- Devel - libvirt List Archives

[libvirt] [PATCH] python: Fix Create*WithFiles filefd passing
by Martin Kletzander 22 Oct '13

22 Oct '13

From: Marian Neagul <marian(a)info.uvt.ro> Commit d76227be added functions virDomainCreateWithFiles and virDomainCreateXMLWithFiles, but there was a little piece missing in python bindings. This patch fixes proper passing of file descriptors in the overwrites of these functions. Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com> --- Notes: I checked that Marian's patch works as expected and that it really fixes the issue. python/libvirt-override.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/python/libvirt-override.c b/python/libvirt-override.c index 4800d1d..2e58bf9 100644 --- a/python/libvirt-override.c +++ b/python/libvirt-override.c @@ -7157,6 +7157,8 @@ libvirt_virDomainCreateWithFiles(PyObject *self ATTRIBUTE_UNUSED, PyObject *args if (libvirt_intUnwrap(pyfd, &fd) < 0) goto cleanup; + + files[i] = fd; } LIBVIRT_BEGIN_ALLOW_THREADS; @@ -7201,6 +7203,8 @@ libvirt_virDomainCreateXMLWithFiles(PyObject *self ATTRIBUTE_UNUSED, PyObject *a if (libvirt_intUnwrap(pyfd, &fd) < 0) goto cleanup; + + files[i] = fd; } LIBVIRT_BEGIN_ALLOW_THREADS; -- 1.8.4

3 3

[libvirt] [PATCH] maint: typo fixes
by Eric Blake 22 Oct '13

22 Oct '13

* ChangeLog-old: s/guarentee/guarantee/ Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing as trivial. ChangeLog-old | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ChangeLog-old b/ChangeLog-old index e07b11c..ba738c4 100644 --- a/ChangeLog-old +++ b/ChangeLog-old @@ -13920,7 +13920,7 @@ Tue Feb 27 10:20:43 EST 2007 Daniel P. Berrange <berrange(a)redhat.com> * src/xend_internal.c: Only hardcode port = 5900+domid if running against old XenD < 3.0.3, because in newer XenD - port is guarenteed to be available in XenStore if the VNC + port is guaranteed to be available in XenStore if the VNC server is running. Mon Feb 26 15:33:08 IST 2007 Mark McLoughlin <markmc(a)redhat.com> @@ -15228,7 +15228,7 @@ Sun Sep 3 12:34:23 EDT 2006 Daniel Berrange <berrange(a)redhat.com> iterating over list of ids/names, because it is not neccessarily the same as the value returned by virConnectNumOfDomains. Use qsort to sort active domains by Id, and inactive domains by name, since - there is no guarenteed sort ordering when listing domains. For inactive + there is no guaranteed sort ordering when listing domains. For inactive domains display a '-' instead of '-1' to make it clear they have no sensible ID number. -- 1.8.3.1

1 0

[libvirt] [PATCH] rpc: Retrieve peer PID via new getsockopt() for Mac
by Doug Goldstein 22 Oct '13

22 Oct '13

While LOCAL_PEERCRED on the BSDs does not return the pid information of the peer, Mac OS X 10.8 added LOCAL_PEERPID to retrieve the pid so we should use that when its available to get that information. --- v3: * Consider EOPNOTSUPP as non-fatal in case they've built on 10.8 but run the binary on 10.6 as that's the error you'll get back in that case. v2: * Make LOCAL_PEERPID call non-fatal in case the user built the binary on a system that supports it but then runs it on a kernel that does not support it --- src/rpc/virnetsocket.c | 26 +++++++++++++++++++++++++- 1 file changed, 25 insertions(+), 1 deletion(-) diff --git a/src/rpc/virnetsocket.c b/src/rpc/virnetsocket.c index b2ebefe..f73bc50 100644 --- a/src/rpc/virnetsocket.c +++ b/src/rpc/virnetsocket.c @@ -1197,12 +1197,36 @@ int virNetSocketGetUNIXIdentity(virNetSocketPtr sock, goto cleanup; } - /* PID and process creation time are not supported on BSDs */ + /* PID and process creation time are not supported on BSDs by + * LOCAL_PEERCRED. + */ *pid = -1; *timestamp = -1; *uid = cr.cr_uid; *gid = cr.cr_gid; +# ifdef LOCAL_PEERPID + /* Exists on Mac OS X 10.8 for retrieving the peer's PID */ + cr_len = sizeof(*pid); + + if (getsockopt(sock->fd, VIR_SOL_PEERCRED, LOCAL_PEERPID, pid, &cr_len) < 0) { + /* Ensure this is set to something sane as there are no guarentees + * as to what its set to now. + */ + *pid = -1; + + /* If this was built on a system with LOCAL_PEERPID defined but + * the kernel doesn't support it we'll get back EOPNOTSUPP so + * treat all errors but EOPNOTSUPP as fatal + */ + if (errno != EOPNOTSUPP) { + virReportSystemError(errno, "%s", + _("Failed to get client socket PID")); + goto cleanup; + } + } +# endif + ret = 0; cleanup: -- 1.8.1.5

