v2.5.1 -> v2.5.1.1

- me: revert the "kill(-1..)" change.  POSIX isn't that clear on the
issue anyway, and the new behaviour breaks things.
- Jens Axboe: more bio updates
- Al Viro: rd_load cleanups. hpfs mount fix, mount cleanups
- Ingo Molnar: more raid updates
- Jakub Jelinek: fix Linux/x86 confusion about arg passing of "save_v86_state" and "do_signal"
- Trond Myklebust: fix NFS client race conditions

Documentation/DocBook/kernel-hacking.tmpl

@@ -18,8 +18,8 @@
   </authorgroup>
 
   <copyright>
-   <year>2000</year>
-   <holder>Paul Russell</holder>
+   <year>2001</year>
+   <holder>Rusty Russell</holder>
   </copyright>
 
   <legalnotice>
@@ -651,6 +651,29 @@ printk(KERN_INFO "my ip: %d.%d.%d.%d\n", NIPQUAD(ipaddress));
    </para> 
   </sect1>
  
+  <sect1 id="routines-endian">
+   <title><function>cpu_to_be32()</function>/<function>be32_to_cpu()</function>/<function>cpu_to_le32()</function>/<function>le32_to_cpu()</function>
+     <filename class=headerfile>include/asm/byteorder.h</filename> 
+   </title>
+
+   <para>
+    The <function>cpu_to_be32()</function> family (where the "32" can
+    be replaced by 64 or 16, and the "be" can be replaced by "le") are
+    the general way to do endian conversions in the kernel: they
+    return the converted value.  All variations supply the reverse as
+    well: <function>be32_to_cpu()</function>, etc.
+   </para>
+
+   <para>
+    There are two major variations of these functions: the pointer
+    variation, such as <function>cpu_to_be32p()</function>, which take
+    a pointer to the given type, and return the converted value.  The
+    other variation is the "in-situ" family, such as
+    <function>cpu_to_be32s()</function>, which convert value referred
+    to by the pointer, and return void.
+   </para> 
+  </sect1>
+
   <sect1 id="routines-local-irqs">
    <title><function>local_irq_save()</function>/<function>local_irq_restore()</function>
     <filename class=headerfile>include/asm/system.h</filename>

MAINTAINERS

@@ -1322,8 +1322,10 @@ W:	http://www.torque.net/sg
 S:	Maintained
 
 SCSI SUBSYSTEM
+P:	Jens Axboe
+M:	axboe@suse.de
 L:	linux-scsi@vger.kernel.org
-S:	Unmaintained
+S:	Maintained
 
 SCSI TAPE DRIVER
 P:	Kai Mkisara

Makefile

 VERSION = 2
 PATCHLEVEL = 5
-SUBLEVEL = 1
-EXTRAVERSION =
+SUBLEVEL = 2
+EXTRAVERSION =-pre1
 
 KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION)
 

arch/i386/kernel/signal.c

@@ -28,7 +28,7 @@
 
 #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
 
-asmlinkage int FASTCALL(do_signal(struct pt_regs *regs, sigset_t *oldset));
+int FASTCALL(do_signal(struct pt_regs *regs, sigset_t *oldset));
 
 int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from)
 {

arch/i386/kernel/vm86.c

@@ -62,7 +62,7 @@
         ( (unsigned)( & (((struct kernel_vm86_regs *)0)->VM86_REGS_PART2) ) )
 #define VM86_REGS_SIZE2 (sizeof(struct kernel_vm86_regs) - VM86_REGS_SIZE1)
 
-asmlinkage struct pt_regs * FASTCALL(save_v86_state(struct kernel_vm86_regs * regs));
+struct pt_regs * FASTCALL(save_v86_state(struct kernel_vm86_regs * regs));
 struct pt_regs * save_v86_state(struct kernel_vm86_regs * regs)
 {
 	struct tss_struct *tss;

arch/ppc/kernel/m8xx_setup.c

@@ -55,12 +55,6 @@ unsigned char __res[sizeof(bd_t)];
 
 extern void m8xx_ide_init(void);
 
-#ifdef CONFIG_BLK_DEV_RAM
-extern int rd_doload;		/* 1 = load ramdisk, 0 = don't load */
-extern int rd_prompt;		/* 1 = prompt for ramdisk, 0 = don't prompt */
-extern int rd_image_start;	/* starting block # of image */
-#endif
-
 extern unsigned long find_available_memory(void);
 extern void m8xx_cpm_reset(uint);
 

arch/ppc/kernel/prep_setup.c

@@ -111,12 +111,6 @@ extern unsigned long Hash_size, Hash_mask;
 extern int probingmem;
 extern unsigned long loops_per_jiffy;
 
-#ifdef CONFIG_BLK_DEV_RAM
-extern int rd_doload;		/* 1 = load ramdisk, 0 = don't load */
-extern int rd_prompt;		/* 1 = prompt for ramdisk, 0 = don't prompt */
-extern int rd_image_start;	/* starting block # of image */
-#endif
-
 #ifdef CONFIG_SOUND_MODULE
 EXPORT_SYMBOL(ppc_cs4232_dma);
 EXPORT_SYMBOL(ppc_cs4232_dma2);

drivers/block/block_ioctl.c

@@ -35,7 +35,7 @@ int blk_do_rq(request_queue_t *q, struct request *rq)
 	DECLARE_COMPLETION(wait);
 	int err = 0;
 
-	rq->flags |= REQ_BARRIER;
+	rq->flags |= REQ_NOMERGE;
 	rq->waiting = &wait;
 	elv_add_request(q, rq, 1);
 	generic_unplug_device(q);
@@ -81,3 +81,5 @@ int block_ioctl(kdev_t dev, unsigned int cmd, unsigned long arg)
 #endif
 	return err;
 }
+
+EXPORT_SYMBOL(block_ioctl);

drivers/block/elevator.c

@@ -53,7 +53,7 @@ inline int bio_rq_in_between(struct bio *bio, struct request *rq,
 	 * if .next is a valid request
 	 */
 	next = rq->queuelist.next;
-	if (next == head)
+	if (unlikely(next == head))
 		return 0;
 
 	next_rq = list_entry(next, struct request, queuelist);
@@ -121,20 +121,52 @@ inline int elv_rq_merge_ok(struct request *rq, struct bio *bio)
 	return 0;
 }
 
-int elevator_linus_merge(request_queue_t *q, struct request **req,
-			 struct list_head *head, struct bio *bio)
+inline int elv_try_merge(struct request *__rq, struct bio *bio)
 {
 	unsigned int count = bio_sectors(bio);
-	struct list_head *entry = &q->queue_head;
-	int ret = ELEVATOR_NO_MERGE;
+
+	if (!elv_rq_merge_ok(__rq, bio))
+		return ELEVATOR_NO_MERGE;
+
+	/*
+	 * we can merge and sequence is ok, check if it's possible
+	 */
+	if (__rq->sector + __rq->nr_sectors == bio->bi_sector) {
+		return ELEVATOR_BACK_MERGE;
+	} else if (__rq->sector - count == bio->bi_sector) {
+		__rq->elevator_sequence -= count;
+		return ELEVATOR_FRONT_MERGE;
+	}
+
+	return ELEVATOR_NO_MERGE;
+}
+
+int elevator_linus_merge(request_queue_t *q, struct request **req,
+			 struct bio *bio)
+{
+	struct list_head *entry;
 	struct request *__rq;
+	int ret;
+
+	/*
+	 * give a one-shot try to merging with the last touched
+	 * request
+	 */
+	if (q->last_merge) {
+		__rq = list_entry_rq(q->last_merge);
+		BUG_ON(__rq->flags & REQ_STARTED);
+
+		if ((ret = elv_try_merge(__rq, bio))) {
+			*req = __rq;
+			return ret;
+		}
+	}
 
 	entry = &q->queue_head;
-	while ((entry = entry->prev) != head) {
+	ret = ELEVATOR_NO_MERGE;
+	while ((entry = entry->prev) != &q->queue_head) {
 		__rq = list_entry_rq(entry);
 
-		prefetch(list_entry_rq(entry->prev));
-
 		/*
 		 * simply "aging" of requests in queue
 		 */
@@ -144,26 +176,15 @@ int elevator_linus_merge(request_queue_t *q, struct request **req,
 			break;
 		if (!(__rq->flags & REQ_CMD))
 			continue;
+		if (__rq->elevator_sequence < 0)
+			break;
 
 		if (!*req && bio_rq_in_between(bio, __rq, &q->queue_head))
 			*req = __rq;
-		if (!elv_rq_merge_ok(__rq, bio))
-			continue;
 
-		if (__rq->elevator_sequence < count)
-			break;
-
-		/*
-		 * we can merge and sequence is ok, check if it's possible
-		 */
-		if (__rq->sector + __rq->nr_sectors == bio->bi_sector) {
-			ret = ELEVATOR_BACK_MERGE;
-			*req = __rq;
-			break;
-		} else if (__rq->sector - count == bio->bi_sector) {
-			ret = ELEVATOR_FRONT_MERGE;
-			__rq->elevator_sequence -= count;
+		if ((ret = elv_try_merge(__rq, bio))) {
 			*req = __rq;
+			q->last_merge = &__rq->queuelist;
 			break;
 		}
 	}
@@ -183,7 +204,6 @@ void elevator_linus_merge_cleanup(request_queue_t *q, struct request *req, int c
 	entry = &req->queuelist;
 	while ((entry = entry->next) != &q->queue_head) {
 		struct request *tmp;
-		prefetch(list_entry_rq(entry->next));
 		tmp = list_entry_rq(entry);
 		tmp->elevator_sequence -= count;
 	}
@@ -199,12 +219,20 @@ void elv_add_request_fn(request_queue_t *q, struct request *rq,
 			       struct list_head *insert_here)
 {
 	list_add(&rq->queuelist, insert_here);
+
+	/*
+	 * new merges must not precede this barrier
+	 */
+	if (rq->flags & REQ_BARRIER)
+		q->last_merge = NULL;
+	else if (!q->last_merge)
+		q->last_merge = &rq->queuelist;
 }
 
 struct request *elv_next_request_fn(request_queue_t *q)
 {
 	if (!blk_queue_empty(q))
-		return list_entry(q->queue_head.next, struct request, queuelist);
+		return list_entry_rq(q->queue_head.next);
 
 	return NULL;
 }
@@ -222,17 +250,24 @@ void elv_linus_exit(request_queue_t *q, elevator_t *e)
  * See if we can find a request that this buffer can be coalesced with.
  */
 int elevator_noop_merge(request_queue_t *q, struct request **req,
-			struct list_head *head, struct bio *bio)
+			struct bio *bio)
 {
-	unsigned int count = bio_sectors(bio);
 	struct list_head *entry = &q->queue_head;
 	struct request *__rq;
+	int ret;
 
-	entry = &q->queue_head;
-	while ((entry = entry->prev) != head) {
-		__rq = list_entry_rq(entry);
+	if (q->last_merge) {
+		__rq = list_entry_rq(q->last_merge);
+		BUG_ON(__rq->flags & REQ_STARTED);
+
+		if ((ret = elv_try_merge(__rq, bio))) {
+			*req = __rq;
+			return ret;
+		}
+	}
 
-		prefetch(list_entry_rq(entry->prev));
+	while ((entry = entry->prev) != &q->queue_head) {
+		__rq = list_entry_rq(entry);
 
 		if (__rq->flags & (REQ_BARRIER | REQ_STARTED))
 			break;
@@ -240,18 +275,10 @@ int elevator_noop_merge(request_queue_t *q, struct request **req,
 		if (!(__rq->flags & REQ_CMD))
 			continue;
 
-		if (!elv_rq_merge_ok(__rq, bio))
-			continue;
-
-		/*
-		 * we can merge and sequence is ok, check if it's possible
-		 */
-		if (__rq->sector + __rq->nr_sectors == bio->bi_sector) {
-			*req = __rq;
-			return ELEVATOR_BACK_MERGE;
-		} else if (__rq->sector - count == bio->bi_sector) {
+		if ((ret = elv_try_merge(__rq, bio))) {
 			*req = __rq;
-			return ELEVATOR_FRONT_MERGE;
+			q->last_merge = &__rq->queuelist;
+			return ret;
 		}
 	}
 
@@ -267,6 +294,7 @@ int elevator_init(request_queue_t *q, elevator_t *e, elevator_t type)
 	*e = type;
 
 	INIT_LIST_HEAD(&q->queue_head);
+	q->last_merge = NULL;
 
 	if (e->elevator_init_fn)
 		return e->elevator_init_fn(q, e);

drivers/block/ll_rw_blk.c

@@ -117,6 +117,11 @@ inline request_queue_t *blk_get_queue(kdev_t dev)
 		return &blk_dev[MAJOR(dev)].request_queue;
 }
 
+void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn)
+{
+	q->prep_rq_fn = pfn;
+}
+
 /**
  * blk_queue_make_request - define an alternate make_request function for a device
  * @q:  the request queue for the device to be affected
@@ -179,7 +184,6 @@ void blk_queue_bounce_limit(request_queue_t *q, u64 dma_addr)
 	if (dma_addr == BLK_BOUNCE_ISA) {
 		init_emergency_isa_pool();
 		q->bounce_gfp = GFP_NOIO | GFP_DMA;
-		printk("isa pfn %lu, max low %lu, max %lu\n", bounce_pfn, blk_max_low_pfn, blk_max_pfn);
 	} else
 		q->bounce_gfp = GFP_NOHIGHIO;
 
@@ -319,7 +323,7 @@ void blk_dump_rq_flags(struct request *rq, char *msg)
 /*
  * standard prep_rq_fn that builds 10 byte cmds
  */
-static int ll_10byte_cmd_build(request_queue_t *q, struct request *rq)
+int ll_10byte_cmd_build(request_queue_t *q, struct request *rq)
 {
 	int hard_sect = get_hardsect_size(rq->rq_dev);
 	sector_t block = rq->hard_sector / (hard_sect >> 9);
@@ -477,7 +481,7 @@ int blk_rq_map_sg(request_queue_t *q, struct request *rq, struct scatterlist *sg
 				sg[nsegs - 1].length += nbytes;
 			} else {
 new_segment:
-				sg[nsegs].address = NULL;
+				memset(&sg[nsegs],0,sizeof(struct scatterlist));
 				sg[nsegs].page = bvec->bv_page;
 				sg[nsegs].length = nbytes;
 				sg[nsegs].offset = bvec->bv_offset;
@@ -540,11 +544,11 @@ static int ll_back_merge_fn(request_queue_t *q, struct request *req,
 {
 	if (req->nr_sectors + bio_sectors(bio) > q->max_sectors) {
 		req->flags |= REQ_NOMERGE;
+		q->last_merge = NULL;
 		return 0;
 	}
 
-	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail),
-				  __BVEC_START(bio)))
+	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(req->biotail), __BVEC_START(bio)))
 		return ll_new_mergeable(q, req, bio);
 
 	return ll_new_hw_segment(q, req, bio);
@@ -555,11 +559,11 @@ static int ll_front_merge_fn(request_queue_t *q, struct request *req,
 {
 	if (req->nr_sectors + bio_sectors(bio) > q->max_sectors) {
 		req->flags |= REQ_NOMERGE;
+		q->last_merge = NULL;
 		return 0;
 	}
 
-	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio),
-				  __BVEC_START(req->bio)))
+	if (BIOVEC_VIRT_MERGEABLE(__BVEC_END(bio), __BVEC_START(req->bio)))
 		return ll_new_mergeable(q, req, bio);
 
 	return ll_new_hw_segment(q, req, bio);
@@ -568,7 +572,7 @@ static int ll_front_merge_fn(request_queue_t *q, struct request *req,
 static int ll_merge_requests_fn(request_queue_t *q, struct request *req,
 				struct request *next)
 {
-	int total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
+	int total_phys_segments = req->nr_phys_segments +next->nr_phys_segments;
 	int total_hw_segments = req->nr_hw_segments + next->nr_hw_segments;
 
 	/*
@@ -599,6 +603,9 @@ static int ll_merge_requests_fn(request_queue_t *q, struct request *req,
 		return 0;
 
 	/* Merge is OK... */
+	if (q->last_merge == &next->queuelist)
+		q->last_merge = NULL;
+
 	req->nr_phys_segments = total_phys_segments;
 	req->nr_hw_segments = total_hw_segments;
 	return 1;
@@ -799,7 +806,7 @@ int blk_init_queue(request_queue_t *q, request_fn_proc *rfn, spinlock_t *lock)
 	q->back_merge_fn       	= ll_back_merge_fn;
 	q->front_merge_fn      	= ll_front_merge_fn;
 	q->merge_requests_fn	= ll_merge_requests_fn;
-	q->prep_rq_fn		= ll_10byte_cmd_build;
+	q->prep_rq_fn		= NULL;
 	q->plug_tq.sync		= 0;
 	q->plug_tq.routine	= &generic_unplug_device;
 	q->plug_tq.data		= q;
@@ -1039,13 +1046,11 @@ static inline void attempt_back_merge(request_queue_t *q, struct request *rq)
 		attempt_merge(q, rq);
 }
 
-static inline void attempt_front_merge(request_queue_t *q,
-				       struct list_head *head,
-				       struct request *rq)
+static inline void attempt_front_merge(request_queue_t *q, struct request *rq)
 {
 	struct list_head *prev = rq->queuelist.prev;
 
-	if (prev != head)
+	if (prev != &q->queue_head)
 		attempt_merge(q, blkdev_entry_to_request(prev));
 }
 
@@ -1081,7 +1086,7 @@ static int __make_request(request_queue_t *q, struct bio *bio)
 {
 	struct request *req, *freereq = NULL;
 	int el_ret, latency = 0, rw, nr_sectors, cur_nr_sectors, barrier;
-	struct list_head *head, *insert_here;
+	struct list_head *insert_here = &q->queue_head;
 	elevator_t *elevator = &q->elevator;
 	sector_t sector;
 
@@ -1102,24 +1107,18 @@ static int __make_request(request_queue_t *q, struct bio *bio)
 	latency = elevator_request_latency(elevator, rw);
 	barrier = test_bit(BIO_RW_BARRIER, &bio->bi_rw);
 
