Source-Changes-HG archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

[src/trunk]: src/share/examples/rump/umserv Add a simple example of how to ru...



details:   https://anonhg.NetBSD.org/src/rev/915ef2daa166
branches:  trunk
changeset: 750270:915ef2daa166
user:      pooka <pooka%NetBSD.org@localhost>
date:      Tue Dec 22 18:36:02 2009 +0000

description:
Add a simple example of how to run the kernel umass/USB in userspace
and attach it back to the host kernel using pud(4).  The result is
pretty much indistinguisable from if kernel drivers were used.

It's possible to e.g. mount a file system backed by the userspace
block device driver:

golem> disklabel ./rumpsd0d
[...]
5 partitions:
#        size    offset     fstype [fsize bsize cpg/sgs]
 d:   1994752         0     unused      0     0        # (Cyl.      0 -   3895)
 e:   1994512       240      MSDOS                     # (Cyl.      0*-   3895)

golem> s mount_msdos ./rumpsd0a /mnt
mount_msdos: "./rumpsd0a" is a non-resolved or relative path.
mount_msdos: using "/usr/home/pooka/src/umserv/rumpsd0a" instead.
mount_msdos: /usr/home/pooka/src/umserv/rumpsd0a on /mnt: Device not configured

golem> s mount_msdos ./rumpsd0e /mnt
mount_msdos: "./rumpsd0e" is a non-resolved or relative path.
mount_msdos: using "/usr/home/pooka/src/umserv/rumpsd0e" instead.
golem> df -h /mnt
Filesystem                               Size       Used      Avail %Cap Mounted on
/usr/home/pooka/src/umserv/rumpsd0e      974M       5.2M       968M   0% /mnt

etcetc.

diffstat:

 share/examples/rump/umserv/Makefile |   10 +
 share/examples/rump/umserv/umserv.c |  224 ++++++++++++++++++++++++++++++++++++
 2 files changed, 234 insertions(+), 0 deletions(-)

diffs (242 lines):