3 2

[libvirt] [PATCH] nodeinfo: fix physical memory size on Mac OS X
by Ryota Ozaki 22 Oct '13

22 Oct '13

HW_PHYSMEM is available on Mac OS X as well as FreeBSD, however, its resulting value for Mac OS X is 32 bits. Mac OS X provides HW_MEMSIZE that is 64 bits version of HW_PHYSMEM. We have to use it. I tested the patch on Mac OS X 10.6.8, 10.7.4, 10.8.5 and FreeBSD 9.2. Signed-off-by: Ryota Ozaki <ozaki.ryota(a)gmail.com> --- src/nodeinfo.c | 38 ++++++++++++++++++++++++++++---------- 1 file changed, 28 insertions(+), 10 deletions(-) diff --git a/src/nodeinfo.c b/src/nodeinfo.c index 70814c2..a0fdfe7 100644 --- a/src/nodeinfo.c +++ b/src/nodeinfo.c @@ -73,6 +73,33 @@ appleFreebsdNodeGetCPUCount(void) return ncpu; } + +/* VIR_HW_PHYSMEM - the resulting value of HW_PHYSMEM of FreeBSD + * is 64 bits while that of Mac OS X is still 32 bits. + * Mac OS X provides HW_MEMSIZE for 64 bits version of HW_PHYSMEM + * since 10.6.8 (Snow Leopard) at least. + */ +# ifdef HW_MEMSIZE +# define VIR_HW_PHYSMEM HW_MEMSIZE +# else +# define VIR_HW_PHYSMEM HW_PHYSMEM +# endif +static int +appleFreebsdNodeGetMemorySize(unsigned long *memory) +{ + int mib[2] = { CTL_HW, VIR_HW_PHYSMEM }; + unsigned long physmem; + size_t len = sizeof(physmem); + + if (sysctl(mib, 2, &physmem, &len, NULL, 0) == -1) { + virReportSystemError(errno, "%s", _("cannot obtain memory size")); + return -1; + } + + *memory = (unsigned long)(physmem / 1024); + + return 0; +} #endif #ifdef __linux__ @@ -914,17 +941,8 @@ cleanup: nodeinfo->mhz = cpu_freq / 1000000; # endif - /* get memory information */ - int mib[2] = { CTL_HW, HW_PHYSMEM }; - unsigned long physmem; - size_t len = sizeof(physmem); - - if (sysctl(mib, 2, &physmem, &len, NULL, 0) == -1) { - virReportSystemError(errno, "%s", _("cannot obtain memory size")); + if (appleFreebsdNodeGetMemorySize(&nodeinfo->memory) < 0) return -1; - } - - nodeinfo->memory = (unsigned long)(physmem / 1024); return 0; } -- 1.8.4

2 4

[libvirt] [PATCH] xenapi: fix the coding stype in xenapi_driver.c
by Hongwei Bi 22 Oct '13

