libvirt/src/libvirt-lxc.c

319 lines
8.9 KiB
C
Raw Normal View History

/*
* libvirt-lxc.c: Interfaces for the libvirt library to handle lxc-specific
* APIs.
*
* Copyright (C) 2012-2014 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
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see
* <http://www.gnu.org/licenses/>.
*/
#include <config.h>
#include "viralloc.h"
#include "virerror.h"
#include "virfile.h"
#include "virlog.h"
#include "virprocess.h"
#include "viruuid.h"
#include "datatypes.h"
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
#ifdef WITH_SELINUX
# include <selinux/selinux.h>
#endif
#ifdef WITH_APPARMOR
# include <sys/apparmor.h>
#endif
#include "vircgroup.h"
#define VIR_FROM_THIS VIR_FROM_NONE
VIR_LOG_INIT("libvirt-lxc");
/**
* virDomainLxcOpenNamespace:
* @domain: a domain object
* @fdlist: pointer to an array to be filled with FDs
* @flags: currently unused, pass 0
*
* This API is LXC specific, so it will only work with hypervisor
* connections to the LXC driver.
*
* Open the namespaces associated with the container @domain.
* The @fdlist array will be allocated to a suitable size,
* and filled with file descriptors for the namespaces. It
* is the caller's responsibility to close the file descriptors
*
* The returned file descriptors are intended to be used with
* the setns() system call.
*
* Returns the number of opened file descriptors, or -1 on error
*/
int
virDomainLxcOpenNamespace(virDomainPtr domain,
int **fdlist,
unsigned int flags)
{
virConnectPtr conn;
VIR_DOMAIN_DEBUG(domain, "fdlist=%p flags=0x%x", fdlist, flags);
virResetLastError();
virCheckDomainReturn(domain, -1);
conn = domain->conn;
virCheckNonNullArgGoto(fdlist, error);
2013-12-20 14:02:49 +00:00
virCheckReadOnlyGoto(conn->flags, error);
if (conn->driver->domainLxcOpenNamespace) {
int ret;
ret = conn->driver->domainLxcOpenNamespace(domain,
fdlist,
flags);
if (ret < 0)
goto error;
return ret;
}
maint: improve VIR_ERR_NO_SUPPORT usage We weren't very consistent in our use of VIR_ERR_NO_SUPPORT; many users just passed __FUNCTION__ on, while others passed "%s" to silence over-eager compilers that warn about __FUNCTION__ not containing any %. It's nicer to route all these uses through a single macro, so that if we ever need to change the reporting, we can do it in one place. I verified that 'virsh -c test:///default qemu-monitor-command test foo' gives the same error message before and after this patch: error: this function is not supported by the connection driver: virDomainQemuMonitorCommand Note that in libvirt.c, we were inconsistent on whether virDomain* API used virLibConnError() (with VIR_FROM_NONE) or virLibDomainError() (with VIR_FROM_DOMAIN); this patch unifies these errors to all use VIR_FROM_NONE, on the grounds that it is unlikely that a caller learning that a call is unimplemented can do anything in particular with extra knowledge of which error domain it belongs to. One particular change to note is virDomainOpenGraphics which was trying to fail with VIR_ERR_NO_SUPPORT after a failed VIR_DRV_SUPPORTS_FEATURE check; all other places that fail a feature check report VIR_ERR_ARGUMENT_UNSUPPORTED. * src/util/virerror.h (virReportUnsupportedError): New macro. * src/libvirt-qemu.c: Use new macro. * src/libvirt-lxc.c: Likewise. * src/lxc/lxc_driver.c: Likewise. * src/security/security_manager.c: Likewise. * src/util/virinitctl.c: Likewise. * src/libvirt.c: Likewise. (virDomainOpenGraphics): Use correct error for unsupported feature. Signed-off-by: Eric Blake <eblake@redhat.com>
2013-12-20 01:38:59 +00:00
virReportUnsupportedError();
error:
virDispatchError(conn);
return -1;
}
/**
* virDomainLxcEnterNamespace:
* @domain: a domain object
* @nfdlist: number of FDs in @fdlist
* @fdlist: list of namespace file descriptors
* @noldfdlist: filled with number of old FDs
* @oldfdlist: pointer to hold list of old namespace file descriptors
* @flags: currently unused, pass 0
*
* This API is LXC specific, so it will only work with hypervisor
* connections to the LXC driver.
*
* Attaches the process to the namespaces associated
* with the FDs in @fdlist
*
* If @oldfdlist is non-NULL, it will be populated with file
* descriptors representing the old namespace. This allows
* the caller to switch back to its current namespace later
*
* Returns 0 on success, -1 on error
*/
int
virDomainLxcEnterNamespace(virDomainPtr domain,
unsigned int nfdlist,
int *fdlist,
unsigned int *noldfdlist,
int **oldfdlist,
unsigned int flags)
{
size_t i;
VIR_DOMAIN_DEBUG(domain, "nfdlist=%d, fdlist=%p, "
"noldfdlist=%p, oldfdlist=%p, flags=0x%x",
nfdlist, fdlist, noldfdlist, oldfdlist, flags);
virResetLastError();
virCheckFlagsGoto(0, error);
if (noldfdlist && oldfdlist) {
size_t nfds;
if (virProcessGetNamespaces(getpid(),
&nfds,
oldfdlist) < 0)
goto error;
*noldfdlist = nfds;
}
if (virProcessSetNamespaces(nfdlist, fdlist) < 0) {
if (oldfdlist && noldfdlist) {
for (i = 0; i < *noldfdlist; i++)
VIR_FORCE_CLOSE((*oldfdlist)[i]);
VIR_FREE(*oldfdlist);
*noldfdlist = 0;
}
goto error;
}
return 0;
error:
virDispatchError(domain->conn);
return -1;
}
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
/**
* virDomainLxcEnterSecurityLabel:
* @model: the security model to set
* @label: the security label to apply
* @oldlabel: filled with old security label
* @flags: currently unused, pass 0
*
* This API is LXC specific, so it will only work with hypervisor
* connections to the LXC driver.
