/* libguestfs
- * Copyright (C) 2009-2010 Red Hat Inc.
+ * Copyright (C) 2009-2011 Red Hat Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
#include <sys/select.h>
#include <dirent.h>
#include <signal.h>
+#include <assert.h>
#include <rpc/types.h>
#include <rpc/xdr.h>
#include "c-ctype.h"
#include "glthread/lock.h"
-#include "ignore-value.h"
#include "guestfs.h"
#include "guestfs-internal.h"
#include "guestfs-internal-actions.h"
#include "guestfs_protocol.h"
-static int check_peer_euid (guestfs_h *g, int sock, uid_t *rtn);
+static int launch_appliance (guestfs_h *g);
+static int64_t timeval_diff (const struct timeval *x, const struct timeval *y);
+static int connect_unix_socket (guestfs_h *g, const char *sock);
static int qemu_supports (guestfs_h *g, const char *option);
/* Add a string to the current command line. */
return 0;
}
+size_t
+guestfs___checkpoint_cmdline (guestfs_h *g)
+{
+ return g->cmdline_size;
+}
+
+void
+guestfs___rollback_cmdline (guestfs_h *g, size_t pos)
+{
+ size_t i;
+
+ assert (g->cmdline_size >= pos);
+
+ for (i = pos; i < g->cmdline_size; ++i)
+ free (g->cmdline[i]);
+
+ g->cmdline_size = pos;
+}
+
+/* Internal command to return the command line. */
+char **
+guestfs__debug_cmdline (guestfs_h *g)
+{
+ size_t i;
+ char **r;
+
+ if (g->cmdline == NULL) {
+ r = safe_malloc (g, sizeof (char *) * 1);
+ r[0] = NULL;
+ return r;
+ }
+
+ r = safe_malloc (g, sizeof (char *) * (g->cmdline_size + 1));
+ r[0] = safe_strdup (g, g->qemu); /* g->cmdline[0] is always NULL */
+
+ for (i = 1; i < g->cmdline_size; ++i)
+ r[i] = safe_strdup (g, g->cmdline[i]);
+
+ r[g->cmdline_size] = NULL;
+
+ return r; /* caller frees */
+}
+
int
guestfs__config (guestfs_h *g,
const char *qemu_param, const char *qemu_value)
return 0;
}
-int
-guestfs__add_drive_with_if (guestfs_h *g, const char *filename,
- const char *drive_if)
+/* cache=off improves reliability in the event of a host crash.
+ *
+ * However this option causes qemu to try to open the file with
+ * O_DIRECT. This fails on some filesystem types (notably tmpfs).
+ * So we check if we can open the file with or without O_DIRECT,
+ * and use cache=off (or not) accordingly.
+ */
+static int
+test_cache_off (guestfs_h *g, const char *filename)
{
- size_t len = strlen (filename) + 64;
- char buf[len];
-
- if (strchr (filename, ',') != NULL) {
- error (g, _("filename cannot contain ',' (comma) character"));
- return -1;
+ int fd = open (filename, O_RDONLY|O_DIRECT);
+ if (fd >= 0) {
+ close (fd);
+ return 1;
}
- /* cache=off improves reliability in the event of a host crash.
- *
- * However this option causes qemu to try to open the file with
- * O_DIRECT. This fails on some filesystem types (notably tmpfs).
- * So we check if we can open the file with or without O_DIRECT,
- * and use cache=off (or not) accordingly.
- *
- * This test also checks for the presence of the file, which
- * is a documented semantic of this interface.
- */
- int fd = open (filename, O_RDONLY|O_DIRECT);
+ fd = open (filename, O_RDONLY);
if (fd >= 0) {
close (fd);
- snprintf (buf, len, "file=%s,cache=off,if=%s", filename, drive_if);
- } else {
- fd = open (filename, O_RDONLY);
- if (fd >= 0) {
- close (fd);
- snprintf (buf, len, "file=%s,if=%s", filename, drive_if);
- } else {
- perrorf (g, "%s", filename);
- return -1;
- }
+ return 0;
}
- return guestfs__config (g, "-drive", buf);
+ perrorf (g, "%s", filename);
+ return -1;
+}
+
+/* Check string parameter matches ^[-_[:alnum:]]+$ (in C locale). */
+static int
+valid_format_iface (const char *str)
+{
+ size_t len = strlen (str);
+
+ if (len == 0)
+ return 0;
+
+ while (len > 0) {
+ char c = *str++;
+ len--;
+ if (c != '-' && c != '_' && !c_isalnum (c))
+ return 0;
+ }
+ return 1;
}
int
-guestfs__add_drive_ro_with_if (guestfs_h *g, const char *filename,
- const char *drive_if)
+guestfs__add_drive_opts (guestfs_h *g, const char *filename,
+ const struct guestfs_add_drive_opts_argv *optargs)
{
+ int readonly;
+ const char *format;
+ const char *iface;
+
if (strchr (filename, ',') != NULL) {
error (g, _("filename cannot contain ',' (comma) character"));
return -1;
}
- if (access (filename, F_OK) == -1) {
- perrorf (g, "%s", filename);
+ readonly = optargs->bitmask & GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK
+ ? optargs->readonly : 0;
+ format = optargs->bitmask & GUESTFS_ADD_DRIVE_OPTS_FORMAT_BITMASK
+ ? optargs->format : NULL;
+ iface = optargs->bitmask & GUESTFS_ADD_DRIVE_OPTS_IFACE_BITMASK
+ ? optargs->iface : DRIVE_IF;
+
+ if (format && !valid_format_iface (format)) {
+ error (g, _("%s parameter is empty or contains disallowed characters"),
+ "format");
+ return -1;
+ }
+ if (!valid_format_iface (iface)) {
+ error (g, _("%s parameter is empty or contains disallowed characters"),
+ "iface");
return -1;
}
- size_t len = strlen (filename) + 64;
+ /* For writable files, see if we can use cache=off. This also
+ * checks for the existence of the file. For readonly we have
+ * to do the check explicitly.
