Re: [libvirt] [PATCH 05/42] rpc: Don't check the output of virGetUserDirectory()

On Thu, Dec 19, 2019 at 05:25:59PM +0100, Pavel Hrdina wrote: > On Thu, Dec 19, 2019 at 05:21:21PM +0100, Fabiano Fidêncio wrote: > > On Thu, Dec 19, 2019 at 5:07 PM Pavel Hrdina <phrdina(a)redhat.com&gt; wrote: > > > > > > On Thu, Dec 19, 2019 at 11:04:10AM +0100, Fabiano Fidêncio wrote: > > > > virGetUserDirectory() *never* *ever* returns NULL, making the checks for > > > > it completely unnecessary. > > > > > > > > Signed-off-by: Fabiano Fidêncio <fidencio(a)redhat.com&gt; > > > > --- > > > > src/rpc/virnetclient.c | 12 ++++-------- > > > > src/rpc/virnettlscontext.c | 12 ------------ > > > > 2 files changed, 4 insertions(+), 20 deletions(-) > > > > > > [...] > > > > > > > diff --git a/src/rpc/virnettlscontext.c b/src/rpc/virnettlscontext.c > > > > index ec9dd35c46..08944f6771 100644 > > > > --- a/src/rpc/virnettlscontext.c > > > > +++ b/src/rpc/virnettlscontext.c > > > > @@ -805,9 +805,6 @@ static int virNetTLSContextLocateCredentials(const char *pkipath, > > > > */ > > > > userdir = virGetUserDirectory(); > > > > > > > > - if (!userdir) > > > > - goto error; > > > > - > > > > user_pki_path = g_strdup_printf("%s/.pki/libvirt", userdir); > > > > > > > > VIR_DEBUG("Trying to find TLS user credentials in %s", user_pki_path); > > > > @@ -864,15 +861,6 @@ static int virNetTLSContextLocateCredentials(const char *pkipath, > > > > VIR_FREE(userdir); > > > > > > > > return 0; > > > > - > > > > - error: > > > > - VIR_FREE(*cacert); > > > > - VIR_FREE(*cacrl); > > > > - VIR_FREE(*key); > > > > - VIR_FREE(*cert); > > > > - VIR_FREE(user_pki_path); > > > > - VIR_FREE(userdir); > > > > - return -1; > > > > > > This doesn't look right. Looks like some leftover from rebase where > > > you wanted to use g_autofree. > > > > No, actually not. The only usage of `goto error` was when checking the > > result of virGetUserDirectory(). > > By removing the check, we have also to remove the label. > > Label yes, the VIR_FREE no as it would leak the memory. > There's "return 0;" right before that label.