--- /opt/kernel/linux-2.4.0-test11-pre5/drivers/block/loop.c Tue Nov 14 22:53:17 2000 +++ drivers/block/loop.c Thu Nov 16 18:09:56 2000 @@ -31,6 +31,9 @@ * max_loop=<1-255> to the kernel on boot. * Erik I. Bolsų, , Oct 31, 1999 * + * Rip out request handling and turn loop into a block remapper. + * Jens Axboe , Nov 2000 + * * Still To Fix: * - Advisory locking is ignored here. * - Should use an own CAP_* category instead of CAP_SYS_ADMIN @@ -56,7 +59,9 @@ #include #include #include - +#include +#include +#include #include #include @@ -66,15 +71,6 @@ #define MAJOR_NR LOOP_MAJOR -#define DEVICE_NAME "loop" -#define DEVICE_REQUEST do_lo_request -#define DEVICE_NR(device) (MINOR(device)) -#define DEVICE_ON(device) -#define DEVICE_OFF(device) -#define DEVICE_NO_RANDOM -#define TIMEOUT_VALUE (6 * HZ) -#include - #include static int max_loop = 8; static struct loop_device *loop_dev; @@ -82,24 +78,21 @@ static int *loop_blksizes; static devfs_handle_t devfs_handle; /* For the directory */ -#define FALSE 0 -#define TRUE (!FALSE) - -/* - * Transfer functions - */ static int transfer_none(struct loop_device *lo, int cmd, char *raw_buf, - char *loop_buf, int size, int real_block) + char *loop_buf, int size, int real_block) { - if (cmd == READ) - memcpy(loop_buf, raw_buf, size); - else - memcpy(raw_buf, loop_buf, size); + if (raw_buf != loop_buf) { + if (cmd == READ) + memcpy(loop_buf, raw_buf, size); + else + memcpy(raw_buf, loop_buf, size); + } + return 0; } static int transfer_xor(struct loop_device *lo, int cmd, char *raw_buf, - char *loop_buf, int size, int real_block) + char *loop_buf, int size, int real_block) { char *in, *out, *key; int i, keysize; @@ -111,17 +104,18 @@ in = loop_buf; out = raw_buf; } + key = lo->lo_encrypt_key; keysize = lo->lo_encrypt_key_size; - for (i=0; i < size; i++) + for (i = 0; i < size; i++) *out++ = *in++ ^ key[(i & 511) % keysize]; return 0; } static int none_status(struct loop_device *lo, struct loop_info *info) { - return 0; -} + return 0; +} static int xor_status(struct loop_device *lo, struct loop_info *info) { @@ -133,7 +127,7 @@ struct loop_func_table none_funcs = { number: LO_CRYPT_NONE, transfer: transfer_none, - init: none_status + init: none_status, }; struct loop_func_table xor_funcs = { @@ -168,8 +162,7 @@ loop_sizes[lo->lo_number] = size; } -static int lo_send(struct loop_device *lo, char *data, int len, loff_t pos, - int blksize) +static int lo_send(struct loop_device *lo, char *data, int len, loff_t pos) { struct file *file = lo->lo_backing_file; /* kudos to NFsckingS */ struct address_space *mapping = lo->lo_dentry->d_inode->i_mapping; @@ -182,7 +175,7 @@ index = pos >> PAGE_CACHE_SHIFT; offset = pos & (PAGE_CACHE_SIZE - 1); while (len > 0) { - int IV = index * (PAGE_CACHE_SIZE/blksize) + offset/blksize; + int IV = index * (PAGE_CACHE_SIZE/lo->lo_blksize) + offset/lo->lo_blksize; size = PAGE_CACHE_SIZE - offset; if (size > len) size = len; @@ -190,10 +183,13 @@ page = grab_cache_page(mapping, index); if (!page) goto fail; + if (!PageLocked(page)) + PAGE_BUG(page); if (aops->prepare_write(file, page, offset, offset+size)) goto unlock; kaddr = page_address(page); - if ((lo->transfer)(lo, WRITE, kaddr+offset, data, size, IV)) + flush_dcache_page(page); + if (lo_do_transfer(lo, WRITE, kaddr + offset, data, size, IV)) goto write_fail; if (aops->commit_write(file, page, offset, offset+size)) goto unlock; @@ -203,6 +199,7 @@ index++; pos += size; UnlockPage(page); + deactivate_page(page); page_cache_release(page); } return 