22 Oct '13

Signed-off-by: Hongwei Bi <hwbi2008(a)gmail.com> --- src/xenapi/xenapi_driver.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/xenapi/xenapi_driver.c b/src/xenapi/xenapi_driver.c index 4b522c0..82af6f8 100644 --- a/src/xenapi/xenapi_driver.c +++ b/src/xenapi/xenapi_driver.c @@ -437,7 +437,8 @@ xenapiConnectGetCapabilities(virConnectPtr conn) virCapsPtr caps = ((struct _xenapiPrivate *)(conn->privateData))->caps; if (caps) { char *xml = virCapabilitiesFormatXML(caps); - if (!xml) goto cleanup; + if (!xml) + goto cleanup; return xml; } cleanup: -- 1.8.1.2

2 2

[libvirt] [PATCH] virsh: Fix job watching when STDIN is not a tty
by Peter Krempa 22 Oct '13

22 Oct '13

In commit b46c4787dde79b015dad67dedda4ccf6ff1a3082 I changed the code to watch long running jobs in virsh. Unfortunately I didn't take into account that poll may get a hangup if the terminal is not a TTY and will be closed. This patch avoids polling the STDIN fd when there's no TTY. --- tools/virsh-domain.c | 7 ++++++- tools/virsh.c | 7 +++++++ tools/virsh.h | 2 ++ 3 files changed, 15 insertions(+), 1 deletion(-) diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c index b75f331..5aabccd 100644 --- a/tools/virsh-domain.c +++ b/tools/virsh-domain.c @@ -3529,6 +3529,7 @@ vshWatchJob(vshControl *ctl, bool functionReturn = false; sigset_t sigmask, oldsigmask; bool jobStarted = false; + nfds_t npollfd = 2; sigemptyset(&sigmask); sigaddset(&sigmask, SIGINT); @@ -3539,9 +3540,13 @@ vshWatchJob(vshControl *ctl, sigemptyset(&sig_action.sa_mask); sigaction(SIGINT, &sig_action, &old_sig_action); + /* don't poll on STDIN if we are not using a terminal */ + if (!vshTTYAvailable(ctl)) + npollfd = 1; + GETTIMEOFDAY(&start); while (1) { - ret = poll((struct pollfd *)&pollfd, 2, 500); + ret = poll((struct pollfd *)&pollfd, npollfd, 500); if (ret > 0) { if (pollfd[1].revents & POLLIN && saferead(STDIN_FILENO, &retchar, sizeof(retchar)) > 0) { diff --git a/tools/virsh.c b/tools/virsh.c index a76229a..8425f53 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -2226,6 +2226,13 @@ vshTTYIsInterruptCharacter(vshControl *ctl ATTRIBUTE_UNUSED, } +bool +vshTTYAvailable(vshControl *ctl) +{ + return ctl->istty; +} + + int vshTTYDisableInterrupt(vshControl *ctl ATTRIBUTE_UNUSED) { diff --git a/tools/virsh.h b/tools/virsh.h index f978d94..b843788 100644 --- a/tools/virsh.h +++ b/tools/virsh.h @@ -365,6 +365,8 @@ bool vshTTYIsInterruptCharacter(vshControl *ctl, const char chr); int vshTTYDisableInterrupt(vshControl *ctl); int vshTTYRestore(vshControl *ctl); int vshTTYMakeRaw(vshControl *ctl, bool report_errors); +bool vshTTYAvailable(vshControl *ctl); + /* allocation wrappers */ void *_vshMalloc(vshControl *ctl, size_t sz, const char *filename, int line); -- 1.8.3.2

