* [PATCH] Loop device - Tracking page writes made to a loop device through mmap
@ 2007-03-13 20:21 Kandan Venkataraman
2007-03-13 20:32 ` Christoph Hellwig
0 siblings, 1 reply; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-13 20:21 UTC (permalink / raw)
To: linux-kernel; +Cc: akpm
[-- Attachment #1: Type: TEXT/PLAIN, Size: 6852 bytes --]
All comments have been taken care of.
Description:
A file_operations structure variable called loop_fops is initialised with
the default block device file operations (def_blk_fops).
The mmap operation is overriden with a new function called loop_file_mmap.
A vm_operations structure variable called loop_file_vm_ops is initialised
with the default operations for a disk file.
The page_mkwrite operation in this variable is initialised to a new
function called loop_track_pgwrites.
In the function lo_open, the file operations pointer of the device file is
initialised with the address of loop_fops.
The function loop_file_mmap simply calls generic_file_mmap and then
initialises the vm_ops of the vma with address of loop_file_vm_ops.
The function loop_track_pgwrites stores the page offset of the page that
is being written to, in a red-black tree within the loop device.
A flag lo_track_pgwrite has been added to the struct loop_device to turn
on/off tracking of page writes.
Three new ioctls have been added.
The ioctl cmd LOOP_SET_TRACK_PGWRITE sets/unsets the flag lo_track_pgwrite
in the struct loop_device.
The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that
have been written to.
The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree.
The existing interface has been left untouched. Thus applications such as
losetup would work as is.
The functionality added would allow us to have a read only version and a
write version of memory by doing the following:
Associate a normal file as backing storage for the loop device and mmap
to the loop device. Call this mmapped address space as area1.
Mmap to a normal file of identical size. Call this mmapped address space
as area2.
Changes made to area1 can be periodically copied to area2 using the ioctl
cmds (retreive dirty page offsets and copy the dirty pages from area1 to
area2). This facility would provide a quick way of updating the read only
version.
Motivation for new ioctls:
Imagine a business server application which processes messages from
clients as they come in (say over a TCP connection).
Some of those messages may be transactions, i.e. they cause data changes
in the application.
Rest of those messages may be queries i.e. they get information from the
application.
The application can consist of two processes. One process will handle the
transactions.
The other process will handle the queries. Each process will have its own
copy of the business data.
The process handling transactions can mmap to the loop device for its copy
of the memory. The loop device must have a normal file for its backing
storage.
The process handling queries can mmap to another normal file for its copy
of the memory. Both these memories have identical data at the beginning.
Queries and transactions can now be handled simultaneously by the
respective processes.
The query process can update its memory periodically by obtaining the
changes that have have happened to the loop device.
By using the ioctl call to retrieve the dirty page offsets, only the dirty
pages need to be copied over to the query process's copy of memory. We can
infact have multiple processes to handle queries sharing the same memory.
During this copy over, the transaction process will hold off processing
transactions till the update is complete.
This would be very useful for high speed in-memory transaction systems,
where the query load can be passed of to other processes. Example of such
systems would be a stock trading system, where clients buy and sell
stock(equity, options etc).
At the same time lot of clients would be downloading market data and this
can be done independently of the transactions.
This new facility will provide a way of tracking changes made to business
data, independent of the application domain.
Test program:
Before you run the test program, please create the backing storage file
for the loop device as follows
dd if=/dev/zero of=/root/file bs=4K count=10
Set bs to be whatever pagesize is in your machine. In my machine it was
4K.
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <string.h>
#include <assert.h>
#include <signal.h>
#include <errno.h>
#include <linux/loop.h>
int main()
{
int maxPages = 10;
char* start = 0;
int fd;
int dfd;
int *array = 0;
int pageSize;
int elemsPerPage;
struct loop_info64 info;
struct loop_pgoff_array pgarray;
pgarray.max = maxPages;
pgarray.pgoff = calloc(maxPages, sizeof(long));
if (pgarray.pgoff == NULL) {
fprintf(stderr, "can't create pgarray\n");
exit(1);
}
pageSize = getpagesize();
elemsPerPage = pageSize/sizeof(int);
/* open the device file */
if ((fd = open ("/dev/loop0", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out5;
}
/* open the disk file to set as backing storage*/
if ((dfd = open ("/root/file", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out4;
}
if (ioctl(fd, LOOP_SET_FD, dfd) < 0) {
perror("ioctl: LOOP_SET_FD");
goto out3;
}
if ((start = mmap(0, maxPages * pageSize, PROT_READ | PROT_WRITE,
MAP_SHARED, fd, 0)) == MAP_FAILED) {
perror("mmap error");
goto out2;
}
if (ioctl(fd, LOOP_SET_TRACK_PGWRITE, 1) < 0) {
perror("ioctl: LOOP_SET_TRACK_PGWRITE");
goto out1;
}
if (ioctl(fd, LOOP_CLR_PGWRITES, 0) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
array = (int *)start;
array[0] = 5;
fprintf(stderr, "value = %d\n", array[0]);
array[1] = 9;
fprintf(stderr, "value = %d\n", array[1]);
array[elemsPerPage] = 14;
fprintf(stderr, "value = %d\n", array[elemsPerPage]);
array[3*elemsPerPage+60] = 35;
fprintf(stderr, "value = %d\n", array[3*elemsPerPage+60]);
if (ioctl(fd, LOOP_GET_PGWRITES, &pgarray) < 0) {
perror("ioctl: LOOP_GET_PGWRITES");
goto out1;
}
int i;
for (i= 0; i < pgarray.num; i++)
fprintf(stderr, "offset %ld\n", pgarray.pgoff[i]);
out1:
munmap(start, maxPages * pageSize);
out2:
ioctl(fd, LOOP_CLR_FD, 0);
out3:
close(dfd);
out4:
close(fd);
out5:
return 0;
}
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
[-- Attachment #2: Type: TEXT/PLAIN, Size: 8883 bytes --]
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-11 07:31:26.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char *cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,85 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node,
+ struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+static int loop_set_track_pgwrite(struct loop_device *lo, unsigned long arg)
+{
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (arg)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
+ return 0;
+}
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo,
+ struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
+ array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +775,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +884,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +1000,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -1159,6 +1248,15 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_SET_TRACK_PGWRITE:
+ err = loop_set_track_pgwrite(lo, arg);
+ break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1322,10 +1420,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1553,24 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
+ * reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1580,12 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name,
+ sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1639,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1660,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-11 07:31:41.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -111,6 +119,13 @@ struct loop_info64 {
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +172,8 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_SET_TRACK_PGWRITE 0x4C07
+#define LOOP_GET_PGWRITES 0x4C08
+#define LOOP_CLR_PGWRITES 0x4C09
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-13 20:21 [PATCH] Loop device - Tracking page writes made to a loop device through mmap Kandan Venkataraman
@ 2007-03-13 20:32 ` Christoph Hellwig
2007-03-14 22:57 ` Kandan Venkataraman
0 siblings, 1 reply; 14+ messages in thread
From: Christoph Hellwig @ 2007-03-13 20:32 UTC (permalink / raw)
To: Kandan Venkataraman; +Cc: linux-kernel, akpm
On Wed, Mar 14, 2007 at 07:21:41AM +1100, Kandan Venkataraman wrote:
> All comments have been taken care of.
>
> Description:
>
> A file_operations structure variable called loop_fops is initialised with
> the default block device file operations (def_blk_fops).
> The mmap operation is overriden with a new function called loop_file_mmap.
NACK. block device driver should never ever play around with file
operations themselves. If you want functionality like the one you
have please don't overload the loop driver, but start a new (character)
driver doing specificaly what you want. And even then I'm not sure
we'd want functionality like this in the mainline tree, but at least
we can have an open discussion if it's done properly.
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-13 20:32 ` Christoph Hellwig
@ 2007-03-14 22:57 ` Kandan Venkataraman
0 siblings, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-14 22:57 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: linux-kernel, akpm
On Tue, 13 Mar 2007, Christoph Hellwig wrote:
> On Wed, Mar 14, 2007 at 07:21:41AM +1100, Kandan Venkataraman wrote:
>> All comments have been taken care of.
