)))
+				continue;
+			file = fget(i);
 			mask = POLLNVAL;
 			if (file) {
 				mask = DEFAULT_POLLMASK;
 				if (file->f_op && file->f_op->poll)
 					mask = file->f_op->poll(file, wait);
 				fput(file);
-				} else { 
-					/* This error will shadow all other results. 
-					 * This matches previous linux behaviour */
-					retval = -EBADF; 
-					goto out; 
 			}
 			if ((mask & POLLIN_SET) && ISSET(bit, __IN(fds,off))) {
 				SET(bit, __RES_IN(fds,off));
 				retval++;
 				wait = NULL;
 			}
-				if ((mask & POLLOUT_SET) && ISSET(bit,__OUT(fds,off))) {
+			if ((mask & POLLOUT_SET) && ISSET(bit, __OUT(fds,off))) {
 				SET(bit, __RES_OUT(fds,off));
 				retval++;
 				wait = NULL;
@@ -176,50 +215,34 @@ int do_select(int n, fd_set_bits *fds, long *timeout)
 				wait = NULL;
 			}
 		}
-		}
-
 		wait = NULL;
 		if (retval || !__timeout || signal_pending(current))
 			break;
-		if (table.error) {
+		if(table.error) {
 			retval = table.error;
 			break;
 		}
 		__timeout = schedule_timeout(__timeout);
 	}
-
-out:	
 	current->state = TASK_RUNNING;
 
 	poll_freewait(&table);
 
 	/*
-	 * Update the caller timeout.
+	 * Up-to-date the caller timeout.
 	 */
 	*timeout = __timeout;
 	return retval;
 }
 
-/*
- * We do a VERIFY_WRITE here even though we are only reading this time:
- * we'll write to it eventually..
- */
+static void *select_bits_alloc(int size)
+{
+	return kmalloc(6 * size, GFP_KERNEL);
+}
 
-static int get_fd_set(unsigned long nr, void *ufdset, unsigned long *fdset)
+static void select_bits_free(void *bits, int size)
 {
-	unsigned long rounded = FDS_BYTES(nr);
-	if (ufdset) {
-		int error = verify_area(VERIFY_WRITE, ufdset, rounded);
-		if (!error && __copy_from_user(fdset, ufdset, rounded))
-			error = -EFAULT;
-		if (nr % __NFDBITS) {
-			unsigned long mask = ~(~0UL << (nr % __NFDBITS)); 
-			fdset[nr/__NFDBITS] &= mask; 
-		} 
-		return error;
-	}
-	memset(fdset, 0, rounded);
-	return 0;
+	kfree(bits);
 }
 
 /*
@@ -237,10 +260,9 @@ asmlinkage long
 sys_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp)
 {
 	fd_set_bits fds;
-	unsigned long *bits;
+	char *bits;
 	long timeout;
 	int ret, size, max_fdset;
-	unsigned long stack_bits[FDS_LONGS(FAST_SELECT_MAX) * 6]; 
 
 	timeout = MAX_SCHEDULE_TIMEOUT;
 	if (tvp) {
@@ -275,27 +297,25 @@ sys_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp)
 	 * since we used fdset we need to allocate memory in units of
 	 * long-words. 
 	 */
-	size = FDS_LONGS(n);
-	bits = stack_bits;
-	if (n >= FAST_SELECT_MAX) { 
 	ret = -ENOMEM;
-		bits = kmalloc(sizeof(unsigned long)*6*size, GFP_KERNEL);
+	size = FDS_BYTES(n);
+	bits = select_bits_alloc(size);
 	if (!bits)
 		goto out_nofds;
-	} 
-
-	fds.in      = bits;
-	fds.out     = bits +   size;
-	fds.ex      = bits + 2*size;
-	fds.res_in  = bits + 3*size;
-	fds.res_out = bits + 4*size;
-	fds.res_ex  = bits + 5*size;
+	fds.in      = (unsigned long *)  bits;
+	fds.out     = (unsigned long *) (bits +   size);
+	fds.ex      = (unsigned long *) (bits + 2*size);
+	fds.res_in  = (unsigned long *) (bits + 3*size);
+	fds.res_out = (unsigned long *) (bits + 4*size);
+	fds.res_ex  = (unsigned long *) (bits + 5*size);
 
 	if ((ret = get_fd_set(n, inp, fds.in)) ||
 	    (ret = get_fd_set(n, outp, fds.out)) ||
 	    (ret = get_fd_set(n, exp, fds.ex)))
 		goto out;
-	memset(fds.res_in, 0, 3*size); 
+	zero_fd_set(n, fds.res_in);
+	zero_fd_set(n, fds.res_out);
+	zero_fd_set(n, fds.res_ex);
 
 	ret = do_select(n, &fds, &timeout);
 