2 2

[libvirt] [PATCH] Also undef WITH_GNUTLS_GCRYPT
by Guido Günther 22 Oct '13

22 Oct '13

If we disallow gnutls we need to disallow gcrypt to since WITH_GNUTLS_GCRYPT only works with WITH_GNUTLS per configure.ac. Othwise we fail to build .libs/libvirt_setuid_rpc_client_la-libvirt.o like: libvirt.c:314:15: error: variable 'virTLSThreadImpl' has initializer but incomplete type static struct gcry_thread_cbs virTLSThreadImpl = { ^ libvirt.c:319:5: error: 'GCRY_THREAD_OPTION_PTHREAD' undeclared here (not in a function) GCRY_THREAD_OPTION_PTHREAD, ^ ...and so on. See: http://honk.sigxcpu.org:8001/job/libvirt-build/1769/console --- config-post.h | 1 + 1 file changed, 1 insertion(+) diff --git a/config-post.h b/config-post.h index d371e8c..8367200 100644 --- a/config-post.h +++ b/config-post.h @@ -34,6 +34,7 @@ # undef WITH_CURL # undef WITH_DTRACE_PROBES # undef WITH_GNUTLS +# undef WITH_GNUTLS_GCRYPT # undef WITH_MACVTAP # undef WITH_NUMACTL # undef WITH_SASL -- 1.8.4.rc3

2 1

[libvirt] [PATCH] Get rid of shadowed booleans
by Michal Privoznik 22 Oct '13

22 Oct '13

There are still two places where we are using 1bit width unsigned integer to store a boolean. There's no real need for this and these occurrences can be replaced with 'bool'. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/conf/interface_conf.h | 4 ++-- src/rpc/virnetserver.c | 8 ++++---- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/src/conf/interface_conf.h b/src/conf/interface_conf.h index 4857822..0e22575 100644 --- a/src/conf/interface_conf.h +++ b/src/conf/interface_conf.h @@ -164,7 +164,7 @@ typedef virInterfaceObj *virInterfaceObjPtr; struct _virInterfaceObj { virMutex lock; - unsigned int active:1; /* 1 if interface is active (up) */ + bool active; /* true if interface is active (up) */ virInterfaceDefPtr def; /* The interface definition */ }; @@ -175,7 +175,7 @@ struct _virInterfaceObjList { virInterfaceObjPtr *objs; }; -static inline int +static inline bool virInterfaceObjIsActive(const virInterfaceObj *iface) { return iface->active; diff --git a/src/rpc/virnetserver.c b/src/rpc/virnetserver.c index 2306e10..8907768 100644 --- a/src/rpc/virnetserver.c +++ b/src/rpc/virnetserver.c @@ -95,7 +95,7 @@ struct _virNetServer { unsigned int keepaliveCount; bool keepaliveRequired; - unsigned int quit :1; + bool quit; #ifdef WITH_GNUTLS virNetTLSContextPtr tls; @@ -1035,7 +1035,7 @@ static void virNetServerAutoShutdownTimer(int timerid ATTRIBUTE_UNUSED, if (!srv->autoShutdownInhibitions) { VIR_DEBUG("Automatic shutdown triggered"); - srv->quit = 1; + srv->quit = true; } virObjectUnlock(srv); @@ -1074,7 +1074,7 @@ void virNetServerRun(virNetServerPtr srv) virNetServerMDNSStart(srv->mdns) < 0) goto cleanup; - srv->quit = 0; + srv->quit = false; if (srv->autoShutdownTimeout && (timerid = virEventAddTimeout(-1, @@ -1162,7 +1162,7 @@ void virNetServerQuit(virNetServerPtr srv) virObjectLock(srv); VIR_DEBUG("Quit requested %p", srv); - srv->quit = 1; + srv->quit = true; virObjectUnlock(srv); } -- 1.8.1.5

2 1

[libvirt] [PATCH 0/2] Fix 'make check' on RHEL-5.10
by Michal Privoznik 22 Oct '13

22 Oct '13

There are two tests failing for me on RHEL-5.10 and current git HEAD. Fix them. Michal Privoznik (2): tests: Use lv_abs_top_builddir instead of bare abs_top_builddir vircgroupmock: Mock access() to some more files tests/Makefile.am | 2 +- tests/vircgroupmock.c | 8 ++++++++ 2 files changed, 9 insertions(+), 1 deletion(-) -- 1.8.1.5