>>
>> Description:
>>
>> A file_operations structure variable called loop_fops is initialised with
>> the default block device file operations (def_blk_fops).
>> The mmap operation is overriden with a new function called loop_file_mmap.
>
> NACK. block device driver should never ever play around with file
> operations themselves. If you want functionality like the one you
> have please don't overload the loop driver, but start a new (character)
> driver doing specificaly what you want. And even then I'm not sure
> we'd want functionality like this in the mainline tree, but at least
> we can have an open discussion if it's done properly.
>
The reason I chose the loop device was the following:
The loop device has backing storage and makes use of the page cache. The
nopage method for all block device files and most disk files is
implemented by filemap_nopage. This function uses the page cache. Thus,
this allows me to mmap a large address space.
If I wrote a character device, I would have to store the contents in
kernel memory as the function filemap_nopage is not suitable for character
devices. This would severly restrict the size of my mmap.
I am out my depth here, any pointers on how to proceed would be very
helpful
Thanks
Kandan
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH] Loop device - Tracking page writes made to a loop device through mmap
@ 2007-03-08 3:01 Kandan Venkataraman
0 siblings, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-08 3:01 UTC (permalink / raw)
To: linux-kernel; +Cc: akpm
[-- Attachment #1: Type: TEXT/PLAIN, Size: 6359 bytes --]
All comments have been taken care of.
Description:
A file_operations structure variable called loop_fops is initialised with
the default block device file operations (def_blk_fops).
The mmap operation is overriden with a new function called loop_file_mmap.
A vm_operations structure variable called loop_file_vm_ops is initialised
with the default operations for a disk file.
The page_mkwrite operation in this variable is initialised to a new
function called loop_track_pgwrites.
In the function lo_open, the file operations pointer of the device file is
initialised with the address of loop_fops.
The function loop_file_mmap simply calls generic_file_mmap and then
initialises the vm_ops of the vma with address of loop_file_vm_ops.
The function loop_track_pgwrites stores the page offset of the page that
is being written to, in a red-black tree within the loop device.
A flag lo_track_pgwrite has been added to the structs loop_device and
loop_info64 to turn on/off tracking of page writes.
Two new ioctls have been added.
The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that
have been written to.
The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree
This functionality would allow us to have a read only version and a write
version of memory by doing the following:
Associate a normal file as backing storage for the loop device and mmap
to the loop device. Call this mmapped address space as area1.
Mmap to a normal file of identical size. Call this mmapped address space
as area2.
Changes made to area1 can be periodically copied to area2 using the ioctl
cmds (retreive dirty page offsets and copy the dirty pages from area1 to
area2). This facility would provide a quick way of updating the read only
version.
Motivation for new ioctls:
Imagine a business server application which processes messages from
clients as they come in (say over a TCP connection).
Some of those messages may be transactions, i.e. they cause data changes
in the application.
Rest of those messages may be queries i.e. they get information from the
application.
The application can consist of two processes. One process will handle the
transactions.
The other process will handle the queries. Each process will have its own
copy of the business data.
The process handling transactions can mmap to the loop device for its copy
of the memory. The loop device must have a normal file for its backing
storage.
The process handling queries can mmap to another normal file for its copy
of the memory. Both these memories have identical data at the beginning.
Queries and transactions can now be handled simultaneously by the
respective processes.
The query process can update its memory periodically by obtaining the
changes that have have happened to the loop device.
By using the ioctl call to retrieve the dirty page offsets, only the dirty
pages need to be copied over to the query process's copy of memory. We can
infact have multiple processes to handle queries sharing the same memory.
During this copy over, the transaction process will hold off processing
transactions till the update is complete.
This would be very useful for high speed in-memory transaction systems,
where the query load can be passed of to other processes. Example of such
systems would be a stock trading system, where clients buy and sell
stock(equity, options etc).
At the same time lot of clients would be downloading market data and this
can be done independently of the transactions.
This new facility will provide a way of tracking changes made to business
data, independent of the application domain.
Test program:
Before you run the test program, please create the backing storage file
for the loop device as follows
dd if=/dev/zero of=/root/file bs=4K count=10
Set bs to be whatever pagesize is in your machine. In my machine it was
4K.
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <string.h>
#include <assert.h>
#include <signal.h>
#include <errno.h>
#include <linux/loop.h>
int main()
{
int maxPages = 10;
char* start = 0;
int fd;
int dfd;
int *array = 0;
int pageSize;
int elemsPerPage;
struct loop_info64 info;
struct loop_pgoff_array pgarray;
pgarray.max = maxPages;
pgarray.pgoff = calloc(maxPages, sizeof(long));
if (pgarray.pgoff == NULL) {
fprintf(stderr, "can't create pgarray\n");
exit(1);
}
pageSize = getpagesize();
elemsPerPage = pageSize/sizeof(int);
/* open the device file */
if ((fd = open ("/dev/loop0", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out5;
}
/* open the disk file to set as backing storage*/
if ((dfd = open ("/root/file", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out4;
}
if (ioctl(fd, LOOP_SET_FD, dfd) < 0) {
perror("ioctl: LOOP_SET_FD");
goto out3;
}
if ((start = mmap(0, maxPages * pageSize, PROT_READ | PROT_WRITE,
MAP_SHARED, fd, 0)) == MAP_FAILED) {
perror("mmap error");
goto out2;
}
if (ioctl(fd, LOOP_GET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
info.lo_track_pgwrite = 1;
if (ioctl(fd, LOOP_SET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_SET_STATUS64");
goto out1;
}
if (ioctl(fd, LOOP_CLR_PGWRITES, 0) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
array = (int *)start;
array[0] = 5;
fprintf(stderr, "value = %d\n", array[0]);
array[1] = 9;
fprintf(stderr, "value = %d\n", array[1]);
array[elemsPerPage] = 14;
fprintf(stderr, "value = %d\n", array[elemsPerPage]);
array[3*elemsPerPage+60] = 35;
fprintf(stderr, "value = %d\n", array[3*elemsPerPage+60]);
if (ioctl(fd, LOOP_GET_PGWRITES, &pgarray) < 0) {
perror("ioctl: LOOP_GET_PGWRITES");
goto out1;
}
int i;
for (i= 0; i < pgarray.num; i++)
fprintf(stderr, "offset %ld\n", pgarray.pgoff[i]);
out1:
munmap(start, maxPages * pageSize);
out2:
ioctl(fd, LOOP_CLR_FD, 0);
out3:
close(dfd);
out4:
close(fd);
out5:
return 0;
}
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
[-- Attachment #2: Type: TEXT/PLAIN, Size: 10588 bytes --]
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-03-03 16:26:03.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-03 16:44:38.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char *cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,70 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node,
+ struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo,
+ struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
+ array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +760,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +869,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +985,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1043,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1093,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1119,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1243,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1295,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1404,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1417,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1550,24 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
+ * reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1577,12 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name,
+ sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1636,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1657,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-03-03 16:25:42.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-03 16:43:49.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-02 16:52 ` Randy Dunlap
2007-03-02 23:21 ` Kandan Venkataraman
2007-03-02 23:28 ` Kandan Venkataraman
@ 2007-03-03 5:46 ` Kandan Venkataraman
2 siblings, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-03 5:46 UTC (permalink / raw)
To: Randy Dunlap; +Cc: linux-kernel, kandan.venkataraman
[-- Attachment #1: Type: TEXT/PLAIN, Size: 165 bytes --]
There were a couple more white spaces, instead of tabs, hopefully this is
the last of them:-)
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
[-- Attachment #2: Type: TEXT/PLAIN, Size: 10588 bytes --]
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-03-03 16:26:03.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-03 16:44:38.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char *cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,70 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node,
+ struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo,
+ struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
+ array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +760,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +869,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +985,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1043,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1093,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1119,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1243,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1295,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1404,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1417,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1550,24 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
+ * reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1577,12 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name,
+ sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1636,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1657,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-03-03 16:25:42.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-03 16:43:49.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-02 16:52 ` Randy Dunlap
2007-03-02 23:21 ` Kandan Venkataraman
@ 2007-03-02 23:28 ` Kandan Venkataraman
2007-03-03 5:46 ` Kandan Venkataraman
2 siblings, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-02 23:28 UTC (permalink / raw)
To: Randy Dunlap; +Cc: linux-kernel, kandan.venkataraman
[-- Attachment #1: Type: TEXT/PLAIN, Size: 228 bytes --]
The patch file seems to start in the same line as the sign off
So I will introduce a new line at the end of my mail, hopefully that will
fix that problem.