+ */
+ int use_cache_off = readonly ? 0 : test_cache_off (g, filename);
+ if (use_cache_off == -1)
+ return -1;
+
+ if (readonly) {
+ if (access (filename, F_OK) == -1) {
+ perrorf (g, "%s", filename);
+ return -1;
+ }
+ }
+
+ /* Construct the final -drive parameter. */
+ size_t len = 64 + strlen (filename) + strlen (iface);
+ if (format) len += strlen (format);
char buf[len];
- snprintf (buf, len, "file=%s,snapshot=on,if=%s", filename, drive_if);
+ snprintf (buf, len, "file=%s%s%s%s%s,if=%s",
+ filename,
+ readonly ? ",snapshot=on" : "",
+ use_cache_off ? ",cache=off" : "",
+ format ? ",format=" : "",
+ format ? format : "",
+ iface);
return guestfs__config (g, "-drive", buf);
}
int
guestfs__add_drive (guestfs_h *g, const char *filename)
{
- return guestfs__add_drive_with_if (g, filename, DRIVE_IF);
+ struct guestfs_add_drive_opts_argv optargs = {
+ .bitmask = 0,
+ };
+
+ return guestfs__add_drive_opts (g, filename, &optargs);
}
int
guestfs__add_drive_ro (guestfs_h *g, const char *filename)
{
- return guestfs__add_drive_ro_with_if (g, filename, DRIVE_IF);
+ struct guestfs_add_drive_opts_argv optargs = {
+ .bitmask = GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK,
+ .readonly = 1,
+ };
+
+ return guestfs__add_drive_opts (g, filename, &optargs);
+}
+
+int
+guestfs__add_drive_with_if (guestfs_h *g, const char *filename,
+ const char *iface)
+{
+ struct guestfs_add_drive_opts_argv optargs = {
+ .bitmask = GUESTFS_ADD_DRIVE_OPTS_IFACE_BITMASK,
+ .iface = iface,
+ };
+
+ return guestfs__add_drive_opts (g, filename, &optargs);
+}
+
+int
+guestfs__add_drive_ro_with_if (guestfs_h *g, const char *filename,
+ const char *iface)
+{
+ struct guestfs_add_drive_opts_argv optargs = {
+ .bitmask = GUESTFS_ADD_DRIVE_OPTS_IFACE_BITMASK
+ | GUESTFS_ADD_DRIVE_OPTS_READONLY_BITMASK,
+ .iface = iface,
+ .readonly = 1,
+ };
+
+ return guestfs__add_drive_opts (g, filename, &optargs);
}
int
return guestfs__config (g, "-cdrom", filename);
}
-/* Returns true iff file is contained in dir. */
-static int
-dir_contains_file (const char *dir, const char *file)
-{
- int dirlen = strlen (dir);
- int filelen = strlen (file);
- int len = dirlen+filelen+2;
- char path[len];
-
- snprintf (path, len, "%s/%s", dir, file);
- return access (path, F_OK) == 0;
-}
-
-/* Returns true iff every listed file is contained in 'dir'. */
-static int
-dir_contains_files (const char *dir, ...)
-{
- va_list args;
- const char *file;
-
- va_start (args, dir);
- while ((file = va_arg (args, const char *)) != NULL) {
- if (!dir_contains_file (dir, file)) {
- va_end (args);
- return 0;
- }
- }
- va_end (args);
- return 1;
-}
-
-static void print_timestamped_message (guestfs_h *g, const char *fs, ...);
-static int build_supermin_appliance (guestfs_h *g, const char *path, char **kernel, char **initrd);
static int is_openable (guestfs_h *g, const char *path, int flags);
-static void print_cmdline (guestfs_h *g);
-
-static const char *kernel_name = "vmlinuz." REPO "." host_cpu;
-static const char *initrd_name = "initramfs." REPO "." host_cpu ".img";
int
guestfs__launch (guestfs_h *g)
{
- const char *tmpdir;
- char dir_template[PATH_MAX];
- int r, pmore;
- size_t len;
- int wfd[2], rfd[2];
- int tries;
- char *path, *pelem, *pend;
- char *kernel = NULL, *initrd = NULL;
- int null_vmchannel_sock;
- char unixsock[256];
- struct sockaddr_un addr;
-
/* Configured? */
- if (!g->cmdline) {
- error (g, _("you must call guestfs_add_drive before guestfs_launch"));
- return -1;
- }
-
if (g->state != CONFIG) {
error (g, _("the libguestfs handle has already been launched"));
return -1;
}
- /* Start the clock ... */
- gettimeofday (&g->launch_t, NULL);
-
/* Make the temporary directory. */
-#ifdef P_tmpdir
- tmpdir = P_tmpdir;
-#else
- tmpdir = "/tmp";
-#endif
-
- tmpdir = getenv ("TMPDIR") ? : tmpdir;
- snprintf (dir_template, sizeof dir_template, "%s/libguestfsXXXXXX", tmpdir);
-
if (!g->tmpdir) {
+ TMP_TEMPLATE_ON_STACK (dir_template);
g->tmpdir = safe_strdup (g, dir_template);
if (mkdtemp (g->tmpdir) == NULL) {
perrorf (g, _("%s: cannot create temporary directory"), dir_template);
- goto cleanup0;
+ return -1;
}
}
- /* Allow anyone to read the temporary directory. There are no
- * secrets in the kernel or initrd files. The socket in this
+ /* Allow anyone to read the temporary directory. The socket in this
* directory won't be readable but anyone can see it exists if they
* want. (RHBZ#610880).
