- Devel - libvirt List Archives

[libvirt] [PATCH v4 1/4] vcpupin: introduce a new libvirt API
by Taku Izumi 09 Jun '11

09 Jun '11

This patch introduces a new libvirt API (virDomainPinVcpuFlags). Signd-off-by: Taku Izumi <izumi.taku(a)jp.fujitsu.com> --- include/libvirt/libvirt.h.in | 5 ++ src/driver.h | 7 +++ src/libvirt.c | 76 +++++++++++++++++++++++++++++++++++++++++++ src/libvirt_public.syms | 1 4 files changed, 89 insertions(+) Index: libvirt/include/libvirt/libvirt.h.in =================================================================== --- libvirt.orig/include/libvirt/libvirt.h.in +++ libvirt/include/libvirt/libvirt.h.in @@ -1049,6 +1049,11 @@ int virDomainPinVcpu unsigned int vcpu, unsigned char *cpumap, int maplen); +int virDomainPinVcpuFlags (virDomainPtr domain, + unsigned int vcpu, + unsigned char *cpumap, + int maplen, + unsigned int flags); /** * VIR_USE_CPU: Index: libvirt/src/driver.h =================================================================== --- libvirt.orig/src/driver.h +++ libvirt/src/driver.h @@ -230,6 +230,12 @@ typedef int unsigned char *cpumap, int maplen); typedef int + (*virDrvDomainPinVcpuFlags) (virDomainPtr domain, + unsigned int vcpu, + unsigned char *cpumap, + int maplen, + unsigned int flags); +typedef int (*virDrvDomainGetVcpus) (virDomainPtr domain, virVcpuInfoPtr info, int maxinfo, @@ -671,6 +677,7 @@ struct _virDriver { virDrvDomainSetVcpusFlags domainSetVcpusFlags; virDrvDomainGetVcpusFlags domainGetVcpusFlags; virDrvDomainPinVcpu domainPinVcpu; + virDrvDomainPinVcpuFlags domainPinVcpuFlags; virDrvDomainGetVcpus domainGetVcpus; virDrvDomainGetMaxVcpus domainGetMaxVcpus; virDrvDomainGetSecurityLabel domainGetSecurityLabel; Index: libvirt/src/libvirt.c =================================================================== --- libvirt.orig/src/libvirt.c +++ libvirt/src/libvirt.c @@ -6709,6 +6709,82 @@ error: } /** + * virDomainPinVcpuFlags: + * @domain: pointer to domain object, or NULL for Domain0 + * @vcpu: virtual CPU number + * @cpumap: pointer to a bit map of real CPUs (in 8-bit bytes) (IN) + * Each bit set to 1 means that corresponding CPU is usable. + * Bytes are stored in little-endian order: CPU0-7, 8-15... + * In each byte, lowest CPU number is least significant bit. + * @maplen: number of bytes in cpumap, from 1 up to size of CPU map in + * underlying virtualization system (Xen...). + * If maplen < size, missing bytes are set to zero. + * If maplen > size, failure code is returned. + * @flags: bitwise-OR of virDomainModificationImpac + * + * Dynamically change the real CPUs which can be allocated to a virtual CPU. + * This function requires privileged access to the hypervisor. + * + * @flags may include VIR_DOMAIN_PIN_VCPU_LIVE or VIR_DOMAIN_PIN_VCPU_CONFIG. + * Both flags may be set. + * If VIR_DOMAIN_PIN_VCPU_LIVE is set, the change affects a running domain + * and may fail if domain is not alive. + * If VIR_DOMAIN_PIN_VCPU_CONFIG is set, the change affects persistent state, + * and will fail for transient domains. If neither flag is specified (that is, + * @flags is VIR_DOMAIN_PIN_VCPU_CURRENT), then an inactive domain modifies + * persistent setup, while an active domain is hypervisor-dependent on whether + * just live or both live and persistent state is changed. + * Not all hypervisors can support all flag combinations. + * + * Returns 0 in case of success, -1 in case of failure. + * + */ +int +virDomainPinVcpuFlags(virDomainPtr domain, unsigned int vcpu, + unsigned char *cpumap, int maplen, unsigned int flags) +{ + virConnectPtr conn; + + VIR_DOMAIN_DEBUG(domain, "vcpu=%u, cpumap=%p, maplen=%d, flags=%u", + vcpu, cpumap, maplen, flags); + + virResetLastError(); + + if (!VIR_IS_CONNECTED_DOMAIN(domain)) { + virLibDomainError(VIR_ERR_INVALID_DOMAIN, __FUNCTION__); + virDispatchError(NULL); + return -1; + } + + if (domain->conn->flags & VIR_CONNECT_RO) { + virLibDomainError(VIR_ERR_OPERATION_DENIED, __FUNCTION__); + goto error; + } + + if ((vcpu > 32000) || (cpumap == NULL) || (maplen < 1)) { + virLibDomainError(VIR_ERR_INVALID_ARG, __FUNCTION__); + goto error; + } + + conn = domain->conn; + + if (conn->driver->domainPinVcpuFlags) { + int ret; + ret = conn->driver->domainPinVcpuFlags (domain, vcpu, cpumap, maplen, flags); + if (ret < 0) + goto error; + return ret; + } + + virLibConnError(VIR_ERR_NO_SUPPORT, __FUNCTION__); + +error: + virDispatchError(domain->conn); + return -1; + +} + +/** * virDomainGetVcpus: * @domain: pointer to domain object, or NULL for Domain0 * @info: pointer to an array of virVcpuInfo structures (OUT) Index: libvirt/src/libvirt_public.syms =================================================================== --- libvirt.orig/src/libvirt_public.syms +++ libvirt/src/libvirt_public.syms @@ -448,6 +448,7 @@ LIBVIRT_0.9.2 { virInterfaceChangeBegin; virInterfaceChangeCommit; virInterfaceChangeRollback; + virDomainPinVcpuFlags; } LIBVIRT_0.9.0; # .... define new API here using predicted next version number ....