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
[-- Attachment #2: Type: TEXT/PLAIN, Size: 10612 bytes --]
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-03-03 07:59:47.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-03 08:02:04.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char *cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,70 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node,
+ struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo,
+ struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
+ array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +760,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +869,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +985,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1043,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1093,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1119,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1243,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1295,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1404,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1417,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1550,24 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
+ * reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1577,12 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name,
+ sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1636,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1657,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-03-03 07:57:21.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-03 08:02:25.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-02 16:52 ` Randy Dunlap
@ 2007-03-02 23:21 ` Kandan Venkataraman
2007-03-02 23:28 ` Kandan Venkataraman
2007-03-03 5:46 ` Kandan Venkataraman
2 siblings, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-02 23:21 UTC (permalink / raw)
To: Randy Dunlap; +Cc: linux-kernel, kandan.venkataraman
[-- Attachment #1: Type: TEXT/PLAIN, Size: 206 bytes --]
I have included the patch as a text file. This is the only way I could get
the patch to work from an email.
Thank you for your patience
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
[-- Attachment #2: Type: TEXT/PLAIN, Size: 10612 bytes --]
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-03-03 07:59:47.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-03 08:02:04.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char *cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,70 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node,
+ struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo,
+ struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
+ array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +760,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +869,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +985,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1043,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1093,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1119,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1243,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1295,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1404,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1417,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1550,24 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
+ * reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1577,12 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name,
+ sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1636,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1657,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-03-03 07:57:21.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-03 08:02:25.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-02 11:13 ` Kandan Venkataraman
@ 2007-03-02 16:52 ` Randy Dunlap
2007-03-02 23:21 ` Kandan Venkataraman
` (2 more replies)
0 siblings, 3 replies; 14+ messages in thread
From: Randy Dunlap @ 2007-03-02 16:52 UTC (permalink / raw)
To: Kandan Venkataraman; +Cc: linux-kernel, kandan.venkataraman
On Fri, 2 Mar 2007 22:13:46 +1100 (EST) Kandan Venkataraman wrote:
> I am resending the message. The first few lines in the diff of the
> original message seemed to have an extra space added by the time it got to
> the mailing list. Hopefully this does not happen the second time around.
>
> Also, I missed out on putting a tab space on one line.
>
>
> I have incorporated all the changes you mentioned, except for one. Thank you very much for taking the time to
> review the code. I still retreive def_blk_fops as I did before, but I have put this in a separate function for now.
OK, that's a good compromise, at least for now.
> I have included a test program. Before you run the test program, please create the backing storage file
> for the loop device as follows
>
> dd if=/dev/zero of=/root/file bs=4K count=10
>
> Set bs to be whatever pagesize is in your machine. In my machine it was 4K.
Thanks.
[snipped]
> Now I will explain what kind of software would find the new ioctls useful.
That's a good explanation, thanks. And your original patch description
was also thorough (IMO).
> diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
> --- linux-2.6.19.2/drivers/block/loop.c 2007-03-02 22:05:06.000000000 +1100
> +++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-02 22:03:49.000000000 +1100
> @@ -74,12 +74,16 @@
> #include <linux/highmem.h>
> #include <linux/gfp.h>
> #include <linux/kthread.h>
> +#include <linux/mm.h>
That looks fishy. Have you tried to apply the patch from this email?
I'm not finding patch (program) happy with the patch file.
(too much leading whitespace on non-patched lines)
>
> #include <asm/uaccess.h>
>
> static int max_loop = 8;
> static struct loop_device *loop_dev;
> static struct gendisk **disks;
> +static kmem_cache_t *pgoff_elem_cache;
> +static char* cache_name = "loop_pgoff_elem_cache";
stataic char *cache_name
> +static struct file_operations loop_fops;
>
> /*
> * Transfer functions
> +static int loop_get_pgwrites(struct loop_device *lo, struct loop_pgoff_array __user *arg)
> +{
> + struct file *filp = lo->lo_backing_file;
> + struct loop_pgoff_array array;
> + loff_t i = 0;
> + struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
> +
> + if (lo->lo_state != Lo_bound)
> + return -ENXIO;
> +
> + if (filp == NULL || !lo->lo_track_pgwrite)
> + return -EINVAL;
> +
> + if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
> + return -EFAULT;
> +
> + while (i < array.max && rb_node != NULL) {
> +
> + if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset, array.pgoff + i))
Still need to break (split) several long lines.
> + return -EFAULT;
> +
> + ++i;
> + rb_node = rb_next(rb_node);
> + }
> + array.num = i;
> +
> + if (copy_to_user(arg, &array, sizeof(array)))
> + return -EFAULT;
> +
> + return 0;
> +}
>
> /*
> * loop_change_fd switched the backing store of a loopback device to
> @@ -1322,10 +1414,67 @@ static long lo_compat_ioctl(struct file
> }
> #endif
>
> +static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
> +{
> + /* This is used for a general mmap of a disk file */
> + int err = generic_file_mmap(file, vma);
> +
> + if (err)
> + return err;
indentation
> +
> + vma->vm_ops = &loop_file_vm_ops;
> + return 0;
> +}
> +
---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-01 21:58 ` Randy Dunlap
2007-03-02 10:31 ` Kandan Venkataraman
@ 2007-03-02 11:13 ` Kandan Venkataraman
2007-03-02 16:52 ` Randy Dunlap
1 sibling, 1 reply; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-02 11:13 UTC (permalink / raw)
To: Randy Dunlap; +Cc: linux-kernel, kandan.venkataraman
I am resending the message. The first few lines in the diff of the
original message seemed to have an extra space added by the time it got to
the mailing list. Hopefully this does not happen the second time around.
Also, I missed out on putting a tab space on one line.
I have incorporated all the changes you mentioned, except for one. Thank you very much for taking the time to
review the code. I still retreive def_blk_fops as I did before, but I have put this in a separate function for now.
I have included a test program. Before you run the test program, please create the backing storage file
for the loop device as follows
dd if=/dev/zero of=/root/file bs=4K count=10
Set bs to be whatever pagesize is in your machine. In my machine it was 4K.
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <string.h>
#include <assert.h>
#include <signal.h>
#include <errno.h>
#include <linux/loop.h>
int main()
{
int maxPages = 10;
char* start = 0;
int fd;
int dfd;
int *array = 0;
int pageSize;
int elemsPerPage;
struct loop_info64 info;
struct loop_pgoff_array pgarray;
pgarray.max = maxPages;
pgarray.pgoff = calloc(maxPages, sizeof(long));
if (pgarray.pgoff == NULL) {
fprintf(stderr, "can't create pgarray\n");
exit(1);
}
pageSize = getpagesize();
elemsPerPage = pageSize/sizeof(int);
/* open the device file */
if ((fd = open ("/dev/loop0", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out5;
}
/* open the disk file to set as backing storage*/
if ((dfd = open ("/root/file", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out4;
}
if (ioctl(fd, LOOP_SET_FD, dfd) < 0) {
perror("ioctl: LOOP_SET_FD");
goto out3;
}
if ((start = mmap(0, maxPages * pageSize, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0)) == MAP_FAILED) {
perror("mmap error");
goto out2;
}
if (ioctl(fd, LOOP_GET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
info.lo_track_pgwrite = 1;
if (ioctl(fd, LOOP_SET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_SET_STATUS64");
goto out1;
}
if (ioctl(fd, LOOP_CLR_PGWRITES, 0) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
array = (int *)start;
array[0] = 5;
fprintf(stderr, "value = %d\n", array[0]);
array[1] = 9;
fprintf(stderr, "value = %d\n", array[1]);
array[elemsPerPage] = 14;
fprintf(stderr, "value = %d\n", array[elemsPerPage]);
array[3*elemsPerPage+60] = 35;
fprintf(stderr, "value = %d\n", array[3*elemsPerPage+60]);
if (ioctl(fd, LOOP_GET_PGWRITES, &pgarray) < 0) {
perror("ioctl: LOOP_GET_PGWRITES");
goto out1;
}
int i;
for (i= 0; i < pgarray.num; i++)
fprintf(stderr, "offset %ld\n", pgarray.pgoff[i]);
out1:
munmap(start, maxPages * pageSize);
out2:
ioctl(fd, LOOP_CLR_FD, 0);
out3:
close(dfd);
out4:
close(fd);
out5:
return 0;
}
Now I will explain what kind of software would find the new ioctls useful.
