[libvirt] [PATCH v6 00/33] BaselineHypervisorCPU using QEMU QMP exchanges

Some architectures (S390) depend on QEMU to compute baseline CPU model and expand a models feature set. Interacting with QEMU requires starting the QEMU process and completing one or more query-cpu-model-baseline QMP exchanges with QEMU in addition to a query-cpu-model-expansion QMP exchange to expand all features in the model. See "s390x CPU models: exposing features" patch set on Qemu-devel for discussion of QEMU aspects. This is part of resolution of: https://bugzilla.redhat.com/show_bug.cgi?id=1511999 ----- v6 addresses these issues from v5: qemu_process: Move process code from qemu_capabilities to qemu_process [x] Remove exception statement from commit message qemu_process: Use qemuProcessQmp prefix [x] s/ProcessQmp/ProcessQMP/ in this and all subsequent patches qemu_process: Refer to proc not cmd in process code [x] Back port indentation fix in qemuProcessQMPRun qemu_capabilities: Stop QEMU process before freeing [x] Back port !proc check in qemuProcessQmpStop qemu_process: Use qemuProcessQmp struct for a single process [x] Remove extra empty lines qemu_process: All ProcessQMP errors are fatal [x] Split into multiple patches qemu_process: Introduce qemuProcessQmpStart [x] Log function parameters separately in the first function debug message [x] Removed NULLSTR check on proc-binary (would fail previously on proc activate) [x] Back port referring to stderr rather than qmperr in comment [ ] Combine qemuProcessQmpStop and qemuProcessQmpFree (See reply to v5 email) emu_process: Collect monitor code in single function [x] Back port qemuMonitorOpen indent fix to earlier patch qemu_process: Don't open monitor if process failed [x] Drop patch qemu_process: Cleanup qemuProcessQmp alloc function [x] s/Qemu/QEMU/ in comments for qemuProcessQMPNew [x] Remove unneeded use of NULLSTR and NULL check before calling qemuProcessQmpFree qemu_process: Cleanup qemuProcessQmpStop function [x] Remove lines from from commit message [x] Remove capitalization from comment [x] Fix func def to match coding style [x] Back port !proc check to earlier patch qemu_process: Catch process free before process stop [x] Remove capitalization from comment [x] Don't do proc NULL check in VIR_DEBUG [ ] Combine qemuProcessQmpStop and qemuProcessQmpFree (See reply to v5 email) qemu_monitor: Make monitor callbacks optional [x] Drop patch qemu_process: Enter QMP command mode when starting QEMU Process [ ] Don't call qemuMonitorSetCapabilities in qemuProcessQMPConnectMonitor (See reply to v5 email) qemu_process: Use unique directories for QMP processes [x] s/qemu./qmp-/ for QMP process directory name prefix [x] Catch mkdtemp returns NULL qemu_process: Stop locking QMP process monitor immediately [x] Drop patch qemu_monitor: Introduce qemuMonitorCPUModelInfoNew [x] Use qemuMonitorCPUModelInfoNew to set model name [x] Drop variable initialization [x] Use qemuMonitorCPUModelInfoNew in qemuMonitorJSONGetCPUModelExpansion (Pull forward) [ ] Remove orig NULL check (Unit tests fail if I do this) qemu_monitor: Introduce qemuMonitorCPUModelInfo / JSON conversion [x] Make variable initialization easier to read in PropAdd [x] Require at least one "prop" (don't allow zero in this commit) qemu_capabilities: Introduce virQEMuCapsMigratablePropsDiff [x] Remove template txt in commit message [x] Change function name [x] Distinguish nmProp from mProp [x] Don't set cpuData->info to NULL [ ] Don't use tmp (defer to next patch set... out of time) qemu_monitor: qemuMonitorGetCPUModelExpansion inputs and outputs CPUModelInfo [x] Remove extra line from commit message [x] Change parameter order [x] Don't use VIR_STEAL_PTR qemu_capabilities: Introduce CPUModelInfo to virCPUDef function [x] Change comment [x] Change parameter order [x] Separate lines [x] Move