@@ -306,8 +326,8 @@ sys_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp)
 			usec = timeout % HZ;
 			usec *= (1000000/HZ);
 		}
-		__put_user(sec, &tvp->tv_sec);
-		__put_user(usec, &tvp->tv_usec);
+		put_user(sec, &tvp->tv_sec);
+		put_user(usec, &tvp->tv_usec);
 	}
 
 	if (ret < 0)
@@ -324,10 +344,8 @@ sys_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp)
 	set_fd_set(n, exp, fds.res_ex);
 
 out:
-	if (n >= FAST_SELECT_MAX) 
-		kfree(bits);
+	select_bits_free(bits, size);
 out_nofds:
-
 	return ret;
 }
 
@@ -392,39 +410,9 @@ static int do_poll(unsigned int nfds, unsigned int nchunks, unsigned int nleft,
 	return count;
 }
 
-static int fast_poll(poll_table *table, poll_table *wait, struct pollfd *ufds, 
-		     unsigned int nfds, long timeout)
-{ 
-	poll_table *pt = wait; 
-	struct pollfd fds[FAST_POLL_MAX];
-	int count, i; 
-
-	if (copy_from_user(fds, ufds, nfds * sizeof(struct pollfd)))
-		return -EFAULT; 
-	for (;;) { 
-		set_current_state(TASK_INTERRUPTIBLE);
-		count = 0; 
-		do_pollfd(nfds, fds, &pt, &count); 
-		pt = NULL;
-		if (count || !timeout || signal_pending(current))
-			break;
-		count = wait->error; 
-		if (count) 
-			break; 		
-		timeout = schedule_timeout(timeout);
-	} 
-	current->state = TASK_RUNNING;
-	for (i = 0; i < nfds; i++) 
-		__put_user(fds[i].revents, &ufds[i].revents);
-	poll_freewait(table);	
-	if (!count && signal_pending(current)) 
-		return -EINTR; 
-	return count; 
-} 
-
 asmlinkage long sys_poll(struct pollfd * ufds, unsigned int nfds, long timeout)
 {
-	int i, j, err, fdcount;
+	int i, j, fdcount, err;
 	struct pollfd **fds;
 	poll_table table, *wait;
 	int nchunks, nleft;
@@ -441,32 +429,31 @@ asmlinkage long sys_poll(struct pollfd * ufds, unsigned int nfds, long timeout)
 			timeout = MAX_SCHEDULE_TIMEOUT;
 	}
 
-
 	poll_initwait(&table);
 	wait = &table;
 	if (!timeout)
 		wait = NULL;
 
-	if (nfds < FAST_POLL_MAX) 
-		return fast_poll(&table, wait, ufds, nfds, timeout); 
-
 	err = -ENOMEM;
+	fds = NULL;
+	if (nfds != 0) {
 		fds = (struct pollfd **)kmalloc(
 			(1 + (nfds - 1) / POLLFD_PER_PAGE) * sizeof(struct pollfd *),
 			GFP_KERNEL);
 		if (fds == NULL)
 			goto out;
+	}
 