*/
if (chmod (g->tmpdir, 0755) == -1)
- fprintf (stderr, "chmod: %s: %m (ignored)\n", g->tmpdir);
-
- /* First search g->path for the supermin appliance, and try to
- * synthesize a kernel and initrd from that. If it fails, we
- * try the path search again looking for a backup ordinary
- * appliance.
- */
- pelem = path = safe_strdup (g, g->path);
- do {
- pend = strchrnul (pelem, ':');
- pmore = *pend == ':';
- *pend = '\0';
- len = pend - pelem;
-
- /* Empty element of "." means cwd. */
- if (len == 0 || (len == 1 && *pelem == '.')) {
- if (g->verbose)
- fprintf (stderr,
- "looking for supermin appliance in current directory\n");
- if (dir_contains_files (".",
- "supermin.d", "kmod.whitelist", NULL)) {
- if (build_supermin_appliance (g, ".", &kernel, &initrd) == -1)
- return -1;
- break;
- }
- }
- /* Look at <path>/supermin* etc. */
- else {
- if (g->verbose)
- fprintf (stderr, "looking for supermin appliance in %s\n", pelem);
-
- if (dir_contains_files (pelem,
- "supermin.d", "kmod.whitelist", NULL)) {
- if (build_supermin_appliance (g, pelem, &kernel, &initrd) == -1)
- return -1;
- break;
- }
- }
+ warning (g, "chmod: %s: %m (ignored)", g->tmpdir);
- pelem = pend + 1;
- } while (pmore);
-
- free (path);
-
- if (kernel == NULL || initrd == NULL) {
- /* Search g->path for the kernel and initrd. */
- pelem = path = safe_strdup (g, g->path);
- do {
- pend = strchrnul (pelem, ':');
- pmore = *pend == ':';
- *pend = '\0';
- len = pend - pelem;
-
- /* Empty element or "." means cwd. */
- if (len == 0 || (len == 1 && *pelem == '.')) {
- if (g->verbose)
- fprintf (stderr,
- "looking for appliance in current directory\n");
- if (dir_contains_files (".", kernel_name, initrd_name, NULL)) {
- kernel = safe_strdup (g, kernel_name);
- initrd = safe_strdup (g, initrd_name);
- break;
- }
- }
- /* Look at <path>/kernel etc. */
- else {
- if (g->verbose)
- fprintf (stderr, "looking for appliance in %s\n", pelem);
-
- if (dir_contains_files (pelem, kernel_name, initrd_name, NULL)) {
- kernel = safe_malloc (g, len + strlen (kernel_name) + 2);
- initrd = safe_malloc (g, len + strlen (initrd_name) + 2);
- sprintf (kernel, "%s/%s", pelem, kernel_name);
- sprintf (initrd, "%s/%s", pelem, initrd_name);
- break;
- }
- }
+ /* Launch the appliance or attach to an existing daemon. */
+ switch (g->attach_method) {
+ case ATTACH_METHOD_APPLIANCE:
+ return launch_appliance (g);
- pelem = pend + 1;
- } while (pmore);
+ case ATTACH_METHOD_UNIX:
+ return connect_unix_socket (g, g->attach_method_arg);
- free (path);
+ default:
+ abort ();
}
+}
- if (kernel == NULL || initrd == NULL) {
- error (g, _("cannot find %s or %s on LIBGUESTFS_PATH (current path = %s)"),
- kernel_name, initrd_name, g->path);
- goto cleanup0;
+static int
+launch_appliance (guestfs_h *g)
+{
+ int r;
+ int wfd[2], rfd[2];
+ char guestfsd_sock[256];
+ struct sockaddr_un addr;
+
+ /* At present you must add drives before starting the appliance. In
+ * future when we enable hotplugging you won't need to do this.
+ */
+ if (!g->cmdline) {
+ error (g, _("you must call guestfs_add_drive before guestfs_launch"));
+ return -1;
}
+ /* Start the clock ... */
+ gettimeofday (&g->launch_t, NULL);
+ guestfs___launch_send_progress (g, 0);
+
+ /* Locate and/or build the appliance. */
+ char *kernel = NULL, *initrd = NULL, *appliance = NULL;
+ if (guestfs___build_appliance (g, &kernel, &initrd, &appliance) == -1)
+ return -1;
+
+ guestfs___launch_send_progress (g, 3);
+
if (g->verbose)
- print_timestamped_message (g, "begin testing qemu features");
+ guestfs___print_timestamped_message (g, "begin testing qemu features");
/* Get qemu help text and version. */
if (qemu_supports (g, NULL) == -1)
goto cleanup0;
- /* Choose which vmchannel implementation to use. */
- if (CAN_CHECK_PEER_EUID && qemu_supports (g, "-net user")) {
- /* The "null vmchannel" implementation. Requires SLIRP (user mode
- * networking in qemu) but no other vmchannel support. The daemon
- * will connect back to a random port number on localhost.
- */
- struct sockaddr_in addr;
- socklen_t addrlen = sizeof addr;
+ /* Using virtio-serial, we need to create a local Unix domain socket
+ * for qemu to connect to.