Imagine a business server application which processes messages from clients as they come in (say over a TCP connection).
Some of those messages may be transactions, i.e. they cause data changes in the application.
Rest of those messages may be queries i.e. they get information from the application.
The application can consist of two processes. One process will handle the transactions.
The other process will handle the queries. Each process will have its own copy of the business data.
The process handling transactions can mmap to the loop device for its copy of the memory. The loop device must have a normal
file for its backing storage.
The process handling queries can mmap to another normal file for its copy of the memory. Both these memories have identical
data at the beginning.
Queries and transactions can now be handled simultaneously by the respective processes.
The query process can update its memory periodically by obtaining the changes that have have happened to the loop device.
By using the ioctl call to retrieve the dirty page offsets, only the dirty pages need to be copied over to the
query process's copy of memory. We can infact have multiple processes to handle queries sharing the same memory.
During this copy over, the transaction process will hold off processing transactions till the update is complete.
This would be very useful for high speed in-memory transaction systems, where the query load can be passed
of to other processes. Example of such systems would be a stock trading system, where clients buy and sell
stock(equity, options etc).
At the same time lot of clients would be downloading market data and this can be done independently of the transactions.
This new facility will provide a way of tracking changes made to business data, independent of the application domain.
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-03-02 22:05:06.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-02 22:03:49.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char* cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,67 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node, struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo, struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset, array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +757,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +866,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +982,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1040,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1090,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1116,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1240,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1292,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1401,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1414,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1547,23 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1573,11 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1631,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1652,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-03-02 22:04:43.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-02 21:58:57.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-01 21:58 ` Randy Dunlap
@ 2007-03-02 10:31 ` Kandan Venkataraman
2007-03-02 11:13 ` Kandan Venkataraman
1 sibling, 0 replies; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-02 10:31 UTC (permalink / raw)
To: Randy Dunlap; +Cc: linux-kernel, kandan.venkataraman
I have incorporated all the changes you mentioned, except for one. Thank you very much for taking the time to
review the code. I still retreive def_blk_fops as I did before, but I have put this in a separate function for now.
I have included a test program. Before you run the test program, please create the backing storage file
for the loop device as follows
dd if=/dev/zero of=/root/file bs=4K count=10
Set bs to be whatever pagesize is in your machine. In my machine it was 4K.
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <string.h>
#include <assert.h>
#include <signal.h>
#include <errno.h>
#include <linux/loop.h>
int main()
{
int maxPages = 10;
char* start = 0;
int fd;
int dfd;
int *array = 0;
int pageSize;
int elemsPerPage;
struct loop_info64 info;
struct loop_pgoff_array pgarray;
pgarray.max = maxPages;
pgarray.pgoff = calloc(maxPages, sizeof(long));
if (pgarray.pgoff == NULL) {
fprintf(stderr, "can't create pgarray\n");
exit(1);
}
pageSize = getpagesize();
elemsPerPage = pageSize/sizeof(int);
/* open the device file */
if ((fd = open ("/dev/loop0", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out5;
}
/* open the disk file to set as backing storage*/
if ((dfd = open ("/root/file", O_RDWR, S_IRWXU)) < 0) {
fprintf(stderr, "can't create device file for writing\n");
goto out4;
}
if (ioctl(fd, LOOP_SET_FD, dfd) < 0) {
perror("ioctl: LOOP_SET_FD");
goto out3;
}
if ((start = mmap(0, maxPages * pageSize, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0)) == MAP_FAILED) {
perror("mmap error");
goto out2;
}
if (ioctl(fd, LOOP_GET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
info.lo_track_pgwrite = 1;
if (ioctl(fd, LOOP_SET_STATUS64, &info) < 0) {
perror("ioctl: LOOP_SET_STATUS64");
goto out1;
}
if (ioctl(fd, LOOP_CLR_PGWRITES, 0) < 0) {
perror("ioctl: LOOP_CLR_PGWRITES");
goto out1;
}
array = (int *)start;
array[0] = 5;
fprintf(stderr, "value = %d\n", array[0]);
array[1] = 9;
fprintf(stderr, "value = %d\n", array[1]);
array[elemsPerPage] = 14;
fprintf(stderr, "value = %d\n", array[elemsPerPage]);
array[3*elemsPerPage+60] = 35;
fprintf(stderr, "value = %d\n", array[3*elemsPerPage+60]);
if (ioctl(fd, LOOP_GET_PGWRITES, &pgarray) < 0) {
perror("ioctl: LOOP_GET_PGWRITES");
goto out1;
}
int i;
for (i= 0; i < pgarray.num; i++)
fprintf(stderr, "offset %ld\n", pgarray.pgoff[i]);
out1:
munmap(start, maxPages * pageSize);
out2:
ioctl(fd, LOOP_CLR_FD, 0);
out3:
close(dfd);
out4:
close(fd);
out5:
return 0;
}
Now I will explain what kind of software would find the new ioctls useful.
Imagine a business server application which processes messages from clients as they come in (say over a TCP connection).
Some of those messages may be transactions, i.e. they cause data changes in the application.
Rest of those messages may be queries i.e. they get information from the application.
The application can consist of two processes. One process will handle the transactions.
The other process will handle the queries. Each process will have its own copy of the business data.
The process handling transactions can mmap to the loop device for its copy of the memory. The loop device must have a normal
file for its backing storage.
The process handling queries can mmap to another normal file for its copy of the memory. Both these memories have identical
data at the beginning.
Queries and transactions can now be handled simultaneously by the respective processes.
The query process can update its memory periodically by obtaining the changes that have have happened to the loop device.
By using the ioctl call to retrieve the dirty page offsets, only the dirty pages need to be copied over to the
query process's copy of memory. We can infact have multiple processes to handle queries sharing the same memory.
During this copy over, the transaction process will hold off processing transactions till the update is complete.
This would be very useful for high speed in-memory transaction systems, where the query load can be passed
of to other processes. Example of such systems would be a stock trading system, where clients buy and sell
stock(equity, options etc).
At the same time lot of clients would be downloading market data and this can be done independently of the transactions.
This new facility will provide a way of tracking changes made to business data, independent of the application domain.
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-03-02 20:20:55.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char* cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,67 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node, struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo, struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL || !lo->lo_track_pgwrite)
+ return -EINVAL;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset, array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +757,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +866,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +982,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1040,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1090,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1116,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1240,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1292,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1313,6 +1401,10 @@ static long lo_compat_ioctl(struct file
case LOOP_CHANGE_FD:
err = lo_ioctl(inode, file, cmd, arg);
break;
+ case LOOP_GET_PGWRITES:
+ case LOOP_CLR_PGWRITES:
+ err = -EINVAL;
+ break;
default:
err = -ENOIOCTLCMD;
break;
@@ -1322,10 +1414,67 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node **p = &rb_root->rb_node;
+ struct rb_node *parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p) {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1398,10 +1547,23 @@ int loop_unregister_transfer(int number)
EXPORT_SYMBOL(loop_register_transfer);
EXPORT_SYMBOL(loop_unregister_transfer);
+static const struct file_operations *get_def_blk_fops(void)
+{
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+
+ return inode.i_fop;
+}
+
static int __init loop_init(void)
{
int i;
+ loop_fops = *(get_def_blk_fops());
+ loop_fops.mmap = loop_file_mmap;
+
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
" 1 and 256), using default (8)\n");
@@ -1411,6 +1573,11 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct pgoff_elem), 0,
+ SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1631,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1652,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-03-02 20:21:35.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,11 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +113,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +173,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-01 5:25 Kandan Venkataraman
@ 2007-03-01 21:58 ` Randy Dunlap
2007-03-02 10:31 ` Kandan Venkataraman
2007-03-02 11:13 ` Kandan Venkataraman
0 siblings, 2 replies; 14+ messages in thread
From: Randy Dunlap @ 2007-03-01 21:58 UTC (permalink / raw)
To: Kandan Venkataraman; +Cc: linux-kernel, kandan.venkataraman
On Thu, 1 Mar 2007 16:25:16 +1100 Kandan Venkataraman wrote:
Sorry, I missed seeing your reposted patch when I replied to the
earlier patch.