-again:
-	req = NULL;
-	head = &q->queue_head;
-
 	spin_lock_irq(q->queue_lock);
 
-	insert_here = head->prev;
 	if (blk_queue_empty(q) || barrier) {
 		blk_plug_device(q);
 		goto get_rq;
-	} else if ((req = __elv_next_request(q))) {
-		if (req->flags & REQ_STARTED)
-			head = head->next;
+	}
 
+again:
 	req = NULL;
-	}
+	insert_here = q->queue_head.prev;
 
-	el_ret = elevator->elevator_merge_fn(q, &req, head, bio);
+	el_ret = elevator->elevator_merge_fn(q, &req, bio);
 	switch (el_ret) {
 		case ELEVATOR_BACK_MERGE:
 			BUG_ON(req->flags & REQ_STARTED);
@@ -1157,7 +1156,7 @@ static int __make_request(request_queue_t *q, struct bio *bio)
 			req->sector = req->hard_sector = sector;
 			req->nr_sectors = req->hard_nr_sectors += nr_sectors;
 			drive_stat_acct(req, nr_sectors, 0);
-			attempt_front_merge(q, head, req);
+			attempt_front_merge(q, req);
 			goto out;
 
 		/*
@@ -1188,7 +1187,6 @@ static int __make_request(request_queue_t *q, struct bio *bio)
 		req = freereq;
 		freereq = NULL;
 	} else if ((req = get_request(q, rw)) == NULL) {
-
 		spin_unlock_irq(q->queue_lock);
 
 		/*
@@ -1200,6 +1198,7 @@ static int __make_request(request_queue_t *q, struct bio *bio)
 		}
 
 		freereq = get_request_wait(q, rw);
+		spin_lock_irq(q->queue_lock);
 		goto again;
 	}
 
@@ -1755,3 +1754,6 @@ EXPORT_SYMBOL(submit_bio);
 EXPORT_SYMBOL(blk_queue_assign_lock);
 EXPORT_SYMBOL(blk_phys_contig_segment);
 EXPORT_SYMBOL(blk_hw_contig_segment);
+
+EXPORT_SYMBOL(ll_10byte_cmd_build);
+EXPORT_SYMBOL(blk_queue_prep_rq);

drivers/block/loop.c

@@ -1007,6 +1007,7 @@ int __init loop_init(void)
 		goto out_mem;
 
 	blk_queue_make_request(BLK_DEFAULT_QUEUE(MAJOR_NR), loop_make_request);
+	blk_queue_bounce_limit(BLK_DEFAULT_QUEUE(MAJOR_NR), BLK_BOUNCE_HIGH);
 
 	for (i = 0; i < max_loop; i++) {
 		struct loop_device *lo = &loop_dev[i];

drivers/block/rd.c

@@ -43,26 +43,12 @@
  */
 
 #include <linux/config.h>
-#include <linux/sched.h>
-#include <linux/fs.h>
-#include <linux/kernel.h>
-#include <linux/hdreg.h>
 #include <linux/string.h>
-#include <linux/mm.h>
-#include <linux/mman.h>
 #include <linux/slab.h>
-#include <linux/ioctl.h>
-#include <linux/fd.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/devfs_fs_kernel.h>
-#include <linux/smp_lock.h>
-
-#include <asm/system.h>
 #include <asm/uaccess.h>
-#include <asm/byteorder.h>
-
-extern void wait_for_keypress(void);
 
 /*
  * 35 has been officially registered as the RAMDISK major number, but
@@ -79,6 +65,8 @@ extern void wait_for_keypress(void);
 #ifdef CONFIG_BLK_DEV_INITRD
 static int initrd_users;
 static spinlock_t initrd_users_lock = SPIN_LOCK_UNLOCKED;
+unsigned long initrd_start, initrd_end;
+int initrd_below_start_ok;
 #endif
 
 /* Various static variables go here.  Most are used only in the RAM disk code.
@@ -111,70 +99,6 @@ int rd_size = CONFIG_BLK_DEV_RAM_SIZE;		/* Size of the RAM disks */
  */
 int rd_blocksize = BLOCK_SIZE;			/* blocksize of the RAM disks */
 
-#ifndef MODULE
-
-int rd_doload;			/* 1 = load RAM disk, 0 = don't load */
-int rd_prompt = 1;		/* 1 = prompt for RAM disk, 0 = don't prompt */
-int rd_image_start;		/* starting block # of image */
-#ifdef CONFIG_BLK_DEV_INITRD
-unsigned long initrd_start, initrd_end;
-int mount_initrd = 1;		/* zero if initrd should not be mounted */
-int initrd_below_start_ok;
-
-static int __init no_initrd(char *str)
-{
-	mount_initrd = 0;
-	return 1;
-}
-
-__setup("noinitrd", no_initrd);
-
-#endif
-
-static int __init ramdisk_start_setup(char *str)
-{
-	rd_image_start = simple_strtol(str,NULL,0);
-	return 1;
-}
-
-static int __init load_ramdisk(char *str)
-{
-	rd_doload = simple_strtol(str,NULL,0) & 3;
-	return 1;
-}
-
-static int __init prompt_ramdisk(char *str)
-{
-	rd_prompt = simple_strtol(str,NULL,0) & 1;
-	return 1;
-}
-
-static int __init ramdisk_size(char *str)
-{
-	rd_size = simple_strtol(str,NULL,0);
-	return 1;
-}
-
-static int __init ramdisk_size2(char *str)
-{
-	return ramdisk_size(str);
-}
-
-static int __init ramdisk_blocksize(char *str)
-{
-	rd_blocksize = simple_strtol(str,NULL,0);
-	return 1;
-}
-
-__setup("ramdisk_start=", ramdisk_start_setup);
-__setup("load_ramdisk=", load_ramdisk);
-__setup("prompt_ramdisk=", prompt_ramdisk);
-__setup("ramdisk=", ramdisk_size);
-__setup("ramdisk_size=", ramdisk_size2);
-__setup("ramdisk_blocksize=", ramdisk_blocksize);
-
-#endif
-
 /*
  * Copyright (C) 2000 Linus Torvalds.
  *               2000 Transmeta Corp.
@@ -492,7 +416,7 @@ static void __exit rd_cleanup (void)
 }
 
 /* This is the registration and initialization section of the RAM disk driver */
-int __init rd_init (void)
+static int __init rd_init (void)
 {
 	int		i;
 
@@ -548,7 +472,28 @@ int __init rd_init (void)
 module_init(rd_init);
 module_exit(rd_cleanup);
 
-/* loadable module support */
+/* options - nonmodular */
+#ifndef MODULE
+static int __init ramdisk_size(char *str)
+{
+	rd_size = simple_strtol(str,NULL,0);
+	return 1;
+}
+static int __init ramdisk_size2(char *str)	/* kludge */
+{
+	return ramdisk_size(str);
+}
+static int __init ramdisk_blocksize(char *str)
+{
+	rd_blocksize = simple_strtol(str,NULL,0);
+	return 1;
+}
+__setup("ramdisk=", ramdisk_size);
+__setup("ramdisk_size=", ramdisk_size2);
+__setup("ramdisk_blocksize=", ramdisk_blocksize);
+#endif
+
+/* options - modular */
 MODULE_PARM     (rd_size, "1i");
 MODULE_PARM_DESC(rd_size, "Size of each RAM disk in kbytes.");
 MODULE_PARM     (rd_blocksize, "i");

drivers/ide/amd74xx.c

@@ -459,6 +459,8 @@ void __init ide_init_amd74xx (ide_hwif_t *hwif)
 	hwif->tuneproc = &amd74xx_tune_drive;
 	hwif->speedproc = &amd74xx_tune_chipset;
 
+	hwif->highmem = 1;
+
 #ifndef CONFIG_BLK_DEV_IDEDMA
 	hwif->drives[0].autotune = 1;
 	hwif->drives[1].autotune = 1;

drivers/ide/ide-cd.c

@@ -594,7 +594,7 @@ static int cdrom_decode_status (ide_startstop_t *startstop, ide_drive_t *drive,
 		cdrom_end_request (1, drive);
 		*startstop = ide_error (drive, "request sense failure", stat);
 		return 1;
-	} else if (rq->flags & REQ_PC) {
+	} else if (rq->flags & (REQ_PC | REQ_BLOCK_PC)) {
 		/* All other functions, except for READ. */
 		struct completion *wait = NULL;
 		pc = (struct packet_command *) rq->special;
@@ -2675,6 +2675,8 @@ int ide_cdrom_setup (ide_drive_t *drive)
 	set_blocksize(MKDEV(HWIF(drive)->major, minor), CD_FRAMESIZE);
 	blk_queue_hardsect_size(&drive->queue, CD_FRAMESIZE);
 
+	blk_queue_prep_rq(&drive->queue, ll_10byte_cmd_build);
+
 	drive->special.all	= 0;
 	drive->ready_stat	= 0;
 

drivers/ide/ide-floppy.c

@@ -336,23 +336,7 @@ typedef struct {
 #define	IDEFLOPPY_IOCTL_FORMAT_START		0x4602
 #define IDEFLOPPY_IOCTL_FORMAT_GET_PROGRESS	0x4603
 
-/*
- *	Special requests for our block device strategy routine.
- */
-#define	IDEFLOPPY_FIRST_RQ		90
-
-/*
- * 	IDEFLOPPY_PC_RQ is used to queue a packet command in the request queue.
- */
-#define	IDEFLOPPY_PC_RQ			90
-
-#define IDEFLOPPY_LAST_RQ		90
-
-/*
- *	A macro which can be used to check if a given request command
- *	originated in the driver or in the buffer cache layer.
- */
-#define IDEFLOPPY_RQ_CMD(cmd) 		((cmd >= IDEFLOPPY_FIRST_RQ) && (cmd <= IDEFLOPPY_LAST_RQ))
+#define IDEFLOPPY_RQ			(REQ_SPECIAL)
 
 /*
  *	Error codes which are returned in rq->errors to the higher part
@@ -696,7 +680,7 @@ static void idefloppy_end_request (byte uptodate, ide_hwgroup_t *hwgroup)
 	/* Why does this happen? */
 	if (!rq)
 		return;
-	if (!IDEFLOPPY_RQ_CMD (rq->cmd)) {
+	if (rq->flags & IDEFLOPPY_RQ) {
 		ide_end_request (uptodate, hwgroup);
 		return;
 	}
@@ -776,7 +760,7 @@ static void idefloppy_queue_pc_head (ide_drive_t *drive,idefloppy_pc_t *pc,struc
 {
 	ide_init_drive_cmd (rq);
 	rq->buffer = (char *) pc;
-	rq->cmd = IDEFLOPPY_PC_RQ;
+	rq->flags = IDEFLOPPY_RQ;
 	(void) ide_do_drive_cmd (drive, rq, ide_preempt);
 }
 
@@ -1192,6 +1176,7 @@ static void idefloppy_create_rw_cmd (idefloppy_floppy_t *floppy, idefloppy_pc_t
 {
 	int block = sector / floppy->bs_factor;
 	int blocks = rq->nr_sectors / floppy->bs_factor;
+	int cmd = rq_data_dir(rq);
 	
 #if IDEFLOPPY_DEBUG_LOG
 	printk ("create_rw1%d_cmd: block == %d, blocks == %d\n",
@@ -1200,18 +1185,18 @@ static void idefloppy_create_rw_cmd (idefloppy_floppy_t *floppy, idefloppy_pc_t
 
 	idefloppy_init_pc (pc);
 	if (test_bit (IDEFLOPPY_USE_READ12, &floppy->flags)) {
-		pc->c[0] = rq->cmd == READ ? IDEFLOPPY_READ12_CMD : IDEFLOPPY_WRITE12_CMD;
+		pc->c[0] = cmd == READ ? IDEFLOPPY_READ12_CMD : IDEFLOPPY_WRITE12_CMD;
 		put_unaligned (htonl (blocks), (unsigned int *) &pc->c[6]);
 	} else {
-		pc->c[0] = rq->cmd == READ ? IDEFLOPPY_READ10_CMD : IDEFLOPPY_WRITE10_CMD;
+		pc->c[0] = cmd == READ ? IDEFLOPPY_READ10_CMD : IDEFLOPPY_WRITE10_CMD;
 		put_unaligned (htons (blocks), (unsigned short *) &pc->c[7]);
 	}
 	put_unaligned (htonl (block), (unsigned int *) &pc->c[2]);
 	pc->callback = &idefloppy_rw_callback;
 	pc->rq = rq;
 	pc->b_data = rq->buffer;
-	pc->b_count = rq->cmd == READ ? 0 : rq->bio->bi_size;
-	if (rq->cmd == WRITE)
+	pc->b_count = cmd == READ ? 0 : rq->bio->bi_size;
+	if (rq->flags & REQ_RW)
 		set_bit (PC_WRITING, &pc->flags);
 	pc->buffer = NULL;
 	pc->request_transfer = pc->buffer_size = blocks * floppy->block_size;
@@ -1227,8 +1212,8 @@ static ide_startstop_t idefloppy_do_request (ide_drive_t *drive, struct request
 	idefloppy_pc_t *pc;
 
 #if IDEFLOPPY_DEBUG_LOG
-	printk (KERN_INFO "rq_status: %d, rq_dev: %u, cmd: %d, errors: %d\n",rq->rq_status,(unsigned int) rq->rq_dev,rq->cmd,rq->errors);
-	printk (KERN_INFO "sector: %ld, nr_sectors: %ld, current_nr_sectors: %ld\n",rq->sector,rq->nr_sectors,rq->current_nr_sectors);
+	printk (KERN_INFO "rq_status: %d, rq_dev: %u, flags: %lx, errors: %d\n",rq->rq_status,(unsigned int) rq->rq_dev,rq->flags,rq->errors);
+	printk (KERN_INFO "sector: %ld, nr_sectors: %ld, current_nr_sectors: %d\n",rq->sector,rq->nr_sectors,rq->current_nr_sectors);
 #endif /* IDEFLOPPY_DEBUG_LOG */
 
 	if (rq->errors >= ERROR_MAX) {
@@ -1240,9 +1225,7 @@ static ide_startstop_t idefloppy_do_request (ide_drive_t *drive, struct request
 		idefloppy_end_request (0, HWGROUP(drive));
 		return ide_stopped;
 	}
-	switch (rq->cmd) {
-		case READ:
-		case WRITE:
+	if (rq->flags & REQ_CMD) {
 		if (rq->sector % floppy->bs_factor || rq->nr_sectors % floppy->bs_factor) {
 			printk ("%s: unsupported r/w request size\n", drive->name);
 			idefloppy_end_request (0, HWGROUP(drive));
@@ -1250,12 +1233,10 @@ static ide_startstop_t idefloppy_do_request (ide_drive_t *drive, struct request
 		}
 		pc = idefloppy_next_pc_storage (drive);
 		idefloppy_create_rw_cmd (floppy, pc, rq, block);
-			break;
-		case IDEFLOPPY_PC_RQ:
+	} else if (rq->flags & IDEFLOPPY_RQ) {
 		pc = (idefloppy_pc_t *) rq->buffer;
-			break;
-		default:
-			printk (KERN_ERR "ide-floppy: unsupported command %x in request queue\n", rq->cmd);
+	} else {
+		blk_dump_rq_flags(rq, "ide-floppy: unsupported command in queue");
 		idefloppy_end_request (0,HWGROUP (drive));
 		return ide_stopped;
 	}
@@ -1273,7 +1254,7 @@ static int idefloppy_queue_pc_tail (ide_drive_t *drive,idefloppy_pc_t *pc)
 
 	ide_init_drive_cmd (&rq);
 	rq.buffer = (char *) pc;
-	rq.cmd = IDEFLOPPY_PC_RQ;
+	rq.flags = IDEFLOPPY_RQ;
 	return ide_do_drive_cmd (drive, &rq, ide_wait);
 }
 

drivers/md/md.c

@@ -66,7 +66,7 @@ static mdk_personality_t *pers[MAX_PERSONALITY];
 
 /*
  * Current RAID-1,4,5 parallel reconstruction 'guaranteed speed limit'
- * is 100 KB/sec, so the extra system load does not show up that much.
+ * is 1000 KB/sec, so the extra system load does not show up that much.
  * Increase it if you want to have more _guaranteed_ speed. Note that
  * the RAID driver will use the maximum available bandwith if the IO
  * subsystem is idle. There is also an 'absolute maximum' reconstruction
@@ -76,8 +76,8 @@ static mdk_personality_t *pers[MAX_PERSONALITY];
  * you can change it via /proc/sys/dev/raid/speed_limit_min and _max.
  */
 
-static int sysctl_speed_limit_min = 100;
-static int sysctl_speed_limit_max = 100000;
+static int sysctl_speed_limit_min = 1000;
+static int sysctl_speed_limit_max = 200000;
 
 static struct ctl_table_header *raid_table_header;
 
@@ -3336,7 +3336,7 @@ void md_done_sync(mddev_t *mddev, int blocks, int ok)
 int md_do_sync(mddev_t *mddev, mdp_disk_t *spare)
 {
 	mddev_t *mddev2;
-	unsigned int max_sectors, currspeed,
+	unsigned int max_sectors, currspeed = 0,
 		j, window, err, serialize;
 	unsigned long mark[SYNC_MARKS];
 	unsigned long mark_cnt[SYNC_MARKS];
@@ -3376,8 +3376,7 @@ int md_do_sync(mddev_t *mddev, mdp_disk_t *spare)
 	max_sectors = mddev->sb->size << 1;
 
 	printk(KERN_INFO "md: syncing RAID array md%d\n", mdidx(mddev));
-	printk(KERN_INFO "md: minimum _guaranteed_ reconstruction speed: %d KB/sec/disc.\n",
-						sysctl_speed_limit_min);
+	printk(KERN_INFO "md: minimum _guaranteed_ reconstruction speed: %d KB/sec/disc.\n", sysctl_speed_limit_min);
 	printk(KERN_INFO "md: using maximum available idle IO bandwith "
 	       "(but not more than %d KB/sec) for reconstruction.\n",
 	       sysctl_speed_limit_max);
@@ -3409,7 +3408,7 @@ int md_do_sync(mddev_t *mddev, mdp_disk_t *spare)
 	for (j = 0; j < max_sectors;) {
 		int sectors;
 