+ */
+ snprintf (guestfsd_sock, sizeof guestfsd_sock, "%s/guestfsd.sock", g->tmpdir);
+ unlink (guestfsd_sock);
- g->sock = socket (AF_INET, SOCK_STREAM, IPPROTO_TCP);
- if (g->sock == -1) {
- perrorf (g, "socket");
- goto cleanup0;
- }
- addr.sin_family = AF_INET;
- addr.sin_port = htons (0);
- addr.sin_addr.s_addr = htonl (INADDR_LOOPBACK);
- if (bind (g->sock, (struct sockaddr *) &addr, addrlen) == -1) {
- perrorf (g, "bind");
- goto cleanup0;
- }
+ g->sock = socket (AF_UNIX, SOCK_STREAM, 0);
+ if (g->sock == -1) {
+ perrorf (g, "socket");
+ goto cleanup0;
+ }
- if (listen (g->sock, 256) == -1) {
- perrorf (g, "listen");
- goto cleanup0;
- }
+ if (fcntl (g->sock, F_SETFL, O_NONBLOCK) == -1) {
+ perrorf (g, "fcntl");
+ goto cleanup0;
+ }
- if (getsockname (g->sock, (struct sockaddr *) &addr, &addrlen) == -1) {
- perrorf (g, "getsockname");
- goto cleanup0;
- }
+ addr.sun_family = AF_UNIX;
+ strncpy (addr.sun_path, guestfsd_sock, UNIX_PATH_MAX);
+ addr.sun_path[UNIX_PATH_MAX-1] = '\0';
- if (fcntl (g->sock, F_SETFL, O_NONBLOCK) == -1) {
- perrorf (g, "fcntl");
- goto cleanup0;
- }
+ if (bind (g->sock, &addr, sizeof addr) == -1) {
+ perrorf (g, "bind");
+ goto cleanup0;
+ }
- null_vmchannel_sock = ntohs (addr.sin_port);
- if (g->verbose)
- fprintf (stderr, "null_vmchannel_sock = %d\n", null_vmchannel_sock);
- } else {
- /* Using some vmchannel impl. We need to create a local Unix
- * domain socket for qemu to use.
- */
- snprintf (unixsock, sizeof unixsock, "%s/sock", g->tmpdir);
- unlink (unixsock);
- null_vmchannel_sock = 0;
+ if (listen (g->sock, 1) == -1) {
+ perrorf (g, "listen");
+ goto cleanup0;
}
if (!g->direct) {
}
if (g->verbose)
- print_timestamped_message (g, "finished testing qemu features");
+ guestfs___print_timestamped_message (g, "finished testing qemu features");
r = fork ();
if (r == -1) {
if (r == 0) { /* Child (qemu). */
char buf[256];
- const char *vmchannel = NULL;
/* Set up the full command line. Do this in the subprocess so we
* don't need to worry about cleaning up.
*/
g->cmdline[0] = g->qemu;
- /* qemu sometimes needs this option to enable hardware
- * virtualization, but some versions of 'qemu-kvm' will use KVM
- * regardless (even where this option appears in the help text).
- * It is rumoured that there are versions of qemu where supplying
- * this option when hardware virtualization is not available will
- * cause qemu to fail, so we we have to check at least that
- * /dev/kvm is openable. That's not reliable, since /dev/kvm
- * might be openable by qemu but not by us (think: SELinux) in
- * which case the user would not get hardware virtualization,
- * although at least shouldn't fail. A giant clusterfuck with the
- * qemu command line, again.
+ if (qemu_supports (g, "-nodefconfig"))
+ add_cmdline (g, "-nodefconfig");
+
+ /* The qemu -machine option (added 2010-12) is a bit more sane
+ * since it falls back through various different acceleration
+ * modes, so try that first (thanks Markus Armbruster).
*/
- if (qemu_supports (g, "-enable-kvm") &&
- is_openable (g, "/dev/kvm", O_RDWR))
- add_cmdline (g, "-enable-kvm");
+ if (qemu_supports (g, "-machine")) {
+ add_cmdline (g, "-machine");
+ add_cmdline (g, "accel=kvm:tcg");
+ } else {
+ /* qemu sometimes needs this option to enable hardware
+ * virtualization, but some versions of 'qemu-kvm' will use KVM
+ * regardless (even where this option appears in the help text).
+ * It is rumoured that there are versions of qemu where supplying
+ * this option when hardware virtualization is not available will
+ * cause qemu to fail, so we we have to check at least that
+ * /dev/kvm is openable. That's not reliable, since /dev/kvm
+ * might be openable by qemu but not by us (think: SELinux) in
+ * which case the user would not get hardware virtualization,
+ * although at least shouldn't fail. A giant clusterfuck with the
+ * qemu command line, again.
+ */
+ if (qemu_supports (g, "-enable-kvm") &&
+ is_openable (g, "/dev/kvm", O_RDWR))
+ add_cmdline (g, "-enable-kvm");
+ }
/* Newer versions of qemu (from around 2009/12) changed the
* behaviour of monitors so that an implicit '-monitor stdio' is
add_cmdline (g, "-nodefaults");
add_cmdline (g, "-nographic");
- add_cmdline (g, "-serial");
- add_cmdline (g, "stdio");
snprintf (buf, sizeof buf, "%d", g->memsize);
add_cmdline (g, "-m");
if (qemu_supports (g, "-rtc-td-hack"))
add_cmdline (g, "-rtc-td-hack");
- /* If qemu has SLIRP (user mode network) enabled then we can get
- * away with "no vmchannel", where we just connect back to a random
- * host port.
- */
- if (null_vmchannel_sock) {
- add_cmdline (g, "-net");
- add_cmdline (g, "user,vlan=0,net=" NETWORK);
-
- snprintf (buf, sizeof buf,
- "guestfs_vmchannel=tcp:" ROUTER ":%d",
- null_vmchannel_sock);
- vmchannel = strdup (buf);
- }
-
- /* New-style -net user,guestfwd=... syntax for guestfwd. See:
- *
- * http://git.savannah.gnu.org/cgit/qemu.git/commit/?id=c92ef6a22d3c71538fcc48fb61ad353f7ba03b62
- *
- * The original suggested format doesn't work, see:
- *
- * http://lists.gnu.org/archive/html/qemu-devel/2009-07/msg01654.html
- *
- * However Gerd Hoffman privately suggested to me using -chardev
- * instead, which does work.