> The patch is for tracking writes made to a loop device made through mmap.
How did you test this? (what program(s))
> Two new ioctls have been added.
>
> The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that have been written to.
> The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree
I don't quite understand: what software would use these ioctls?
> This functionality would allow us to have a read only version and a write version of memory by doing the following:
> Associate a normal file as backing storage for the loop device and mmap to the loop device. Call this mmapped address space as area1.
> Mmap to a normal file of identical size. Call this mmapped address space as area2.
>
> Changes made to area1 can be periodically copied to area2 using the ioctl cmds (retreive dirty page offsets and copy the dirty pages from area1 to area2). This facility would provide a quick way of updating the read only version.
>
> Please CC your reply to kandan.venkataraman@omxgroup.com
>
> The following patch is against vanilla linux-2.6.19.2
>
> Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
>
>
> diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
> --- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000 +1100
> +++ linux-2.6.19.2-new/drivers/block/loop.c 2007-02-27 17:23:18.000000000 +1100
> @@ -74,12 +74,16 @@
> #include <linux/highmem.h>
> #include <linux/gfp.h>
> #include <linux/kthread.h>
> +#include <linux/mm.h>
>
> #include <asm/uaccess.h>
>
> static int max_loop = 8;
> static struct loop_device *loop_dev;
> static struct gendisk **disks;
> +static kmem_cache_t *pgoff_elem_cache;
> +static char* cache_name = "loop_pgoff_elem_cache";
Use style/format as 2 lines above:
static char *cache_name = "loop_pgoff_elem_cache";
> +static struct file_operations loop_fops;
>
> /*
> * Transfer functions
> @@ -646,6 +650,73 @@ static void do_loop_switch(struct loop_d
> complete(&p->wait);
> }
>
> +
> +
> +static int loop_get_pgwrites(struct loop_device *lo, struct loop_pgoff_array __user *arg)
> +{
> + struct file *filp = lo->lo_backing_file;
> + struct loop_pgoff_array array;
> + loff_t i = 0;
> + struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
> +
> + if (lo->lo_state != Lo_bound)
> + return -ENXIO;
> +
> + if (filp == NULL)
> + return -EINVAL;
> +
> + if (!lo->lo_track_pgwrite)
> + return 0;
Indent above with 2 tabs, please.
> +
> + if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
> + return -EFAULT;
> +
> + while (i < array.max && rb_node != NULL) {
> +
> + if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset, array.pgoff + i))
> + return -EFAULT;
> +
> + ++i;
> + rb_node = rb_next(rb_node);
Indent with tabs, not spaces.
> + }
> + array.num = i;
> +
> + if (copy_to_user(arg, &array, sizeof(array)))
> + return -EFAULT;
indentation...
> +
> + return 0;
> +}
>
> /*
> * loop_change_fd switched the backing store of a loopback device to
> @@ -969,6 +1046,14 @@ loop_set_status(struct loop_device *lo,
> return -EFBIG;
> }
>
> + if (info->lo_track_pgwrite)
> + lo->lo_track_pgwrite = 1;
> + else {
> + if (lo->lo_track_pgwrite)
> + pgoff_tree_clear(&lo->pgoff_tree);
> + lo->lo_track_pgwrite = 0;
> + }
indentation
> +
> memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
> memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
> lo->lo_file_name[LO_NAME_SIZE-1] = 0;
> @@ -1322,10 +1416,68 @@ static long lo_compat_ioctl(struct file
> }
> #endif
>
> +static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
> +{
> + struct rb_node ** p = &rb_root->rb_node;
> + struct rb_node * parent = NULL;
No space after "**" and "*" above.
> + struct pgoff_elem *pgoff_elem;
> +
> + while (*p)
> + {
Don't put { on separate line, just put it after while:
while (*p) {
> + parent = *p;
> + pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
> +
> + if (offset < pgoff_elem->offset)
> + p = &(*p)->rb_left;
> + else if (offset > pgoff_elem->offset)
> + p = &(*p)->rb_right;
> + else
> + return 0;
> + }
> +
> + pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
> + if (!pgoff_elem)
> + return -ENOMEM;
indentation.
> + pgoff_elem->offset = offset;
> +
> + rb_link_node(&pgoff_elem->node, parent, p);
> + rb_insert_color(&pgoff_elem->node, rb_root);
> +
> + return 0;
> +}
> +
> +
> +struct vm_operations_struct loop_file_vm_ops = {
> + .nopage = filemap_nopage,
> + .populate = filemap_populate,
> + .page_mkwrite = loop_track_pgwrites
> +};
> +
> +static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
> +{
> + /* This is used for a general mmap of a disk file */
> + int err = generic_file_mmap(file, vma);
> +
> + if (err)
> + return err;
> +
Odd/inconsistent indentation above/below. Above lines should be
indented with one tab (2 tabs on 'return').
> + vma->vm_ops = &loop_file_vm_ops;
> + return 0;
> +}
> +
> @@ -1401,6 +1553,12 @@ EXPORT_SYMBOL(loop_unregister_transfer);
> static int __init loop_init(void)
> {
> int i;
> + struct inode inode;
> +
> + /* a roundabout way to retrieve def_blk_fops but avoids undefined reference warning */
Should there be a clean kernel API instead of this roundabout way?
> + init_special_inode(&inode, S_IFBLK, 0);
> + loop_fops = *(inode.i_fop);
> + loop_fops.mmap = loop_file_mmap;
>
> if (max_loop < 1 || max_loop > 256) {
> printk(KERN_WARNING "loop: invalid max_loop (must be between"
> @@ -1411,6 +1569,10 @@ static int __init loop_init(void)
> if (register_blkdev(LOOP_MAJOR, "loop"))
> return -EIO;
>
> + pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct pgoff_elem), 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
line too long.
> + if (!pgoff_elem_cache)
> + goto out_mem0;
indentation.
> +
> loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
> if (!loop_dev)
> goto out_mem1;
> diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
> --- linux-2.6.19.2/include/linux/loop.h 2007-01-11 06:10:37.000000000 +1100
> +++ linux-2.6.19.2-new/include/linux/loop.h 2007-02-27 17:23:28.000000000 +1100
> @@ -66,6 +69,12 @@ struct loop_device {
> request_queue_t *lo_queue;
> };
>
> +struct pgoff_elem {
> +
no blank line here.
> + struct rb_node node;
> + unsigned long offset;
> +};
> +
> #endif /* __KERNEL__ */
>
> /*
> @@ -105,12 +114,20 @@ struct loop_info64 {
> __u32 lo_encrypt_type;
> __u32 lo_encrypt_key_size; /* ioctl w/o */
> __u32 lo_flags; /* ioctl r/o */
> + __u32 lo_track_pgwrite;
Make the fields align. And that line ends with trailing space.
(There are 2 new patch lines like that; please don't do that.)