-		sectors = mddev->pers->sync_request(mddev, j);
+		sectors = mddev->pers->sync_request(mddev, j, currspeed < sysctl_speed_limit_min);
 		if (sectors < 0) {
 			err = sectors;
 			goto out;

drivers/md/raid1.c

@@ -28,7 +28,6 @@
 #define MD_DRIVER
 #define MD_PERSONALITY
 
-#define MAX_WORK_PER_DISK 128
 /*
  * Number of guaranteed r1bios in case of extreme VM load:
  */
@@ -38,24 +37,6 @@ static mdk_personality_t raid1_personality;
 static spinlock_t retry_list_lock = SPIN_LOCK_UNLOCKED;
 static LIST_HEAD(retry_list_head);
 
-static inline void check_all_w_bios_empty(r1bio_t *r1_bio)
-{
-	int i;
-
-	return;
-	for (i = 0; i < MD_SB_DISKS; i++)
-		if (r1_bio->write_bios[i])
-			BUG();
-}
-
-static inline void check_all_bios_empty(r1bio_t *r1_bio)
-{
-	return;
-	if (r1_bio->read_bio)
-		BUG();
-	check_all_w_bios_empty(r1_bio);
-}
-
 static void * r1bio_pool_alloc(int gfp_flags, void *data)
 {
 	r1bio_t *r1_bio;
@@ -69,11 +50,11 @@ static void * r1bio_pool_alloc(int gfp_flags, void *data)
 
 static void r1bio_pool_free(void *r1_bio, void *data)
 {
-	check_all_bios_empty(r1_bio);
 	kfree(r1_bio);
 }
 
 #define RESYNC_BLOCK_SIZE (64*1024)
+#define RESYNC_SECTORS (RESYNC_BLOCK_SIZE >> 9)
 #define RESYNC_PAGES ((RESYNC_BLOCK_SIZE + PAGE_SIZE-1) / PAGE_SIZE)
 #define RESYNC_WINDOW (2048*1024)
 
@@ -86,7 +67,6 @@ static void * r1buf_pool_alloc(int gfp_flags, void *data)
 	int i, j;
 
 	r1_bio = mempool_alloc(conf->r1bio_pool, gfp_flags);
-	check_all_bios_empty(r1_bio);
 
 	bio = bio_alloc(gfp_flags, RESYNC_PAGES);
 	if (!bio)
@@ -131,7 +111,6 @@ static void r1buf_pool_free(void *__r1_bio, void *data)
 	r1bio_t *r1bio = __r1_bio;
 	struct bio *bio = r1bio->master_bio;
 
-	check_all_bios_empty(r1bio);
 	if (atomic_read(&bio->bi_cnt) != 1)
 		BUG();
 	for (i = 0; i < RESYNC_PAGES; i++) {
@@ -163,13 +142,25 @@ static void put_all_bios(conf_t *conf, r1bio_t *r1_bio)
 		}
 		*bio = NULL;
 	}
-	check_all_bios_empty(r1_bio);
 }
 
 static inline void free_r1bio(r1bio_t *r1_bio)
 {
+	unsigned long flags;
+
 	conf_t *conf = mddev_to_conf(r1_bio->mddev);
 
+	/*
+	 * Wake up any possible resync thread that waits for the device
+	 * to go idle.
+	 */
+	spin_lock_irqsave(&conf->resync_lock, flags);
+	if (!--conf->nr_pending) {
+		wake_up(&conf->wait_idle);
+		wake_up(&conf->wait_resume);
+	}
+	spin_unlock_irqrestore(&conf->resync_lock, flags);
+
 	put_all_bios(conf, r1_bio);
 	mempool_free(r1_bio, conf->r1bio_pool);
 }
@@ -178,7 +169,14 @@ static inline void put_buf(r1bio_t *r1_bio)
 {
 	conf_t *conf = mddev_to_conf(r1_bio->mddev);
 	struct bio *bio = r1_bio->master_bio;
+	unsigned long flags;
 
+	spin_lock_irqsave(&conf->resync_lock, flags);
+	if (!--conf->nr_pending) {
+		wake_up(&conf->wait_idle);
+		wake_up(&conf->wait_resume);
+	}
+	spin_unlock_irqrestore(&conf->resync_lock, flags);
 	/*
 	 * undo any possible partial request fixup magic:
 	 */
@@ -222,37 +220,6 @@ static void reschedule_retry(r1bio_t *r1_bio)
 	md_wakeup_thread(conf->thread);
 }
 
-
-static void inline raid_request_done(unsigned long sector, conf_t *conf, int phase)
-{
-	unsigned long flags;
-	spin_lock_irqsave(&conf->segment_lock, flags);
-	if (sector < conf->start_active)
-		conf->cnt_done--;
-	else if (sector >= conf->start_future && conf->phase == phase)
-		conf->cnt_future--;
-	else if (!--conf->cnt_pending)
-		wake_up(&conf->wait_ready);
-
-	spin_unlock_irqrestore(&conf->segment_lock, flags);
-}
-
-static void inline sync_request_done(sector_t sector, conf_t *conf)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&conf->segment_lock, flags);
-	if (sector >= conf->start_ready)
-		--conf->cnt_ready;
-	else if (sector >= conf->start_active) {
-		if (!--conf->cnt_active) {
-			conf->start_active = conf->start_ready;
-			wake_up(&conf->wait_done);
-		}
-	}
-	spin_unlock_irqrestore(&conf->segment_lock, flags);
-}
-
 /*
  * raid_end_bio_io() is called when we have finished servicing a mirrored
  * operation and are ready to return a success/failure code to the buffer
@@ -262,19 +229,29 @@ static int raid_end_bio_io(r1bio_t *r1_bio, int uptodate, int nr_sectors)
 {
 	struct bio *bio = r1_bio->master_bio;
 
-	raid_request_done(bio->bi_sector, mddev_to_conf(r1_bio->mddev),
-			test_bit(R1BIO_SyncPhase, &r1_bio->state));
-
 	bio_endio(bio, uptodate, nr_sectors);
 	free_r1bio(r1_bio);
 
 	return 0;
 }
 
+/*
+ * Update disk head position estimator based on IRQ completion info.
+ */
+static void inline update_head_pos(int disk, r1bio_t *r1_bio)
+{
+	conf_t *conf = mddev_to_conf(r1_bio->mddev);
+
+	conf->mirrors[disk].head_position =
+		r1_bio->sector + (r1_bio->master_bio->bi_size >> 9);
+	atomic_dec(&conf->mirrors[disk].nr_pending);
+}
+
 static int end_request(struct bio *bio, int nr_sectors)
 {
 	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
 	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
+	int i;
 
 	/*
 	 * this branch is our 'one mirror IO has finished' event handler:
@@ -287,20 +264,16 @@ static int end_request(struct bio *bio, int nr_sectors)
 		 * we will return a good error code for to the higher
 		 * levels even if IO on some other mirrored buffer fails.
 		 *
-		 * The 'master' represents the complex operation to
+		 * The 'master' represents the composite IO operation to
 		 * user-side. So if something waits for IO, then it will
 		 * wait for the 'master' bio.
 		 */
 		set_bit(R1BIO_Uptodate, &r1_bio->state);
 
-	/*
-	 * We split up the read and write side, imho they are
-	 * conceptually different.
-	 */
-
 	if ((r1_bio->cmd == READ) || (r1_bio->cmd == READA)) {
 		if (!r1_bio->read_bio)
 			BUG();
+		update_head_pos(r1_bio->read_disk, r1_bio);
 		/*
 		 * we have only one bio on the read side
 		 */
@@ -322,6 +295,14 @@ static int end_request(struct bio *bio, int nr_sectors)
 	/*
 	 * WRITE:
 	 *
+	 * First, find the disk this bio belongs to.
+	 */
+	for (i = 0; i < MD_SB_DISKS; i++)
+		if (r1_bio->write_bios[i] == bio) {
+			update_head_pos(i, r1_bio);
+			break;
+		}
+	/*
 	 * Let's see if all mirrored write operations have finished
 	 * already.
 	 */
@@ -332,44 +313,49 @@ static int end_request(struct bio *bio, int nr_sectors)
 
 /*
  * This routine returns the disk from which the requested read should
- * be done. It bookkeeps the last read position for every disk
- * in array and when new read requests come, the disk which last
- * position is nearest to the request, is chosen.
+ * be done. There is a per-array 'next expected sequential IO' sector
+ * number - if this matches on the next IO then we use the last disk.
+ * There is also a per-disk 'last know head position' sector that is
+ * maintained from IRQ contexts, both the normal and the resync IO
+ * completion handlers update this position correctly. If there is no
+ * perfect sequential match then we pick the disk whose head is closest.
  *
- * TODO: now if there are 2 mirrors in the same 2 devices, performance
- * degrades dramatically because position is mirror, not device based.
- * This should be changed to be device based. Also atomic sequential
- * reads should be somehow balanced.
+ * If there are 2 mirrors in the same 2 devices, performance degrades
+ * because position is mirror, not device based.
  */
-
 static int read_balance(conf_t *conf, struct bio *bio, r1bio_t *r1_bio)
 {
-	const int sectors = bio->bi_size >> 9;
 	const unsigned long this_sector = r1_bio->sector;
-	unsigned long new_distance, current_distance;
 	int new_disk = conf->last_used, disk = new_disk;
+	const int sectors = bio->bi_size >> 9;
+	sector_t new_distance, current_distance;
 
 	/*
-	 * Check if it is sane at all to balance
+	 * Check if it if we can balance. We can balance on the whole
+	 * device if no resync is going on, or below the resync window.
+	 * We take the first readable disk when above the resync window.
 	 */
-
-	if (conf->resync_mirrors)
+	if (conf->resync_mirrors && (this_sector + sectors >= conf->next_resync)) {
+		/* make sure that disk is operational */
+		new_disk = 0;
+		while (!conf->mirrors[new_disk].operational || conf->mirrors[new_disk].write_only) {
+			new_disk++;
+			if (new_disk == conf->raid_disks) {
+				new_disk = 0;
+				break;
+			}
+		}
 		goto rb_out;
+	}
 
 
-	/* make sure that disk is operational */
-	while( !conf->mirrors[new_disk].operational) {
-		if (new_disk <= 0) new_disk = conf->raid_disks;
+	/* make sure the disk is operational */
+	while (!conf->mirrors[new_disk].operational) {
+		if (new_disk <= 0)
+			new_disk = conf->raid_disks;
 		new_disk--;
 		if (new_disk == disk) {
-			/*
-			 * This means no working disk was found
-			 * Nothing much to do, lets not change anything
-			 * and hope for the best...
-			 */
-
 			new_disk = conf->last_used;
-
 			goto rb_out;
 		}
 	}
@@ -377,36 +363,16 @@ static int read_balance(conf_t *conf, struct bio *bio, r1bio_t *r1_bio)
 	/* now disk == new_disk == starting point for search */
 
 	/*
-	 * Don't touch anything for sequential reads.
+	 * Don't change to another disk for sequential reads:
 	 */
-	if (this_sector == conf->mirrors[new_disk].head_position)
+	if (conf->next_seq_sect == this_sector)
 		goto rb_out;
-
-	/*
-	 * If reads have been done only on a single disk
-	 * for a time, lets give another disk a change.
-	 * This is for kicking those idling disks so that
-	 * they would find work near some hotspot.
-	 */
-	if (conf->sect_count >= conf->mirrors[new_disk].sect_limit) {
-		conf->sect_count = 0;
-
-		do {
-			if (new_disk <= 0)
-				new_disk = conf->raid_disks;
-			new_disk--;
-			if (new_disk == disk)
-				break;
-		} while ((conf->mirrors[new_disk].write_only) ||
-			(!conf->mirrors[new_disk].operational));
-
+	if (this_sector == conf->mirrors[new_disk].head_position)
 		goto rb_out;
-	}
 
-	current_distance = abs(this_sector -
-				conf->mirrors[disk].head_position);
+	current_distance = abs(this_sector - conf->mirrors[disk].head_position);
 
-	/* Find the disk which is closest */
+	/* Find the disk whose head is closest */
 
 	do {
 		if (disk <= 0)
@@ -417,44 +383,56 @@ static int read_balance(conf_t *conf, struct bio *bio, r1bio_t *r1_bio)
 				(!conf->mirrors[disk].operational))
 			continue;
 
-		new_distance = abs(this_sector -
-					conf->mirrors[disk].head_position);
-
+		if (!atomic_read(&conf->mirrors[disk].nr_pending)) {
+			new_disk = disk;
+			break;
+		}
+		new_distance = abs(this_sector - conf->mirrors[disk].head_position);
 		if (new_distance < current_distance) {
-			conf->sect_count = 0;
 			current_distance = new_distance;
 			new_disk = disk;
 		}
 	} while (disk != conf->last_used);
 
 rb_out:
-	conf->mirrors[new_disk].head_position = this_sector + sectors;
+	r1_bio->read_disk = new_disk;
+	conf->next_seq_sect = this_sector + sectors;
 
 	conf->last_used = new_disk;
-	conf->sect_count += sectors;
 
 	return new_disk;
 }
 
 /*
- * Wait if the reconstruction state machine puts up a bar for
- * new requests in this sector range:
+ * Throttle resync depth, so that we can both get proper overlapping of
+ * requests, but are still able to handle normal requests quickly.
  */
-static inline void new_request(conf_t *conf, r1bio_t *r1_bio)
+#define RESYNC_DEPTH 32
+
+static void device_barrier(conf_t *conf, sector_t sect)
+{
+	spin_lock_irq(&conf->resync_lock);
+	wait_event_lock_irq(conf->wait_idle, !waitqueue_active(&conf->wait_resume), conf->resync_lock);
+	
+	if (!conf->barrier++) {
+		wait_event_lock_irq(conf->wait_idle, !conf->nr_pending, conf->resync_lock);
+		if (conf->nr_pending)
+			BUG();
+	}
+	wait_event_lock_irq(conf->wait_resume, conf->barrier < RESYNC_DEPTH, conf->resync_lock);
+	conf->next_resync = sect;
+	spin_unlock_irq(&conf->resync_lock);
+}
+
+static void resume_device(conf_t *conf)
 {
-	spin_lock_irq(&conf->segment_lock);
-	wait_event_lock_irq(conf->wait_done,
-			r1_bio->sector < conf->start_active ||
-			r1_bio->sector >= conf->start_future,
-			conf->segment_lock);
-	if (r1_bio->sector < conf->start_active)
-		conf->cnt_done++;
-	else {
-		conf->cnt_future++;
-		if (conf->phase)
-			set_bit(R1BIO_SyncPhase, &r1_bio->state);
-	}
-	spin_unlock_irq(&conf->segment_lock);
+	spin_lock_irq(&conf->resync_lock);
+	if (!conf->barrier)
+		BUG();
+	--conf->barrier;
+	wake_up(&conf->wait_resume);
+	wake_up(&conf->wait_idle);
+	spin_unlock_irq(&conf->resync_lock);
 }
 
 static int make_request(mddev_t *mddev, int rw, struct bio * bio)
@@ -465,6 +443,16 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 	struct bio *read_bio;
 	int i, sum_bios = 0, disks = MD_SB_DISKS;
 
+	/*
+	 * Register the new request and wait if the reconstruction
+	 * thread has put up a bar for new requests.
+	 * Continue immediately if no resync is active currently.
+	 */
+	spin_lock_irq(&conf->resync_lock);
+	wait_event_lock_irq(conf->wait_resume, !conf->barrier, conf->resync_lock);
+	conf->nr_pending++;
+	spin_unlock_irq(&conf->resync_lock);
+
 	/*
 	 * make_request() can abort the operation when READA is being
 	 * used and no empty request is available.
@@ -475,7 +463,6 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 		rw = READ;
 
 	r1_bio = mempool_alloc(conf->r1bio_pool, GFP_NOIO);
-	check_all_bios_empty(r1_bio);
 
 	r1_bio->master_bio = bio;
 
@@ -483,8 +470,6 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 	r1_bio->sector = bio->bi_sector;
 	r1_bio->cmd = rw;
 
-	new_request(conf, r1_bio);
-
 	if (rw == READ) {
 		/*
 		 * read balancing logic:
@@ -503,15 +488,13 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 		read_bio->bi_private = r1_bio;
 
 		generic_make_request(read_bio);
+		atomic_inc(&conf->mirrors[r1_bio->read_disk].nr_pending);
 		return 0;
 	}
 
 	/*
 	 * WRITE:
 	 */
-
-	check_all_w_bios_empty(r1_bio);
-
 	for (i = 0; i < disks; i++) {
 		struct bio *mbio;
 		if (!conf->mirrors[i].operational)
@@ -542,14 +525,13 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 
 	/*
 	 * We have to be a bit careful about the semaphore above, thats
-	 * why we start the requests separately. Since kmalloc() could
-	 * fail, sleep and make_request() can sleep too, this is the
-	 * safer solution. Imagine, end_request decreasing the semaphore
-	 * before we could have set it up ... We could play tricks with
-	 * the semaphore (presetting it and correcting at the end if
-	 * sum_bios is not 'n' but we have to do end_request by hand if
-	 * all requests finish until we had a chance to set up the
-	 * semaphore correctly ... lots of races).
+	 * why we start the requests separately. Since generic_make_request()
+	 * can sleep, this is the safer solution. Imagine, end_request
+	 * decreasing the semaphore before we could have set it up ...
+	 * We could play tricks with the semaphore (presetting it and
+	 * correcting at the end if sum_bios is not 'n' but we have to
+	 * do end_request by hand if all requests finish until we had a
+	 * chance to set up the semaphore correctly ... lots of races).
 	 */
 	for (i = 0; i < disks; i++) {
 		struct bio *mbio;
@@ -558,6 +540,7 @@ static int make_request(mddev_t *mddev, int rw, struct bio * bio)
 			continue;
 
 		generic_make_request(mbio);
+		atomic_inc(&conf->mirrors[i].nr_pending);
 	}
 	return 0;
 }
@@ -610,8 +593,7 @@ static void mark_disk_bad(mddev_t *mddev, int failed)
 	md_wakeup_thread(conf->thread);
 	if (!mirror->write_only)
 		conf->working_disks--;
-	printk(DISK_FAILED, partition_name(mirror->dev),
-				conf->working_disks);
+	printk(DISK_FAILED, partition_name(mirror->dev), conf->working_disks);
 }
 
 static int error(mddev_t *mddev, kdev_t dev)
@@ -644,12 +626,6 @@ static int error(mddev_t *mddev, kdev_t dev)
 	return 0;
 }
 