- */
- else if (qemu_supports (g, "-chardev") && qemu_supports (g, "guestfwd")) {
- snprintf (buf, sizeof buf,
- "socket,id=guestfsvmc,path=%s,server,nowait", unixsock);
-
- add_cmdline (g, "-chardev");
- add_cmdline (g, buf);
-
- snprintf (buf, sizeof buf,
- "user,vlan=0,net=" NETWORK ","
- "guestfwd=tcp:" GUESTFWD_ADDR ":" GUESTFWD_PORT
- "-chardev:guestfsvmc");
-
- add_cmdline (g, "-net");
- add_cmdline (g, buf);
-
- vmchannel = "guestfs_vmchannel=tcp:" GUESTFWD_ADDR ":" GUESTFWD_PORT;
- }
+ /* Create the virtio serial bus. */
+ add_cmdline (g, "-device");
+ add_cmdline (g, "virtio-serial");
- /* Not guestfwd. HOPEFULLY this qemu uses the older -net channel
- * syntax, or if not then we'll get a quick failure.
+#if 0
+ /* Use virtio-console (a variant form of virtio-serial) for the
+ * guest's serial console.
*/
- else {
- snprintf (buf, sizeof buf,
- "channel," GUESTFWD_PORT ":unix:%s,server,nowait", unixsock);
-
- add_cmdline (g, "-net");
- add_cmdline (g, buf);
- add_cmdline (g, "-net");
- add_cmdline (g, "user,vlan=0,net=" NETWORK);
+ add_cmdline (g, "-chardev");
+ add_cmdline (g, "stdio,id=console");
+ add_cmdline (g, "-device");
+ add_cmdline (g, "virtconsole,chardev=console,name=org.libguestfs.console.0");
+#else
+ /* When the above works ... until then: */
+ add_cmdline (g, "-serial");
+ add_cmdline (g, "stdio");
+#endif
- vmchannel = "guestfs_vmchannel=tcp:" GUESTFWD_ADDR ":" GUESTFWD_PORT;
+ /* Set up virtio-serial for the communications channel. */
+ add_cmdline (g, "-chardev");
+ snprintf (buf, sizeof buf, "socket,path=%s,id=channel0", guestfsd_sock);
+ add_cmdline (g, buf);
+ add_cmdline (g, "-device");
+ add_cmdline (g, "virtserialport,chardev=channel0,name=org.libguestfs.channel.0");
+
+ /* Enable user networking. */
+ if (g->enable_network) {
+ add_cmdline (g, "-netdev");
+ add_cmdline (g, "user,id=usernet,net=169.254.0.0/16");
+ add_cmdline (g, "-device");
+ add_cmdline (g, NET_IF ",netdev=usernet");
}
- add_cmdline (g, "-net");
- add_cmdline (g, "nic,model=" NET_IF ",vlan=0");
#define LINUX_CMDLINE \
"panic=1 " /* force kernel to panic if daemon exits */ \
snprintf (buf, sizeof buf,
LINUX_CMDLINE
"%s " /* (selinux) */
- "%s " /* (vmchannel) */
"%s " /* (verbose) */
"TERM=%s " /* (TERM environment variable) */
"%s", /* (append) */
g->selinux ? "selinux=1 enforcing=0" : "selinux=0",
- vmchannel ? vmchannel : "",
g->verbose ? "guestfs_verbose=1" : "",
getenv ("TERM") ? : "linux",
g->append ? g->append : "");
add_cmdline (g, "-kernel");
- add_cmdline (g, (char *) kernel);
+ add_cmdline (g, kernel);
add_cmdline (g, "-initrd");
- add_cmdline (g, (char *) initrd);
+ add_cmdline (g, initrd);
add_cmdline (g, "-append");
add_cmdline (g, buf);
+ /* Add the ext2 appliance drive (last of all). */
+ if (appliance) {
+ const char *cachemode = "";
+ if (qemu_supports (g, "cache=")) {
+ if (qemu_supports (g, "unsafe"))
+ cachemode = ",cache=unsafe";
+ else if (qemu_supports (g, "writeback"))
+ cachemode = ",cache=writeback";
+ }
+
+ char buf2[PATH_MAX + 64];
+ add_cmdline (g, "-drive");
+ snprintf (buf2, sizeof buf2, "file=%s,snapshot=on,if=" DRIVE_IF "%s",
+ appliance, cachemode);
+ add_cmdline (g, buf2);
+ }
+
/* Finish off the command line. */
incr_cmdline_size (g);
g->cmdline[g->cmdline_size-1] = NULL;
if (g->verbose)
- print_cmdline (g);
+ guestfs___print_timestamped_argv (g, (const char **)g->cmdline);
if (!g->direct) {
/* Set up stdin, stdout. */
close (rfd[1]);
}
-#if 0
- /* Set up a new process group, so we can signal this process
- * and all subprocesses (eg. if qemu is really a shell script).
- */
- setpgid (0, 0);
-#endif
+ /* Put qemu in a new process group. */
+ if (g->pgroup)
+ setpgid (0, 0);
setenv ("LC_ALL", "C", 1);
kernel = NULL;
free (initrd);
initrd = NULL;
+ free (appliance);
+ appliance = NULL;
/* Fork the recovery process off which will kill qemu if the parent
* process fails to do so (eg. if the parent segfaults).
pid_t qemu_pid = g->pid;
pid_t parent_pid = getppid ();
+ /* It would be nice to be able to put this in the same process
+ * group as qemu (ie. setpgid (0, qemu_pid)). However this is
+ * not possible because we don't have any guarantee here that
+ * the qemu process has started yet.