2 1

[libvirt] [PATCH 4/4] vcpupin: add vcpupin resetting feature to qemu driver
by Taku Izumi 09 Jun '11

09 Jun '11

Pinning to all physical cpus means resetting, hence it is preferable to delete vcpupin setting of XML. This patch changes qemu driver to delete vcpupin setting by invoking virDomainVcpupinDel API when pinning the specified virtual cpu to all host physical cpus. Signed-off-by: Taku Izumi <izumi.taku(a)jp.fujitsu.com> --- src/qemu/qemu_driver.c | 64 +++++++++++++++++++++++++++++++++++-------------- 1 file changed, 46 insertions(+), 18 deletions(-) Index: libvirt/src/qemu/qemu_driver.c =================================================================== --- libvirt.orig/src/qemu/qemu_driver.c +++ libvirt/src/qemu/qemu_driver.c @@ -2894,6 +2894,7 @@ qemudDomainPinVcpuFlags(virDomainPtr dom int ret = -1; bool isActive; qemuDomainObjPrivatePtr priv; + bool canResetting = true; virCheckFlags(VIR_DOMAIN_AFFECT_LIVE | VIR_DOMAIN_AFFECT_CONFIG, -1); @@ -2944,15 +2945,23 @@ qemudDomainPinVcpuFlags(virDomainPtr dom goto cleanup; } - if (flags & VIR_DOMAIN_AFFECT_LIVE) { - - if (nodeGetInfo(dom->conn, &nodeinfo) < 0) - goto cleanup; + if (nodeGetInfo(dom->conn, &nodeinfo) < 0) + goto cleanup; + hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo); + maxcpu = maplen * 8; + if (maxcpu > hostcpus) + maxcpu = hostcpus; + /* pinning to all physical cpus means resetting, + * so check if we can reset setting. + */ + for (int pcpu = 0; pcpu < hostcpus; pcpu++) { + if ((cpumap[pcpu/8] & (1 << (pcpu % 8))) == 0) { + canResetting = false; + break; + } + } - hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo); - maxcpu = maplen * 8; - if (maxcpu > hostcpus) - maxcpu = hostcpus; + if (flags & VIR_DOMAIN_AFFECT_LIVE) { if (priv->vcpupids != NULL) { if (virProcessInfoSetAffinity(priv->vcpupids[vcpu], @@ -2964,23 +2973,42 @@ qemudDomainPinVcpuFlags(virDomainPtr dom goto cleanup; } - if (virDomainVcpupinAdd(vm->def, cpumap, maplen, vcpu) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("failed to update or add vcpupin xml of " - "a running domain")); - goto cleanup; + if (canResetting) { + if (virDomainVcpupinDel(vm->def, vcpu) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("failed to delete vcpupin xml of " + "a running domain")); + goto cleanup; + } + } else { + if (virDomainVcpupinAdd(vm->def, cpumap, maplen, vcpu) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("failed to update or add vcpupin xml of " + "a running domain")); + goto cleanup; + } } } if (flags & VIR_DOMAIN_AFFECT_CONFIG) { - if (virDomainVcpupinAdd(persistentDef, cpumap, maplen, vcpu) < 0) { - qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("failed to update or add vcpupin xml of " - "a persistent domain")); - goto cleanup; + if (canResetting) { + if (virDomainVcpupinDel(persistentDef, vcpu) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("failed to delete vcpupin xml of " + "a persistent domain")); + goto cleanup; + } + } else { + if (virDomainVcpupinAdd(persistentDef, cpumap, maplen, vcpu) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("failed to update or add vcpupin xml of " + "a persistent domain")); + goto cleanup; + } } + ret = virDomainSaveConfig(driver->configDir, persistentDef); goto cleanup; }