-#undef LAST_DISK
-#undef NO_SPARE_DISK
-#undef DISK_FAILED
-#undef START_SYNCING
-
-
 static void print_conf(conf_t *conf)
 {
 	int i;
@@ -674,29 +650,13 @@ static void print_conf(conf_t *conf)
 
 static void close_sync(conf_t *conf)
 {
-	mddev_t *mddev = conf->mddev;
-	/*
-	 * If reconstruction was interrupted, we need to close the "active"
-	 * and "pending" holes.
-	 * we know that there are no active rebuild requests,
-	 * os cnt_active == cnt_ready == 0
-	 */
-	spin_lock_irq(&conf->segment_lock);
-	conf->start_active = conf->start_pending;
-	conf->start_ready = conf->start_pending;
-	wait_event_lock_irq(conf->wait_ready, !conf->cnt_pending, conf->segment_lock);
-	conf->start_active = conf->start_ready = conf->start_pending = conf->start_future;
-	conf->start_future = mddev->sb->size+1;
-	conf->cnt_pending = conf->cnt_future;
-	conf->cnt_future = 0;
-	conf->phase = conf->phase ^1;
-	wait_event_lock_irq(conf->wait_ready, !conf->cnt_pending, conf->segment_lock);
-	conf->start_active = conf->start_ready = conf->start_pending = conf->start_future = 0;
-	conf->phase = 0;
-	conf->cnt_future = conf->cnt_done;;
-	conf->cnt_done = 0;
-	spin_unlock_irq(&conf->segment_lock);
-	wake_up(&conf->wait_done);
+	spin_lock_irq(&conf->resync_lock);
+	wait_event_lock_irq(conf->wait_resume, !conf->barrier, conf->resync_lock);
+	spin_unlock_irq(&conf->resync_lock);
+
+	if (conf->barrier) BUG();
+	if (waitqueue_active(&conf->wait_idle)) BUG();
+	if (waitqueue_active(&conf->wait_resume)) BUG();
 }
 
 static int diskop(mddev_t *mddev, mdp_disk_t **d, int state)
@@ -975,9 +935,9 @@ static int end_sync_read(struct bio *bio, int nr_sectors)
 	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
 	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
 
-	check_all_w_bios_empty(r1_bio);
 	if (r1_bio->read_bio != bio)
 		BUG();
+	update_head_pos(r1_bio->read_disk, r1_bio);
 	/*
 	 * we have read a block, now it needs to be re-written,
 	 * or re-read if the read failed.
@@ -997,13 +957,21 @@ static int end_sync_write(struct bio *bio, int nr_sectors)
 	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
 	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
 	mddev_t *mddev = r1_bio->mddev;
+	int i;
 
 	if (!uptodate)
 		md_error(mddev, bio->bi_dev);
 
+	for (i = 0; i < MD_SB_DISKS; i++)
+		if (r1_bio->write_bios[i] == bio) {
+			update_head_pos(i, r1_bio);
+			break;
+		}
+
 	if (atomic_dec_and_test(&r1_bio->remaining)) {
-		sync_request_done(r1_bio->sector, mddev_to_conf(mddev));
+		conf_t *conf = mddev_to_conf(mddev);
 		md_done_sync(mddev, r1_bio->master_bio->bi_size >> 9, uptodate);
+		resume_device(conf);
 		put_buf(r1_bio);
 	}
 	return 0;
@@ -1029,11 +997,11 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
 		 */
 		printk(IO_ERROR, partition_name(bio->bi_dev), r1_bio->sector);
 		md_done_sync(mddev, r1_bio->master_bio->bi_size >> 9, 0);
+		resume_device(conf);
+		put_buf(r1_bio);
 		return;
 	}
 
-	check_all_w_bios_empty(r1_bio);
-
 	for (i = 0; i < disks ; i++) {
 		if (!conf->mirrors[i].operational)
 			continue;
@@ -1071,8 +1039,8 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
 		 * must be done
 		 */
 		printk(IO_ERROR, partition_name(bio->bi_dev), r1_bio->sector);
-		sync_request_done(r1_bio->sector, conf);
 		md_done_sync(mddev, r1_bio->master_bio->bi_size >> 9, 0);
+		resume_device(conf);
 		put_buf(r1_bio);
 		return;
 	}
@@ -1083,6 +1051,7 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
 
 		md_sync_acct(mbio->bi_dev, mbio->bi_size >> 9);
 		generic_make_request(mbio);
+		atomic_inc(&conf->mirrors[i].nr_pending);
 	}
 }
 
@@ -1101,6 +1070,7 @@ static void raid1d(void *data)
 	struct bio *bio;
 	unsigned long flags;
 	mddev_t *mddev;
+	conf_t *conf;
 	kdev_t dev;
 
 
@@ -1111,9 +1081,9 @@ static void raid1d(void *data)
 		r1_bio = list_entry(head->prev, r1bio_t, retry_list);
 		list_del(head->prev);
 		spin_unlock_irqrestore(&retry_list_lock, flags);
-		check_all_w_bios_empty(r1_bio);
 
 		mddev = r1_bio->mddev;
+		conf = mddev_to_conf(mddev);
 		if (mddev->sb_dirty) {
 			printk(KERN_INFO "raid1: dirty sb detected, updating.\n");
 			mddev->sb_dirty = 0;
@@ -1139,13 +1109,12 @@ static void raid1d(void *data)
 			bio->bi_rw = r1_bio->cmd;
 
 			generic_make_request(bio);
+			atomic_inc(&conf->mirrors[r1_bio->read_disk].nr_pending);
 			break;
 		}
 	}
 	spin_unlock_irqrestore(&retry_list_lock, flags);
 }
-#undef IO_ERROR
-#undef REDIRECT_SECTOR
 
 /*
  * Private kernel thread to reconstruct mirrors after an unclean
@@ -1177,101 +1146,27 @@ static int init_resync(conf_t *conf)
 {
 	int buffs;
 
-	conf->start_active = 0;
-	conf->start_ready = 0;
-	conf->start_pending = 0;
-	conf->start_future = 0;
-	conf->phase = 0;
-
 	buffs = RESYNC_WINDOW / RESYNC_BLOCK_SIZE;
 	if (conf->r1buf_pool)
 		BUG();
 	conf->r1buf_pool = mempool_create(buffs, r1buf_pool_alloc, r1buf_pool_free, conf);
 	if (!conf->r1buf_pool)
 		return -ENOMEM;
-	conf->window = 2048;
-	conf->cnt_future += conf->cnt_done+conf->cnt_pending;
-	conf->cnt_done = conf->cnt_pending = 0;
-	if (conf->cnt_ready || conf->cnt_active)
-		MD_BUG();
+	conf->next_resync = 0;
 	return 0;
 }
 
-static void wait_sync_pending(conf_t *conf, sector_t sector_nr)
-{
-	spin_lock_irq(&conf->segment_lock);
-	while (sector_nr >= conf->start_pending) {
-//		printk("wait .. sect=%lu start_active=%d ready=%d pending=%d future=%d, cnt_done=%d active=%d ready=%d pending=%d future=%d\n", sector_nr, conf->start_active, conf->start_ready, conf->start_pending, conf->start_future, conf->cnt_done, conf->cnt_active, conf->cnt_ready, conf->cnt_pending, conf->cnt_future);
-		wait_event_lock_irq(conf->wait_done, !conf->cnt_active,
-					conf->segment_lock);
-		wait_event_lock_irq(conf->wait_ready, !conf->cnt_pending,
-					conf->segment_lock);
-		conf->start_active = conf->start_ready;
-		conf->start_ready = conf->start_pending;
-		conf->start_pending = conf->start_future;
-		conf->start_future = conf->start_future+conf->window;
-
-		// Note: falling off the end is not a problem
-		conf->phase = conf->phase ^1;
-		conf->cnt_active = conf->cnt_ready;
-		conf->cnt_ready = 0;
-		conf->cnt_pending = conf->cnt_future;
-		conf->cnt_future = 0;
-		wake_up(&conf->wait_done);
-	}
-	conf->cnt_ready++;
-	spin_unlock_irq(&conf->segment_lock);
-}
-
 /*
  * perform a "sync" on one "block"
  *
  * We need to make sure that no normal I/O request - particularly write
  * requests - conflict with active sync requests.
- * This is achieved by conceptually dividing the block space into a
- * number of sections:
- *  DONE: 0 .. a-1     These blocks are in-sync
- *  ACTIVE: a.. b-1    These blocks may have active sync requests, but
- *                     no normal IO requests
- *  READY: b .. c-1    These blocks have no normal IO requests - sync
- *                     request may be happening
- *  PENDING: c .. d-1  These blocks may have IO requests, but no new
- *                     ones will be added
- *  FUTURE:  d .. end  These blocks are not to be considered yet. IO may
- *                     be happening, but not sync
  *
- * We keep a
- *   phase    which flips (0 or 1) each time d moves and
- * a count of:
- *   z =  active io requests in FUTURE since d moved - marked with
- *        current phase
- *   y =  active io requests in FUTURE before d moved, or PENDING -
- *        marked with previous phase
- *   x =  active sync requests in READY
- *   w =  active sync requests in ACTIVE
- *   v =  active io requests in DONE
- *
- * Normally, a=b=c=d=0 and z= active io requests
- *   or a=b=c=d=END and v= active io requests
- * Allowed changes to a,b,c,d:
- * A:  c==d &&  y==0 -> d+=window, y=z, z=0, phase=!phase
- * B:  y==0 -> c=d
- * C:   b=c, w+=x, x=0
- * D:  w==0 -> a=b
- * E: a==b==c==d==end -> a=b=c=d=0, z=v, v=0
- *
- * At start of sync we apply A.
- * When y reaches 0, we apply B then A then being sync requests
- * When sync point reaches c-1, we wait for y==0, and W==0, and
- * then apply apply B then A then D then C.
- * Finally, we apply E
- *
- * The sync request simply issues a "read" against a working drive
- * This is marked so that on completion the raid1d thread is woken to
- * issue suitable write requests
+ * This is achieved by tracking pending requests and a 'barrier' concept
+ * that can be installed to exclude normal IO requests.
  */
 
-static int sync_request(mddev_t *mddev, sector_t sector_nr)
+static int sync_request(mddev_t *mddev, sector_t sector_nr, int go_faster)
 {
 	conf_t *conf = mddev_to_conf(mddev);
 	mirror_info_t *mirror;
@@ -1283,8 +1178,13 @@ static int sync_request(mddev_t *mddev, sector_t sector_nr)
 	if (!sector_nr)
 		if (init_resync(conf))
 			return -ENOMEM;
-
-	wait_sync_pending(conf, sector_nr);
+	/*
+	 * If there is non-resync activity waiting for us then
+	 * put in a delay to throttle resync.
+	 */
+	if (!go_faster && waitqueue_active(&conf->wait_resume))
+		schedule_timeout(HZ);
+	device_barrier(conf, sector_nr + RESYNC_SECTORS);
 
 	/*
 	 * If reconstructing, and >1 working disc,
@@ -1302,10 +1202,13 @@ static int sync_request(mddev_t *mddev, sector_t sector_nr)
 	}
 	conf->last_used = disk;
 
-	mirror = conf->mirrors+conf->last_used;
+	mirror = conf->mirrors + conf->last_used;
 
 	r1_bio = mempool_alloc(conf->r1buf_pool, GFP_NOIO);
-	check_all_bios_empty(r1_bio);
+
+	spin_lock_irq(&conf->resync_lock);
+	conf->nr_pending++;
+	spin_unlock_irq(&conf->resync_lock);
 
 	r1_bio->mddev = mddev;
 	r1_bio->sector = sector_nr;
@@ -1344,6 +1247,7 @@ static int sync_request(mddev_t *mddev, sector_t sector_nr)
 	md_sync_acct(read_bio->bi_dev, nr_sectors);
 
 	generic_make_request(read_bio);
+	atomic_inc(&conf->mirrors[conf->last_used].nr_pending);
 
 	return nr_sectors;
 }
@@ -1447,7 +1351,6 @@ static int run(mddev_t *mddev)
 			disk->number = descriptor->number;
 			disk->raid_disk = disk_idx;
 			disk->dev = rdev->dev;
-			disk->sect_limit = MAX_WORK_PER_DISK;
 			disk->operational = 0;
 			disk->write_only = 0;
 			disk->spare = 0;
@@ -1479,7 +1382,6 @@ static int run(mddev_t *mddev)
 			disk->number = descriptor->number;
 			disk->raid_disk = disk_idx;
 			disk->dev = rdev->dev;
-			disk->sect_limit = MAX_WORK_PER_DISK;
 			disk->operational = 1;
 			disk->write_only = 0;
 			disk->spare = 0;
@@ -1494,7 +1396,6 @@ static int run(mddev_t *mddev)
 			disk->number = descriptor->number;
 			disk->raid_disk = disk_idx;
 			disk->dev = rdev->dev;
-			disk->sect_limit = MAX_WORK_PER_DISK;
 			disk->operational = 0;
 			disk->write_only = 0;
 			disk->spare = 1;
@@ -1507,9 +1408,9 @@ static int run(mddev_t *mddev)
 	conf->mddev = mddev;
 	conf->device_lock = SPIN_LOCK_UNLOCKED;
 
-	conf->segment_lock = SPIN_LOCK_UNLOCKED;
-	init_waitqueue_head(&conf->wait_done);
-	init_waitqueue_head(&conf->wait_ready);
+	conf->resync_lock = SPIN_LOCK_UNLOCKED;
+	init_waitqueue_head(&conf->wait_idle);
+	init_waitqueue_head(&conf->wait_resume);
 
 	if (!conf->working_disks) {
 		printk(NONE_OPERATIONAL, mdidx(mddev));
@@ -1611,17 +1512,6 @@ static int run(mddev_t *mddev)
 	return -EIO;
 }
 
-#undef INVALID_LEVEL
-#undef NO_SB
-#undef ERRORS
-#undef NOT_IN_SYNC
-#undef INCONSISTENT
-#undef ALREADY_RUNNING
-#undef OPERATIONAL
-#undef SPARE
-#undef NONE_OPERATIONAL
-#undef ARRAY_IS_ACTIVE
-
 static int stop_resync(mddev_t *mddev)
 {
 	conf_t *conf = mddev_to_conf(mddev);

drivers/mtd/nftlmount.c

@@ -4,7 +4,7 @@
  * Author: Fabrice Bellard (fabrice.bellard@netgem.com) 
  * Copyright (C) 2000 Netgem S.A.
  *
- * $Id: nftlmount.c,v 1.23 2001/09/19 21:42:32 dwmw2 Exp $
+ * $Id: nftlmount.c,v 1.25 2001/11/30 16:46:27 dwmw2 Exp $
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -39,7 +39,7 @@
 
 #define SECTORSIZE 512
 
-char nftlmountrev[]="$Revision: 1.23 $";
+char nftlmountrev[]="$Revision: 1.25 $";
 
 /* find_boot_record: Find the NFTL Media Header and its Spare copy which contains the
  *	various device information of the NFTL partition and Bad Unit Table. Update
@@ -94,11 +94,11 @@ static int find_boot_record(struct NFTLrecord *nftl)
 			continue;
 		}
 
-#if 1 /* Some people seem to have devices without ECC or erase marks
+#if 0 /* Some people seem to have devices without ECC or erase marks
 	 on the Media Header blocks. There are enough other sanity
 	 checks in here that we can probably do without it.
       */
-		if (le16_to_cpu ((h1.EraseMark | h1.EraseMark1) != ERASE_MARK)) {
+		if (le16_to_cpu(h1.EraseMark | h1.EraseMark1) != ERASE_MARK) {
 			printk(KERN_NOTICE "ANAND header found at 0x%x in mtd%d, but erase mark not present (0x%04x,0x%04x instead)\n",
 			       block * nftl->EraseSize, nftl->mtd->index, 
 			       le16_to_cpu(h1.EraseMark), le16_to_cpu(h1.EraseMark1));

drivers/scsi/qlogicfas.c

@@ -467,10 +467,11 @@ Scsi_Cmnd	   *icmd;
 static void	    do_ql_ihandl(int irq, void *dev_id, struct pt_regs * regs)
 {
 	unsigned long flags;
+	struct Scsi_Host *host = dev_id;
 
-	spin_lock_irqsave(&io_request_lock, flags);
+	spin_lock_irqsave(&host->host_lock, flags);
 	ql_ihandl(irq, dev_id, regs);
-	spin_unlock_irqrestore(&io_request_lock, flags);
+	spin_unlock_irqrestore(&host->host_lock, flags);
 }
 #endif
 

drivers/scsi/scsi.h

@@ -548,10 +548,8 @@ extern void print_status (int status);
  */
 
 struct scsi_device {
-/* private: */
 	/*
-	 * This information is private to the scsi mid-layer.  Wrapping it in a
-	 * struct private is a way of marking it in a sort of C++ type of way.
+	 * This information is private to the scsi mid-layer.
 	 */
 	struct scsi_device *next;	/* Used for linked list */
 	struct scsi_device *prev;	/* Used for linked list */
@@ -563,7 +561,6 @@ struct scsi_device {
 	volatile unsigned short device_busy;	/* commands actually active on low-level */
 	Scsi_Cmnd *device_queue;	/* queue of SCSI Command structures */
 
-/* public: */
 	unsigned int id, lun, channel;
 
 	unsigned int manufacturer;	/* Manufacturer of device, for using 
@@ -681,11 +678,7 @@ struct scsi_request {
  */
 struct scsi_cmnd {
 	int     sc_magic;
-/* private: */
-	/*
-	 * This information is private to the scsi mid-layer.  Wrapping it in a
-	 * struct private is a way of marking it in a sort of C++ type of way.
-	 */
+
 	struct Scsi_Host *host;
 	unsigned short state;
 	unsigned short owner;
@@ -727,8 +720,6 @@ struct scsi_cmnd {
 	struct scsi_cmnd *bh_next;	/* To enumerate the commands waiting 
 					   to be processed. */
 
