From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:40813) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YLiDJ-0006uV-SI for qemu-devel@nongnu.org; Wed, 11 Feb 2015 20:05:09 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YLiDF-0007vw-Mz for qemu-devel@nongnu.org; Wed, 11 Feb 2015 20:05:05 -0500 Message-ID: <54DBFC1D.6020901@huawei.com> Date: Thu, 12 Feb 2015 09:04:29 +0800 From: Gonglei MIME-Version: 1.0 References: <1423635971-9928-1-git-send-email-arei.gonglei@huawei.com> <87386c8uy4.fsf@blackfin.pond.sub.org> In-Reply-To: <87386c8uy4.fsf@blackfin.pond.sub.org> Content-Type: text/plain; charset="windows-1252" Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH] monitor: Fix Resource leak List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Markus Armbruster Cc: qemu-trivial@nongnu.org, qemu-devel@nongnu.org, lcapitulino@redhat.com On 2015/2/11 21:00, Markus Armbruster wrote: > writes: > >> From: Gonglei >> >> Coverity spot: >> qemu_using_spice allocates memory that is stored into err, >> but not freed before return. >> >> Cc:Markus Armbruster >> Signed-off-by: Gonglei >> --- >> monitor.c | 3 ++- >> 1 file changed, 2 insertions(+), 1 deletion(-) >> >> diff --git a/monitor.c b/monitor.c >> index c3cc060..137d23f 100644 >> --- a/monitor.c >> +++ b/monitor.c >> @@ -1095,12 +1095,13 @@ static int client_migrate_info(Monitor *mon, const QDict *qdict, >> const char *subject = qdict_get_try_str(qdict, "cert-subject"); >> int port = qdict_get_try_int(qdict, "port", -1); >> int tls_port = qdict_get_try_int(qdict, "tls-port", -1); >> - Error *err; >> + Error *err = NULL; >> int ret; >> >> if (strcmp(protocol, "spice") == 0) { >> if (!qemu_using_spice(&err)) { >> qerror_report_err(err); >> + error_free(err); >> return -1; >> } > > Your commit message is incomplete. The resource leak is real, but it's > the less serious bug here. The serious one is missing initialization of > err. > > If using_spice, qemu_using_spice() returns true without touching err. > All fine. > > Else, if err is null by chance, qemu_using_spice()'s error_set() creates > an error object and stores it in err. We leak it. > > If err is not null, error_set() fails its assertion. > Thanks for your explanation, I haven't a global inspection TBH. > Broken in commit b25d81b by yours truly. Thanks for cleaning up my > mess! > > Let's fix up the commit message: > > monitor: Fix missing err = NULL in client_migrate_info() > > When SPICE isn't used, we either fail an assertion in error_set(), > or leak an error object. Broken in commit b25d81b. > OK, will fix. > With such a fixup: > > Reviewed-by: Markus Armbruster > Thanks. Regards, -Gonglei