1 0

[libvirt] [PATCH 3/4] vcpupin: add virDomainVcpupinDel function
by Taku Izumi 09 Jun '11

09 Jun '11

This patch add the private API (virDomainVcpupinDel). This API can delete the vcpupin setting of a specified virtual cpu. Signed-off-by: Taku Izumi <izumi.taku(a)jp.fujitsu.com> --- src/conf/domain_conf.c | 39 +++++++++++++++++++++++++++++++++++++++ src/conf/domain_conf.h | 2 ++ src/libvirt_private.syms | 1 + 3 files changed, 42 insertions(+) Index: libvirt/src/conf/domain_conf.c =================================================================== --- libvirt.orig/src/conf/domain_conf.c +++ libvirt/src/conf/domain_conf.c @@ -8126,6 +8126,45 @@ cleanup: return -1; } +int +virDomainVcpupinDel(virDomainDefPtr def, int vcpu) +{ + int n; + bool deleted = false; + virDomainVcpupinDefPtr *vcpupin_list = def->cputune.vcpupin; + + /* No vcpupin exists yet */ + if (!def->cputune.nvcpupin) { + return 0; + } + + for (n = 0; n < def->cputune.nvcpupin; n++) { + if (vcpupin_list[n]->vcpuid == vcpu) { + VIR_FREE(vcpupin_list[n]->cpumask); + VIR_FREE(vcpupin_list[n]); + memmove(&vcpupin_list[n], + &vcpupin_list[n+1], + (def->cputune.nvcpupin - n - 1) * sizeof(virDomainVcpupinDef *)); + deleted = true; + break; + } + } + + if (!deleted) + return 0; + + if (--def->cputune.nvcpupin == 0) { + virDomainVcpupinDefFree(def->cputune.vcpupin, 0); + } else { + if (VIR_REALLOC_N(def->cputune.vcpupin, def->cputune.nvcpupin) < 0) { + virReportOOMError(); + return -1; + } + } + + return 0; +} + static int virDomainLifecycleDefFormat(virBufferPtr buf, int type, Index: libvirt/src/conf/domain_conf.h =================================================================== --- libvirt.orig/src/conf/domain_conf.h +++ libvirt/src/conf/domain_conf.h @@ -1381,6 +1381,8 @@ int virDomainVcpupinAdd(virDomainDefPtr int maplen, int vcpu); +int virDomainVcpupinDel(virDomainDefPtr def, int vcpu); + int virDomainDiskIndexByName(virDomainDefPtr def, const char *name); int virDomainDiskInsert(virDomainDefPtr def, virDomainDiskDefPtr disk); Index: libvirt/src/libvirt_private.syms =================================================================== --- libvirt.orig/src/libvirt_private.syms +++ libvirt/src/libvirt_private.syms @@ -360,6 +360,7 @@ virDomainTimerTickpolicyTypeToString; virDomainTimerTrackTypeFromString; virDomainTimerTrackTypeToString; virDomainVcpupinAdd; +virDomainVcpupinDel; virDomainVcpupinFindByVcpu; virDomainVcpupinIsDuplicate; virDomainVideoDefFree;

1 0

[libvirt] [PATCH 2/4] vcpupin: add reset option to virsh vcpupin command
by Taku Izumi 09 Jun '11