+ */
+ if (g->pgroup)
+ setpgid (0, 0);
+
/* Writing to argv is hideously complicated and error prone. See:
* http://anoncvs.postgresql.org/cvsweb.cgi/pgsql/src/backend/utils/misc/ps_status.c?rev=1.33.2.1;content-type=text%2Fplain
*/
}
}
- if (null_vmchannel_sock) {
- int sock = -1;
- uid_t uid;
-
- /* Null vmchannel implementation: We listen on g->sock for a
- * connection. The connection could come from any local process
- * so we must check it comes from the appliance (or at least
- * from our UID) for security reasons.
- */
- while (sock == -1) {
- sock = guestfs__accept_from_daemon (g);
- if (sock == -1)
- goto cleanup1;
-
- if (check_peer_euid (g, sock, &uid) == -1)
- goto cleanup1;
- if (uid != geteuid ()) {
- fprintf (stderr,
- "libguestfs: warning: unexpected connection from UID %d to port %d\n",
- uid, null_vmchannel_sock);
- close (sock);
- sock = -1;
- continue;
- }
- }
-
- if (fcntl (sock, F_SETFL, O_NONBLOCK) == -1) {
- perrorf (g, "fcntl");
- goto cleanup1;
- }
-
- close (g->sock);
- g->sock = sock;
- } else {
- /* Other vmchannel. Open the Unix socket.
- *
- * The vmchannel implementation that got merged with qemu sucks in
- * a number of ways. Both ends do connect(2), which means that no
- * one knows what, if anything, is connected to the other end, or
- * if it becomes disconnected. Even worse, we have to wait some
- * indeterminate time for qemu to create the socket and connect to
- * it (which happens very early in qemu's start-up), so any code
- * that uses vmchannel is inherently racy. Hence this silly loop.
- */
- g->sock = socket (AF_UNIX, SOCK_STREAM, 0);
- if (g->sock == -1) {
- perrorf (g, "socket");
- goto cleanup1;
- }
+ g->state = LAUNCHING;
- if (fcntl (g->sock, F_SETFL, O_NONBLOCK) == -1) {
- perrorf (g, "fcntl");
- goto cleanup1;
- }
+ /* Wait for qemu to start and to connect back to us via
+ * virtio-serial and send the GUESTFS_LAUNCH_FLAG message.
+ */
+ r = guestfs___accept_from_daemon (g);
+ if (r == -1)
+ goto cleanup1;
- addr.sun_family = AF_UNIX;
- strncpy (addr.sun_path, unixsock, UNIX_PATH_MAX);
- addr.sun_path[UNIX_PATH_MAX-1] = '\0';
-
- tries = 100;
- /* Always sleep at least once to give qemu a small chance to start up. */
- usleep (10000);
- while (tries > 0) {
- r = connect (g->sock, (struct sockaddr *) &addr, sizeof addr);
- if ((r == -1 && errno == EINPROGRESS) || r == 0)
- goto connected;
-
- if (errno != ENOENT)
- perrorf (g, "connect");
- tries--;
- usleep (100000);
- }
+ close (g->sock); /* Close the listening socket. */
+ g->sock = r; /* This is the accepted data socket. */
- error (g, _("failed to connect to vmchannel socket"));
+ if (fcntl (g->sock, F_SETFL, O_NONBLOCK) == -1) {
+ perrorf (g, "fcntl");
goto cleanup1;
-
- connected: ;
}
- g->state = LAUNCHING;
-
- /* Wait for qemu to start and to connect back to us via vmchannel and
- * send the GUESTFS_LAUNCH_FLAG message.
- */
uint32_t size;
void *buf = NULL;
- r = guestfs__recv_from_daemon (g, &size, &buf);
+ r = guestfs___recv_from_daemon (g, &size, &buf);
free (buf);
if (r == -1) return -1;
}
if (g->verbose)
- print_timestamped_message (g, "appliance is up");
+ guestfs___print_timestamped_message (g, "appliance is up");
/* This is possible in some really strange situations, such as
* guestfsd starts up OK but then qemu immediately exits. Check for
goto cleanup1;
}
+ guestfs___launch_send_progress (g, 12);
+
return 0;
cleanup1:
}
if (g->pid > 0) kill (g->pid, 9);
if (g->recoverypid > 0) kill (g->recoverypid, 9);
- waitpid (g->pid, NULL, 0);
+ if (g->pid > 0) waitpid (g->pid, NULL, 0);
if (g->recoverypid > 0) waitpid (g->recoverypid, NULL, 0);
g->fd[0] = -1;
g->fd[1] = -1;
g->state = CONFIG;
free (kernel);
free (initrd);
+ free (appliance);
return -1;
}
-/* This function is used to print the qemu command line before it gets
- * executed, when in verbose mode.
+/* Alternate attach method: instead of launching the appliance,
+ * connect to an existing unix socket.
*/
-static void
-print_cmdline (guestfs_h *g)
+static int
+connect_unix_socket (guestfs_h *g, const char *sockpath)
{
- int i = 0;
- int needs_quote;
+ int r;
+ struct sockaddr_un addr;
- while (g->cmdline[i]) {
- if (g->cmdline[i][0] == '-') /* -option starts a new line */
- fprintf (stderr, " \\\n ");
+ /* Start the clock ... */
+ gettimeofday (&g->launch_t, NULL);
- if (i > 0) fputc (' ', stderr);
+ /* Set these to nothing so we don't try to kill random processes or
+ * read from random file descriptors.
