[PATCH] nbd in 2.5.3 does not work, and can cause severe damage when read-write

From: Petr Vandrovec (vandrove@vc.cvut.cz)
Date: Thu Jan 31 2002 - 08:24:46 EST


Hi Linus,
    I've got strange idea and tried to build diskless machine around
2.5.3... Besides problem with segfaulting crc32 (it is initialized after
net/ipv4/ipconfig.c due to lib/lib.a being a library... I had to hardcode
lib/crc32.o before --start-group in main Makefile, but it is another
story) there is bad problem with NBD caused by BIO changes:

(1) request flags were immediately put into on-wire request format.
    In the past, we had 0=READ, !0=WRITE. Now only REQ_RW bit determines
    direction. As nbd-server from nbd distribution package treats any
    non-zero value as write, it performs writes instead of read. Fortunately
    it will die due to other consistency checks on incoming request, but...

(2) nbd servers handle only up to 10240 byte requests. So setting max_sectors
    to 20 is needed, as otherwise nbd server commits suicide. Maximum request size
    should be handshaked during nbd initialization, but currently just use
    hardwired 20 sectors, so it will behave like it did in the past.

                                                Thanks,
                                                        Petr Vandrovec
                                                        vandrove@vc.cvut.cz

diff -urdN linux/drivers/block/nbd.c linux/drivers/block/nbd.c
--- linux/drivers/block/nbd.c Thu Jan 10 18:15:38 2002
+++ linux/drivers/block/nbd.c Thu Jan 31 00:24:50 2002
@@ -155,14 +155,15 @@
         unsigned long size = req->nr_sectors << 9;
 
         DEBUG("NBD: sending control, ");
+
+ rw = rq_data_dir(req);
+
         request.magic = htonl(NBD_REQUEST_MAGIC);
- request.type = htonl(req->flags);
+ request.type = htonl((rw & WRITE) ? 1 : 0);
         request.from = cpu_to_be64( (u64) req->sector << 9);
         request.len = htonl(size);
         memcpy(request.handle, &req, sizeof(req));
 
- rw = rq_data_dir(req);
-
         result = nbd_xmit(1, sock, (char *) &request, sizeof(request), rw & WRITE ? MSG_MORE : 0);
         if (result <= 0)
                 FAIL("Sendmsg failed for control.");
@@ -517,6 +518,7 @@
         blksize_size[MAJOR_NR] = nbd_blksizes;
         blk_size[MAJOR_NR] = nbd_sizes;
         blk_init_queue(BLK_DEFAULT_QUEUE(MAJOR_NR), do_nbd_request, &nbd_lock);
+ blk_queue_max_sectors(BLK_DEFAULT_QUEUE(MAJOR_NR), 20);
         for (i = 0; i < MAX_NBD; i++) {
                 nbd_dev[i].refcnt = 0;
                 nbd_dev[i].file = NULL;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Thu Jan 31 2002 - 21:01:34 EST