-/* public: */
-
 	unsigned int target;
 	unsigned int lun;
 	unsigned int channel;

drivers/scsi/scsi_merge.c

@@ -59,12 +59,10 @@
  */
 int scsi_init_io(Scsi_Cmnd *SCpnt)
 {
-	struct request     *req;
+	struct request     *req = &SCpnt->request;
 	struct scatterlist *sgpnt;
 	int count, gfp_mask;
 
-	req = &SCpnt->request;
-
 	/*
 	 * First we need to know how many scatter gather segments are needed.
 	 */
@@ -85,14 +83,13 @@ int scsi_init_io(Scsi_Cmnd *SCpnt)
 	BUG_ON(!sgpnt);
 
 	SCpnt->request_buffer = (char *) sgpnt;
-	SCpnt->request_bufflen = 0;
+	SCpnt->request_bufflen = req->nr_sectors << 9;
 	req->buffer = NULL;
 
 	/* 
 	 * Next, walk the list, and fill in the addresses and sizes of
 	 * each segment.
 	 */
-	SCpnt->request_bufflen = req->nr_sectors << 9;
 	count = blk_rq_map_sg(req->q, req, SCpnt->request_buffer);
 
 	/*
@@ -142,8 +139,7 @@ void scsi_initialize_merge_fn(Scsi_Device * SDpnt)
 			bounce_limit = BLK_BOUNCE_ANY;
 		else
 			bounce_limit = SHpnt->pci_dev->dma_mask;
-	}
-	if (SHpnt->unchecked_isa_dma)
+	} else if (SHpnt->unchecked_isa_dma)
 		bounce_limit = BLK_BOUNCE_ISA;
 
 	blk_queue_bounce_limit(q, bounce_limit);

drivers/scsi/sr.c

@@ -287,9 +287,6 @@ static int sr_init_command(Scsi_Cmnd * SCpnt)
 		return 0;
 	}
 
-	if (rq_data_dir(&SCpnt->request) == WRITE && !scsi_CDs[dev].device->writeable)
-		return 0;
-
 	/*
 	 * we do lazy blocksize switching (when reading XA sectors,
 	 * see CDROMREADMODE2 ioctl) 

fs/bio.c

@@ -142,6 +142,7 @@ struct bio *bio_alloc(int gfp_mask, int nr_iovecs)
 		bio->bi_io_vec = bvl;
 		return bio;
 	}
+
 	mempool_free(bio, bio_pool);
 	return NULL;
 }
@@ -311,28 +312,6 @@ struct bio *bio_copy(struct bio *bio, int gfp_mask, int copy)
 	return NULL;
 }
 
-#ifdef BIO_PAGEIO
-static int bio_end_io_page(struct bio *bio)
-{
-	struct page *page = bio_page(bio);
-
-	if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
-		SetPageError(page);
-	if (!PageError(page))
-		SetPageUptodate(page);
-
-	/*
-	 * Run the hooks that have to be done when a page I/O has completed.
-	 */
-	if (PageTestandClearDecrAfter(page))
-		atomic_dec(&nr_async_pages);
-
-	UnlockPage(page);
-	bio_put(bio);
-	return 1;
-}
-#endif
-
 static int bio_end_io_kio(struct bio *bio, int nr_sectors)
 {
 	struct kiobuf *kio = (struct kiobuf *) bio->bi_private;

fs/hpfs/super.c

@@ -410,6 +410,8 @@ struct super_block *hpfs_read_super(struct super_block *s, void *options,
 	/*s->s_hpfs_mounting = 1;*/
 	dev = s->s_dev;
 	set_blocksize(dev, 512);
+	s->s_blocksize = 512;
+	s->s_blocksize_bits = 9;
 	s->s_hpfs_fs_size = -1;
 	if (!(bootblock = hpfs_map_sector(s, 0, &bh0, 0))) goto bail1;
 	if (!(superblock = hpfs_map_sector(s, 16, &bh1, 1))) goto bail2;
@@ -436,8 +438,6 @@ struct super_block *hpfs_read_super(struct super_block *s, void *options,
 
 	/* Fill superblock stuff */
 	s->s_magic = HPFS_SUPER_MAGIC;
-	s->s_blocksize = 512;
-	s->s_blocksize_bits = 9;
 	s->s_op = &hpfs_sops;
 
 	s->s_hpfs_root = superblock->root;

fs/namespace.c

@@ -19,7 +19,7 @@
 
 #include <asm/uaccess.h>
 
-struct vfsmount *do_kern_mount(char *type, int flags, char *name, void *data);
+struct vfsmount *do_kern_mount(const char *type, int flags, char *name, void *data);
 int do_remount_sb(struct super_block *sb, int flags, void * data);
 void kill_super(struct super_block *sb);
 
@@ -622,9 +622,18 @@ static int do_move_mount(struct nameidata *nd, char *old_name)
 static int do_add_mount(struct nameidata *nd, char *type, int flags,
 			int mnt_flags, char *name, void *data)
 {
-	struct vfsmount *mnt = do_kern_mount(type, flags, name, data);
-	int err = PTR_ERR(mnt);
+	struct vfsmount *mnt;
+	int err;
+
+	if (!type || !memchr(type, 0, PAGE_SIZE))
+		return -EINVAL;
+
+	/* we need capabilities... */
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
 
+	mnt = do_kern_mount(type, flags, name, data);
+	err = PTR_ERR(mnt);
 	if (IS_ERR(mnt))
 		goto out;
 

fs/nfs/file.c

@@ -161,15 +161,10 @@ static int nfs_prepare_write(struct file *file, struct page *page, unsigned offs
 static int nfs_commit_write(struct file *file, struct page *page, unsigned offset, unsigned to)
 {
 	long status;
-	loff_t pos = ((loff_t)page->index<<PAGE_CACHE_SHIFT) + to;
-	struct inode *inode = page->mapping->host;
 
 	lock_kernel();
 	status = nfs_updatepage(file, page, offset, to-offset);
 	unlock_kernel();
-	/* most likely it's already done. CHECKME */
-	if (pos > inode->i_size)
-		inode->i_size = pos;
 	return status;
 }
 

fs/nfs/inode.c

@@ -107,17 +107,10 @@ nfs_read_inode(struct inode * inode)
 	inode->i_rdev = 0;
 	/* We can't support UPDATE_ATIME(), since the server will reset it */
 	inode->i_flags |= S_NOATIME;
-	NFS_FILEID(inode) = 0;
-	NFS_FSID(inode) = 0;
-	NFS_FLAGS(inode) = 0;
 	INIT_LIST_HEAD(&inode->u.nfs_i.read);
 	INIT_LIST_HEAD(&inode->u.nfs_i.dirty);
 	INIT_LIST_HEAD(&inode->u.nfs_i.commit);
 	INIT_LIST_HEAD(&inode->u.nfs_i.writeback);
-	inode->u.nfs_i.nread = 0;
-	inode->u.nfs_i.ndirty = 0;
-	inode->u.nfs_i.ncommit = 0;
-	inode->u.nfs_i.npages = 0;
 	NFS_CACHEINV(inode);
 	NFS_ATTRTIMEO(inode) = NFS_MINATTRTIMEO(inode);
 	NFS_ATTRTIMEO_UPDATE(inode) = jiffies;
@@ -655,19 +648,6 @@ nfs_fill_inode(struct inode *inode, struct nfs_fh *fh, struct nfs_fattr *fattr)
 			inode->i_op = &nfs_symlink_inode_operations;
 		else
 			init_special_inode(inode, inode->i_mode, fattr->rdev);
-		/*
-		 * Preset the size and mtime, as there's no need
-		 * to invalidate the caches.
-		 */ 
-		inode->i_size  = nfs_size_to_loff_t(fattr->size);
-		inode->i_mtime = nfs_time_to_secs(fattr->mtime);
-		inode->i_atime = nfs_time_to_secs(fattr->atime);
-		inode->i_ctime = nfs_time_to_secs(fattr->ctime);
-		NFS_CACHE_CTIME(inode) = fattr->ctime;
-		NFS_CACHE_MTIME(inode) = fattr->mtime;
-		NFS_CACHE_ISIZE(inode) = fattr->size;
-		NFS_ATTRTIMEO(inode) = NFS_MINATTRTIMEO(inode);
-		NFS_ATTRTIMEO_UPDATE(inode) = jiffies;
 		memcpy(&inode->u.nfs_i.fh, fh, sizeof(inode->u.nfs_i.fh));
 	}
 	nfs_refresh_inode(inode, fattr);
@@ -697,6 +677,9 @@ nfs_find_actor(struct inode *inode, unsigned long ino, void *opaque)
 		return 0;
 	if (memcmp(&inode->u.nfs_i.fh, fh, sizeof(inode->u.nfs_i.fh)) != 0)
 		return 0;
+	/* Force an attribute cache update if inode->i_count == 0 */
+	if (!atomic_read(&inode->i_count))
+		NFS_CACHEINV(inode);
 	return 1;
 }
 
@@ -797,7 +780,9 @@ printk("nfs_notify_change: revalidate failed, error=%d\n", error);
 	if (!S_ISREG(inode->i_mode))
 		attr->ia_valid &= ~ATTR_SIZE;
 
+	filemap_fdatasync(inode->i_mapping);
 	error = nfs_wb_all(inode);
+	filemap_fdatawait(inode->i_mapping);
 	if (error)
 		goto out;
 
@@ -825,6 +810,8 @@ printk("nfs_notify_change: revalidate failed, error=%d\n", error);
 		fattr.pre_ctime = NFS_CACHE_CTIME(inode);
 		fattr.valid |= NFS_ATTR_WCC;
 	}
+	/* Force an attribute cache update */
+	NFS_CACHEINV(inode);
 	error = nfs_refresh_inode(inode, &fattr);
 out:
 	return error;
@@ -965,6 +952,34 @@ __nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 	return status;
 }
 
+/*
+ * nfs_fattr_obsolete - Test if attribute data is newer than cached data
+ * @inode: inode
+ * @fattr: attributes to test
+ *
+ * Avoid stuffing the attribute cache with obsolete information.
+ * We always accept updates if the attribute cache timed out, or if
+ * fattr->ctime is newer than our cached value.
+ * If fattr->ctime matches the cached value, we still accept the update
+ * if it increases the file size.
+ */
+static inline
+int nfs_fattr_obsolete(struct inode *inode, struct nfs_fattr *fattr)
+{
+	s64 cdif;
+
+	if (time_after(jiffies, NFS_READTIME(inode)+NFS_ATTRTIMEO(inode)))
+		goto out_valid;
+	if ((cdif = (s64)fattr->ctime - (s64)NFS_CACHE_CTIME(inode)) > 0)
+		goto out_valid;
+	/* Ugh... */
+	if (cdif == 0 && fattr->size > NFS_CACHE_ISIZE(inode))
+		goto out_valid;
+	return -1;
+ out_valid:
+	return 0;
+}
+
 /*
  * Many nfs protocol calls return the new file attributes after
  * an operation.  Here we update the inode to reflect the state
@@ -982,6 +997,7 @@ __nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
 {
 	__u64		new_size, new_mtime;
 	loff_t		new_isize;
+	time_t		new_atime;
 	int		invalid = 0;
 
 	dfprintk(VFS, "NFS: refresh_inode(%x/%ld ct=%d info=0x%x)\n",
@@ -1007,6 +1023,11 @@ __nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
 	new_size = fattr->size;
  	new_isize = nfs_size_to_loff_t(fattr->size);
 
+	new_atime = nfs_time_to_secs(fattr->atime);
+	/* Avoid races */
+	if (nfs_fattr_obsolete(inode, fattr))
+		goto out_nochange;
+
 	/*
 	 * Update the read time so we don't revalidate too often.
 	 */
@@ -1056,7 +1077,7 @@ __nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
 	NFS_CACHE_CTIME(inode) = fattr->ctime;
 	inode->i_ctime = nfs_time_to_secs(fattr->ctime);
 
-	inode->i_atime = nfs_time_to_secs(fattr->atime);
+	inode->i_atime = new_atime;
 
 	NFS_CACHE_MTIME(inode) = new_mtime;
 	inode->i_mtime = nfs_time_to_secs(new_mtime);
@@ -1093,7 +1114,10 @@ __nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
 	if (invalid)
 		nfs_zap_caches(inode);
 	return 0;
-
+ out_nochange:
+	if (new_atime - inode->i_atime > 0)
+		inode->i_atime = new_atime;
+	return 0;
  out_changed:
 	/*
 	 * Big trouble! The inode has become a different object.

fs/nfs/nfs2xdr.c

@@ -270,14 +270,12 @@ nfs_xdr_readres(struct rpc_rqst *req, u32 *p, struct nfs_readres *res)
 
 	count = ntohl(*p++);
 	hdrlen = (u8 *) p - (u8 *) iov->iov_base;
-	recvd = req->rq_rlen - hdrlen;
-	if (p != iov[req->rq_rnr-1].iov_base) {
-		/* Unexpected reply header size. Punt.
-		 * XXX: Move iovec contents to align data on page
-		 * boundary and adjust RPC header size guess */
-		printk(KERN_WARNING "NFS: Odd RPC header size in read reply: %d\n", hdrlen);
-		return -errno_NFSERR_IO;
+	if (iov->iov_len > hdrlen) {
+		dprintk("NFS: READ header is short. iovec will be shifted.\n");
+		xdr_shift_iovec(iov, req->rq_rnr, iov->iov_len - hdrlen);
 	}
+
+	recvd = req->rq_rlen - hdrlen;
 	if (count > recvd) {
 		printk(KERN_WARNING "NFS: server cheating in read reply: "
 			"count %d > recvd %d\n", count, recvd);
@@ -448,27 +446,23 @@ static int
 nfs_xdr_readdirres(struct rpc_rqst *req, u32 *p, struct nfs_readdirres *res)
 {
 	struct iovec		*iov = req->rq_rvec;
+	int			 hdrlen;
 	int			 status, nr;
 	u32			*end, *entry, len;
 
 	if ((status = ntohl(*p++)))
 		return -nfs_stat_to_errno(status);
-	if ((void *) p != ((u8 *) iov->iov_base+iov->iov_len)) {
-		/* Unexpected reply header size. Punt. */
-		printk(KERN_WARNING "NFS: Odd RPC header size in readdirres reply\n");
-		return -errno_NFSERR_IO;
+
+	hdrlen = (u8 *) p - (u8 *) iov->iov_base;
+	if (iov->iov_len > hdrlen) {
+		dprintk("NFS: READDIR header is short. iovec will be shifted.\n");
+		xdr_shift_iovec(iov, req->rq_rnr, iov->iov_len - hdrlen);
 	}
 
+
 	/* Get start and end address of XDR data */
 	p   = (u32 *) iov[1].iov_base;
 	end = (u32 *) ((u8 *) p + iov[1].iov_len);
-
-	/* Get start and end of dirent buffer */
-	if (res->buffer != p) {
-		printk(KERN_ERR "NFS: Bad result buffer in readdir\n");
-		return -errno_NFSERR_IO;
-	}
-
 	for (nr = 0; *p++; nr++) {
 		entry = p - 1;
 		if (p + 2 > end)
@@ -598,13 +592,21 @@ nfs_xdr_readlinkargs(struct rpc_rqst *req, u32 *p, struct nfs_readlinkargs *args
 static int
 nfs_xdr_readlinkres(struct rpc_rqst *req, u32 *p, struct nfs_readlinkres *res)
 {
+	struct iovec *iov = req->rq_rvec;
 	u32	*strlen;
 	char	*string;
+	int	hdrlen;
 	int	status;
 	unsigned int len;
 
 	if ((status = ntohl(*p++)))
 		return -nfs_stat_to_errno(status);
+	hdrlen = (u8 *) p - (u8 *) iov->iov_base;
+	if (iov->iov_len > hdrlen) {
+		dprintk("NFS: READLINK header is short. iovec will be shifted.\n");
+		xdr_shift_iovec(iov, req->rq_rnr, iov->iov_len - hdrlen);
+	}
+
 	strlen = (u32*)res->buffer;
 	/* Convert length of symlink */
 	len = ntohl(*strlen);

fs/nfs/read.c

@@ -397,7 +397,7 @@ nfs_readpage_result(struct rpc_task *task)
 {
 	struct nfs_read_data	*data = (struct nfs_read_data *) task->tk_calldata;
 	struct inode		*inode = data->inode;
-	int			count = data->res.count;
+	unsigned int		count = data->res.count;
 
 	dprintk("NFS: %4d nfs_readpage_result, (status %d)\n",
 		task->tk_pid, task->tk_status);
@@ -408,9 +408,15 @@ nfs_readpage_result(struct rpc_task *task)
 		struct page *page = req->wb_page;
 		nfs_list_remove_request(req);
 
-		if (task->tk_status >= 0 && count >= 0) {
-			SetPageUptodate(page);
+		if (task->tk_status >= 0) {
+			if (count < PAGE_CACHE_SIZE) {
+				char *p = kmap(page);
+				memset(p + count, 0, PAGE_CACHE_SIZE - count);
+				kunmap(page);
+				count = 0;
+			} else
 				count -= PAGE_CACHE_SIZE;
+			SetPageUptodate(page);
 		} else
 			SetPageError(page);
 		flush_dcache_page(page);

fs/nfs/write.c

@@ -213,6 +213,7 @@ nfs_writepage_async(struct file *file, struct inode *inode, struct page *page,
 		    unsigned int offset, unsigned int count)
 {
 	struct nfs_page	*req;
+	loff_t		end;
 	int		status;
 
 	req = nfs_update_request(file, inode, page, offset, count);
@@ -223,6 +224,10 @@ nfs_writepage_async(struct file *file, struct inode *inode, struct page *page,
 		req->wb_cred = get_rpccred(NFS_I(inode)->mm_cred);
 	nfs_unlock_request(req);
 	nfs_strategy(inode);
+	end = ((loff_t)page->index<<PAGE_CACHE_SHIFT) + (loff_t)(offset + count);
+	if (inode->i_size < end)
+		inode->i_size = end;
+
  out:
 	return status;
 }
@@ -781,6 +786,7 @@ nfs_updatepage(struct file *file, struct page *page, unsigned int offset, unsign
 	struct dentry	*dentry = file->f_dentry;
 	struct inode	*inode = dentry->d_inode;
 	struct nfs_page	*req;
+	loff_t		end;
 	int		status = 0;
 
 	dprintk("NFS:      nfs_updatepage(%s/%s %d@%Ld)\n",
@@ -812,6 +818,10 @@ nfs_updatepage(struct file *file, struct page *page, unsigned int offset, unsign
 		goto done;
 
 	status = 0;
+	end = ((loff_t)page->index<<PAGE_CACHE_SHIFT) + (loff_t)(offset + count);
+	if (inode->i_size < end)
+		inode->i_size = end;
+
 	/* If we wrote past the end of the page.
 	 * Call the strategy routine so it can send out a bunch
 	 * of requests.