> __u8 lo_file_name[LO_NAME_SIZE];
> __u8 lo_crypt_name[LO_NAME_SIZE];
> __u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
> __u64 lo_init[2];
> };
>
> +struct loop_pgoff_array {
> + __u64 max; /* size of array passed by user */
> + __u64 num; /* number of entries filled in by driver */
> + __u64 *pgoff; /* array of page offsets of pages written to by mmap */
> +};
> +
> +
> /*
> * Loop filter types
> */
> @@ -157,5 +174,7 @@ int loop_unregister_transfer(int number)
> #define LOOP_SET_STATUS64 0x4C04
> #define LOOP_GET_STATUS64 0x4C05
> #define LOOP_CHANGE_FD 0x4C06
> +#define LOOP_GET_PGWRITES 0x4C07
> +#define LOOP_CLR_PGWRITES 0x4C08
>
> #endif
> -
---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH] Loop device - Tracking page writes made to a loop device through mmap
2007-03-01 1:40 Kandan Venkataraman
@ 2007-03-01 19:52 ` Randy Dunlap
0 siblings, 0 replies; 14+ messages in thread
From: Randy Dunlap @ 2007-03-01 19:52 UTC (permalink / raw)
To: Kandan Venkataraman; +Cc: linux-kernel
On Thu, 1 Mar 2007 02:40:41 +0100 Kandan Venkataraman wrote:
> The patch is for tracking writes made to a loop device made through
> mmap.
>
> A file_operations structure variable called loop_fops is initialised
> with the default block device file operations (def_blk_fops).
> The mmap operation is overriden with a new function called
> loop_file_mmap.
>
> A vm_operations structure variable called loop_file_vm_ops is
> initialised with the default operations for a disk file.
> The page_mkwrite operation in this variable is initialised to a new
> function called loop_track_pgwrites.
>
> In the function lo_open, the file operations pointer of the device file
> is initialised with the address of loop_fops.
>
> The function loop_file_mmap simply calls generic_file_mmap and then
> initialises the vm_ops of the vma with address of loop_file_vm_ops.
>
> The function loop_track_pgwrites stores the page offset of the page that
> is being written to, in a red-black tree within the loop device.
>
> A flag lo_track_pgwrite has been added to the structs loop_device and
> loop_info64 to turn on/off tracking of page writes.
>
> Two new ioctls have been added.
>
> The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that
> have been written to.
> The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree
>
> This functionality would allow us to have a read only version and a
> write version of memory by doing the following:
> Associate a normal file as backing storage for the loop device and mmap
> to the loop device. Call this mmapped address space as area1.
> Mmap to a normal file of identical size. Call this mmapped address space
> as area2.
>
> Changes made to area1 can be periodically copied to area2 using the
> ioctl cmds (retreive dirty page offsets and copy the dirty pages from
> area1 to area2). This facility would provide a quick way of updating the
> read only version.
>
> The following patch is against vanilla linux-2.6.19.2
Patches should be made against current mainline (Linus's tree)
unless (1) they are specifically for one of the stable trees
or (2) they are specifically for the -mm patchset.
> Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
>
>
> diff -uprN linux-2.6.19.2/drivers/block/loop.c
> linux-2.6.19.2-new/drivers/block/loop.c
> --- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000
> +1100
Long lines above & below were split somewhere along the way
(likely by some mail s/w)...
> +++ linux-2.6.19.2-new/drivers/block/loop.c 2007-02-27
> 17:23:18.000000000 +1100
> @@ -74,12 +74,16 @@
> #include <linux/highmem.h>
> #include <linux/gfp.h>
> #include <linux/kthread.h>
> +#include <linux/mm.h>
>
> #include <asm/uaccess.h>
>
> static int max_loop = 8;
> static struct loop_device *loop_dev;
> static struct gendisk **disks;
> +static kmem_cache_t *pgoff_elem_cache;
> +static char* cache_name = "loop_pgoff_elem_cache";
> +static struct file_operations loop_fops;
>
> /*
> * Transfer functions
> @@ -646,6 +650,73 @@ static void do_loop_switch(struct loop_d
> complete(&p->wait);
> }
>
> +static void pgoff_tree_clear(struct rb_root *rb_root)
> +{
> + struct rb_node *rb_node = rb_root->rb_node;
> +
> + while (rb_node != NULL) {
> +
> + rb_erase(rb_node, rb_root);
> + kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node, struct
> pgoff_elem, node));
another long line (bad) split above. There are more of these,
but I won't continue to point out all of them.
And all tabs from the source have been converted to spaces, again
by some mail s/w somewhere along the way... so the patch does not
apply cleanly.
> + rb_node = rb_root->rb_node;
> + }
> +
> + *rb_root = RB_ROOT;
> +}
> +
> +
> +static int loop_get_pgwrites(struct loop_device *lo, struct
> loop_pgoff_array __user *arg)
Please try to keep source lines < 80 columns. It's OK to split
a function declaration into multiple lines, e.g.:
static int loop_get_pgwrites(struct loop_device *lo,
struct loop_pgoff_array __user *arg)
> +{
> +}
> @@ -1401,6 +1553,12 @@ EXPORT_SYMBOL(loop_unregister_transfer);
> static int __init loop_init(void)
> {
> int i;
> + struct inode inode;
> +
> + /* a roundabout way to retrieve def_blk_fops but avoids undefined
> reference warning */
Does there need to be a legitimate API for retrieving this pointer?
> + init_special_inode(&inode, S_IFBLK, 0);
> + loop_fops = *(inode.i_fop);
> + loop_fops.mmap = loop_file_mmap;
>
> if (max_loop < 1 || max_loop > 256) {
> printk(KERN_WARNING "loop: invalid max_loop (must be between"
> @@ -1411,6 +1569,10 @@ static int __init loop_init(void)
> if (register_blkdev(LOOP_MAJOR, "loop"))
> return -EIO;
>
> + pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct
> pgoff_elem), 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
Line too long.
> + if (!pgoff_elem_cache)
> + goto out_mem0;
> +
> loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
> if (!loop_dev)
> goto out_mem1;
---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH] Loop device - Tracking page writes made to a loop device through mmap
@ 2007-03-01 5:25 Kandan Venkataraman
2007-03-01 21:58 ` Randy Dunlap
0 siblings, 1 reply; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-01 5:25 UTC (permalink / raw)
To: linux-kernel
The patch is for tracking writes made to a loop device made through mmap.
A file_operations structure variable called loop_fops is initialised with the default block device file operations (def_blk_fops).
The mmap operation is overriden with a new function called loop_file_mmap.
A vm_operations structure variable called loop_file_vm_ops is initialised with the default operations for a disk file.
The page_mkwrite operation in this variable is initialised to a new function called loop_track_pgwrites.
In the function lo_open, the file operations pointer of the device file is initialised with the address of loop_fops.
The function loop_file_mmap simply calls generic_file_mmap and then initialises the vm_ops of the vma with address of loop_file_vm_ops.
The function loop_track_pgwrites stores the page offset of the page that is being written to, in a red-black tree within the loop device.
A flag lo_track_pgwrite has been added to the structs loop_device and loop_info64 to turn on/off tracking of page writes.
Two new ioctls have been added.
The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that have been written to.
The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree
This functionality would allow us to have a read only version and a write version of memory by doing the following:
Associate a normal file as backing storage for the loop device and mmap to the loop device. Call this mmapped address space as area1.
Mmap to a normal file of identical size. Call this mmapped address space as area2.
Changes made to area1 can be periodically copied to area2 using the ioctl cmds (retreive dirty page offsets and copy the dirty pages from area1 to area2). This facility would provide a quick way of updating the read only version.
Please CC your reply to kandan.venkataraman@omxgroup.com
The following patch is against vanilla linux-2.6.19.2
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
diff -uprN linux-2.6.19.2/drivers/block/loop.c linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-02-27 17:23:18.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char* cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,73 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node, struct pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL)
+ return -EINVAL;
+
+ if (!lo->lo_track_pgwrite)
+ return 0;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo, struct loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL)
+ return -EINVAL;
+
+ if (!lo->lo_track_pgwrite)
+ return 0;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset, array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +763,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +872,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
@@ -913,6 +988,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1046,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1096,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1122,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1246,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1298,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1322,10 +1416,68 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long offset)
+{
+ struct rb_node ** p = &rb_root->rb_node;
+ struct rb_node * parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p)
+ {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page *page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct * vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1401,6 +1553,12 @@ EXPORT_SYMBOL(loop_unregister_transfer);
static int __init loop_init(void)
{
int i;
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+ loop_fops = *(inode.i_fop);
+ loop_fops.mmap = loop_file_mmap;
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
@@ -1411,6 +1569,10 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct pgoff_elem), 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1626,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1647,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-01-11 06:10:37.000000000 +1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-02-27 17:23:28.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,12 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +114,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +174,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH] Loop device - Tracking page writes made to a loop device through mmap
@ 2007-03-01 1:40 Kandan Venkataraman
2007-03-01 19:52 ` Randy Dunlap
0 siblings, 1 reply; 14+ messages in thread
From: Kandan Venkataraman @ 2007-03-01 1:40 UTC (permalink / raw)
To: linux-kernel
The patch is for tracking writes made to a loop device made through
mmap.