+ */
+ g->pid = 0;
+ g->recoverypid = 0;
+ g->fd[0] = -1;
+ g->fd[1] = -1;
- /* Does it need shell quoting? This only deals with simple cases. */
- needs_quote = strcspn (g->cmdline[i], " ") != strlen (g->cmdline[i]);
+ if (g->verbose)
+ guestfs___print_timestamped_message (g, "connecting to %s", sockpath);
- if (needs_quote) fputc ('\'', stderr);
- fprintf (stderr, "%s", g->cmdline[i]);
- if (needs_quote) fputc ('\'', stderr);
- i++;
+ g->sock = socket (AF_UNIX, SOCK_STREAM, 0);
+ if (g->sock == -1) {
+ perrorf (g, "socket");
+ return -1;
}
- fputc ('\n', stderr);
-}
+ addr.sun_family = AF_UNIX;
+ strncpy (addr.sun_path, sockpath, UNIX_PATH_MAX);
+ addr.sun_path[UNIX_PATH_MAX-1] = '\0';
-/* This function does the hard work of building the supermin appliance
- * on the fly. 'path' is the directory containing the control files.
- * 'kernel' and 'initrd' are where we will return the names of the
- * kernel and initrd (only initrd is built). The work is done by
- * an external script. We just tell it where to put the result.
- */
-static int
-build_supermin_appliance (guestfs_h *g, const char *path,
- char **kernel, char **initrd)
-{
- char cmd[4096];
- int r, len;
+ g->state = LAUNCHING;
- if (g->verbose)
- print_timestamped_message (g, "begin building supermin appliance");
+ if (connect (g->sock, &addr, sizeof addr) == -1) {
+ perrorf (g, "bind");
+ goto cleanup;
+ }
- len = strlen (g->tmpdir);
- *kernel = safe_malloc (g, len + 8);
- snprintf (*kernel, len+8, "%s/kernel", g->tmpdir);
- *initrd = safe_malloc (g, len + 8);
- snprintf (*initrd, len+8, "%s/initrd", g->tmpdir);
+ if (fcntl (g->sock, F_SETFL, O_NONBLOCK) == -1) {
+ perrorf (g, "fcntl");
+ goto cleanup;
+ }
- /* Set a sensible umask in the subprocess, so kernel and initrd
- * output files are world-readable (RHBZ#610880).
- */
- snprintf (cmd, sizeof cmd,
- "umask 0002; "
- "febootstrap-supermin-helper%s "
- "-k '%s/kmod.whitelist' "
- "'%s/supermin.d' "
- host_cpu " "
- "%s %s",
- g->verbose ? " --verbose" : "",
- path,
- path,
- *kernel, *initrd);
- if (g->verbose)
- print_timestamped_message (g, "%s", cmd);
-
- r = system (cmd);
- if (r == -1 || WEXITSTATUS(r) != 0) {
- error (g, _("external command failed: %s"), cmd);
- free (*kernel);
- free (*initrd);
- *kernel = *initrd = NULL;
- return -1;
+ uint32_t size;
+ void *buf = NULL;
+ r = guestfs___recv_from_daemon (g, &size, &buf);
+ free (buf);
+
+ if (r == -1) return -1;
+
+ if (size != GUESTFS_LAUNCH_FLAG) {
+ error (g, _("guestfs_launch failed, unexpected initial message from guestfsd"));
+ goto cleanup;
}
if (g->verbose)
- print_timestamped_message (g, "finished building supermin appliance");
+ guestfs___print_timestamped_message (g, "connected");
+
+ if (g->state != READY) {
+ error (g, _("contacted guestfsd, but state != READY"));
+ goto cleanup;
+ }
return 0;
+
+ cleanup:
+ close (g->sock);
+ return -1;
+}
+
+/* launch (of the ordinary appliance) generates approximate progress
+ * messages. Currently these are defined as follows:
+ *
+ * 0 / 12: launch clock starts
+ * 3 / 12: appliance created
+ * 6 / 12: detected that guest kernel started
+ * 9 / 12: detected that /init script is running
+ * 12 / 12: launch completed successfully
+ *
+ * Notes:
+ * (1) This is not a documented ABI and the behaviour may be changed
+ * or removed in future.
+ * (2) Messages are only sent if more than 5 seconds has elapsed
+ * since the launch clock started.
+ * (3) There is a gross hack in proto.c to make this work.
+ */
+void
+guestfs___launch_send_progress (guestfs_h *g, int perdozen)
+{
+ struct timeval tv;
+
+ gettimeofday (&tv, NULL);
+ if (timeval_diff (&g->launch_t, &tv) >= 5000) {
+ guestfs_progress progress_message =
+ { .proc = 0, .serial = 0, .position = perdozen, .total = 12 };
+
+ guestfs___progress_message_callback (g, &progress_message);
+ }
+}
+
+/* Return the location of the tmpdir (eg. "/tmp") and allow users
+ * to override it at runtime using $TMPDIR.
+ * http://www.pathname.com/fhs/pub/fhs-2.3.html#TMPTEMPORARYFILES
+ */
+const char *
+guestfs_tmpdir (void)
+{
+ const char *tmpdir;
+
+#ifdef P_tmpdir
+ tmpdir = P_tmpdir;
+#else
+ tmpdir = "/tmp";
+#endif
+
+ const char *t = getenv ("TMPDIR");
+ if (t) tmpdir = t;
+
+ return tmpdir;
+}
+
+/* Return the location of the persistent tmpdir (eg. "/var/tmp") and
+ * allow users to override it at runtime using $TMPDIR.
+ * http://www.pathname.com/fhs/pub/fhs-2.3.html#VARTMPTEMPORARYFILESPRESERVEDBETWEE
+ */
+const char *
+guestfs___persistent_tmpdir (void)
+{
+ const char *tmpdir;
+
+ tmpdir = "/var/tmp";
+
+ const char *t = getenv ("TMPDIR");
+ if (t) tmpdir = t;
+
+ return tmpdir;
}
/* Compute Y - X and return the result in milliseconds.
return msec;
}
-static void
-print_timestamped_message (guestfs_h *g, const char *fs, ...)
