On 4/2/24 00:14, Jim Fehlig wrote: > When performing an install, it's common for tooling such as virt-install > to remove the install kernel/initrd once they are successfully booted and > the domain has been redefined to boot without them. After the installation > is complete and the domain is rebooted/shutdown, the DAC and selinux > security drivers attempt to restore labels on the now deleted files. It's > harmles wrt functionality, but results in error messages such as > > Mar 08 12:40:37 virtqemud[5639]: internal error: child reported (status=125): unable to stat: /var/lib/libvirt/boot/vir> > Mar 08 12:40:37 virtqemud[5639]: unable to stat: /var/lib/libvirt/boot/virtinst-yvp19moo-linux: No such file or directo> > Mar 08 12:40:37 virtqemud[5639]: Unable to run security manager transaction > > Add a check for file existence to the virSecurity*RestoreFileLabel functions, > and avoid relabeling if the file is no longer available. Skipping the restore > caused failures in qemusecuritytest, which mocks stat, chown, etc as part of > ensuring the security drivers properly restore labels. virFileExists is now > mocked in qemusecuritymock.c to return true when passed a file previously > seen by the mocked stat, chown, etc functions. > > Signed-off-by: Jim Fehlig <jfehlig(a)suse.com&gt; > --- > src/security/security_dac.c | 3 +++ > src/security/security_selinux.c | 2 ++ > tests/qemusecuritymock.c | 18 ++++++++++++++++++ > 3 files changed, 23 insertions(+) > Looks better now, thanks! > diff --git a/src/security/security_dac.c b/src/security/security_dac.c > index 567be4bd23..4e850e219e 100644 > --- a/src/security/security_dac.c > +++ b/src/security/security_dac.c > @@ -825,6 +825,9 @@ virSecurityDACRestoreFileLabelInternal(virSecurityManager *mgr, > virStorageSourceIsLocalStorage(src)) > path = src->path; > > + if (!virFileExists(path)) > + return 0; > + > /* Be aware that this function might run in a separate process. > * Therefore, any driver state changes would be thrown away. */ > > diff --git a/src/security/security_selinux.c b/src/security/security_selinux.c > index b49af26e49..aaec34ff8b 100644 > --- a/src/security/security_selinux.c > +++ b/src/security/security_selinux.c > @@ -1488,6 +1488,8 @@ virSecuritySELinuxRestoreFileLabel(virSecurityManager *mgr, > */ > if (!path) > return 0; > + if (!virFileExists(path)) > + return 0; > > VIR_INFO("Restoring SELinux context on '%s'", path); > > diff --git a/tests/qemusecuritymock.c b/tests/qemusecuritymock.c > index 80d59536b1..bb0a85544b 100644 > --- a/tests/qemusecuritymock.c > +++ b/tests/qemusecuritymock.c > @@ -382,6 +382,24 @@ int virFileUnlock(int fd G_GNUC_UNUSED, > } > > > +bool virFileExists(const char *path G_GNUC_UNUSED) This argument is used, so please drop the unused annotation. > +{ > + VIR_LOCK_GUARD lock = virLockGuardLock(&m); > + > + if (getenv(ENVVAR) == NULL) > + return access(path, F_OK) == 0; I wonder whether we should call real virFileExists() here. That way, if a test would chain mocks they can work (though both would probably be mocking access()). I'll leave it up to you whether you want to squash in the following:

On 4/8/24 10:48 AM, Jim Fehlig wrote: > On 4/8/24 6:18 AM, Michal Prívozník wrote: >> On 4/2/24 00:14, Jim Fehlig wrote: >>> When performing an install, it's common for tooling such as >>> virt-install >>> to remove the install kernel/initrd once they are successfully >>> booted and >>> the domain has been redefined to boot without them. After the >>> installation >>> is complete and the domain is rebooted/shutdown, the DAC and selinux >>> security drivers attempt to restore labels on the now deleted files. >>> It's >>> harmles wrt functionality, but results in error messages such as >>> >>> Mar 08 12:40:37 virtqemud[5639]: internal error: child reported >>> (status=125): unable to stat: /var/lib/libvirt/boot/vir> >>> Mar 08 12:40:37 virtqemud[5639]: unable to stat: >>> /var/lib/libvirt/boot/virtinst-yvp19moo-linux: No such file or directo> >>> Mar 08 12:40:37 virtqemud[5639]: Unable to run security manager >>> transaction >>> >>> Add a check for file existence to the virSecurity*RestoreFileLabel >>> functions, >>> and avoid relabeling if the file is no longer available. Skipping >>> the restore >>> caused failures in qemusecuritytest, which mocks stat, chown, etc as >>> part of >>> ensuring the security drivers properly restore labels. virFileExists >>> is now >>> mocked in qemusecuritymock.c to return true when passed a file >>> previously >>> seen by the mocked stat, chown, etc functions. >>> >>> Signed-off-by: Jim Fehlig <jfehlig(a)suse.com&gt; >>> --- >>>   src/security/security_dac.c     |  3 +++ >>>   src/security/security_selinux.c |  2 ++ >>>   tests/qemusecuritymock.c        | 18 ++++++++++++++++++ >>>   3 files changed, 23 insertions(+) >>> >> >> >> Looks better now, thanks! >> >>> diff --git a/src/security/security_dac.c b/src/security/security_dac.c >>> index 567be4bd23..4e850e219e 100644 >>> --- a/src/security/security_dac.c >>> +++ b/src/security/security_dac.c >>> @@ -825,6 +825,9 @@ >>> virSecurityDACRestoreFileLabelInternal(virSecurityManager *mgr, >>>           virStorageSourceIsLocalStorage(src)) >>>           path = src->path; >>> +    if (!virFileExists(path)) >>> +        return 0; >>> + >>>       /* Be aware that this function might run in a separate process. >>>        * Therefore, any driver state changes would be thrown away. */ >>> diff --git a/src/security/security_selinux.c >>> b/src/security/security_selinux.c >>> index b49af26e49..aaec34ff8b 100644 >>> --- a/src/security/security_selinux.c >>> +++ b/src/security/security_selinux.c >>> @@ -1488,6 +1488,8 @@ >>> virSecuritySELinuxRestoreFileLabel(virSecurityManager *mgr, >>>        */ >>>       if (!path) >>>           return 0; >>> +    if (!virFileExists(path)) >>> +        return 0; >>>       VIR_INFO("Restoring SELinux context on '%s'", path); >>> diff --git a/tests/qemusecuritymock.c b/tests/qemusecuritymock.c >>> index 80d59536b1..bb0a85544b 100644 >>> --- a/tests/qemusecuritymock.c >>> +++ b/tests/qemusecuritymock.c >>> @@ -382,6 +382,24 @@ int virFileUnlock(int fd G_GNUC_UNUSED, >>>   } >>> +bool virFileExists(const char *path G_GNUC_UNUSED) >> >> This argument is used, so please drop the unused annotation. >> >>> +{ >>> +    VIR_LOCK_GUARD lock = virLockGuardLock(&m); >>> + >>> +    if (getenv(ENVVAR) == NULL) >>> +        return access(path, F_OK) == 0; >> >> I wonder whether we should call real virFileExists() here. That way, if >> a test would chain mocks they can work (though both would probably be >> mocking access()). I'll leave it up to you whether you want to squash in >> the following: > > I considered calling the real virFileExists, but was being lazy since > the impl is a one-liner :-). Agree it's better, for consistency if > nothing else. I've squashed in your suggestion and pushed the patch. Weird. Calling the real virFileExists results in a segfault in some of the CI jobs, e.g. https://gitlab.com/libvirt/libvirt/-/jobs/6574951832

I wasn't able to quickly reproduce it on my development setup and don't have time ATM to investigate further. Calling access(2) instead of the real virFileExists, as the original patch did, avoids the CI failures