A file_operations structure variable called loop_fops is initialised
with the default block device file operations (def_blk_fops).
The mmap operation is overriden with a new function called
loop_file_mmap.
A vm_operations structure variable called loop_file_vm_ops is
initialised with the default operations for a disk file.
The page_mkwrite operation in this variable is initialised to a new
function called loop_track_pgwrites.
In the function lo_open, the file operations pointer of the device file
is initialised with the address of loop_fops.
The function loop_file_mmap simply calls generic_file_mmap and then
initialises the vm_ops of the vma with address of loop_file_vm_ops.
The function loop_track_pgwrites stores the page offset of the page that
is being written to, in a red-black tree within the loop device.
A flag lo_track_pgwrite has been added to the structs loop_device and
loop_info64 to turn on/off tracking of page writes.
Two new ioctls have been added.
The ioctl cmd LOOP_GET_PGWRITES retrieves the page offsets of pages that
have been written to.
The ioctl cmd LOOP_CLR_PGWRITES empties the red-black tree
This functionality would allow us to have a read only version and a
write version of memory by doing the following:
Associate a normal file as backing storage for the loop device and mmap
to the loop device. Call this mmapped address space as area1.
Mmap to a normal file of identical size. Call this mmapped address space
as area2.
Changes made to area1 can be periodically copied to area2 using the
ioctl cmds (retreive dirty page offsets and copy the dirty pages from
area1 to area2). This facility would provide a quick way of updating the
read only version.
The following patch is against vanilla linux-2.6.19.2
Signed-off-by: Kandan Venkataraman kandan.venkataraman@omxgroup.com
diff -uprN linux-2.6.19.2/drivers/block/loop.c
linux-2.6.19.2-new/drivers/block/loop.c
--- linux-2.6.19.2/drivers/block/loop.c 2007-01-11 06:10:37.000000000
+1100
+++ linux-2.6.19.2-new/drivers/block/loop.c 2007-02-27
17:23:18.000000000 +1100
@@ -74,12 +74,16 @@
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/kthread.h>
+#include <linux/mm.h>
#include <asm/uaccess.h>
static int max_loop = 8;
static struct loop_device *loop_dev;
static struct gendisk **disks;
+static kmem_cache_t *pgoff_elem_cache;
+static char* cache_name = "loop_pgoff_elem_cache";
+static struct file_operations loop_fops;
/*
* Transfer functions
@@ -646,6 +650,73 @@ static void do_loop_switch(struct loop_d
complete(&p->wait);
}
+static void pgoff_tree_clear(struct rb_root *rb_root)
+{
+ struct rb_node *rb_node = rb_root->rb_node;
+
+ while (rb_node != NULL) {
+
+ rb_erase(rb_node, rb_root);
+ kmem_cache_free(pgoff_elem_cache, rb_entry(rb_node, struct
pgoff_elem, node));
+ rb_node = rb_root->rb_node;
+ }
+
+ *rb_root = RB_ROOT;
+}
+
+
+static int loop_clr_pgwrites(struct loop_device *lo)
+{
+ struct file *filp = lo->lo_backing_file;
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL)
+ return -EINVAL;
+
+ if (!lo->lo_track_pgwrite)
+ return 0;
+
+ pgoff_tree_clear(&lo->pgoff_tree);
+
+ return 0;
+}
+
+static int loop_get_pgwrites(struct loop_device *lo, struct
loop_pgoff_array __user *arg)
+{
+ struct file *filp = lo->lo_backing_file;
+ struct loop_pgoff_array array;
+ loff_t i = 0;
+ struct rb_node *rb_node = rb_first(&lo->pgoff_tree);
+
+ if (lo->lo_state != Lo_bound)
+ return -ENXIO;
+
+ if (filp == NULL)
+ return -EINVAL;
+
+ if (!lo->lo_track_pgwrite)
+ return 0;
+
+ if (copy_from_user(&array, arg, sizeof (struct loop_pgoff_array)))
+ return -EFAULT;
+
+ while (i < array.max && rb_node != NULL) {
+
+ if (put_user(rb_entry(rb_node, struct pgoff_elem, node)->offset,
array.pgoff + i))
+ return -EFAULT;
+
+ ++i;
+ rb_node = rb_next(rb_node);
+ }
+ array.num = i;
+
+ if (copy_to_user(arg, &array, sizeof(array)))
+ return -EFAULT;
+
+ return 0;
+}
/*
* loop_change_fd switched the backing store of a loopback device to
@@ -692,6 +763,8 @@ static int loop_change_fd(struct loop_de
if (get_loop_size(lo, file) != get_loop_size(lo, old_file))
goto out_putf;
+ pgoff_tree_clear(&lo->pgoff_tree);
+
/* and ... switch */
error = loop_switch(lo, file);
if (error)
@@ -799,6 +872,8 @@ static int loop_set_fd(struct loop_devic
lo->transfer = transfer_none;
lo->ioctl = NULL;
lo->lo_sizelimit = 0;
+ lo->lo_track_pgwrite = 0;
+ lo->pgoff_tree = RB_ROOT;
lo->old_gfp_mask = mapping_gfp_mask(mapping);
mapping_set_gfp_mask(mapping, lo->old_gfp_mask &
~(__GFP_IO|__GFP_FS));
@@ -913,6 +988,8 @@ static int loop_clr_fd(struct loop_devic
lo->lo_sizelimit = 0;
lo->lo_encrypt_key_size = 0;
lo->lo_flags = 0;
+ lo->lo_track_pgwrite = 0;
+ pgoff_tree_clear(&lo->pgoff_tree);
lo->lo_thread = NULL;
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
@@ -969,6 +1046,14 @@ loop_set_status(struct loop_device *lo,
return -EFBIG;
}
+ if (info->lo_track_pgwrite)
+ lo->lo_track_pgwrite = 1;
+ else {
+ if (lo->lo_track_pgwrite)
+ pgoff_tree_clear(&lo->pgoff_tree);
+ lo->lo_track_pgwrite = 0;
+ }
+
memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
memcpy(lo->lo_crypt_name, info->lo_crypt_name, LO_NAME_SIZE);
lo->lo_file_name[LO_NAME_SIZE-1] = 0;
@@ -1011,6 +1096,7 @@ loop_get_status(struct loop_device *lo,
info->lo_offset = lo->lo_offset;
info->lo_sizelimit = lo->lo_sizelimit;
info->lo_flags = lo->lo_flags;
+ info->lo_track_pgwrite = lo->lo_track_pgwrite;
memcpy(info->lo_file_name, lo->lo_file_name, LO_NAME_SIZE);
memcpy(info->lo_crypt_name, lo->lo_crypt_name, LO_NAME_SIZE);
info->lo_encrypt_type =
@@ -1036,6 +1122,7 @@ loop_info64_from_old(const struct loop_i
info64->lo_encrypt_type = info->lo_encrypt_type;
info64->lo_encrypt_key_size = info->lo_encrypt_key_size;
info64->lo_flags = info->lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info->lo_init[0];
info64->lo_init[1] = info->lo_init[1];
if (info->lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1159,6 +1246,12 @@ static int lo_ioctl(struct inode * inode
case LOOP_GET_STATUS64:
err = loop_get_status64(lo, (struct loop_info64 __user *) arg);
break;
+ case LOOP_GET_PGWRITES:
+ err = loop_get_pgwrites(lo, (struct loop_pgoff_array __user *) arg);
+ break;
+ case LOOP_CLR_PGWRITES:
+ err = loop_clr_pgwrites(lo);
+ break;
default:
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
@@ -1205,6 +1298,7 @@ loop_info64_from_compat(const struct com
info64->lo_encrypt_type = info.lo_encrypt_type;
info64->lo_encrypt_key_size = info.lo_encrypt_key_size;
info64->lo_flags = info.lo_flags;
+ info64->lo_track_pgwrite = 0;
info64->lo_init[0] = info.lo_init[0];
info64->lo_init[1] = info.lo_init[1];
if (info.lo_encrypt_type == LO_CRYPT_CRYPTOAPI)
@@ -1322,10 +1416,68 @@ static long lo_compat_ioctl(struct file
}
#endif
+static int pgoff_tree_insert(struct rb_root *rb_root, unsigned long
offset)
+{
+ struct rb_node ** p = &rb_root->rb_node;
+ struct rb_node * parent = NULL;
+ struct pgoff_elem *pgoff_elem;
+
+ while (*p)
+ {
+ parent = *p;
+ pgoff_elem = rb_entry(parent, struct pgoff_elem, node);
+
+ if (offset < pgoff_elem->offset)
+ p = &(*p)->rb_left;
+ else if (offset > pgoff_elem->offset)
+ p = &(*p)->rb_right;
+ else
+ return 0;
+ }
+
+ pgoff_elem = kmem_cache_alloc(pgoff_elem_cache, GFP_KERNEL);
+ if (!pgoff_elem)
+ return -ENOMEM;
+ pgoff_elem->offset = offset;
+
+ rb_link_node(&pgoff_elem->node, parent, p);
+ rb_insert_color(&pgoff_elem->node, rb_root);
+
+ return 0;
+}
+
+static int loop_track_pgwrites(struct vm_area_struct *vma, struct page
*page)
+{
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+
+ return pgoff_tree_insert(&lo->pgoff_tree, page->index);
+}
+
+struct vm_operations_struct loop_file_vm_ops = {
+ .nopage = filemap_nopage,
+ .populate = filemap_populate,
+ .page_mkwrite = loop_track_pgwrites
+};
+
+static int loop_file_mmap(struct file * file, struct vm_area_struct *
vma)
+{
+ /* This is used for a general mmap of a disk file */
+ int err = generic_file_mmap(file, vma);
+
+ if (err)
+ return err;
+
+ vma->vm_ops = &loop_file_vm_ops;
+ return 0;
+}
+
static int lo_open(struct inode *inode, struct file *file)
{
struct loop_device *lo = inode->i_bdev->bd_disk->private_data;
+ file->f_op = &loop_fops;
mutex_lock(&lo->lo_ctl_mutex);
lo->lo_refcnt++;
mutex_unlock(&lo->lo_ctl_mutex);
@@ -1401,6 +1553,12 @@ EXPORT_SYMBOL(loop_unregister_transfer);
static int __init loop_init(void)
{
int i;
+ struct inode inode;
+
+ /* a roundabout way to retrieve def_blk_fops but avoids undefined
reference warning */
+ init_special_inode(&inode, S_IFBLK, 0);
+ loop_fops = *(inode.i_fop);
+ loop_fops.mmap = loop_file_mmap;
if (max_loop < 1 || max_loop > 256) {
printk(KERN_WARNING "loop: invalid max_loop (must be between"
@@ -1411,6 +1569,10 @@ static int __init loop_init(void)
if (register_blkdev(LOOP_MAJOR, "loop"))
return -EIO;
+ pgoff_elem_cache = kmem_cache_create(cache_name, sizeof(struct
pgoff_elem), 0, SLAB_HWCACHE_ALIGN, NULL, NULL);
+ if (!pgoff_elem_cache)
+ goto out_mem0;
+
loop_dev = kmalloc(max_loop * sizeof(struct loop_device), GFP_KERNEL);
if (!loop_dev)
goto out_mem1;
@@ -1464,6 +1626,8 @@ out_mem3:
out_mem2:
kfree(loop_dev);
out_mem1:
+ kmem_cache_destroy(pgoff_elem_cache);
+out_mem0:
unregister_blkdev(LOOP_MAJOR, "loop");
printk(KERN_ERR "loop: ran out of memory\n");
return -ENOMEM;
@@ -1483,6 +1647,7 @@ static void loop_exit(void)
kfree(disks);
kfree(loop_dev);
+ kmem_cache_destroy(pgoff_elem_cache);
}
module_init(loop_init);
diff -uprN linux-2.6.19.2/include/linux/loop.h
linux-2.6.19.2-new/include/linux/loop.h
--- linux-2.6.19.2/include/linux/loop.h 2007-01-11 06:10:37.000000000
+1100
+++ linux-2.6.19.2-new/include/linux/loop.h 2007-02-27
17:23:28.000000000 +1100
@@ -18,6 +18,7 @@
#include <linux/blkdev.h>
#include <linux/spinlock.h>
#include <linux/mutex.h>
+#include <linux/rbtree.h>
/* Possible states of device */
enum {
@@ -34,6 +35,8 @@ struct loop_device {
loff_t lo_offset;
loff_t lo_sizelimit;
int lo_flags;
+ int lo_track_pgwrite;
+ struct rb_root pgoff_tree;
int (*transfer)(struct loop_device *, int cmd,
struct page *raw_page, unsigned raw_off,
struct page *loop_page, unsigned loop_off,
@@ -66,6 +69,12 @@ struct loop_device {
request_queue_t *lo_queue;
};
+struct pgoff_elem {
+
+ struct rb_node node;
+ unsigned long offset;
+};
+
#endif /* __KERNEL__ */
/*
@@ -105,12 +114,20 @@ struct loop_info64 {
__u32 lo_encrypt_type;
__u32 lo_encrypt_key_size; /* ioctl w/o */
__u32 lo_flags; /* ioctl r/o */
+ __u32 lo_track_pgwrite;
__u8 lo_file_name[LO_NAME_SIZE];
__u8 lo_crypt_name[LO_NAME_SIZE];
__u8 lo_encrypt_key[LO_KEY_SIZE]; /* ioctl w/o */
__u64 lo_init[2];
};
+struct loop_pgoff_array {
+ __u64 max; /* size of array passed by user */
+ __u64 num; /* number of entries filled in by driver */
+ __u64 *pgoff; /* array of page offsets of pages written to by mmap */
+};
+
+
/*
* Loop filter types
*/
@@ -157,5 +174,7 @@ int loop_unregister_transfer(int number)
#define LOOP_SET_STATUS64 0x4C04
#define LOOP_GET_STATUS64 0x4C05
#define LOOP_CHANGE_FD 0x4C06
+#define LOOP_GET_PGWRITES 0x4C07
+#define LOOP_CLR_PGWRITES 0x4C08
#endif
********************************************************************************
This e-mail and the information it contains may be privileged and/or
confidential. It is for the intended addressee(s) only.
The unauthorised use, disclosure or copying of this e-mail, or any information it contains, is prohibited.
If you are not an intended recipient, please contact the sender and delete the material from your computer.
********************************************************************************
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2007-03-14 22:57 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-03-13 20:21 [PATCH] Loop device - Tracking page writes made to a loop device through mmap Kandan Venkataraman
2007-03-13 20:32 ` Christoph Hellwig
2007-03-14 22:57 ` Kandan Venkataraman
-- strict thread matches above, loose matches on Subject: below --
2007-03-08 3:01 Kandan Venkataraman
2007-03-01 5:25 Kandan Venkataraman
2007-03-01 21:58 ` Randy Dunlap
2007-03-02 10:31 ` Kandan Venkataraman
2007-03-02 11:13 ` Kandan Venkataraman
2007-03-02 16:52 ` Randy Dunlap
2007-03-02 23:21 ` Kandan Venkataraman
2007-03-02 23:28 ` Kandan Venkataraman
2007-03-03 5:46 ` Kandan Venkataraman
2007-03-01 1:40 Kandan Venkataraman
2007-03-01 19:52 ` Randy Dunlap
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).