Commit 50109b5a authored by Anton Ivanov's avatar Anton Ivanov Committed by Richard Weinberger
Browse files

um: Add support for DISCARD in the UBD Driver 



Support for DISCARD and WRITE_ZEROES in the ubd driver using
fallocate.

DISCARD is enabled by default and can be disabled using a new
UBD command line flag.

If the underlying fs on which the UBD image is stored does not
support DISCARD the support for both DISCARD and WRITE_ZEROES
is turned off.

Signed-off-by: default avatarAnton Ivanov <anton.ivanov@cambridgegreys.com>
Signed-off-by: default avatarRichard Weinberger <richard@nod.at>
parent a41421ed

arch/um/drivers/ubd_kern.c

+54 −11
Original line number Diff line number Diff line
@@ -154,6 +154,7 @@ struct ubd { 
	struct openflags openflags;

	unsigned shared:1;

	unsigned no_cow:1;

	unsigned no_trim:1;

	struct cow cow;

	struct platform_device pdev;

	struct request_queue *queue;
@@ -177,6 +178,7 @@ struct ubd { 
	.boot_openflags =	OPEN_FLAGS, \

	.openflags =		OPEN_FLAGS, \

	.no_cow =               0, \

	.no_trim =		0, \

	.shared =		0, \

	.cow =			DEFAULT_COW, \

	.lock =			__SPIN_LOCK_UNLOCKED(ubd_devs.lock), \
@@ -323,7 +325,7 @@ static int ubd_setup_common(char *str, int *index_out, char **error_out) 
		*index_out = n;



	err = -EINVAL;

	for (i = 0; i < sizeof("rscd="); i++) {

	for (i = 0; i < sizeof("rscdt="); i++) {

		switch (*str) {

		case 'r':

			flags.w = 0;
@@ -337,12 +339,15 @@ static int ubd_setup_common(char *str, int *index_out, char **error_out) 
		case 'c':

			ubd_dev->shared = 1;

			break;

		case 't':

			ubd_dev->no_trim = 1;

			break;

		case '=':

			str++;

			goto break_loop;

		default:

			*error_out = "Expected '=' or flag letter "

				"(r, s, c, or d)";

				"(r, s, c, t or d)";

			goto out;

		}

		str++;
@@ -415,6 +420,7 @@ __uml_help(ubd_setup, 
"    'c' will cause the device to be treated as being shared between multiple\n"

"    UMLs and file locking will be turned off - this is appropriate for a\n"

"    cluster filesystem and inappropriate at almost all other times.\n\n"

"    't' will disable trim/discard support on the device (enabled by default).\n\n"

);



static int udb_setup(char *str)
@@ -513,9 +519,17 @@ static void ubd_handler(void) 
		for (count = 0; count < n/sizeof(struct io_thread_req *); count++) {

			struct io_thread_req *io_req = (*irq_req_buffer)[count];



			if ((io_req->error == BLK_STS_NOTSUPP) && (req_op(io_req->req) == REQ_OP_DISCARD)) {

				blk_queue_max_discard_sectors(io_req->req->q, 0);

				blk_queue_max_write_zeroes_sectors(io_req->req->q, 0);

				blk_queue_flag_clear(QUEUE_FLAG_DISCARD, io_req->req->q);

			}

			if ((io_req->error) || (io_req->buffer == NULL))

				blk_mq_end_request(io_req->req, io_req->error);

			else {

				if (!blk_update_request(io_req->req, io_req->error, io_req->length))

					__blk_mq_end_request(io_req->req, io_req->error);



			}

			kfree(io_req);

		}

	}
@@ -829,6 +843,13 @@ static int ubd_open_dev(struct ubd *ubd_dev) 
		if(err < 0) goto error;

		ubd_dev->cow.fd = err;