diff -r b93529a2bab9 -r 915ef2daa166 share/examples/rump/umserv/Makefile
--- /dev/null   Thu Jan 01 00:00:00 1970 +0000
+++ b/share/examples/rump/umserv/Makefile       Tue Dec 22 18:36:02 2009 +0000
@@ -0,0 +1,10 @@
+PROG=umserv
+NOMAN=
+LDADD+=-lrumpdev_usbhc -lrumpdev_umass -lrumpdev_usb -lrumpdev_disk -lrumpdev
+LDADD+=-lrumpvfs
+LDADD+=-lrump -lrumpuser -lpthread
+
+WARNS=4
+DBG=-g
+
+.include <bsd.prog.mk>
diff -r b93529a2bab9 -r 915ef2daa166 share/examples/rump/umserv/umserv.c
--- /dev/null   Thu Jan 01 00:00:00 1970 +0000
+++ b/share/examples/rump/umserv/umserv.c       Tue Dec 22 18:36:02 2009 +0000
@@ -0,0 +1,224 @@
+/*     $NetBSD: umserv.c,v 1.1 2009/12/22 18:36:02 pooka Exp $ */
+
+/*
+ * Copyright (c) 2009 Antti Kantee.  All Rights Reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+/*
+ * This is a [quick, simple & dirty] userspace sd@umass server.
+ * We probe USB devices using rump and attach them to the host kernel
+ * using pud(4).  The resulting block devices can be e.g. read
+ * and/or mounted.
+ *
+ * Since there is no devfs support in NetBSD, we create crudo & cotto
+ * device nodes in the current directory.  Operating on these in the
+ * host OS will direct operations to this userspace server, e.g.:
+ *   golem> disklabel ./rumpsd0d
+ *   golem> mount_msdos ./rumpsd0e /mnt
+ * will cause file system access to /mnt be backed by the umass server
+ * in userspace.  Due to the relatively experimental nature of this
+ * server, rump file servers are recommended for mounting experiments.
+ */
+
+#include <sys/types.h>
+#include <sys/syslimits.h>
+
+#include <dev/pud/pud_msgif.h>
+
+#include <rump/rump.h>
+
+#include <assert.h>
+#include <err.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <paths.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+/*
+ * No devfs?  No problem.  We just hack a bit & wait for the dust to settle.
+ */
+#define NODEBASE "rumpsd0"
+#define MYMAJOR 411
+static int
+makenodes(void)
+{
+       char path[PATH_MAX];
+       struct stat sb;
+       int i, j, rv;
+
+       for (i = 0; i < 2; i++) {
+               int minnum = 0;
+
+               for (j = 0; j < 8; j++, minnum++) {
+                       sprintf(path, "%s%s%c",
+                           i == 0 ? "" : "r", NODEBASE, minnum + 'a');
+                       if (stat(path, &sb) == 0)
+                               continue;
+                       rv = mknod(path, (i == 0 ? S_IFBLK : S_IFCHR) | 0666,
+                           makedev(MYMAJOR, minnum));
+                       if (rv != 0 && !(rv == -1 && errno == EEXIST))
+                               return rv;
+               }
+       }
+
+       return 0;
+}
+
+int
+main(int argc, char *argv[])
+{
+       char path[PATH_MAX];
+       struct pud_conf_reg pcr;
+       struct pud_req *pdr;
+       struct vnode *devvps[8], *devvp;
+       kauth_cred_t rootcred;
+       ssize_t n;
+       int fd, rv, i;
+
+       if (makenodes() == -1)
+               err(1, "makenodes");
+
+       fd = open(_PATH_PUD, O_RDWR);
+       if (fd == -1)
+               err(1, "open");
+
+#define PDRSIZE (64*1024+1024)
+       pdr = malloc(PDRSIZE);
+       if (pdr == NULL)
+               err(1, "malloc");
+
+       memset(&pcr, 0, sizeof(pcr));
+       pcr.pm_pdr.pdr_pth.pth_framelen = sizeof(struct pud_conf_reg);
+       pcr.pm_version = PUD_DEVELVERSION | PUD_VERSION;
+       pcr.pm_pdr.pdr_reqclass = PUD_REQ_CONF;
+       pcr.pm_pdr.pdr_reqtype = PUD_CONF_REG;
+
+       pcr.pm_regdev = makedev(MYMAJOR, 0);
+       pcr.pm_flags = PUD_CONFFLAG_BDEV;
+       strlcpy(pcr.pm_devname, "youmass", sizeof(pcr.pm_devname));
+
+       n = write(fd, &pcr, pcr.pm_pdr.pdr_pth.pth_framelen);
+       if (n == -1)
+               err(1, "configure"); /* XXX: doubles as protocol error */
+
+       rump_boot_sethowto(RUMP_AB_VERBOSE);
+       rump_init();
+
+       /*
+        * We use the raw devices.  This avoids undesired block caching
+        * in read/write.  It's also simpler.  It's even mostly correct.
+        *
+        * As is probably obvious by now, we execute ops through specfs.
+        * Why?  Can't say it wasn't because specfs-via-vnodes was
+        * already supported by rump.  But, that's mostly how the
+        * kernel does it (cf. mounting file system etc).
+        */
+       for (i = 0; i < 8; i++) {
+               sprintf(path, "/dev/rsd0%c", 'a' + i);
+               if ((rv = rump_pub_namei(RUMP_NAMEI_LOOKUP, 0, path,
+                   NULL, &devvps[i], NULL)) != 0)
+                       errx(1, "raw device lookup failed %d", rv);
+       }
+
+       rootcred = rump_pub_cred_create(0, 0, 0, NULL);
+
+       /* process requests ad infinitum */
+       for (;;) {
+               struct pud_creq_open *pr_open;
+               struct pud_creq_close *pr_close;
+               struct pud_req_readwrite *pr_rw;
+               struct pud_req_ioctl *pr_ioctl;
+               struct uio *uio;
+               size_t reslen;
+               int minordev;
+
+               n = read(fd, pdr, PDRSIZE);
+               if (n == -1)
+                       err(1, "read");
+
+               minordev = minor(pdr->pdr_dev);
+               if (minordev < 0 || minordev >= 8) {
+                       rv = ENXIO;
+                       goto sendresponse;
+               }
+               devvp = devvps[minordev];
+
+               switch (pdr->pdr_reqtype) {
+               case PUD_BDEV_OPEN:
+                       pr_open = (void *)pdr;
+                       RUMP_VOP_LOCK(devvp, RUMP_LK_EXCLUSIVE);
+                       rv = RUMP_VOP_OPEN(devvp, pr_open->pm_fmt, rootcred);
+                       RUMP_VOP_UNLOCK(devvp, 0);
+                       break;
+
+               case PUD_BDEV_CLOSE:
+                       pr_close = (void *)pdr;
+                       RUMP_VOP_LOCK(devvp, RUMP_LK_EXCLUSIVE);
+                       rv = RUMP_VOP_CLOSE(devvp, pr_close->pm_fmt, rootcred);
+                       RUMP_VOP_UNLOCK(devvp, 0);
+                       break;
+
+               case PUD_BDEV_IOCTL:
+                       pr_ioctl = (void *)pdr;
+                       rv = RUMP_VOP_IOCTL(devvp, pr_ioctl->pm_iocmd,
+                           pr_ioctl->pm_data, pr_ioctl->pm_flag, rootcred);
+                       break;
+
+               case PUD_BDEV_STRATREAD:
+                       pr_rw = (void *)pdr;
+                       assert(pr_rw->pm_resid <= 64*1024);
+                       uio = rump_pub_uio_setup(&pr_rw->pm_data[0],
+                           pr_rw->pm_resid, pr_rw->pm_offset, RUMPUIO_READ);
+                       RUMP_VOP_LOCK(devvp, RUMP_LK_SHARED);
+                       rv = RUMP_VOP_READ(devvp, uio, 0, rootcred);
+                       RUMP_VOP_UNLOCK(devvp, 0);
+                       reslen = rump_pub_uio_free(uio);
+                       pdr->pdr_pth.pth_framelen -= reslen;
+                       pr_rw->pm_resid = reslen;
+                       break;
+
+               case PUD_BDEV_STRATWRITE:
+                       pr_rw = (void *)pdr;
+                       uio = rump_pub_uio_setup(&pr_rw->pm_data[0],
+                           pr_rw->pm_resid, pr_rw->pm_offset, RUMPUIO_WRITE);
+                       RUMP_VOP_LOCK(devvp, RUMP_LK_EXCLUSIVE);
+                       rv = RUMP_VOP_WRITE(devvp, uio, 0, rootcred);
+                       RUMP_VOP_UNLOCK(devvp, 0);
+                       reslen = rump_pub_uio_free(uio);
+                       pr_rw->pm_resid = reslen;
+                       pdr->pdr_pth.pth_framelen=sizeof(struct pud_creq_write);
+                       rv = 0;
+                       break;
+
+               }
+
+ sendresponse:
+               pdr->pdr_rv = rv;
+               n = write(fd, pdr, pdr->pdr_pth.pth_framelen);
+               assert(n == (ssize_t)pdr->pdr_pth.pth_framelen);
+       }
+}



Home | Main Index | Thread Index | Old Index