September 2009 - Devel - libvirt List Archives

[libvirt] [PATCH] esx_vi: remove unreachable code
by Jim Meyering 03 Sep '09

03 Sep '09

This removes dead code and suppresses a warning from clang. However, see below: >From 2ed548cbe948aa7ef37ccd121bd084bd05019ac8 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Thu, 3 Sep 2009 16:35:47 +0200 Subject: [PATCH] esx_vi: remove unreachable code * src/esx/esx_vi.c (esxVI_Enumeration_Deserialize): Remove unreachable code. --- src/esx/esx_vi.c | 2 -- 1 files changed, 0 insertions(+), 2 deletions(-) diff --git a/src/esx/esx_vi.c b/src/esx/esx_vi.c index 0da908b..c689a7a 100644 --- a/src/esx/esx_vi.c +++ b/src/esx/esx_vi.c @@ -857,8 +857,6 @@ esxVI_Enumeration_Deserialize(virConnectPtr conn, failure: goto cleanup; - - result = -1; } -------------------------------------------- This function does not allow the caller to distinguish via its return value whether it has failed. It always returns 0. I see that this function's return value is used in numerous places (each esxVI_.*Deserialize function), so this should be fixed. int esxVI_Enumeration_Deserialize(virConnectPtr conn, const esxVI_Enumeration *enumeration, xmlNodePtr node, int *value) { int i; int result = 0; char *name = NULL; if (value == NULL) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "Invalid argument"); goto failure; } *value = 0; /* undefined */ if (esxVI_String_DeserializeValue(conn, node, &name) < 0) { goto failure; } for (i = 0; enumeration->values[i].name != NULL; ++i) { if (STREQ(name, enumeration->values[i].name)) { *value = enumeration->values[i].value; goto cleanup; } } ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "Unknown value '%s' for %s", name, enumeration->type); cleanup: VIR_FREE(name); return result; failure: goto cleanup; }

2 2

[libvirt] [PATCH 0/1] Support huge pages in guests
by Daniel P. Berrange 03 Sep '09

03 Sep '09

This patch is an update of John Cooper's original work from last month http://www.redhat.com/archives/libvir-list/2009-July/msg00753.html Changes since that patch - We create a directory $MOUNTPOINT/libvirt/qemu inside the hugetlbfs mount point. This is neccessary because QEMU now runs as 'qemu:qemu' in Fedora and thus cannot create files in the root. libvirtd will automatically chown() this subdir to allow QEMU guests access - Don't automatically probe for hugetlbfs mount if runing as an unprivileged libvirtd, since user won't have access to the mount unless administrator has manually created them a subdir. - Change the XML to look like <memoryBacking> <hugepages/> </memoryBacking> This is because I fear we may have more config options for memory backing store in the future, such as whether KVM can use KSM for sharing on the guest. - Add a test case for it in the XML -> ARGV convetor - Document the new XML options - Add the new XML options to the RNG schema - Fix setup & auto-detection of hugetlbfs mount if no qemu.conf file exists at all - Add config parameter to the augeas schema - Move the code which finds a mount point into util.c since its not specific to QEMU

4 5

[libvirt] PATCH 0/0: Support setting a disk drive serial (number) v2
by Daniel P. Berrange 03 Sep '09

03 Sep '09

This is a second version of the patch posted here http://www.redhat.com/archives/libvir-list/2009-August/msg00341.html In this version, I make the QEMU driver refuse to run with any serial that doesn't contain 0-9, a-Z, -, _, since there's no way to escape special chars in QEMU command line args. Hopefully we can lift this restriction in a future version of QEMU I also add a check that serial=s is actually supported by QEMU

3 3

[libvirt] [PATCH] mdns.c: remove dead initialization
by Jim Meyering 03 Sep '09

03 Sep '09

The only following uses of group look like this: (always preceded by initialization): AVAHI_DEBUG("Client collision/connecting %p", mdns->client); group = mdns->group; while (group) { if (group->handle) avahi_entry_group_reset(group->handle); group = group->next; } So the value used in the initialization at the top is never used. >From 515e4a4542ce396b8775c618cbe17a30b78299b4 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Thu, 3 Sep 2009 13:45:22 +0200 Subject: [PATCH] mdns.c: remove dead initialization * qemud/mdns.c (libvirtd_mdns_client_callback): Remove dead initialization. --- qemud/mdns.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/qemud/mdns.c b/qemud/mdns.c index ae6ff08..73b7d11 100644 --- a/qemud/mdns.c +++ b/qemud/mdns.c @@ -178,7 +178,7 @@ static void libvirtd_mdns_create_services(struct libvirtd_mdns_group *group) { static void libvirtd_mdns_client_callback(AvahiClient *c, AvahiClientState state, void *userdata) { struct libvirtd_mdns *mdns = (struct libvirtd_mdns *)userdata; - struct libvirtd_mdns_group *group = mdns->group; + struct libvirtd_mdns_group *group; if (!mdns->client) mdns->client = c; -- 1.6.4.2.395.ge3d52