09 Jun '11

When resetting vcpupin setting, we have to specify all host physical cpus as a cpulist parameter of virsh vcpupin command. It's a little tedious. This patch changes to allow to receive the special keyword 'r' as a cpulist parameter of virsh vcpupin command when resetting vcpupin setting. If you set the following: # virsh vcpupin VM 0 r the vcpu0 will be pinned to all physical cpus. Signed-off-by: Taku Izumi <izumi.taku(a)jp.fujitsu.com> --- tools/virsh.c | 7 ++++++- tools/virsh.pod | 2 ++ 2 files changed, 8 insertions(+), 1 deletion(-) Index: libvirt/tools/virsh.c =================================================================== --- libvirt.orig/tools/virsh.c +++ libvirt/tools/virsh.c @@ -3010,8 +3010,13 @@ cmdVcpupin(vshControl *ctl, const vshCmd /* Parse cpulist */ cur = cpulist; - if (*cur == 0) + if (*cur == 0) { goto parse_error; + } else if (*cur == 'r') { + for (cpu = 0; cpu < maxcpu; cpu++) + VIR_USE_CPU(cpumap, cpu); + *cur = 0; + } while (*cur != 0) { Index: libvirt/tools/virsh.pod =================================================================== --- libvirt.orig/tools/virsh.pod +++ libvirt/tools/virsh.pod @@ -797,6 +797,8 @@ Pin domain VCPUs to host physical CPUs. and I<cpulist> is a list of physical CPU numbers. Its syntax is a comma separated list and a special markup using '-' and '^' (ex. '0-4', '0-3,^2') can also be allowed. The '-' denotes the range and the '^' denotes exclusive. +If you want to reset vcpupin setting, that is, to pin vcpu all physical cpus, +simply specify 'r' as a cpulist. If I<--live> is specified, affect a running guest. If I<--config> is specified, affect the next boot of a persistent guest. If I<--current> is specified, affect the current guest state.

1 0

[libvirt] [PATCH 1/4] vcpupin: improve vcpupin definition of virsh vcpupin
by Taku Izumi 09 Jun '11