0; @@ -213,6 +210,7 @@ kunmap(page); unlock: UnlockPage(page); + deactivate_page(page); page_cache_release(page); fail: return -1; @@ -221,7 +219,6 @@ struct lo_read_data { struct loop_device *lo; char *data; - int blksize; }; static int lo_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size) @@ -230,16 +227,15 @@ unsigned long count = desc->count; struct lo_read_data *p = (struct lo_read_data*)desc->buf; struct loop_device *lo = p->lo; - int IV = page->index * (PAGE_CACHE_SIZE/p->blksize) + offset/p->blksize; + int IV = page->index * (PAGE_CACHE_SIZE/lo->lo_blksize) + offset/lo->lo_blksize; if (size > count) size = count; kaddr = kmap(page); - if ((lo->transfer)(lo,READ,kaddr+offset,p->data,size,IV)) { + if (lo_do_transfer(lo, READ, kaddr + offset, p->data, size, IV)) { size = 0; - printk(KERN_ERR "loop: transfer error block %ld\n", - page->index); + printk(KERN_ERR "loop: transfer error block %ld\n",page->index); desc->error = -EINVAL; } kunmap(page); @@ -250,8 +246,7 @@ return size; } -static int lo_receive(struct loop_device *lo, char *data, int len, loff_t pos, - int blksize) +static int lo_receive(struct loop_device *lo, char *data, int len, loff_t pos) { struct file *file = lo->lo_backing_file; struct lo_read_data cookie; @@ -259,7 +254,6 @@ cookie.lo = lo; cookie.data = data; - cookie.blksize = blksize; desc.written = 0; desc.count = len; desc.buf = (char*)&cookie; @@ -268,135 +262,99 @@ return desc.error; } -static void do_lo_request(request_queue_t * q) +static int do_bh_filebacked(struct loop_device *lo, struct buffer_head *bh, int rw) { - int block, offset, len, blksize, size; - char *dest_addr; - struct loop_device *lo; - struct buffer_head *bh; - struct request *current_request; loff_t pos; + int ret; -repeat: - INIT_REQUEST; - current_request=CURRENT; - blkdev_dequeue_request(current_request); - if (MINOR(current_request->rq_dev) >= max_loop) - goto error_out; - lo = &loop_dev[MINOR(current_request->rq_dev)]; - if (!lo->lo_dentry || !lo->transfer) - goto error_out; - if (current_request->cmd == WRITE) { - if (lo->lo_flags & LO_FLAGS_READ_ONLY) - goto error_out; - } else if (current_request->cmd != READ) { - printk(KERN_ERR "unknown loop device command (%d)?!?", - current_request->cmd); - goto error_out; - } + pos = ((loff_t) bh->b_rsector << 9) + lo->lo_offset; - dest_addr = current_request->buffer; - len = current_request->current_nr_sectors << 9; + if (rw == WRITE) + ret = lo_send(lo, bh->b_data, bh->b_size, pos); + else + ret = lo_receive(lo, bh->b_data, bh->b_size, pos); - blksize = BLOCK_SIZE; - if (blksize_size[MAJOR(lo->lo_device)]) { - blksize = blksize_size[MAJOR(lo->lo_device)][MINOR(lo->lo_device)]; - if (!blksize) - blksize = BLOCK_SIZE; - } + return ret; +} - if (lo->lo_flags & LO_FLAGS_DO_BMAP) - goto file_backed; +/* + * when buffer read has completed, invoke transfer func and orig end_io + */ +static void loop_end_io_transfer(struct buffer_head *bh, int uptodate) +{ + struct loop_device *lo = &loop_dev[MINOR(bh->b_dev)]; + bh_end_io_t *end_io = bh->b_private; + char *buf = bh->b_data; - if (blksize < 512) { - block = current_request->sector * (512/blksize); - offset = 0; - } else { - block = current_request->sector / (blksize >> 9); - offset = (current_request->sector % (blksize >> 9)) << 9; - } - block += lo->lo_offset / blksize; - offset += lo->lo_offset % blksize; - if (offset >= blksize) { - block++; - offset -= blksize; - } - spin_unlock_irq(&io_request_lock); + if (uptodate) + uptodate = !lo_do_transfer(lo, READ, buf, buf, bh->b_size, bh->b_blocknr); - while (len > 0) { + end_io(bh, uptodate); +} - size = blksize - offset; - if (size > len) - size = len; +static int loop_make_request(request_queue_t *q, int rw, struct buffer_head *bh) +{ + struct loop_device *lo; - bh = getblk(lo->lo_device, block, blksize); - if (!bh) { - printk(KERN_ERR "loop: device %s: getblk(-, %d, %d) returned NULL", - kdevname(lo->lo_device), - block, blksize); - goto error_out_lock; - } - if (!buffer_uptodate(bh) && ((current_request->cmd == READ) || - (offset || (len < blksize)))) { - ll_rw_block(READ, 1, &bh); - wait_on_buffer(bh); - if (!buffer_uptodate(bh)) { - brelse(bh); - goto error_out_lock; - } - } + if (MINOR(bh->b_dev) >= max_loop) + goto err; - if ((lo->transfer)(lo, current_request->cmd, - bh->b_data + offset, - dest_addr, size, block)) { - printk(KERN_ERR "loop: transfer error block %d\n", - block); - brelse(bh); - goto error_out_lock; - } + lo = &loop_dev[MINOR(bh->b_dev)]; + if (!lo->lo_dentry) + goto err; - if (current_request->cmd == WRITE) { - mark_buffer_uptodate(bh, 1); - mark_buffer_dirty(bh); - } - brelse(bh); - dest_addr += size; - len -= size; - offset = 0; - block++; + if (rw == WRITE) { + if (lo->lo_flags & LO_FLAGS_READ_ONLY) + goto err; + } else if (rw == READA) { + rw = READ; + } else if (rw != READ) { + printk(KERN_ERR "loop: unknown command (%d)\n", rw); + goto err; + } + +#if CONFIG_HIGHMEM + bh = create_bounce(rw, bh); +#endif + + /* + * file backed + */ + if (lo->lo_flags & LO_FLAGS_DO_BMAP) { + if (do_bh_filebacked(lo, bh, rw)) + goto err; + bh->b_end_io(bh, 1); + return 0; } - goto done; -file_backed: - pos = ((loff_t)current_request->sector << 9) + lo->lo_offset; - spin_unlock_irq(&io_request_lock); - if (current_request->cmd == WRITE) { - if (lo_send(lo, dest_addr, len, pos, blksize)) - goto error_out_lock; + /* + * remap and redirect. for reads, we need the buffer from the + * device before we can feed that through the transfer functions. + * we use the b_end_io handler for this. for writes, we can + * complete the transfer and just let generic_make_request remap + * the buffer + */ + bh->b_rdev = lo->lo_device; + bh->b_rsector += (lo->lo_offset >> 9); + if (rw == READ) { + bh->b_private = bh->b_end_io; + bh->b_end_io = loop_end_io_transfer; } else { - if (lo_receive(lo, dest_addr, len, pos, blksize)) - goto error_out_lock; + if (lo_do_transfer(lo, WRITE, bh->b_data, bh->b_data, bh->b_size, bh->b_rsector)) + goto err; } -done: - spin_lock_irq(&io_request_lock); - current_request->sector += current_request->current_nr_sectors; - current_request->nr_sectors -= current_request->current_nr_sectors; - list_add(¤t_request->queue, &q->queue_head); - end_request(1); - goto repeat; -error_out_lock: - spin_lock_irq(&io_request_lock); -error_out: - list_add(¤t_request->queue, &q->queue_head); - end_request(0); - goto repeat; + return 1; + +err: + buffer_IO_error(bh); + return 0; } static int loop_set_fd(struct loop_device *lo, kdev_t dev, unsigned int arg) { struct file *file; struct inode *inode; - int error; + int error, bs; MOD_INC_USE_COUNT; @@ -482,6 +440,15 @@ if (IS_RDONLY (inode) || is_read_only(lo->lo_device)) lo->lo_flags |= LO_FLAGS_READ_ONLY; + bs = 0; + if (blksize_size[MAJOR(inode->i_rdev)]) + bs = blksize_size[MAJOR(inode->i_rdev)][MINOR(inode->i_rdev)]; + if (!bs) + bs = BLOCK_SIZE; + + lo->lo_blksize = bs; + set_blocksize(dev, bs); + set_device_ro(dev, (lo->lo_flags & LO_FLAGS_READ_ONLY)!