2 1

[libvirt] [PATCH] node_device_conf.c: remove dead initialization
by Jim Meyering 03 Sep '09

03 Sep '09

The following use of "caps" is the assignment in a for-loop, so the initialization is unnecessary. >From 7032a0b370db1a9489cf19fce0cabbb6959320ab Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Thu, 3 Sep 2009 13:38:57 +0200 Subject: [PATCH] node_device_conf.c: remove dead initialization * src/node_device_conf.c (virNodeDeviceDefFormat): Remove dead initialization. --- src/node_device_conf.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/node_device_conf.c b/src/node_device_conf.c index 56a9bb5..5b25bb7 100644 --- a/src/node_device_conf.c +++ b/src/node_device_conf.c @@ -213,7 +213,7 @@ char *virNodeDeviceDefFormat(virConnectPtr conn, const virNodeDeviceDefPtr def) { virBuffer buf = VIR_BUFFER_INITIALIZER; - virNodeDevCapsDefPtr caps = def->caps; + virNodeDevCapsDefPtr caps; char *tmp; virBufferAddLit(&buf, "<device>\n"); -- 1.6.4.2.395.ge3d52

2 1

[libvirt] [PATCH] Fix misc OOM bugs
by Daniel P. Berrange 03 Sep '09

03 Sep '09

* tests/testutils.c: Run test function twice, once to prime it for static allocations, once to count the non-static allocations. * tests/testutilsqemu.c: Initialize variable correctl * src/capabilities.c: Don't free machines variable upon failure since caller must do that * src/xm_internal.c: Add missing check for OOM in building VIF config param --- src/capabilities.c | 17 +++++++++-------- src/xm_internal.c | 3 +++ tests/testutils.c | 12 ++++++++---- tests/testutilsqemu.c | 2 +- 4 files changed, 21 insertions(+), 13 deletions(-) diff --git a/src/capabilities.c b/src/capabilities.c index 193a9fe..289180d 100644 --- a/src/capabilities.c +++ b/src/capabilities.c @@ -354,10 +354,6 @@ virCapabilitiesAddGuest(virCapsPtr caps, if (loader && (guest->arch.defaultInfo.loader = strdup(loader)) == NULL) goto no_memory; - if (nmachines) { - guest->arch.defaultInfo.nmachines = nmachines; - guest->arch.defaultInfo.machines = machines; - } if (VIR_REALLOC_N(caps->guests, caps->nguests + 1) < 0) @@ -365,6 +361,11 @@ virCapabilitiesAddGuest(virCapsPtr caps, caps->guests[caps->nguests] = guest; caps->nguests++; + if (nmachines) { + guest->arch.defaultInfo.nmachines = nmachines; + guest->arch.defaultInfo.machines = machines; + } + return guest; no_memory: @@ -407,10 +408,6 @@ virCapabilitiesAddGuestDomain(virCapsGuestPtr guest, if (loader && (dom->info.loader = strdup(loader)) == NULL) goto no_memory; - if (nmachines) { - dom->info.nmachines = nmachines; - dom->info.machines = machines; - } if (VIR_REALLOC_N(guest->arch.domains, guest->arch.ndomains + 1) < 0) @@ -418,6 +415,10 @@ virCapabilitiesAddGuestDomain(virCapsGuestPtr guest, guest->arch.domains[guest->arch.ndomains] = dom; guest->arch.ndomains++; + if (nmachines) { + dom->info.nmachines = nmachines; + dom->info.machines = machines; + } return dom; diff --git a/src/xm_internal.c b/src/xm_internal.c index 71b852e..66f1e1c 100644 --- a/src/xm_internal.c +++ b/src/xm_internal.c @@ -2048,6 +2048,9 @@ static int xenXMDomainConfigFormatNet(virConnectPtr conn, virBufferVSprintf(&buf, ",vifname=%s", net->ifname); + if (virBufferError(&buf)) + goto cleanup; + if (VIR_ALLOC(val) < 0) { virReportOOMError(conn); goto cleanup; diff --git a/tests/testutils.c b/tests/testutils.c index 7a1dbdc..5072fec 100644 --- a/tests/testutils.c +++ b/tests/testutils.c @@ -367,10 +367,7 @@ int virtTestMain(int argc, } } - if (testOOM) - virAllocTestInit(); - - /* Run once to count allocs, and ensure it passes :-) */ + /* Run once to prime any static allocations & ensure it passes */ ret = (func)(argc, argv); if (ret != EXIT_SUCCESS) goto cleanup; @@ -385,6 +382,13 @@ int virtTestMain(int argc, testOOM++; virSetErrorFunc(NULL, virtTestErrorFuncQuiet); + virAllocTestInit(); + + /* Run again to count allocs, and ensure it passes :-) */ + ret = (func)(argc, argv); + if (ret != EXIT_SUCCESS) + goto cleanup; + approxAlloc = virAllocTestCount(); testCounter++; if (testDebug) diff --git a/tests/testutilsqemu.c b/tests/testutilsqemu.c index 58707e1..6b6a185 100644 --- a/tests/testutilsqemu.c +++ b/tests/testutilsqemu.c @@ -9,7 +9,7 @@ virCapsPtr testQemuCapsInit(void) { struct utsname utsname; virCapsPtr caps; virCapsGuestPtr guest; - virCapsGuestMachinePtr *machines; + virCapsGuestMachinePtr *machines = NULL; int nmachines; static const char *const x86_machines[] = { "pc", "isapc" -- 1.6.2.5