09 Jun '11

When using vcpupin command, we have to speficy comma-separated list as cpulist, but this is tedious in case the number of phsycal cpus is large. This patch improves this by introducing special markup "-" and "^" which are similar to XML schema of "cpuset" attribute. That is: # virsh vcpupin Guest 0 0-15,^8 is identical to # virsh vcpupin Guest 0 0,1,2,3,4,5,6,7,9,10,11,12,13,14,15 NOTE: The expression is sequencially evaluated, so "0-15,^8" is not identical to "^8,0-15". Signed-off-by: Taku Izumi <izumi.taku(a)jp.fujitsu.com> --- tools/virsh.c | 125 +++++++++++++++++++++++++++++++------------------------- tools/virsh.pod | 4 + 2 files changed, 73 insertions(+), 56 deletions(-) Index: libvirt/tools/virsh.c =================================================================== --- libvirt.orig/tools/virsh.c +++ libvirt/tools/virsh.c @@ -2946,8 +2946,9 @@ cmdVcpupin(vshControl *ctl, const vshCmd bool ret = true; unsigned char *cpumap; int cpumaplen; - int i; - enum { expect_num, expect_num_or_comma } state; + int i, cpu, lastcpu, maxcpu; + bool unuse = false; + char *cur; int config = vshCommandOptBool(cmd, "config"); int live = vshCommandOptBool(cmd, "live"); int current = vshCommandOptBool(cmd, "current"); @@ -3003,66 +3004,74 @@ cmdVcpupin(vshControl *ctl, const vshCmd return false; } - /* Check that the cpulist parameter is a comma-separated list of - * numbers and give an intelligent error message if not. - */ - if (cpulist[0] == '\0') { - vshError(ctl, "%s", _("cpulist: Invalid format. Empty string.")); - virDomainFree (dom); - return false; - } + maxcpu = VIR_NODEINFO_MAXCPUS(nodeinfo); + cpumaplen = VIR_CPU_MAPLEN(maxcpu); + cpumap = vshCalloc(ctl, 0, cpumaplen); + + /* Parse cpulist */ + cur = cpulist; + if (*cur == 0) + goto parse_error; + + while (*cur != 0) { + + /* the char '^' denotes exclusive */ + if (*cur == '^') { + cur++; + unuse = true; + } + + /* parse physical CPU number */ + if (!c_isdigit(*cur)) + goto parse_error; + cpu = virParseNumber(&cur); + if (cpu < 0) { + goto parse_error; + } + if (cpu >= maxcpu) { + vshError(ctl, _("Physical CPU %d doesn't exist."), cpu); + goto parse_error; + } + virSkipSpaces(&cur); - state = expect_num; - for (i = 0; cpulist[i]; i++) { - switch (state) { - case expect_num: - if (!c_isdigit (cpulist[i])) { - vshError(ctl, _("cpulist: %s: Invalid format. Expecting " - "digit at position %d (near '%c')."), - cpulist, i, cpulist[i]); - virDomainFree (dom); - return false; + if ((*cur == ',') || (*cur == 0)) { + if (unuse) { + VIR_UNUSE_CPU(cpumap, cpu); + } else { + VIR_USE_CPU(cpumap, cpu); } - state = expect_num_or_comma; - break; - case expect_num_or_comma: - if (cpulist[i] == ',') - state = expect_num; - else if (!c_isdigit (cpulist[i])) { - vshError(ctl, _("cpulist: %s: Invalid format. Expecting " - "digit or comma at position %d (near '%c')."), - cpulist, i, cpulist[i]); - virDomainFree (dom); - return false; + } else if (*cur == '-') { + /* the char '-' denotes range */ + if (unuse) { + goto parse_error; + } + cur++; + virSkipSpaces(&cur); + /* parse the end of range */ + lastcpu = virParseNumber(&cur); + if (lastcpu < cpu) { + goto parse_error; + } + if (lastcpu >= maxcpu) { + vshError(ctl, _("Physical CPU %d doesn't exist."), maxcpu); + goto parse_error; + } + for (i = cpu; i <= lastcpu; i++) { + VIR_USE_CPU(cpumap, i); } + virSkipSpaces(&cur); } - } - if (state == expect_num) { - vshError(ctl, _("cpulist: %s: Invalid format. Trailing comma " - "at position %d."), - cpulist, i); - virDomainFree (dom); - return false; - } - cpumaplen = VIR_CPU_MAPLEN(VIR_NODEINFO_MAXCPUS(nodeinfo)); - cpumap = vshCalloc(ctl, 1, cpumaplen); - - do { - unsigned int cpu = atoi(cpulist); - - if (cpu < VIR_NODEINFO_MAXCPUS(nodeinfo)) { - VIR_USE_CPU(cpumap, cpu); + if (*cur == ',') { + cur++; + virSkipSpaces(&cur); + unuse = false; + } else if (*cur == 0) { + break; } else { - vshError(ctl, _("Physical CPU %d doesn't exist."), cpu); - VIR_FREE(cpumap); - virDomainFree(dom); - return false; + goto parse_error; } - cpulist = strchr(cpulist, ','); - if (cpulist) - cpulist++; - } while (cpulist); + } if (flags == -1) { if (virDomainPinVcpu(dom, vcpu, cpumap, cpumaplen) != 0) { @@ -3074,9 +3083,15 @@ cmdVcpupin(vshControl *ctl, const vshCmd } } +cleanup: VIR_FREE(cpumap); virDomainFree(dom); return ret; + +parse_error: + vshError(ctl, "%s", _("cpulist: Invalid format.")); + ret = false; + goto cleanup; } /* Index: libvirt/tools/virsh.pod =================================================================== --- libvirt.orig/tools/virsh.pod +++ libvirt/tools/virsh.pod @@ -794,7 +794,9 @@ vCPUs, the running time, the affinity to I<--current> Pin domain VCPUs to host physical CPUs. The I<vcpu> number must be provided -and I<cpulist> is a comma separated list of physical CPU numbers. +and I<cpulist> is a list of physical CPU numbers. Its syntax is a comma +separated list and a special markup using '-' and '^' (ex. '0-4', '0-3,^2') can +also be allowed. The '-' denotes the range and the '^' denotes exclusive. If I<--live> is specified, affect a running guest. If I<--config> is specified, affect the next boot of a persistent guest. If I<--current> is specified, affect the current guest state.

1 0

[libvirt] [PATCH 0/4] RFC: vcpupin: some improvements of virsh vcpupin command
by Taku Izumi 09 Jun '11

09 Jun '11

Hi all, This patchset improves virsh vcpupin command like the following: (i) introduce special markup "-" and "^" which are similar to XML schema of "cpuset" attribute # virsh vcpupin VM 0 0-8,^4 --config # virsh dumpxml VM ... <vcpu>4</vcpu> <cputune> <vcpupin vcpu='0' cpuset='0-3,5-8'/> </cputune> ... (ii) allow to receive the special keyword 'r' as a cpulist parameter when resetting vcpupin setting. # virsh vcpupin VM 0 r --config (iii) when resetting, that is, pinning the specified virtual cpu to all host physical cpus, vcpupin setting will be deleted # virsh dumpxml VM ... <vcpu>4</vcpu> <cputune> </cputune> ... The following my patchset is the prerequisite of this: http://www.redhat.com/archives/libvir-list/2011-June/msg00390.html *[PATCH 1/4] vcpupin: improve vcpupin definition of virsh vcpupin *[PATCH 2/4] vcpupin: add reset option to virsh vcpupin command *[PATCH 3/4] vcpupin: add virDomainVcpupinDel function *[PATCH 4/4] vcpupin: add vcpupin resetting feature to qemu driver Best regards, Taku Izumi