+void
+guestfs___print_timestamped_argv (guestfs_h *g, const char * argv[])
+{
+ int i = 0;
+ int needs_quote;
+ char *buf = NULL;
+ size_t len;
+ FILE *fp;
+
+ fp = open_memstream (&buf, &len);
+ if (fp == NULL) {
+ warning (g, "open_memstream: %m");
+ return;
+ }
+
+ struct timeval tv;
+ gettimeofday (&tv, NULL);
+ fprintf (fp, "[%05" PRIi64 "ms] ", timeval_diff (&g->launch_t, &tv));
+
+ while (argv[i]) {
+ if (argv[i][0] == '-') /* -option starts a new line */
+ fprintf (fp, " \\\n ");
+
+ if (i > 0) fputc (' ', fp);
+
+ /* Does it need shell quoting? This only deals with simple cases. */
+ needs_quote = strcspn (argv[i], " ") != strlen (argv[i]);
+
+ if (needs_quote) fputc ('\'', fp);
+ fprintf (fp, "%s", argv[i]);
+ if (needs_quote) fputc ('\'', fp);
+ i++;
+ }
+
+ fclose (fp);
+
+ debug (g, "%s", buf);
+
+ free (buf);
+}
+
+void
+guestfs___print_timestamped_message (guestfs_h *g, const char *fs, ...)
{
va_list args;
char *msg;
gettimeofday (&tv, NULL);
- fprintf (stderr, "[%05" PRIi64 "ms] %s\n",
- timeval_diff (&g->launch_t, &tv), msg);
+ debug (g, "[%05" PRIi64 "ms] %s", timeval_diff (&g->launch_t, &tv), msg);
free (msg);
}
goto error;
snprintf (cmd, sizeof cmd, "LC_ALL=C '%s' -nographic -version 2>/dev/null",
- g->qemu);
+ g->qemu);
fp = popen (cmd, "r");
if (fp) {
{
int fd = open (path, flags);
if (fd == -1) {
- if (g->verbose)
- perror (path);
+ debug (g, "is_openable: %s: %m", path);
return 0;
}
close (fd);
return 1;
}
-/* Check the peer effective UID for a TCP socket. Ideally we'd like
- * SO_PEERCRED for a loopback TCP socket. This isn't possible on
- * Linux (but it is on Solaris!) so we read /proc/net/tcp instead.
- */
-static int
-check_peer_euid (guestfs_h *g, int sock, uid_t *rtn)
-{
-#if CAN_CHECK_PEER_EUID
- struct sockaddr_in peer;
- socklen_t addrlen = sizeof peer;
-
- if (getpeername (sock, (struct sockaddr *) &peer, &addrlen) == -1) {
- perrorf (g, "getpeername");
- return -1;
- }
-
- if (peer.sin_family != AF_INET ||
- ntohl (peer.sin_addr.s_addr) != INADDR_LOOPBACK) {
- error (g, "check_peer_euid: unexpected connection from non-IPv4, non-loopback peer (family = %d, addr = %s)",
- peer.sin_family, inet_ntoa (peer.sin_addr));
- return -1;
- }
-
- struct sockaddr_in our;
- addrlen = sizeof our;
- if (getsockname (sock, (struct sockaddr *) &our, &addrlen) == -1) {
- perrorf (g, "getsockname");
- return -1;
- }
-
- FILE *fp = fopen ("/proc/net/tcp", "r");
- if (fp == NULL) {
- perrorf (g, "/proc/net/tcp");
- return -1;
- }
-
- char line[256];
- if (fgets (line, sizeof line, fp) == NULL) { /* Drop first line. */
- error (g, "unexpected end of file in /proc/net/tcp");
- fclose (fp);
- return -1;
- }
-
- while (fgets (line, sizeof line, fp) != NULL) {
- unsigned line_our_addr, line_our_port, line_peer_addr, line_peer_port;
- int dummy0, dummy1, dummy2, dummy3, dummy4, dummy5, dummy6;
- int line_uid;
-
- if (sscanf (line, "%d:%08X:%04X %08X:%04X %02X %08X:%08X %02X:%08X %08X %d",
- &dummy0,
- &line_our_addr, &line_our_port,
- &line_peer_addr, &line_peer_port,
- &dummy1, &dummy2, &dummy3, &dummy4, &dummy5, &dummy6,
- &line_uid) == 12) {
- /* Note about /proc/net/tcp: local_address and rem_address are
- * always in network byte order. However the port part is
- * always in host byte order.
- *
- * The sockname and peername that we got above are in network
- * byte order. So we have to byte swap the port but not the
- * address part.
- */
- if (line_our_addr == our.sin_addr.s_addr &&
- line_our_port == ntohs (our.sin_port) &&
- line_peer_addr == peer.sin_addr.s_addr &&
- line_peer_port == ntohs (peer.sin_port)) {
- *rtn = line_uid;
- fclose (fp);
- return 0;
- }
- }
- }
-
- error (g, "check_peer_euid: no matching TCP connection found in /proc/net/tcp");
- fclose (fp);
- return -1;
-#else /* !CAN_CHECK_PEER_EUID */
- /* This function exists but should never be called in this
- * configuration.
- */
- abort ();
-#endif /* !CAN_CHECK_PEER_EUID */
-}
-
/* You had to call this function after launch in versions <= 1.0.70,
* but it is now a no-op.
*/
return -1;
}
- if (g->verbose)
- fprintf (stderr, "sending SIGTERM to process %d\n", g->pid);
+ debug (g, "sending SIGTERM to process %d", g->pid);
if (g->pid > 0) kill (g->pid, SIGTERM);
if (g->recoverypid > 0) kill (g->recoverypid, 9);