2 2

[libvirt] [PATCH] remote_internal.c: don't dereference a NULL "conn"
by Jim Meyering 03 Sep '09

03 Sep '09

Here's the test just before the else-if in the patch below: if (conn && conn->driver && STREQ (conn->driver->name, "remote")) { So, in the else-branch, "conn" is guaranteed to be NULL. And dereferenced. This may be only a theoretical risk, but if so, the test of "conn" above should be changed to an assertion, and/or the parameter should get the nonnull attribute. >From a1b1d36d96f6b50ddf514539af85da20ca671bf5 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Wed, 2 Sep 2009 11:54:38 +0200 Subject: [PATCH] remote_internal.c: don't dereference a NULL "conn" * src/remote_internal.c (remoteDevMonOpen): Avoid NULL-dereference. --- src/remote_internal.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/remote_internal.c b/src/remote_internal.c index ea50c11..141fef9 100644 --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -5148,7 +5148,7 @@ remoteDevMonOpen(virConnectPtr conn, conn->devMonPrivateData = priv; remoteDriverUnlock(priv); return VIR_DRV_OPEN_SUCCESS; - } else if (conn->networkDriver && + } else if (conn && conn->networkDriver && STREQ (conn->networkDriver->name, "remote")) { struct private_data *priv = conn->networkPrivateData; remoteDriverLock(priv); -- 1.6.4.2.395.ge3d52

5 9

[libvirt] [PATCH] test.c: don't use undefined local, "def"
by Jim Meyering 03 Sep '09

03 Sep '09

another real bug: >From fd41bd47252654f84684830e02ecbeb3d26e7dee Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Thu, 3 Sep 2009 12:12:19 +0200 Subject: [PATCH] test.c: don't use undefined local, "def" * src/test.c (testOpenVolumesForPool): Upon early virAsprintf or virXPathNodeSet failure, "goto error" would take us to virStorageVolDefFree(def), but with "def" not defined. Initialize it to NULL. --- src/test.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/test.c b/src/test.c index 7c8f85b..895cce1 100644 --- a/src/test.c +++ b/src/test.c @@ -435,7 +435,7 @@ static int testOpenVolumesForPool(virConnectPtr conn, char *vol_xpath; int i, ret, func_ret = -1; xmlNodePtr *vols = NULL; - virStorageVolDefPtr def; + virStorageVolDefPtr def = NULL; /* Find storage volumes */ if (virAsprintf(&vol_xpath, "/node/pool[%d]/volume", poolidx) < 0) { -- 1.6.4.2.395.ge3d52

2 1

[libvirt] Support for XCI (Xen Client)
by Andreas Sommer 03 Sep '09

03 Sep '09

Hi, I just wondered whether the Xen Client project (http://www.xen.org/products/xci.html) has been tested, or is supported by libvirt at all? I'm not sure if it has a different Xen API, but it definitely has different domain configuration files - that's why I ask. Best regards

2 1

[libvirt] [PATCH] Support for defining/dumping video device in VirtualBox
by Pritesh Kothari 03 Sep '09

03 Sep '09

Hi All, I have added support for defining/dumping video device in VirtualBox. The patch for the same is attached here. Also this patch depends on the earlier patch "Support for 3d Acceleration in video tag". Regards, Pritesh

3 3