5:44 a.m.
Moreover, since virAsprintf now does report OOM error, there's no need
to call virReportOOMError in error path.
---
src/qemu/qemu_hotplug.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
index 788ad47..a98de9d 100644
--- a/src/qemu/qemu_hotplug.c
+++ b/src/qemu/qemu_hotplug.c
@@ -3155,10 +3155,8 @@ int qemuDomainDetachChrDevice(virQEMUDriverPtr driver,
if (qemuBuildChrDeviceStr(&devstr, vm->def, chr, priv->qemuCaps) < 0)
return ret;
- if (virAsprintf(&charAlias, "char%s", tmpChr->info.alias) < 0) {
- virReportOOMError();
- return ret;
- }
+ if (virAsprintf(&charAlias, "char%s", tmpChr->info.alias) < 0)
+ goto cleanup;
qemuDomainObjEnterMonitor(driver, vm);
if (devstr && qemuMonitorDelDevice(priv->mon, tmpChr->info.alias) <
0) {
--
1.8.1.5
5:44 a.m.
New subject: [libvirt] [PATCH 2/2] qemuhotplugtest: Resolve some memleaks
If testQemuHotplugAttach succeeds, the vm->def steals the dev pointer.
However, not the envelope, which needs to be freed. In addition,
driver.config is allocated, but never freed.
---
tests/qemuhotplugtest.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/tests/qemuhotplugtest.c b/tests/qemuhotplugtest.c
index d4971c2..5e8c5f0 100644
--- a/tests/qemuhotplugtest.c
+++ b/tests/qemuhotplugtest.c
@@ -89,6 +89,10 @@ testQemuHotplugAttach(virDomainObjPtr vm,
switch (dev->type) {
case VIR_DOMAIN_DEVICE_CHR:
ret = qemuDomainAttachChrDevice(&driver, vm, dev->data.chr);
+ if (!ret) {
+ /* vm->def stolen dev->data.chr so we ought to avoid freeing it */
+ dev->data.chr = NULL;
+ }
break;
default:
if (virTestGetVerbose())
@@ -214,11 +218,6 @@ testQemuHotplug(const void *data)
switch (test->action) {
case ATTACH:
ret = testQemuHotplugAttach(vm, dev);
- if (!ret) {
- /* avoid @dev double free on success,
- * as @dev is part of vm->def now */
- dev = NULL;
- }
break;
case DETACH:
@@ -323,6 +322,7 @@ mymain(void)
virObjectUnref(driver.caps);
virObjectUnref(driver.xmlopt);
+ virObjectUnref(driver.config);
return (ret == 0) ? EXIT_SUCCESS : EXIT_FAILURE;
}
--
1.8.1.5
7:05 a.m.
New subject: [libvirt] [PATCH 2/2] qemuhotplugtest: Resolve some memleaks
On 07/18/13 12:44, Michal Privoznik wrote:
If testQemuHotplugAttach succeeds, the vm->def steals the dev
pointer.
However, not the envelope, which needs to be freed. In addition,
driver.config is allocated, but never freed.
---
tests/qemuhotplugtest.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/tests/qemuhotplugtest.c b/tests/qemuhotplugtest.c
index d4971c2..5e8c5f0 100644
--- a/tests/qemuhotplugtest.c
+++ b/tests/qemuhotplugtest.c
@@ -89,6 +89,10 @@ testQemuHotplugAttach(virDomainObjPtr vm,
switch (dev->type) {
case VIR_DOMAIN_DEVICE_CHR:
ret = qemuDomainAttachChrDevice(&driver, vm, dev->data.chr);
+ if (!ret) {
For integer comparisons I prefer "if (ret == 0)", this looks like
returning a pointer at first glance.
+ /* vm->def stolen dev->data.chr so we ought to
avoid freeing it */
+ dev->data.chr = NULL;
+ }
break;
default:
if (virTestGetVerbose())
ACK anyways.
Peter
7:09 a.m.
s/@charAlias/@devstr/
On 07/18/13 12:44, Michal Privoznik wrote:
Moreover, since virAsprintf now does report OOM error, there's no
need
to call virReportOOMError in error path.
---
src/qemu/qemu_hotplug.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
index 788ad47..a98de9d 100644
--- a/src/qemu/qemu_hotplug.c
+++ b/src/qemu/qemu_hotplug.c
@@ -3155,10 +3155,8 @@ int qemuDomainDetachChrDevice(virQEMUDriverPtr driver,
if (qemuBuildChrDeviceStr(&devstr, vm->def, chr, priv->qemuCaps) < 0)
return ret;
- if (virAsprintf(&charAlias, "char%s", tmpChr->info.alias) < 0)
{
- virReportOOMError();
- return ret;
- }
+ if (virAsprintf(&charAlias, "char%s", tmpChr->info.alias) < 0)
+ goto cleanup;
This actually would leak devstr from the function called above.
qemuDomainObjEnterMonitor(driver, vm);
if (devstr && qemuMonitorDelDevice(priv->mon, tmpChr->info.alias)
< 0) {
ACK when you adjust the commit message
Peter
4146
days inactive
4146
days old
3 comments
2 participants
participants (2)
-
Michal Privoznik -
Peter Krempa