Re: [libvirt] [PATCH] virsh: check actual access to the cert
Subject is wrong - this is remote, not virsh (that is, more than just
virsh is impacted by this change).
On 05/06/2011 10:24 AM, Doug Goldstein wrote:
Instead of calling stat(), check that we'll actually be able to
access
and read the file.
Signed-off-by: Doug Goldstein <cardoe(a)gentoo.org>
---
src/remote/remote_driver.c | 3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/src/remote/remote_driver.c b/src/remote/remote_driver.c
index 9965d38..c4e43aa 100644
--- a/src/remote/remote_driver.c
+++ b/src/remote/remote_driver.c
@@ -1140,8 +1140,7 @@ static gnutls_certificate_credentials_t x509_cred;
static int
check_cert_file(const char *type, const char *file)
{
- struct stat sb;
- if (stat(file, &sb) < 0) {
+ if (access(file, F_OK|R_OK)) {
F_OK|R_OK is forbidden by POSIX (true, F_OK is usually 0, in which case
it happens to work, but POSIX allows an implementation where F_OK is
non-zero, and states that you either use F_OK or a combination of the
other three [RWX]_OK). Here, you only need R_OK (since if you can read
the file, it must exist).
access() is in general not the best function to use from a library, when
compared to faccessat(,AT_EACCESS) [aka eaccess or euidaccess], because
the former doesn't work well in the face of effective uid different from
real uid while the latter does (normally, you want the answer for the
effective uid, not the real uid, if your library was linked into an
application that plays games with ids). But none of the rest of libvirt
makes that distinction, so you're in good company.
So ACK with those nits fixed, and pushed.
--
Eric Blake eblake(a)redhat.com +1-801-349-2682
Libvirt virtualization library http://libvirt.org
Attachments:
- signature.asc (application/pgp-signature — 619 bytes)