fs/super.c

@@ -34,8 +34,6 @@
 #define __NO_VERSION__
 #include <linux/module.h>
 
-int do_remount_sb(struct super_block *sb, int flags, void * data);
-
 LIST_HEAD(super_blocks);
 spinlock_t sb_lock = SPIN_LOCK_UNLOCKED;
 
@@ -377,7 +375,7 @@ static void insert_super(struct super_block *s, struct file_system_type *type)
 	get_filesystem(type);
 }
 
-void put_unnamed_dev(kdev_t dev);	/* should become static */
+static void put_anon_dev(kdev_t dev);
 
 /**
  *	remove_super	-	makes superblock unreachable
@@ -407,7 +405,7 @@ static void remove_super(struct super_block *s)
 	if (bdev)
 		blkdev_put(bdev, BDEV_FS);
 	else
-		put_unnamed_dev(dev);
+		put_anon_dev(dev);
 }
 
 struct vfsmount *alloc_vfsmnt(void);
@@ -530,36 +528,122 @@ asmlinkage long sys_ustat(dev_t dev, struct ustat * ubuf)
 	return err;
 }
 
+/**
+ *	do_remount_sb	-	asks filesystem to change mount options.
+ *	@sb:	superblock in question
+ *	@flags:	numeric part of options
+ *	@data:	the rest of options
+ *
+ *	Alters the mount options of a mounted file system.
+ */
+int do_remount_sb(struct super_block *sb, int flags, void *data)
+{
+	int retval;
+	
+	if (!(flags & MS_RDONLY) && sb->s_dev && is_read_only(sb->s_dev))
+		return -EACCES;
+		/*flags |= MS_RDONLY;*/
+	if (flags & MS_RDONLY)
+		acct_auto_close(sb->s_dev);
+	shrink_dcache_sb(sb);
+	fsync_super(sb);
+	/* If we are remounting RDONLY, make sure there are no rw files open */
+	if ((flags & MS_RDONLY) && !(sb->s_flags & MS_RDONLY))
+		if (!fs_may_remount_ro(sb))
+			return -EBUSY;
+	if (sb->s_op && sb->s_op->remount_fs) {
+		lock_super(sb);
+		retval = sb->s_op->remount_fs(sb, &flags, data);
+		unlock_super(sb);
+		if (retval)
+			return retval;
+	}
+	sb->s_flags = (sb->s_flags & ~MS_RMT_MASK) | (flags & MS_RMT_MASK);
+	return 0;
+}
+
 /*
  * Unnamed block devices are dummy devices used by virtual
  * filesystems which don't use real block-devices.  -- jrs
  */
 
-static unsigned long unnamed_dev_in_use[256/(8*sizeof(unsigned long))];
+enum {Max_anon = 256};
+static unsigned long unnamed_dev_in_use[Max_anon/(8*sizeof(unsigned long))];
+static spinlock_t unnamed_dev_lock = SPIN_LOCK_UNLOCKED;/* protects the above */
 
-kdev_t get_unnamed_dev(void)
+/**
+ *	put_anon_dev	-	release anonymous device number.
+ *	@dev:	device in question
+ */
+static void put_anon_dev(kdev_t dev)
 {
-	int i;
-
-	for (i = 1; i < 256; i++) {
-		if (!test_and_set_bit(i,unnamed_dev_in_use))
-			return MKDEV(UNNAMED_MAJOR, i);
-	}
-	return 0;
+	spin_lock(&unnamed_dev_lock);
+	clear_bit(MINOR(dev), unnamed_dev_in_use);
+	spin_unlock(&unnamed_dev_lock);
 }
 
-void put_unnamed_dev(kdev_t dev)
+/**
+ *	get_anon_super	-	allocate a superblock for non-device fs
+ *	@type:		filesystem type
+ *	@compare:	check if existing superblock is what we want
+ *	@data:		argument for @compare.
+ *
+ *	get_anon_super is a helper for non-blockdevice filesystems.
+ *	It either finds and returns one of the superblocks of given type
+ *	(if it can find one that would satisfy caller) or creates a new
+ *	one.  In the either case we return an active reference to superblock
+ *	with ->s_umount locked.  If superblock is new it gets a new
+ *	anonymous device allocated for it and is inserted into lists -
+ *	other initialization is left to caller.
+ *
+ *	Rather than duplicating all that logics every time when
+ *	we want something that doesn't fit "nodev" and "single" we pull
+ *	the relevant code into common helper and let get_sb_...() call
+ *	it.
+ *
+ *	NB: get_sb_...() is going to become an fs type method, with
+ *	current ->read_super() becoming a callback used by common instances.
+ */
+struct super_block *get_anon_super(struct file_system_type *type,
+	int (*compare)(struct super_block *,void *), void *data)
 {
-	if (!dev || MAJOR(dev) != UNNAMED_MAJOR)
-		return;
-	if (test_and_clear_bit(MINOR(dev), unnamed_dev_in_use))
-		return;
-	printk("VFS: put_unnamed_dev: freeing unused device %s\n",
-			kdevname(dev));
+	struct super_block *s = alloc_super();
+	kdev_t dev;
+	struct list_head *p;
+
+	if (!s)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock(&unnamed_dev_lock);
+	dev = find_first_zero_bit(unnamed_dev_in_use, Max_anon);
+	if (dev == Max_anon) {
+		spin_unlock(&unnamed_dev_lock);
+		destroy_super(s);
+		return ERR_PTR(-EMFILE);
+	}
+	set_bit(dev, unnamed_dev_in_use);
+	spin_unlock(&unnamed_dev_lock);
+
+retry:
+	spin_lock(&sb_lock);
+	if (compare) list_for_each(p, &type->fs_supers) {
+		struct super_block *old;
+		old = list_entry(p, struct super_block, s_instances);
+		if (!compare(old, data))
+			continue;
+		if (!grab_super(old))
+			goto retry;
+		destroy_super(s);
+		return old;
+	}
+
+	s->s_dev = dev;
+	insert_super(s, type);
+	return s;
 }
 
 static struct super_block *get_sb_bdev(struct file_system_type *fs_type,
-	char *dev_name, int flags, void * data)
+	int flags, char *dev_name, void * data)
 {
 	struct inode *inode;
 	struct block_device *bdev;
@@ -598,17 +682,13 @@ static struct super_block *get_sb_bdev(struct file_system_type *fs_type,
 		goto out;
 	check_disk_change(dev);
 	error = -EACCES;
-	if (!(flags & MS_RDONLY) && is_read_only(dev)) {
-		blkdev_put(bdev, BDEV_FS);
-		goto out;
-	}
+	if (!(flags & MS_RDONLY) && is_read_only(dev))
+		goto out1;
 
 	error = -ENOMEM;
 	s = alloc_super();
-	if (!s) {
-		blkdev_put(bdev, BDEV_FS);
-		goto out;
-	}
+	if (!s)
+		goto out1;
 
 	error = -EBUSY;
 restart:
@@ -622,8 +702,7 @@ static struct super_block *get_sb_bdev(struct file_system_type *fs_type,
 		    ((flags ^ old->s_flags) & MS_RDONLY)) {
 			spin_unlock(&sb_lock);
 			destroy_super(s);
-			blkdev_put(bdev, BDEV_FS);
-			goto out;
+			goto out1;
 		}
 		if (!grab_super(old))
 			goto restart;
@@ -636,97 +715,103 @@ static struct super_block *get_sb_bdev(struct file_system_type *fs_type,
 	s->s_bdev = bdev;
 	s->s_flags = flags;
 	insert_super(s, fs_type);
-
-	error = -EINVAL;
-	lock_super(s);
 	if (!fs_type->read_super(s, data, flags & MS_VERBOSE ? 1 : 0))
-		goto out_fail;
+		goto Einval;
 	s->s_flags |= MS_ACTIVE;
-	unlock_super(s);
 	path_release(&nd);
 	return s;
 
-out_fail:
-	unlock_super(s);
+Einval:
 	deactivate_super(s);
 	remove_super(s);
+	error = -EINVAL;
+	goto out;
+out1:
+	blkdev_put(bdev, BDEV_FS);
 out:
 	path_release(&nd);
 	return ERR_PTR(error);
 }
 
 static struct super_block *get_sb_nodev(struct file_system_type *fs_type,
-	int flags, void * data)
+	int flags, char *dev_name, void *data)
 {
-	struct super_block *s = alloc_super();
+	struct super_block *s = get_anon_super(fs_type, NULL, NULL);
 
-	if (!s)
-		return ERR_PTR(-ENOMEM);
-	s->s_dev = get_unnamed_dev();
-	if (!s->s_dev) {
-		destroy_super(s);
-		return ERR_PTR(-EMFILE);
-	}
-	s->s_flags = flags;
-	spin_lock(&sb_lock);
-	insert_super(s, fs_type);
-	lock_super(s);
-	if (!fs_type->read_super(s, data, flags & MS_VERBOSE ? 1 : 0))
-		goto out_fail;
-	s->s_flags |= MS_ACTIVE;
-	unlock_super(s);
+	if (IS_ERR(s))
 		return s;
 
-out_fail:
-	unlock_super(s);
+	s->s_flags = flags;
+	if (!fs_type->read_super(s, data, flags & MS_VERBOSE ? 1 : 0)) {
 		deactivate_super(s);
 		remove_super(s);
 		return ERR_PTR(-EINVAL);
+	}
+	s->s_flags |= MS_ACTIVE;
+	return s;
+}
+
+static int compare_single(struct super_block *s, void *p)
+{
+	return 1;
 }
 
 static struct super_block *get_sb_single(struct file_system_type *fs_type,
-	int flags, void *data)
+	int flags, char *dev_name, void *data)
 {
-	struct super_block * s = alloc_super();
-	if (!s)
-		return ERR_PTR(-ENOMEM);
-	/*
-	 * Get the superblock of kernel-wide instance, but
-	 * keep the reference to fs_type.
-	 */
-retry:
-	spin_lock(&sb_lock);
-	if (!list_empty(&fs_type->fs_supers)) {
-		struct super_block *old;
-		old = list_entry(fs_type->fs_supers.next, struct super_block,
-				s_instances);
-		if (!grab_super(old))
-			goto retry;
-		destroy_super(s);
-		do_remount_sb(old, flags, data);
-		return old;
-	} else {
-		s->s_dev = get_unnamed_dev();
-		if (!s->s_dev) {
-			spin_unlock(&sb_lock);
-			destroy_super(s);
-			return ERR_PTR(-EMFILE);
-		}
-		s->s_flags = flags;
-		insert_super(s, fs_type);
-		lock_super(s);
-		if (!fs_type->read_super(s, data, flags & MS_VERBOSE ? 1 : 0))
-			goto out_fail;
-		s->s_flags |= MS_ACTIVE;
-		unlock_super(s);
-		return s;
+	struct super_block *s = get_anon_super(fs_type, compare_single, NULL);
 
-	out_fail:
-		unlock_super(s);
+	if (IS_ERR(s))
+		return s;
+	if (!s->s_root) {
+		s->s_flags = flags;
+		if (!fs_type->read_super(s, data, flags & MS_VERBOSE ? 1 : 0)) {
 			deactivate_super(s);
 			remove_super(s);
 			return ERR_PTR(-EINVAL);
 		}
+		s->s_flags |= MS_ACTIVE;
+	}
+	do_remount_sb(s, flags, data);
+	return s;
+}
+
+struct vfsmount *
+do_kern_mount(const char *fstype, int flags, char *name, void *data)
+{
+	struct file_system_type *type = get_fs_type(fstype);
+	struct super_block *sb = ERR_PTR(-ENOMEM);
+	struct vfsmount *mnt;
+
+	if (!type)
+		return ERR_PTR(-ENODEV);
+
+	mnt = alloc_vfsmnt();
+	if (!mnt)
+		goto out;
+	set_devname(mnt, name);
+	if (type->fs_flags & FS_REQUIRES_DEV)
+		sb = get_sb_bdev(type, flags, name, data);
+	else if (type->fs_flags & FS_SINGLE)
+		sb = get_sb_single(type, flags, name, data);
+	else
+		sb = get_sb_nodev(type, flags, name, data);
+	if (IS_ERR(sb))
+		goto out_mnt;
+	if (type->fs_flags & FS_NOMOUNT)
+		sb->s_flags |= MS_NOUSER;
+	mnt->mnt_sb = sb;
+	mnt->mnt_root = dget(sb->s_root);
+	mnt->mnt_mountpoint = sb->s_root;
+	mnt->mnt_parent = mnt;
+	up_write(&sb->s_umount);
+	put_filesystem(type);
+	return mnt;
+out_mnt:
+	free_vfsmnt(mnt);
+out:
+	put_filesystem(type);
+	return (struct vfsmount *)sb;
 }
 
 void kill_super(struct super_block *sb)
@@ -739,7 +824,6 @@ void kill_super(struct super_block *sb)
 		return;
 
 	down_write(&sb->s_umount);
-	lock_kernel();
 	sb->s_root = NULL;
 	/* Need to clean after the sucker */
 	if (fs->fs_flags & FS_LITTER)
@@ -748,6 +832,7 @@ void kill_super(struct super_block *sb)
 	dput(root);
 	fsync_super(sb);
 	lock_super(sb);
+	lock_kernel();
 	sb->s_flags &= ~MS_ACTIVE;
 	invalidate_inodes(sb);	/* bad name - it should be evict_inodes() */
 	if (sop) {
@@ -768,96 +853,7 @@ void kill_super(struct super_block *sb)
 	remove_super(sb);
 }
 
-/*
- * Alters the mount flags of a mounted file system. Only the mount point
- * is used as a reference - file system type and the device are ignored.
- */
-
-int do_remount_sb(struct super_block *sb, int flags, void *data)
-{
-	int retval;
-	
-	if (!(flags & MS_RDONLY) && sb->s_dev && is_read_only(sb->s_dev))
-		return -EACCES;
-		/*flags |= MS_RDONLY;*/
-	if (flags & MS_RDONLY)
-		acct_auto_close(sb->s_dev);
-	shrink_dcache_sb(sb);
-	fsync_super(sb);
-	/* If we are remounting RDONLY, make sure there are no rw files open */
-	if ((flags & MS_RDONLY) && !(sb->s_flags & MS_RDONLY))
-		if (!fs_may_remount_ro(sb))
-			return -EBUSY;
-	if (sb->s_op && sb->s_op->remount_fs) {
-		lock_super(sb);
-		retval = sb->s_op->remount_fs(sb, &flags, data);
-		unlock_super(sb);
-		if (retval)
-			return retval;
-	}
-	sb->s_flags = (sb->s_flags & ~MS_RMT_MASK) | (flags & MS_RMT_MASK);
-
-	/*
-	 * We can't invalidate inodes as we can loose data when remounting
-	 * (someone might manage to alter data while we are waiting in lock_super()
-	 * or in foo_remount_fs()))
-	 */
-
-	return 0;
-}
-
-struct vfsmount *do_kern_mount(char *type, int flags, char *name, void *data)
-{
-	struct file_system_type * fstype;
-	struct vfsmount *mnt = NULL;
-	struct super_block *sb;
-
-	if (!type || !memchr(type, 0, PAGE_SIZE))
-		return ERR_PTR(-EINVAL);
-
-	/* we need capabilities... */
-	if (!capable(CAP_SYS_ADMIN))
-		return ERR_PTR(-EPERM);
-
-	/* ... filesystem driver... */
-	fstype = get_fs_type(type);
-	if (!fstype)		
-		return ERR_PTR(-ENODEV);
-
-	/* ... allocated vfsmount... */
-	mnt = alloc_vfsmnt();
-	if (!mnt) {
-		mnt = ERR_PTR(-ENOMEM);
-		goto fs_out;
-	}
-	set_devname(mnt, name);
-	/* get locked superblock */
-	if (fstype->fs_flags & FS_REQUIRES_DEV)
-		sb = get_sb_bdev(fstype, name, flags, data);
-	else if (fstype->fs_flags & FS_SINGLE)
-		sb = get_sb_single(fstype, flags, data);
-	else
-		sb = get_sb_nodev(fstype, flags, data);
-
-	if (IS_ERR(sb)) {
-		free_vfsmnt(mnt);
-		mnt = (struct vfsmount *)sb;
-		goto fs_out;
-	}
-	if (fstype->fs_flags & FS_NOMOUNT)
-		sb->s_flags |= MS_NOUSER;
-
-	mnt->mnt_sb = sb;
-	mnt->mnt_root = dget(sb->s_root);
-	mnt->mnt_mountpoint = mnt->mnt_root;
-	mnt->mnt_parent = mnt;
-	up_write(&sb->s_umount);
-fs_out:
-	put_filesystem(fstype);
-	return mnt;
-}
-
 struct vfsmount *kern_mount(struct file_system_type *type)
 {
-	return do_kern_mount((char *)type->name, 0, (char *)type->name, NULL);
+	return do_kern_mount(type->name, 0, (char *)type->name, NULL);
 }

include/linux/blk.h

@@ -20,7 +20,6 @@ extern void add_blkdev_randomness(int major);
 #define INITRD_MINOR 250 /* shouldn't collide with /dev/ram* too soon ... */
 
 extern unsigned long initrd_start,initrd_end;
-extern int mount_initrd; /* zero if initrd should not be mounted */
 extern int initrd_below_start_ok; /* 1 if it is not an error if initrd_start < memory_start */
 extern int rd_doload;		/* 1 = load ramdisk, 0 = don't load */
 extern int rd_prompt;		/* 1 = prompt for ramdisk, 0 = don't prompt */
@@ -55,6 +54,9 @@ extern inline struct request *elv_next_request(request_queue_t *q)
 	while ((rq = __elv_next_request(q))) {
 		rq->flags |= REQ_STARTED;
 
