On 11/10/21 00:46, Michal Prívozník wrote:

On 11/10/21 12:34 AM, Jim Fehlig wrote:

...

Found via valgrind

==15016== 3,701 bytes in 2 blocks are definitely lost in loss record 975 of 1,009 ==15016== at 0x4C2A2AF: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==15016== by 0x1FCD30CB: libxl_read_file_contents (in /usr/lib64/libxenlight.so.4.12.0) ==15016== by 0x1FCCA58A: ??? (in /usr/lib64/libxenlight.so.4.12.0) ==15016== by 0x1FCCA6C2: libxl_userdata_retrieve (in /usr/lib64/libxenlight.so.4.12.0) ==15016== by 0x1FA42A5A: libxlReconnectDomain (libxl_driver.c:394) ==15016== by 0x53BAC99: virDomainObjListHelper (virdomainobjlist.c:802) ==15016== by 0x530842F: virHashForEach (virhash.c:575) ==15016== by 0x53BC0E0: virDomainObjListForEach (virdomainobjlist.c:817) ==15016== by 0x1FA423C4: libxlReconnectDomains (libxl_driver.c:468) ==15016== by 0x1FA423C4: libxlStateInitialize (libxl_driver.c:778) ==15016== by 0x54E8E9E: virStateInitialize (libvirt.c:657) ==15016== by 0x12DBFA: daemonRunStateInit (remote_daemon.c:797) ==15016== by 0x535BF79: virThreadHelper (virthread.c:206)

Signed-off-by: Jim Fehlig <jfehlig@suse.com> --- src/libxl/libxl_driver.c | 1 + 1 file changed, 1 insertion(+)

diff --git a/src/libxl/libxl_driver.c b/src/libxl/libxl_driver.c index 7ea157f9c4..1a9bf8f186 100644 --- a/src/libxl/libxl_driver.c +++ b/src/libxl/libxl_driver.c @@ -461,6 +461,7 @@ libxlReconnectDomain(virDomainObj *vm,

cleanup: libxl_dominfo_dispose(&d_info); + VIR_FREE(data); virObjectUnlock(vm); virObjectUnref(vm); virObjectUnref(cfg);

Alternatively, data can be declared as g_autofree:

g_autofree uint8_t *data = NULL;

Yes, that is better. I must try harder to break old habits :-).

Regardless of what to decide to go with:

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>

Thanks! Jim