2 4

[libvirt] libvirt VF indexing issue
by Niilona 22 Oct '13

22 Oct '13

libvirt points to wrong VF -channel, when trying to change MAC -address. libvirt v1 1.0.2 linux driver : ixgbe 3.17.3 Symptom description ---------------------------------- 1.) How the VF channels are seen in the Host ( correct in increasing address order ) command : ls -la /sys/bus/pci/devices/0000\:04\:00.0/ .................. lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn0 -> ../0000:04:10.0/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn1 -> ../0000:04:10.2/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn10 -> ../0000:04:12.4/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn11 -> ../0000:04:12.6/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn12 -> ../0000:04:13.0/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn13 -> ../0000:04:13.2/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn2 -> ../0000:04:10.4/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn3 -> ../0000:04:10.6/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn4 -> ../0000:04:11.0/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn5 -> ../0000:04:11.2/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn6 -> ../0000:04:11.4/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn7 -> ../0000:04:11.6/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn8 -> ../0000:04:12.0/ lrwxrwxrwx 1 root root 0 Oct 17 12:25 virtfn9 -> ../0000:04:12.2/ ----------------------- 2.) How the libvirt sees them virsh nodedev-dumpxml pci_0000_04_00_0 <device> <name>pci_0000_04_00_0</name> <parent>pci_0000_00_03_0</parent> <driver> <name>ixgbe</name> </driver> <capability type='pci'> <domain>0</domain> <bus>4</bus> <slot>0</slot> <function>0</function> <product id='0x10fb'>82599EB 10-Gigabit SFI/SFP+ Network Connection</product> <vendor id='0x8086'>Intel Corporation</vendor> <capability type='virt_functions'> <address domain='0x0000' bus='0x04' slot='0x12' function='0x4'/> <address domain='0x0000' bus='0x04' slot='0x12' function='0x6'/> <address domain='0x0000' bus='0x04' slot='0x13' function='0x0'/> <address domain='0x0000' bus='0x04' slot='0x13' function='0x2'/> <address domain='0x0000' bus='0x04' slot='0x10' function='0x0'/> <address domain='0x0000' bus='0x04' slot='0x10' function='0x2'/> <address domain='0x0000' bus='0x04' slot='0x10' function='0x4'/> <address domain='0x0000' bus='0x04' slot='0x10' function='0x6'/> <address domain='0x0000' bus='0x04' slot='0x11' function='0x0'/> <address domain='0x0000' bus='0x04' slot='0x11' function='0x2'/> <address domain='0x0000' bus='0x04' slot='0x11' function='0x4'/> <address domain='0x0000' bus='0x04' slot='0x11' function='0x6'/> <address domain='0x0000' bus='0x04' slot='0x12' function='0x0'/> <address domain='0x0000' bus='0x04' slot='0x12' function='0x2'/> </capability> </capability> </device> 3.) Libvirt sets MAC -address in slot/position 5 instead of 1 ip link show eth10 87: eth10: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 2100 qdisc mq state UP mode DEFAULT qlen 1000 link/ether 00:1b:21:b9:a5:60 brd ff:ff:ff:ff:ff:ff vf 0 MAC 92:e2:6b:53:a7:c2, spoof checking off <------ this is the slot, MAC tried to set vf 1 MAC 0a:04:38:2c:79:35, spoof checking off vf 2 MAC b2:46:b6:91:44:2e, spoof checking off vf 3 MAC de:b6:28:da:20:44, spoof checking off vf 4 MAC 02:00:00:80:00:01, spoof checking off <--------- MAC sets in wrong slot vf 5 MAC 02:00:00:80:00:02, spoof checking off vf 6 MAC 02:00:00:80:00:20, spoof checking off vf 7 MAC 02:00:00:80:00:30, spoof checking off vf 8 MAC 52:54:00:de:f8:08, spoof checking off vf 9 MAC 5e:b0:d0:34:d6:cc, spoof checking off vf 10 MAC d2:88:5c:2d:db:b5, spoof checking off vf 11 MAC 6a:06:8a:9c:31:8b, spoof checking off vf 12 MAC 4a:38:be:8f:b3:bf, spoof checking off vf 13 MAC b6:c1:d4:12:d3:3d, spoof checking off 5.) Question : Why this happens in ixgbe -driver and when we try to use more than 10 VF's/physical port. ?

2 1