*
* Attaches the process to the security label specified
* by @label. @label is interpreted relative to @model
* Depending on the security driver, this may
* not take effect until the next call to exec().
*
* If @oldlabel is not NULL, it will be filled with info
* about the current security label. This may let the
* process be moved back to the previous label if no
* exec() has yet been performed.
*
* Returns 0 on success, -1 on error
*/
int
virDomainLxcEnterSecurityLabel(virSecurityModelPtr model,
virSecurityLabelPtr label,
virSecurityLabelPtr oldlabel,
unsigned int flags)
{
VIR_DEBUG("model=%p, label=%p, oldlabel=%p, flags=0x%x",
model, label, oldlabel, flags);
virResetLastError();
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
virCheckFlagsGoto(0, error);
virCheckNonNullArgGoto(model, error);
virCheckNonNullArgGoto(label, error);
if (oldlabel)
memset(oldlabel, 0, sizeof(*oldlabel));
if (STREQ(model->model, "selinux")) {
#ifdef WITH_SELINUX
if (oldlabel) {
char *ctx;
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
if (getcon(&ctx) < 0) {
virReportSystemError(errno,
_("unable to get PID %d security context"),
getpid());
goto error;
}
if (strlen((char *) ctx) >= VIR_SECURITY_LABEL_BUFLEN) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("security label exceeds "
"maximum length: %d"),
VIR_SECURITY_LABEL_BUFLEN - 1);
freecon(ctx);
goto error;
}
strcpy(oldlabel->label, (char *) ctx);
freecon(ctx);
if ((oldlabel->enforcing = security_getenforce()) < 0) {
virReportSystemError(errno, "%s",
_("error calling security_getenforce()"));
goto error;
}
}
if (setexeccon(label->label) < 0) {
virReportSystemError(errno,
_("Cannot set context %s"),
label->label);
goto error;
}
#else
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
_("Support for SELinux is not enabled"));
goto error;
#endif
} else if (STREQ(model->model, "apparmor")) {
#ifdef WITH_APPARMOR
if (aa_change_profile(label->label) < 0) {
virReportSystemError(errno, _("error changing profile to %s"),
label->label);
goto error;
}
#else
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
_("Support for AppArmor is not enabled"));
goto error;
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
#endif
} else if (STREQ(model->model, "none")) {
/* nothing todo */
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
} else {
virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED,
_("Security model %s cannot be entered"),
model->model);
goto error;
}
return 0;
error:
Apply security label when entering LXC namespaces Add a new virDomainLxcEnterSecurityLabel() function as a counterpart to virDomainLxcEnterNamespaces(), which can change the current calling process to have a new security context. This call runs client side, not in libvirtd so we can't use the security driver infrastructure. When entering a namespace, the process spawned from virsh will default to running with the security label of virsh. The actual desired behaviour is to run with the security label of the container most of the time. So this changes virsh lxc-enter-namespace command to invoke the virDomainLxcEnterSecurityLabel method. The current behaviour is: LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 29 ? 00:00:00 dhclient staff_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 47 ? 00:00:00 ps Note the ps command is running as unconfined_t, After this patch, The new behaviour is this: virsh -c lxc:/// lxc-enter-namespace dan -- /bin/ps -eZ LABEL PID TTY TIME CMD system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 1 pts/0 00:00:00 systemd system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 3 pts/1 00:00:00 sh system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 24 ? 00:00:00 systemd-journal system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 32 ? 00:00:00 dhclient system_u:system_r:svirt_lxc_net_t:s0:c0.c1023 38 ? 00:00:00 ps The '--noseclabel' flag can be used to skip security labelling. Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-03-12 17:24:01 +00:00
virDispatchError(NULL);
return -1;
}
/**
* virDomainLxcEnterCGroup:
* @domain: a domain object
* @flags: currently unused, pass 0
*
* This API is LXC specific, so it will only work with hypervisor
* connections to the LXC driver.
*
* Attaches the process to the control cgroups associated
* with the container @domain.
*
* Returns 0 on success, -1 on error
*/
int virDomainLxcEnterCGroup(virDomainPtr domain,
unsigned int flags)
{
virConnectPtr conn;
virCgroupPtr cgroup = NULL;
VIR_DOMAIN_DEBUG(domain, "flags=0x%x", flags);
virResetLastError();
virCheckDomainReturn(domain, -1);
conn = domain->conn;
virCheckReadOnlyGoto(conn->flags, error);
virCheckFlagsGoto(0, error);
if (virCgroupNewDetect(domain->id, -1, &cgroup) < 0)
goto error;
if (virCgroupAddProcess(cgroup, getpid()) < 0)
goto error;
virCgroupFree(&cgroup);
return 0;
error:
virDispatchError(NULL);
virCgroupFree(&cgroup);
return -1;
}