=0); lo->lo_dentry = dget(file->f_dentry); @@ -736,11 +703,8 @@ /* * And now the modules code and kernel interface. */ -#ifdef MODULE -#define loop_init init_module MODULE_PARM(max_loop, "i"); MODULE_PARM_DESC(max_loop, "Maximum number of loop devices (1-255)"); -#endif int loop_register_transfer(struct loop_func_table *funcs) { @@ -771,10 +735,6 @@ EXPORT_SYMBOL(loop_register_transfer); EXPORT_SYMBOL(loop_unregister_transfer); -static void no_plug_device(request_queue_t *q, kdev_t device) -{ -} - int __init loop_init(void) { int i; @@ -795,8 +755,6 @@ max_loop = 8; } - printk(KERN_INFO "loop: enabling %d loop devices\n", max_loop); - loop_dev = kmalloc (max_loop * sizeof(struct loop_device), GFP_KERNEL); if (!loop_dev) { printk (KERN_ERR "loop: Unable to create loop_dev\n"); @@ -818,9 +776,8 @@ return -ENOMEM; } - blk_init_queue(BLK_DEFAULT_QUEUE(MAJOR_NR), DEVICE_REQUEST); - blk_queue_pluggable(BLK_DEFAULT_QUEUE(MAJOR_NR), no_plug_device); - blk_queue_headactive(BLK_DEFAULT_QUEUE(MAJOR_NR), 0); + blk_queue_make_request(BLK_DEFAULT_QUEUE(MAJOR_NR), loop_make_request); + for (i=0; i < max_loop; i++) { memset(&loop_dev[i], 0, sizeof(struct loop_device)); loop_dev[i].lo_number = i; @@ -832,22 +789,23 @@ for (i=0; i < max_loop; i++) register_disk(NULL, MKDEV(MAJOR_NR,i), 1, &lo_fops, 0); + printk(KERN_INFO "loop: loaded (max %d devices)\n", max_loop); return 0; } -#ifdef MODULE -void cleanup_module(void) +void loop_exit(void) { devfs_unregister (devfs_handle); if (devfs_unregister_blkdev(MAJOR_NR, "loop") != 0) printk(KERN_WARNING "loop: cannot unregister blkdev\n"); - blk_cleanup_queue(BLK_DEFAULT_QUEUE(MAJOR_NR)); kfree (loop_dev); kfree (loop_sizes); kfree (loop_blksizes); } -#endif + +module_init(loop_init); +module_exit(loop_exit); #ifndef MODULE static int __init max_loop_setup(char *str) --- /opt/kernel/linux-2.4.0-test11-pre5/kernel/ksyms.c Tue Nov 14 22:53:19 2000 +++ kernel/ksyms.c Thu Nov 16 15:28:39 2000 @@ -94,6 +94,7 @@ EXPORT_SYMBOL(get_zeroed_page); EXPORT_SYMBOL(__free_pages); EXPORT_SYMBOL(free_pages); +EXPORT_SYMBOL(deactivate_page); #ifndef CONFIG_DISCONTIGMEM EXPORT_SYMBOL(contig_page_data); #endif @@ -120,6 +121,7 @@ EXPORT_SYMBOL(kmap_high); EXPORT_SYMBOL(kunmap_high); EXPORT_SYMBOL(highmem_start_page); +EXPORT_SYMBOL(create_bounce); #endif /* filesystem internal functions */ --- /opt/kernel/linux-2.4.0-test11-pre5/include/linux/loop.h Mon Nov 23 06:29:54 1998 +++ include/linux/loop.h Thu Nov 16 13:23:20 2000 @@ -39,12 +39,23 @@ struct file * lo_backing_file; void *key_data; char key_reserved[48]; /* for use by the filter modules */ + + int lo_blksize; }; typedef int (* transfer_proc_t)(struct loop_device *, int cmd, char *raw_buf, char *loop_buf, int size, int real_block); +extern inline int lo_do_transfer(struct loop_device *lo, int cmd, char *rbuf, + char *lbuf, int size, int rblock) +{ + if (!lo->transfer) + return 0; + + return lo->transfer(lo, cmd, rbuf, lbuf, size, rblock); +} + #endif /* __KERNEL__ */ /* @@ -102,9 +113,8 @@ /* Support for loadable transfer modules */ struct loop_func_table { int number; /* filter type */ - int (*transfer)(struct loop_device *lo, int cmd, - char *raw_buf, char *loop_buf, int size, - int real_block); + int (*transfer)(struct loop_device *lo, int cmd, char *raw_buf, + char *loop_buf, int size, int real_block); int (*init)(struct loop_device *, struct loop_info *); /* release is called from loop_unregister_transfer or clr_fd */ int (*release)(struct loop_device *);