1 0

[libvirt] [PATCH v4 0/4] vcpupin: configure inactive domain's CPU affinity setting
by Taku Izumi 09 Jun '11

09 Jun '11

Hi all, This patchset enables us to configure inactive domains' CPU affinity setting. This is the rebased version, but retains the version number. *[PATCH v4 1/4] vcpupin: introduce a new libvirt API (virDomainPinVcpuFlags) *[PATCH v4 2/4] vcpupin: implement the code to address the new API in the qemu driver *[PATCH v4 3/4] vcpupin: implement the remote protocol to address the new API *[PATCH v4 4/4] vcpupin: add the new option to "virsh vcpupin" command Best regards, Taku Izumi

1 0

[libvirt] Driver development
by Carlos N. A. Corrêa 08 Jun '11

08 Jun '11

Hi, I'm interested in developing a new network driver for libvirt (one for the definition and management of OpenVSwitch datapaths). I've read some of the code already, and by now my general directions are: - Change the function virNetworkDefParseXML in src/conf/network_conf.c to define and parse new configuration directives needed by my driver - Use src/network/bridge_driver.c as a template for my new driver, replacing code from the services provided by the driver (listed on "virNetworkDriver" type) with my own code, for my own net type. Have you any other directions in this matter? Any tips? If it works, would be this code of any use to you? Best regards, Carlos "Bill" Nilton CISSP, RHCE, ITIL Foundations http://carlosnilton.com.br/

2 2

[libvirt] [RFC PATCH] Set and reset MAC for PASSTHROUGH mode
by D. Herrendoerfer 08 Jun '11