+		if (&rq->queuelist == q->last_merge)
+			q->last_merge = NULL;
+
 		if ((rq->flags & REQ_DONTPREP) || !q->prep_rq_fn)
 			break;
 

include/linux/blkdev.h

@@ -129,17 +129,18 @@ enum blk_queue_state {
 
 struct request_queue
 {
-	/*
-	 * the queue request freelist, one for reads and one for writes
-	 */
-	struct request_list	rq[2];
-
 	/*
 	 * Together with queue_head for cacheline sharing
 	 */
 	struct list_head	queue_head;
+	struct list_head	*last_merge;
 	elevator_t		elevator;
 
+	/*
+	 * the queue request freelist, one for reads and one for writes
+	 */
+	struct request_list	rq[2];
+
 	request_fn_proc		*request_fn;
 	merge_request_fn	*back_merge_fn;
 	merge_request_fn	*front_merge_fn;
@@ -213,27 +214,25 @@ struct request_queue
 
 extern unsigned long blk_max_low_pfn, blk_max_pfn;
 
-#define BLK_BOUNCE_HIGH	(blk_max_low_pfn << PAGE_SHIFT)
-#define BLK_BOUNCE_ANY	(blk_max_pfn << PAGE_SHIFT)
+/*
+ * standard bounce addresses:
+ *
+ * BLK_BOUNCE_HIGH	: bounce all highmem pages
+ * BLK_BOUNCE_ANY	: don't bounce anything
+ * BLK_BOUNCE_ISA	: bounce pages above ISA DMA boundary
+ */
+#define BLK_BOUNCE_HIGH		((blk_max_low_pfn + 1) << PAGE_SHIFT)
+#define BLK_BOUNCE_ANY		((blk_max_pfn + 1) << PAGE_SHIFT)
 #define BLK_BOUNCE_ISA		(ISA_DMA_THRESHOLD)
 
-#ifdef CONFIG_HIGHMEM
-
+extern int init_emergency_isa_pool(void);
 extern void create_bounce(unsigned long pfn, int gfp, struct bio **bio_orig);
-extern void init_emergency_isa_pool(void);
 
 extern inline void blk_queue_bounce(request_queue_t *q, struct bio **bio)
 {
 	create_bounce(q->bounce_pfn, q->bounce_gfp, bio);
 }
 
-#else /* CONFIG_HIGHMEM */
-
-#define blk_queue_bounce(q, bio)	do { } while (0)
-#define init_emergency_isa_pool()	do { } while (0)
-
-#endif /* CONFIG_HIGHMEM */
-
 #define rq_for_each_bio(bio, rq)	\
 	if ((rq->bio))			\
 		for (bio = (rq)->bio; bio; bio = bio->bi_next)
@@ -275,9 +274,8 @@ extern void blk_plug_device(request_queue_t *);
 extern void blk_recount_segments(request_queue_t *, struct bio *);
 extern inline int blk_phys_contig_segment(request_queue_t *q, struct bio *, struct bio *);
 extern inline int blk_hw_contig_segment(request_queue_t *q, struct bio *, struct bio *);
-extern void blk_queue_assign_lock(request_queue_t *q, spinlock_t *);
-
 extern int block_ioctl(kdev_t, unsigned int, unsigned long);
+extern int ll_10byte_cmd_build(request_queue_t *, struct request *);
 
 /*
  * Access functions for manipulating queue properties
@@ -292,6 +290,9 @@ extern void blk_queue_max_hw_segments(request_queue_t *q, unsigned short);
 extern void blk_queue_max_segment_size(request_queue_t *q, unsigned int);
 extern void blk_queue_hardsect_size(request_queue_t *q, unsigned short);
 extern void blk_queue_segment_boundary(request_queue_t *q, unsigned long);
+extern void blk_queue_assign_lock(request_queue_t *q, spinlock_t *);
+extern void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn);
+
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);
 extern void generic_unplug_device(void *);

include/linux/blkdev.h.orig

-#ifndef _LINUX_BLKDEV_H
-#define _LINUX_BLKDEV_H
-
-#include <linux/major.h>
-#include <linux/sched.h>
-#include <linux/genhd.h>
-#include <linux/tqueue.h>
-#include <linux/list.h>
-#include <linux/mm.h>
-
-#include <asm/scatterlist.h>
-
-struct request_queue;
-typedef struct request_queue request_queue_t;
-struct elevator_s;
-typedef struct elevator_s elevator_t;
-
-struct request_list {
-	unsigned int count;
-	struct list_head free;
-	wait_queue_head_t wait;
-};
-
-struct request {
-	struct list_head queuelist; /* looking for ->queue? you must _not_
-				     * access it directly, use
-				     * blkdev_dequeue_request! */
-	int elevator_sequence;
-
-	unsigned char cmd[16];
-
-	unsigned long flags;		/* see REQ_ bits below */
-
-	int rq_status;	/* should split this into a few status bits */
-	kdev_t rq_dev;
-	int errors;
-	sector_t sector;
-	unsigned long nr_sectors;
-	unsigned long hard_sector;	/* the hard_* are block layer
-					 * internals, no driver should
-					 * touch them
-					 */
-	unsigned long hard_nr_sectors;
-
-	/* Number of scatter-gather DMA addr+len pairs after
-	 * physical address coalescing is performed.
-	 */
-	unsigned short nr_phys_segments;
-
-	/* Number of scatter-gather addr+len pairs after
-	 * physical and DMA remapping hardware coalescing is performed.
-	 * This is the number of scatter-gather entries the driver
-	 * will actually have to deal with after DMA mapping is done.
-	 */
-	unsigned short nr_hw_segments;
-
-	unsigned int current_nr_sectors;
-	unsigned int hard_cur_sectors;
-	void *special;
-	char *buffer;
-	struct completion *waiting;
-	struct bio *bio, *biotail;
-	request_queue_t *q;
-	struct request_list *rl;
-};
-
-/*
- * first three bits match BIO_RW* bits, important
- */
-enum rq_flag_bits {
-	__REQ_RW,	/* not set, read. set, write */
-	__REQ_RW_AHEAD,	/* READA */
-	__REQ_BARRIER,	/* may not be passed */
-	__REQ_CMD,	/* is a regular fs rw request */
-	__REQ_NOMERGE,	/* don't touch this for merging */
-	__REQ_STARTED,	/* drive already may have started this one */
-	__REQ_DONTPREP,	/* don't call prep for this one */
-	/*
-	 * for IDE
- 	*/
-	__REQ_DRIVE_CMD,
-	__REQ_DRIVE_TASK,
-
-	__REQ_PC,	/* packet command (special) */
-	__REQ_BLOCK_PC,	/* queued down pc from block layer */
-	__REQ_SENSE,	/* sense retrival */
-
-	__REQ_SPECIAL,	/* driver special command */
-
-	__REQ_NR_BITS,	/* stops here */
-};
-
-#define REQ_RW		(1 << __REQ_RW)
-#define REQ_RW_AHEAD	(1 << __REQ_RW_AHEAD)
-#define REQ_BARRIER	(1 << __REQ_BARRIER)
-#define REQ_CMD		(1 << __REQ_CMD)
-#define REQ_NOMERGE	(1 << __REQ_NOMERGE)
-#define REQ_STARTED	(1 << __REQ_STARTED)
-#define REQ_DONTPREP	(1 << __REQ_DONTPREP)
-#define REQ_DRIVE_CMD	(1 << __REQ_DRIVE_CMD)
-#define REQ_DRIVE_TASK	(1 << __REQ_DRIVE_TASK)
-#define REQ_PC		(1 << __REQ_PC)
-#define REQ_SENSE	(1 << __REQ_SENSE)
-#define REQ_BLOCK_PC	(1 << __REQ_BLOCK_PC)
-#define REQ_SPECIAL	(1 << __REQ_SPECIAL)
-
-#include <linux/elevator.h>
-
-typedef int (merge_request_fn) (request_queue_t *, struct request *,
-				struct bio *);
-typedef int (merge_requests_fn) (request_queue_t *, struct request *,
-				 struct request *);
-typedef void (request_fn_proc) (request_queue_t *q);
-typedef request_queue_t * (queue_proc) (kdev_t dev);
-typedef int (make_request_fn) (request_queue_t *q, struct bio *bio);
-typedef int (prep_rq_fn) (request_queue_t *, struct request *);
-typedef void (unplug_device_fn) (void *q);
-
-enum blk_queue_state {
-	Queue_down,
-	Queue_up,
-};
-
-/*
- * Default nr free requests per queue, ll_rw_blk will scale it down
- * according to available RAM at init time
- */
-#define QUEUE_NR_REQUESTS	8192
-
-struct request_queue
-{
-	/*
-	 * the queue request freelist, one for reads and one for writes
-	 */
-	struct request_list	rq[2];
-
-	/*
-	 * Together with queue_head for cacheline sharing
-	 */
-	struct list_head	queue_head;
-	elevator_t		elevator;
-
-	request_fn_proc		*request_fn;
-	merge_request_fn	*back_merge_fn;
-	merge_request_fn	*front_merge_fn;
-	merge_requests_fn	*merge_requests_fn;
-	make_request_fn		*make_request_fn;
-	prep_rq_fn		*prep_rq_fn;
-
-	/*
-	 * The queue owner gets to use this for whatever they like.
-	 * ll_rw_blk doesn't touch it.
-	 */
-	void			*queuedata;
-
-	/*
-	 * queue needs bounce pages for pages above this limit
-	 */
-	unsigned long		bounce_pfn;
-	int			bounce_gfp;
-
-	/*
-	 * This is used to remove the plug when tq_disk runs.
-	 */
-	struct tq_struct	plug_tq;
-
-	/*
-	 * various queue flags, see QUEUE_* below
-	 */
-	unsigned long		queue_flags;
-
-	/*
-	 * protects queue structures from reentrancy
-	 */
-	spinlock_t		*queue_lock;
-
-	/*
-	 * queue settings
-	 */
-	unsigned short		max_sectors;
-	unsigned short		max_phys_segments;
-	unsigned short		max_hw_segments;
-	unsigned short		hardsect_size;
-	unsigned int		max_segment_size;
-
-	unsigned long		seg_boundary_mask;
-
-	wait_queue_head_t	queue_wait;
-};
-
-#define RQ_INACTIVE		(-1)
-#define RQ_ACTIVE		1
-#define RQ_SCSI_BUSY		0xffff
-#define RQ_SCSI_DONE		0xfffe
-#define RQ_SCSI_DISCONNECTING	0xffe0
-
-#define QUEUE_FLAG_PLUGGED	0	/* queue is plugged */
-#define QUEUE_FLAG_NOSPLIT	1	/* can process bio over several goes */
-#define QUEUE_FLAG_CLUSTER	2	/* cluster several segments into 1 */
-
-#define blk_queue_plugged(q)	test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
-#define blk_mark_plugged(q)	set_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
-#define blk_queue_empty(q)	elv_queue_empty(q)
-#define list_entry_rq(ptr)	list_entry((ptr), struct request, queuelist)
-
-#define rq_data_dir(rq)		((rq)->flags & 1)
-
-/*
- * noop, requests are automagically marked as active/inactive by I/O
- * scheduler -- see elv_next_request
- */
-#define blk_queue_headactive(q, head_active)
-
-extern unsigned long blk_max_low_pfn, blk_max_pfn;
-
-#define BLK_BOUNCE_HIGH	(blk_max_low_pfn << PAGE_SHIFT)
-#define BLK_BOUNCE_ANY	(blk_max_pfn << PAGE_SHIFT)
-#define BLK_BOUNCE_ISA	(ISA_DMA_THRESHOLD)
-
-#ifdef CONFIG_HIGHMEM
-
-extern void create_bounce(unsigned long pfn, int gfp, struct bio **bio_orig);
-extern void init_emergency_isa_pool(void);
-
-extern inline void blk_queue_bounce(request_queue_t *q, struct bio **bio)
-{
-	create_bounce(q->bounce_pfn, q->bounce_gfp, bio);
-}
-
-#else /* CONFIG_HIGHMEM */
-
-#define blk_queue_bounce(q, bio)	do { } while (0)
-#define init_emergency_isa_pool()	do { } while (0)
-
-#endif /* CONFIG_HIGHMEM */
-
-#define rq_for_each_bio(bio, rq)	\
-	if ((rq->bio))			\
-		for (bio = (rq)->bio; bio; bio = bio->bi_next)
-
-struct blk_dev_struct {
-	/*
-	 * queue_proc has to be atomic
-	 */
-	request_queue_t		request_queue;
-	queue_proc		*queue;
-	void			*data;
-};
-
-struct sec_size {
-	unsigned block_size;
-	unsigned block_size_bits;
-};
-
-/*
- * Used to indicate the default queue for drivers that don't bother
- * to implement multiple queues.  We have this access macro here
- * so as to eliminate the need for each and every block device
- * driver to know about the internal structure of blk_dev[].
- */
-#define BLK_DEFAULT_QUEUE(_MAJOR)  &blk_dev[_MAJOR].request_queue
-
-extern struct sec_size * blk_sec[MAX_BLKDEV];
-extern struct blk_dev_struct blk_dev[MAX_BLKDEV];
-extern void grok_partitions(kdev_t dev, long size);
-extern int wipe_partitions(kdev_t dev);
-extern void register_disk(struct gendisk *dev, kdev_t first, unsigned minors, struct block_device_operations *ops, long size);
-extern void generic_make_request(struct bio *bio);
-extern inline request_queue_t *blk_get_queue(kdev_t dev);
-extern void blkdev_release_request(struct request *);
-extern void blk_attempt_remerge(request_queue_t *, struct request *);
-extern struct request *blk_get_request(request_queue_t *, int, int);
-extern void blk_put_request(struct request *);
-extern void blk_plug_device(request_queue_t *);
-extern void blk_recount_segments(request_queue_t *, struct bio *);
-extern inline int blk_phys_contig_segment(request_queue_t *q, struct bio *, struct bio *);
-extern inline int blk_hw_contig_segment(request_queue_t *q, struct bio *, struct bio *);
-extern void blk_queue_assign_lock(request_queue_t *q, spinlock_t *);
-
-extern int block_ioctl(kdev_t, unsigned int, unsigned long);
-
-/*
- * Access functions for manipulating queue properties
- */
-extern int blk_init_queue(request_queue_t *, request_fn_proc *, spinlock_t *);
-extern void blk_cleanup_queue(request_queue_t *);
-extern void blk_queue_make_request(request_queue_t *, make_request_fn *);
-extern void blk_queue_bounce_limit(request_queue_t *, u64);
-extern void blk_queue_max_sectors(request_queue_t *q, unsigned short);
-extern void blk_queue_max_phys_segments(request_queue_t *q, unsigned short);
-extern void blk_queue_max_hw_segments(request_queue_t *q, unsigned short);
-extern void blk_queue_max_segment_size(request_queue_t *q, unsigned int);
-extern void blk_queue_hardsect_size(request_queue_t *q, unsigned short);
-extern void blk_queue_segment_boundary(request_queue_t *q, unsigned long);
-extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
-extern void blk_dump_rq_flags(struct request *, char *);
-extern void generic_unplug_device(void *);
-
-extern int * blk_size[MAX_BLKDEV];
-
-extern int * blksize_size[MAX_BLKDEV];
-
-extern int * max_readahead[MAX_BLKDEV];
-
-#define MAX_PHYS_SEGMENTS 128
-#define MAX_HW_SEGMENTS 128
-#define MAX_SECTORS 255
-
-#define MAX_SEGMENT_SIZE	65536
-
-/* read-ahead in pages.. */
-#define MAX_READAHEAD	31
-#define MIN_READAHEAD	3
-
-#define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
-#define blkdev_entry_next_request(entry) blkdev_entry_to_request((entry)->next)
-#define blkdev_entry_prev_request(entry) blkdev_entry_to_request((entry)->prev)
-#define blkdev_next_request(req) blkdev_entry_to_request((req)->queuelist.next)
-#define blkdev_prev_request(req) blkdev_entry_to_request((req)->queuelist.prev)
-
-extern void drive_stat_acct(struct request *, int, int);
-
-extern inline void blk_clear(int major)
-{
-	blk_size[major] = NULL;
-#if 0
-	blk_size_in_bytes[major] = NULL;
-#endif
-	blksize_size[major] = NULL;
-	max_readahead[major] = NULL;
-	read_ahead[major] = 0;
-}
-
-extern inline int get_hardsect_size(kdev_t dev)
-{
-	request_queue_t *q = blk_get_queue(dev);
-	int retval = 512;
-
-	if (q && q->hardsect_size)
-		retval = q->hardsect_size;
-
-	return retval;
-}
-
-#define blk_finished_io(nsects)	do { } while (0)
-#define blk_started_io(nsects)	do { } while (0)
-
-extern inline unsigned int blksize_bits(unsigned int size)
-{
-	unsigned int bits = 8;
-	do {
-		bits++;
-		size >>= 1;
-	} while (size > 256);
-	return bits;
-}
-
-extern inline unsigned int block_size(kdev_t dev)
-{
-	int retval = BLOCK_SIZE;
-	int major = MAJOR(dev);
-
-	if (blksize_size[major]) {
-		int minor = MINOR(dev);
-		if (blksize_size[major][minor])
-			retval = blksize_size[major][minor];
-	}
-	return retval;
-}
-
-#endif

include/linux/elevator.h

 #ifndef _LINUX_ELEVATOR_H
 #define _LINUX_ELEVATOR_H
 
-typedef void (elevator_fn) (struct request *, elevator_t *,
-			    struct list_head *,
-			    struct list_head *, int);
-
 typedef int (elevator_merge_fn) (request_queue_t *, struct request **,
-				 struct list_head *, struct bio *);
+				 struct bio *);
 
 typedef void (elevator_merge_cleanup_fn) (request_queue_t *, struct request *, int);
 