	}

	if (ubd_dev->no_trim == 0) {

		ubd_dev->queue->limits.discard_granularity = SECTOR_SIZE;

		ubd_dev->queue->limits.discard_alignment = SECTOR_SIZE;

		blk_queue_max_discard_sectors(ubd_dev->queue, UBD_MAX_REQUEST);

		blk_queue_max_write_zeroes_sectors(ubd_dev->queue, UBD_MAX_REQUEST);

		blk_queue_flag_set(QUEUE_FLAG_DISCARD, ubd_dev->queue);

	}

	blk_queue_flag_set(QUEUE_FLAG_NONROT, ubd_dev->queue);

	return 0;

 error:
@@ -1372,6 +1393,10 @@ static blk_status_t ubd_queue_rq(struct blk_mq_hw_ctx *hctx, 
	case REQ_OP_WRITE:

		ret = queue_rw_req(hctx, req);

		break;

	case REQ_OP_DISCARD:

	case REQ_OP_WRITE_ZEROES:

		ret = ubd_queue_one_vec(hctx, req, (u64)blk_rq_pos(req) << 9, NULL);

		break;

	default:

		WARN_ON_ONCE(1);

		res = BLK_STS_NOTSUPP;
@@ -1471,11 +1496,13 @@ static int update_bitmap(struct io_thread_req *req) 


static void do_io(struct io_thread_req *req)

{

	char *buf;

	char *buf = NULL;

	unsigned long len;

	int n, nsectors, start, end, bit;

	__u64 off;



	/* FLUSH is really a special case, we cannot "case" it with others */



	if (req_op(req->req) == REQ_OP_FLUSH) {

		/* fds[0] is always either the rw image or our cow file */

		req->error = map_error(-os_sync_file(req->fds[0]));
@@ -1495,9 +1522,11 @@ static void do_io(struct io_thread_req *req) 
		off = req->offset + req->offsets[bit] +

			start * req->sectorsize;

		len = (end - start) * req->sectorsize;

		if (req->buffer != NULL)

			buf = &req->buffer[start * req->sectorsize];



		if (req_op(req->req) == REQ_OP_READ) {

		switch (req_op(req->req)) {

		case REQ_OP_READ:

			n = 0;

			do {

				buf = &buf[n];
@@ -1509,12 +1538,26 @@ static void do_io(struct io_thread_req *req) 
				}

			} while((n < len) && (n != 0));

			if (n < len) memset(&buf[n], 0, len - n);

		} else {

			break;

		case REQ_OP_WRITE:

			n = os_pwrite_file(req->fds[bit], buf, len, off);

			if(n != len){

				req->error = map_error(-n);

				return;

			}

			break;

		case REQ_OP_DISCARD:

		case REQ_OP_WRITE_ZEROES:

			n = os_falloc_punch(req->fds[bit], off, len);

			if (n) {

				req->error = map_error(-n);

				return;

			}

			break;

		default:

			WARN_ON_ONCE(1);

			req->error = BLK_STS_NOTSUPP;

			return;

		}



		start = end;

arch/um/include/shared/os.h

+1 −0
Original line number Diff line number Diff line
@@ -175,6 +175,7 @@ extern int os_fchange_dir(int fd); 
extern unsigned os_major(unsigned long long dev);

extern unsigned os_minor(unsigned long long dev);

extern unsigned long long os_makedev(unsigned major, unsigned minor);

extern int os_falloc_punch(int fd, unsigned long long offset, int count);



/* start_up.c */

extern void os_early_checks(void);

arch/um/os-Linux/file.c

+10 −0
Original line number Diff line number Diff line
@@ -610,3 +610,13 @@ unsigned long long os_makedev(unsigned major, unsigned minor) 
{

	return makedev(major, minor);

}



int os_falloc_punch(int fd, unsigned long long offset, int len)

{

	int n = fallocate(fd, FALLOC_FL_PUNCH_HOLE|FALLOC_FL_KEEP_SIZE, offset, len);



	if (n < 0)

		return -errno;

	return n;

}

CRA Git | Maintained and supported by SUSTech CRA and CCSE