08 Jun '11

Hi to all, there is a problem present in libvirt that when a PASSTHROUGH mode DIRECT NIC connection is made the MAC address of the NIC is not automatically set and reset to the configured VM MAC and back again. The attached patch fixes this problem by setting and resetting the MAC while remembering the previous setting while the VM is running. This also works if libvirtd is restarted while the VM is running. The storing of the previous MAC address is done in a link ( currently in /tmp, but should probably go into /var/run somewhere ). Why a link ? It only uses one inode, no need to waste a whole data block for 16 bytes or less. It's also more readable by simply looking at it with 'ls -l'. Best regards Dirk Signed-off-by: Dirk Herrendoerfer <d.herrendoerfer at herrendoerfer.name> diff --git a/src/libvirt_macvtap.syms b/src/libvirt_macvtap.syms index b48565b..0520cb5 100644 --- a/src/libvirt_macvtap.syms +++ b/src/libvirt_macvtap.syms @@ -6,5 +6,7 @@ # macvtap.h delMacvtap; openMacvtapTap; +get_macaddr; +set_macaddr; vpAssociatePortProfileId; vpDisassociatePortProfileId; diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index ef2d002..4cddbba 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -118,6 +118,7 @@ qemuPhysIfaceConnect(virDomainDefPtr def, int rc; #if WITH_MACVTAP char *res_ifname = NULL; + unsigned char old_macaddr[6]; int vnet_hdr = 0; int err; @@ -125,6 +126,53 @@ qemuPhysIfaceConnect(virDomainDefPtr def, net->model && STREQ(net->model, "virtio")) vnet_hdr = 1; + /** Note: When using PASSTHROUGH mode with MACVTAP devices the link + * devices's MAC address must be set to the VMs MAC address. In + * order to not confuse the first switch or bridge in line this MAC + * address must be reset when the VM is shut down. + * This is especially important when using SRIOV capable cards that + * emulate their switch in firmware. + */ + if ( net->data.direct.mode == VIR_DOMAIN_NETDEV_MACVTAP_MODE_PASSTHRU ) { + rc = get_macaddr(&old_macaddr, 6, net->data.direct.linkdev); + if (rc) { + virReportSystemError(rc, + _("Getting MAC address from '%s' " + "to '%02x:%02x:%02x:%02x:%02x:%02x' failed."), + net->data.direct.linkdev, + old_macaddr[0],old_macaddr[1],old_macaddr[2], + old_macaddr[3],old_macaddr[4],old_macaddr[5]); + } else { + char oldmacname[256], newmacname[256]; + + sprintf(oldmacname,"%02x:%02x:%02x:%02x:%02x:%02x", + old_macaddr[0],old_macaddr[1],old_macaddr[2], + old_macaddr[3],old_macaddr[4],old_macaddr[5]); + + sprintf(newmacname,"/tmp/%s@%02x:%02x:%02x:%02x:%02x:%02x", + net->data.direct.linkdev, + net->mac[0],net->mac[1],net->mac[2], + net->mac[3],net->mac[4],net->mac[5]); + + rc = symlink (oldmacname, newmacname); + if (rc) { + virReportSystemError(rc, + _("MAC link file creation failed for %s."), + net->data.direct.linkdev); + } + } + + rc = set_macaddr(net->mac, 6, net->data.direct.linkdev); + if (rc) { + virReportSystemError(rc, + _("Setting MAC address on '%s' to " + "'%02x:%02x:%02x:%02x:%02x:%02x' failed."), + net->data.direct.linkdev, + net->mac[0],net->mac[1],net->mac[2], + net->mac[3],net->mac[4],net->mac[5]); + } + } + rc = openMacvtapTap(net->ifname, net->mac, net- >data.direct.linkdev, net->data.direct.mode, vnet_hdr, def->uuid, &net->data.direct.virtPortProfile, &res_ifname, diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 1efe024..127acb8 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -2663,6 +2663,51 @@ void qemuProcessStop(struct qemud_driver *driver, if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) { delMacvtap(net->ifname, net->mac, net- >data.direct.linkdev, &net->data.direct.virtPortProfile); + + if ( net->data.direct.mode == VIR_DOMAIN_NETDEV_MACVTAP_MODE_PASSTHRU ) { + char newmacname[256], linkdata[64]; + unsigned int old_macaddr[6]; + + sprintf(newmacname,"/tmp/%s@%02x:%02x:%02x:%02x:%02x: %02x", + net->data.direct.linkdev, + net->mac[0],net->mac[1],net->mac[2], + net->mac[3],net->mac[4],net->mac[5]); + + ret = readlink (newmacname, linkdata, 64); + if ( ret == 17 ) { + linkdata[17] = 0; + + ret = sscanf(linkdata,"%x:%x:%x:%x:%x:%x", + &old_macaddr[0],&old_macaddr[1],&old_macaddr[2], + &old_macaddr[3],&old_macaddr[4],&old_macaddr[5]); + if ( ret == 6 ) { + unsigned char oldmac[6]; + oldmac[0] = (unsigned char)(old_macaddr[0] & 0xFF) ; + oldmac[1] = (unsigned char)(old_macaddr[1] & 0xFF) ; + oldmac[2] = (unsigned char)(old_macaddr[2] & 0xFF) ; + oldmac[3] = (unsigned char)(old_macaddr[3] & 0xFF) ; + oldmac[4] = (unsigned char)(old_macaddr[4] & 0xFF) ; + oldmac[5] = (unsigned char)(old_macaddr[5] & 0xFF) ; + + VIR_WARN(("Resetting MAC address on '%s' " + "to '%02x:%02x:%02x:%02x:%02x: %02x'."), + net->data.direct.linkdev, + oldmac[0],oldmac[1],oldmac[2], + oldmac[3],oldmac[4],oldmac[5]); + /*reset mac and remove link file-ignore restults*/ + ret = set_macaddr(oldmac, 6, net- >data.direct.linkdev); + ret = unlink(newmacname); + } else { + VIR_WARN(("Scanning MAC address from '%s' " + "failed ! Got %i values."), + linkdata, ret); + } + } else { + VIR_WARN(("Reading MAC address from '%s' failed ! " + "Got %i bytes."), + newmacname, ret); + } + } VIR_FREE(net->ifname); } } diff --git a/src/util/macvtap.c b/src/util/macvtap.c index 068638e..1026a84 100644 --- a/src/util/macvtap.c +++ b/src/util/macvtap.c @@ -191,6 +191,155 @@ err_exit: # if WITH_MACVTAP +/** + * get_macaddr: + * Get the MAC address of a network device + * + * @macaddress: Pointer where the MAC address will be stored + * @macaddrsize: Size of the MAC address. + * @srcdev: The interface name of the NIC to get the MAC from + * + * Returns zero in case of success, + * negative value otherwise with error reported. + * + */ +int +get_macaddr(const unsigned char *macaddress, int macaddrsize, + const char *srcdev ) +{ + int sockfd; + int io; + char buffer[1024]; + struct ifreq ifr; + + strcpy(ifr.ifr_name, srcdev); + + sockfd = socket(AF_INET, SOCK_STREAM, 0); + if(sockfd < 0){ + return -1; + } + + io = ioctl(sockfd, SIOCGIFHWADDR, (char *)&ifr); + if(io < 0){ + return -1; + } + + bcopy( ifr.ifr_ifru.ifru_hwaddr.sa_data, macaddress, macaddrsize); + + return 0; +} + +/** + * Set_macaddr: + * Set the MAC address of a network device + * + * @macaddress: MAC address to assign to the NIC + * @macaddrsize: Size of the MAC address. + * @srcdev: The interface name of the NIC + * + * Returns zero in case of success, + * negative value otherwise with error reported. + * + */ +int +set_macaddr(const unsigned char *macaddress, int macaddrsize, + const char *srcdev ) +{ + int rc = 0; + struct nlmsghdr *resp; + struct nlmsgerr *err; + struct ifinfomsg ifinfo = { .ifi_family = AF_UNSPEC }; + int ifindex; + unsigned char *recvbuf = NULL; + unsigned int recvbuflen; + struct nl_msg *nl_msg; + struct nlattr *linkinfo; + + if (ifaceGetIndex(true, srcdev, &ifindex) != 0) + return -1; + + nl_msg = nlmsg_alloc_simple(RTM_SETLINK, NLM_F_REQUEST); + + if (!nl_msg) { + virReportOOMError(); + return -1; + } + + if (nlmsg_append(nl_msg, &ifinfo, sizeof(ifinfo), NLMSG_ALIGNTO) < 0) + goto buffer_too_small; + + if (nla_put_u32(nl_msg, IFLA_LINK, ifindex) < 0) + goto buffer_too_small; + + if (nla_put(nl_msg, IFLA_ADDRESS, macaddrsize, macaddress) < 0) + goto buffer_too_small; + + if (srcdev && + nla_put(nl_msg, IFLA_IFNAME, strlen(srcdev)+1, srcdev) < 0) + goto buffer_too_small; + + if (nlComm(nl_msg, &recvbuf, &recvbuflen, 0) < 0) { + rc = -1; + goto err_exit; + } + + if (recvbuflen < NLMSG_LENGTH(0) || recvbuf == NULL) + goto malformed_resp; + + resp = (struct nlmsghdr *)recvbuf; + + switch (resp->nlmsg_type) { + case NLMSG_ERROR: + err = (struct nlmsgerr *)NLMSG_DATA(resp); + if (resp->nlmsg_len < NLMSG_LENGTH(sizeof(*err))) + goto malformed_resp; + + switch (err->error) { + + case 0: + break; + + case -EEXIST: + rc = -1; + break; + + default: + macvtapError(VIR_ERR_INTERNAL_ERROR, "%s", + _("error setting device mac address")); + rc = -1; + } + break; + + case NLMSG_DONE: + break; + + default: + goto malformed_resp; + } + +err_exit: + nlmsg_free(nl_msg); + + VIR_FREE(recvbuf); + + return rc; + +malformed_resp: + nlmsg_free(nl_msg); + + macvtapError(VIR_ERR_INTERNAL_ERROR, "%s", + _("malformed netlink response message")); + VIR_FREE(recvbuf); + return -1; + +buffer_too_small: + nlmsg_free(nl_msg); + + macvtapError(VIR_ERR_INTERNAL_ERROR, "%s", + _("allocated netlink buffer is too small")); + return -1; +} + static int link_add(const char *type, const unsigned char *macaddress, int macaddrsize, diff --git a/src/util/macvtap.h b/src/util/macvtap.h index a1383c4..d71546a 100644 --- a/src/util/macvtap.h +++ b/src/util/macvtap.h @@ -75,6 +75,12 @@ enum virVMOperationType { # include "internal.h" +int get_macaddr(const unsigned char *macaddress, int macaddrsize, + const char *srcdev ); + +int set_macaddr(const unsigned char *macaddress, int macaddrsize, + const char *srcdev ); + int openMacvtapTap(const char *ifname, const unsigned char *macaddress, const char *linkdev,

2 1

[libvirt] [PATCH] virt-aa-helper: add missing include
by Eric Blake 08 Jun '11

08 Jun '11

Regression introduced in commit 02e8691. * src/security/virt-aa-helper.c (includes): Reflect move of virRun. --- Pushing under the build-breaker rule, for Ubuntu. src/security/virt-aa-helper.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/src/security/virt-aa-helper.c b/src/security/virt-aa-helper.c index 41ba4a3..afc0d33 100644 --- a/src/security/virt-aa-helper.c +++ b/src/security/virt-aa-helper.c @@ -31,6 +31,7 @@ #include "buf.h" #include "util.h" #include "memory.h" +#include "command.h" #include "security_driver.h" #include "security_apparmor.h" -- 1.7.4.4

1 0