@@ -21,8 +17,7 @@ typedef void (elevator_exit_fn) (request_queue_t *, elevator_t *);
 
 struct elevator_s
 {
-	int read_latency;
-	int write_latency;
+	int latency[2];
 
 	elevator_merge_fn *elevator_merge_fn;
 	elevator_merge_cleanup_fn *elevator_merge_cleanup_fn;
@@ -35,11 +30,11 @@ struct elevator_s
 	elevator_exit_fn *elevator_exit_fn;
 };
 
-int elevator_noop_merge(request_queue_t *, struct request **, struct list_head *, struct bio *);
+int elevator_noop_merge(request_queue_t *, struct request **, struct bio *);
 void elevator_noop_merge_cleanup(request_queue_t *, struct request *, int);
 void elevator_noop_merge_req(struct request *, struct request *);
 
-int elevator_linus_merge(request_queue_t *, struct request **, struct list_head *, struct bio *);
+int elevator_linus_merge(request_queue_t *, struct request **, struct bio *);
 void elevator_linus_merge_cleanup(request_queue_t *, struct request *, int);
 void elevator_linus_merge_req(struct request *, struct request *);
 int elv_linus_init(request_queue_t *, elevator_t *);
@@ -69,32 +64,7 @@ extern void elevator_exit(request_queue_t *, elevator_t *);
 #define ELEVATOR_FRONT_MERGE	1
 #define ELEVATOR_BACK_MERGE	2
 
-/*
- * This is used in the elevator algorithm.  We don't prioritise reads
- * over writes any more --- although reads are more time-critical than
- * writes, by treating them equally we increase filesystem throughput.
- * This turns out to give better overall performance.  -- sct
- */
-#define IN_ORDER(s1,s2)				\
-	((((s1)->rq_dev == (s2)->rq_dev &&	\
-	   (s1)->sector < (s2)->sector)) ||	\
-	 (s1)->rq_dev < (s2)->rq_dev)
-
-#define BHRQ_IN_ORDER(bh, rq)			\
-	((((bh)->b_rdev == (rq)->rq_dev &&	\
-	   (bh)->b_rsector < (rq)->sector)) ||	\
-	 (bh)->b_rdev < (rq)->rq_dev)
-
-static inline int elevator_request_latency(elevator_t * elevator, int rw)
-{
-	int latency;
-
-	latency = elevator->read_latency;
-	if (rw != READ)
-		latency = elevator->write_latency;
-
-	return latency;
-}
+#define elevator_request_latency(e, rw)	((e)->latency[(rw) & 1])
 
 /*
  * will change once we move to a more complex data structure than a simple
@@ -116,9 +86,7 @@ struct elv_linus_data {
 
 #define ELEVATOR_NOOP							\
 ((elevator_t) {								\
-	0,				/* read_latency */		\
-	0,				/* write_latency */		\
-									\
+	{ 0, 0},							\
 	elevator_noop_merge,		/* elevator_merge_fn */		\
 	elevator_noop_merge_cleanup,	/* elevator_merge_cleanup_fn */	\
 	elevator_noop_merge_req,	/* elevator_merge_req_fn */	\
@@ -130,9 +98,7 @@ struct elv_linus_data {
 
 #define ELEVATOR_LINUS							\
 ((elevator_t) {								\
-	8192,				/* read passovers */		\
-	16384,				/* write passovers */		\
-									\
+	{ 8192, 16384 },						\
 	elevator_linus_merge,		/* elevator_merge_fn */		\
 	elevator_linus_merge_cleanup,	/* elevator_merge_cleanup_fn */	\
 	elevator_linus_merge_req,	/* elevator_merge_req_fn */	\

include/linux/raid/md_k.h

@@ -240,7 +240,7 @@ struct mdk_personality_s
 
 	int (*stop_resync)(mddev_t *mddev);
 	int (*restart_resync)(mddev_t *mddev);
-	int (*sync_request)(mddev_t *mddev, sector_t sector_nr);
+	int (*sync_request)(mddev_t *mddev, sector_t sector_nr, int go_faster);
 };
 
 

include/linux/raid/raid1.h

@@ -9,8 +9,8 @@ struct mirror_info {
 	int		number;
 	int		raid_disk;
 	kdev_t		dev;
-	int		sect_limit;
-	int		head_position;
+	sector_t	head_position;
+	atomic_t	nr_pending;
 
 	/*
 	 * State bits:
@@ -31,23 +31,21 @@ struct r1_private_data_s {
 	int			raid_disks;
 	int			working_disks;
 	int			last_used;
-	sector_t		next_sect;
-	int			sect_count;
+	sector_t		next_seq_sect;
 	mdk_thread_t		*thread, *resync_thread;
 	int			resync_mirrors;
 	mirror_info_t		*spare;
 	spinlock_t		device_lock;
 
 	/* for use when syncing mirrors: */
-	unsigned long	start_active, start_ready,
-		start_pending, start_future;
-	int	cnt_done, cnt_active, cnt_ready,
-		cnt_pending, cnt_future;
-	int	phase;
-	int	window;
-	wait_queue_head_t	wait_done;
-	wait_queue_head_t	wait_ready;
-	spinlock_t		segment_lock;
+
+	spinlock_t		resync_lock;
+	int nr_pending;
+	int barrier;
+	sector_t		next_resync;
+
+	wait_queue_head_t	wait_idle;
+	wait_queue_head_t	wait_resume;
 
 	mempool_t *r1bio_pool;
 	mempool_t *r1buf_pool;
@@ -62,7 +60,8 @@ typedef struct r1_private_data_s conf_t;
 #define mddev_to_conf(mddev) ((conf_t *) mddev->private)
 
 /*
- * this is our 'private' 'collective' RAID1 buffer head.
+ * this is our 'private' RAID1 bio.
+ *
  * it contains information about what kind of IO operations were started
  * for this RAID1 operation, and about their status:
  */
@@ -83,6 +82,7 @@ struct r1bio_s {
 	 * if the IO is in READ direction, then this bio is used:
 	 */
 	struct bio		*read_bio;
+	int			read_disk;
 	/*
 	 * if the IO is in WRITE direction, then multiple bios are used:
 	 */
@@ -94,5 +94,5 @@ struct r1bio_s {
 
 /* bits for r1bio.state */
 #define	R1BIO_Uptodate	1
-#define	R1BIO_SyncPhase	2
+
 #endif

include/linux/vt_kern.h

@@ -9,6 +9,7 @@
 #include <linux/config.h>
 #include <linux/vt.h>
 #include <linux/kd.h>
+#include <linux/tty.h>
 
 /*
  * Presently, a lot of graphics programs do not restore the contents of

init/do_mounts.c

@@ -3,12 +3,8 @@
 #include <linux/slab.h>
 #include <linux/devfs_fs_kernel.h>
 #include <linux/unistd.h>
-#include <linux/string.h>
 #include <linux/ctype.h>
-#include <linux/init.h>
-#include <linux/smp_lock.h>
 #include <linux/blk.h>
-#include <linux/tty.h>
 #include <linux/fd.h>
 
 #include <linux/nfs_fs.h>
@@ -18,8 +14,6 @@
 #include <linux/ext2_fs.h>
 #include <linux/romfs_fs.h>
 
-#include <asm/uaccess.h>
-
 #define BUILD_CRAMDISK
 
 extern int get_filesystem_list(char * buf);
@@ -38,12 +32,21 @@ asmlinkage long sys_ioctl(int fd, int cmd, unsigned long arg);
 
 #ifdef CONFIG_BLK_DEV_INITRD
 unsigned int real_root_dev;	/* do_proc_dointvec cannot handle kdev_t */
-#endif
-#ifdef CONFIG_BLK_DEV_RAM
-extern int rd_doload;
+static int __initdata mount_initrd = 1;
+
+static int __init no_initrd(char *str)
+{
+	mount_initrd = 0;
+	return 1;
+}
+
+__setup("noinitrd", no_initrd);
 #else
-static int rd_doload = 0;
+static int __initdata mount_initrd = 0;
 #endif
+
+int __initdata rd_doload;	/* 1 = load RAM disk, 0 = don't load */
+
 int root_mountflags = MS_RDONLY | MS_VERBOSE;
 static char root_device_name[64];
 
@@ -52,6 +55,13 @@ kdev_t ROOT_DEV;
 
 static int do_devfs = 0;
 
+static int __init load_ramdisk(char *str)
+{
+	rd_doload = simple_strtol(str,NULL,0) & 3;
+	return 1;
+}
+__setup("load_ramdisk=", load_ramdisk);
+
 static int __init readonly(char *str)
 {
 	if (*str)
@@ -371,6 +381,24 @@ static void __init change_floppy(char *fmt, ...)
 
 #ifdef CONFIG_BLK_DEV_RAM
 
+int __initdata rd_prompt = 1;	/* 1 = prompt for RAM disk, 0 = don't prompt */
+
+static int __init prompt_ramdisk(char *str)
+{
+	rd_prompt = simple_strtol(str,NULL,0) & 1;
+	return 1;
+}
+__setup("prompt_ramdisk=", prompt_ramdisk);
+
+int __initdata rd_image_start;		/* starting block # of image */
+
+static int __init ramdisk_start_setup(char *str)
+{
+	rd_image_start = simple_strtol(str,NULL,0);
+	return 1;
+}
+__setup("ramdisk_start=", ramdisk_start_setup);
+
 static int __init crd_load(int in_fd, int out_fd);
 
 /*
@@ -588,7 +616,6 @@ static int __init rd_load_image(char *from)
 static int __init rd_load_disk(int n)
 {
 #ifdef CONFIG_BLK_DEV_RAM
-	extern int rd_prompt;
 	if (rd_prompt)
 		change_floppy("root floppy disk to be loaded into RAM disk");
 	create_dev("/dev/ram", MKDEV(RAMDISK_MAJOR, n), NULL);
@@ -715,13 +742,10 @@ static int __init initrd_load(void)
  */
 void prepare_namespace(void)
 {
-	int do_initrd = 0;
 	int is_floppy = MAJOR(ROOT_DEV) == FLOPPY_MAJOR;
 #ifdef CONFIG_BLK_DEV_INITRD
 	if (!initrd_start)
 		mount_initrd = 0;
-	if (mount_initrd)
-		do_initrd = 1;
 	real_root_dev = ROOT_DEV;
 #endif
 	sys_mkdir("/dev", 0700);
@@ -732,7 +756,7 @@ void prepare_namespace(void)
 #endif
 
 	create_dev("/dev/root", ROOT_DEV, NULL);
-	if (do_initrd) {
+	if (mount_initrd) {
 		if (initrd_load() && ROOT_DEV != MKDEV(RAMDISK_MAJOR, 0)) {
 			handle_initrd();
 			goto out;

kernel/ksyms.c

@@ -116,11 +116,11 @@ EXPORT_SYMBOL(vmtruncate);
 EXPORT_SYMBOL(find_vma);
 EXPORT_SYMBOL(get_unmapped_area);
 EXPORT_SYMBOL(init_mm);
+EXPORT_SYMBOL(create_bounce);
 #ifdef CONFIG_HIGHMEM
 EXPORT_SYMBOL(kmap_high);
 EXPORT_SYMBOL(kunmap_high);
 EXPORT_SYMBOL(highmem_start_page);
-EXPORT_SYMBOL(create_bounce);
 EXPORT_SYMBOL(kmap_prot);
 EXPORT_SYMBOL(kmap_pte);
 #endif

kernel/signal.c

@@ -649,10 +649,8 @@ kill_proc_info(int sig, struct siginfo *info, pid_t pid)
 /*
  * kill_something_info() interprets pid in interesting ways just like kill(2).
  *
- * POSIX (2001) specifies "If pid is -1, sig shall be sent to all processes
- * (excluding an unspecified set of system processes) for which the process
- * has permission to send that signal."
- * So, probably the process should also signal itself.
+ * POSIX specifies that kill(-1,sig) is unspecified, but what we have
+ * is probably wrong.  Should make it like BSD or SYSV.
  */
 
 static int kill_something_info(int sig, struct siginfo *info, int pid)
@@ -665,7 +663,7 @@ static int kill_something_info(int sig, struct siginfo *info, int pid)
 
 		read_lock(&tasklist_lock);
 		for_each_task(p) {
-			if (p->pid > 1) {
+			if (p->pid > 1 && p != current) {
 				int err = send_sig_info(sig, info, p);
 				++count;
 				if (err != -EPERM)

mm/Makefile

@@ -14,8 +14,6 @@ export-objs := shmem.o filemap.o mempool.o
 obj-y	 := memory.o mmap.o filemap.o mprotect.o mlock.o mremap.o \
 	    vmalloc.o slab.o bootmem.o swap.o vmscan.o page_io.o \
 	    page_alloc.o swap_state.o swapfile.o numa.o oom_kill.o \
-	    shmem.o mempool.o
-
-obj-$(CONFIG_HIGHMEM) += highmem.o
+	    shmem.o highmem.o mempool.o
 
 include $(TOPDIR)/Rules.make

mm/highmem.c

@@ -19,6 +19,19 @@
 #include <linux/mm.h>
 #include <linux/pagemap.h>
 #include <linux/mempool.h>
+#include <linux/blkdev.h>
+
+static mempool_t *page_pool, *isa_page_pool;
+
+static void *page_pool_alloc(int gfp_mask, void *data)
+{
+	return alloc_page(gfp_mask);
+}
+
+static void page_pool_free(void *page, void *data)
+{
+	__free_page(page);
+}
 
 /*
  * Virtual_count is not a pure "count".
@@ -28,6 +41,7 @@
  *    since the last TLB flush - so we can't use it.
  *  n means that there are (n-1) current users of it.
  */
+#ifdef CONFIG_HIGHMEM
 static int pkmap_count[LAST_PKMAP];
 static unsigned int last_pkmap_nr;
 static spinlock_t kmap_lock = SPIN_LOCK_UNLOCKED;
@@ -185,19 +199,6 @@ void kunmap_high(struct page *page)
 }
 
 #define POOL_SIZE	64
-#define ISA_POOL_SIZE	16
-
-static mempool_t *page_pool, *isa_page_pool;
-
-static void *page_pool_alloc(int gfp_mask, void *data)
-{
-	return alloc_page(gfp_mask);
-}
-
-static void page_pool_free(void *page, void *data)
-{
-	__free_page(page);
-}
 
 static __init int init_emergency_pool(void)
 {
@@ -211,11 +212,37 @@ static __init int init_emergency_pool(void)
 	page_pool = mempool_create(POOL_SIZE, page_pool_alloc, page_pool_free, NULL);
 	if (!page_pool)
 		BUG();
-	printk("highmem bounce pool size: %d pages and bhs.\n", POOL_SIZE);
+	printk("highmem bounce pool size: %d pages\n", POOL_SIZE);
 
 	return 0;
 }
 
+__initcall(init_emergency_pool);
+
+/*
+ * highmem version, map in to vec
+ */
+static inline void bounce_copy_vec(struct bio_vec *to, unsigned char *vfrom)
+{
+	unsigned long flags;
+	unsigned char *vto;
+
+	local_irq_save(flags);
+	vto = kmap_atomic(to->bv_page, KM_BOUNCE_READ);
+	memcpy(vto + to->bv_offset, vfrom, to->bv_len);
+	kunmap_atomic(vto, KM_BOUNCE_READ);
+	local_irq_restore(flags);
+}
+
+#else /* CONFIG_HIGHMEM */
+
+#define bounce_copy_vec(to, vfrom)	\
+	memcpy(page_address((to)->bv_page) + (to)->bv_offset, vfrom, (to)->bv_len)
+
+#endif
+
+#define ISA_POOL_SIZE	16
+
 /*
  * gets called "every" time someone init's a queue with BLK_BOUNCE_ISA
  * as the max address, so check if the pool has already been created.
@@ -233,8 +260,6 @@ int init_emergency_isa_pool(void)
 	return 0;
 }
 
-__initcall(init_emergency_pool);
-
 /*
  * Simple bounce buffer support for highmem pages. Depending on the
  * queue gfp mask set, *to may or may not be a highmem page. kmap it
@@ -242,8 +267,7 @@ __initcall(init_emergency_pool);
  */
 static inline void copy_to_high_bio_irq(struct bio *to, struct bio *from)
 {
-	unsigned char *vto, *vfrom;
-	unsigned long flags;
+	unsigned char *vfrom;
 	struct bio_vec *tovec, *fromvec;
 	int i;
 
@@ -258,11 +282,7 @@ static inline void copy_to_high_bio_irq(struct bio *to, struct bio *from)
 
 		vfrom = page_address(fromvec->bv_page) + fromvec->bv_offset;
 
-		local_irq_save(flags);
-		vto = kmap_atomic(tovec->bv_page, KM_BOUNCE_READ);
-		memcpy(vto + tovec->bv_offset, vfrom, tovec->bv_len);
-		kunmap_atomic(vto, KM_BOUNCE_READ);
-		local_irq_restore(flags);
+		bounce_copy_vec(tovec, vfrom);
 	}
 }
 
@@ -336,10 +356,25 @@ void create_bounce(unsigned long pfn, int gfp, struct bio **bio_orig)
 
 	BUG_ON((*bio_orig)->bi_idx);
 
+	/*
+	 * for non-isa bounce case, just check if the bounce pfn is equal
+	 * to or bigger than the highest pfn in the system -- in that case,
+	 * don't waste time iterating over bio segments
+	 */
 	if (!(gfp & GFP_DMA)) {
+		if (pfn >= blk_max_pfn)
+			return;
+
+#ifndef CONFIG_HIGHMEM
+		/*
+		 * should not hit for non-highmem case
+		 */
+		BUG();
+#endif
 		bio_gfp = GFP_NOHIGHIO;
 		pool = page_pool;
 	} else {
+		BUG_ON(!isa_page_pool);
 		bio_gfp = GFP_NOIO;
 		pool = isa_page_pool;
 	}