 	nchunks = 0;
 	nleft = nfds;
-	while (nleft > POLLFD_PER_PAGE) { 
+	while (nleft > POLLFD_PER_PAGE) { /* allocate complete PAGE_SIZE chunks */
 		fds[nchunks] = (struct pollfd *)__get_free_page(GFP_KERNEL);
 		if (fds[nchunks] == NULL)
 			goto out_fds;
 		nchunks++;
 		nleft -= POLLFD_PER_PAGE;
 	}
-	if (nleft) { 
+	if (nleft) { /* allocate last PAGE_SIZE chunk, only nleft elements used */
 		fds[nchunks] = (struct pollfd *)__get_free_page(GFP_KERNEL);
 		if (fds[nchunks] == NULL)
 			goto out_fds;
@@ -476,7 +463,6 @@ asmlinkage long sys_poll(struct pollfd * ufds, unsigned int nfds, long timeout)
 	for (i=0; i < nchunks; i++)
 		if (copy_from_user(fds[i], ufds + i*POLLFD_PER_PAGE, PAGE_SIZE))
 			goto out_fds1;
-	
 	if (nleft) {
 		if (copy_from_user(fds[nchunks], ufds + nchunks*POLLFD_PER_PAGE, 
 				nleft * sizeof(struct pollfd)))
@@ -503,6 +489,7 @@ asmlinkage long sys_poll(struct pollfd * ufds, unsigned int nfds, long timeout)
 out_fds:
 	for (i=0; i < nchunks; i++)
 		free_page((unsigned long)(fds[i]));
+	if (nfds != 0)
 		kfree(fds);
 out:
 	poll_freewait(&table);

include/linux/poll.h

@@ -10,32 +10,13 @@
 #include <linux/mm.h>
 #include <asm/uaccess.h>
 
-#define POLL_INLINE_BYTES 256
-#define FAST_SELECT_MAX  128
-#define FAST_POLL_MAX    128
-#define POLL_INLINE_ENTRIES (1+(POLL_INLINE_BYTES / sizeof(struct poll_table_entry)))
-
-struct poll_table_entry {
-	struct file * filp;
-	wait_queue_t wait;
-	wait_queue_head_t * wait_address;
-};
-
-struct poll_table_page {
-	struct poll_table_page * next;
-	struct poll_table_entry * entry;
-	struct poll_table_entry entries[0];
-};
+struct poll_table_page;
 
 typedef struct poll_table_struct {
 	int error;
 	struct poll_table_page * table;
-	struct poll_table_page inline_page; 
-	struct poll_table_entry inline_table[POLL_INLINE_ENTRIES]; 
 } poll_table;
 
-#define POLL_INLINE_TABLE_LEN (sizeof(poll_table) - offsetof(poll_table, inline_page))
-
 extern void __pollwait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p);
 
 static inline void poll_wait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p)
@@ -49,7 +30,6 @@ static inline void poll_initwait(poll_table* pt)
 	pt->error = 0;
 	pt->table = NULL;
 }
-
 extern void poll_freewait(poll_table* pt);
 
 
@@ -69,6 +49,27 @@ typedef struct {
 #define FDS_LONGS(nr)	(((nr)+FDS_BITPERLONG-1)/FDS_BITPERLONG)
 #define FDS_BYTES(nr)	(FDS_LONGS(nr)*sizeof(long))
 
+/*
+ * We do a VERIFY_WRITE here even though we are only reading this time:
+ * we'll write to it eventually..
+ *
+ * Use "unsigned long" accesses to let user-mode fd_set's be long-aligned.
+ */
+static inline
+int get_fd_set(unsigned long nr, void *ufdset, unsigned long *fdset)
+{
+	nr = FDS_BYTES(nr);
+	if (ufdset) {
+		int error;
+		error = verify_area(VERIFY_WRITE, ufdset, nr);
+		if (!error && __copy_from_user(fdset, ufdset, nr))
+			error = -EFAULT;
+		return error;
+	}
+	memset(fdset, 0, nr);
+	return 0;
+}
+
 static inline
 void set_fd_set(unsigned long nr, void *ufdset, unsigned long *fdset)
 {
@@ -76,6 +77,12 @@ void set_fd_set(unsigned long nr, void *ufdset, unsigned long *fdset)
 		__copy_to_user(ufdset, fdset, FDS_BYTES(nr));
 }
 
+static inline
+void zero_fd_set(unsigned long nr, unsigned long *fdset)
+{
+	memset(fdset, 0, FDS_BYTES(nr));
+}
+
 extern int do_select(int n, fd_set_bits *fds, long *timeout);
 
 #endif /* KERNEL */