debug message before alloc [x] Back port function spacing fix from next patch qemu_capabilities: Introduce virCPUDef to CPUModelInfo function [x] Change comment [x] Move debug message before alloc [x] Remove NULL check on feature->name [x] Document prop value semantics Reviewed in v5 but no changes required: qemu_process: Limit qemuProcessQmpNew to const input strings qemu_process: Use consistent name for stop process function qemu_process: Persist stderr in qemuProcessQmp struct qemu_process: Store libDir in qemuProcessQmp struct qemu_process: Setup paths within qemuProcessQmpInit qemu_process: Stop retaining Monitor config in qemuProcessQmp Not reviewed in v5: qemu_monitor: Support query-cpu-model-baseline QMP command qemu_driver: Consolidate code to baseline using libvirt qemu_driver: Decouple code for baseline using libvirt qemu_driver: Identify using libvirt as a distinct way to compute baseline qemu_driver: Support baseline calculation using QEMU qemu_driver: Support feature expansion via QEMU when baselining cpu qemu_driver: Remove unsupported props in expanded hypervisor baseline output qemu_monitor: Default props to migratable when expanding cpu model ----- v5: Fixes all process issues identified here: https://www.redhat.com/archives/libvir-list/2018-November/msg00349.html - Make the process code generic (not capabilities specific) for use by BaselineHypervisorCPU - Many of the process patches are simple code moves with implementation changes in other distinct patches - A thread safe library function creates a unique directory under libDir for each QEMU process (for QMP messaging) to decouple processes in terms of sockets and file system footprint. The remaining (non-process) patches in v4 address all issues in v1-v4 of 'BaselineHypervisorCPU using QEMU QMP exchanges' Thanks, Chris *** BLURB HERE *** Chris Venteicher (33): qEmu_process: Move process code from qemu_capabilities to qemu_process qemu_process: Use qemuProcessQMP prefix qemu_process: Limit qemuProcessQMPNew to const input strings qemu_process: Refer to proc not cmd in process code qemu_process: Use consistent name for stop process function qemu_capabilities: Stop QEMU process before freeing qemu_process: Use qemuProcessQMP struct for a single process qemu_process: All ProcessQMP errors are fatal qemu_process: Expose process exit status code qemu_process: Persist stderr in qemuProcessQMP struct qemu_process: Introduce qemuProcessQMPStart qemu_process: Collect monitor code in single function qemu_process: Store libDir in qemuProcessQMP struct qemu_process: Setup paths within qemuProcessQMPInit qemu_process: Stop retaining Monitor config in qemuProcessQMP qemu_process: Cleanup qemuProcessQMP alloc function qemu_process: Cleanup qemuProcessQMPStop function qemu_process: Catch process free before process stop qemu_process: Enter QMP command mode when starting QEMU Process qemu_process: Use unique directories for QMP processes qemu_monitor: Introduce qemuMonitorCPUModelInfoNew qemu_monitor: Introduce qemuMonitorCPUModelInfo / JSON conversion qemu_capabilities: Introduce virQEMuCapsMigratablePropsCalc qemu_monitor: qemuMonitorGetCPUModelExpansion inputs and outputs CPUModelInfo qemu_capabilities: Introduce CPUModelInfo to virCPUDef function qemu_capabilities: Introduce virCPUDef to CPUModelInfo function qemu_monitor: Support query-cpu-model-baseline QMP command qemu_driver: Consolidate code to baseline using libvirt qemu_driver: Decouple code for baseline using libvirt qemu_driver: Identify using libvirt as a distinct way to compute baseline qemu_driver: Support baseline calculation using QEMU qemu_driver: Support feature expansion via QEMU when baselining cpu qemu_monitor: Default props to migratable when expanding cpu model src/qemu/qemu_capabilities.c | 631 ++++++++---------- src/qemu/qemu_capabilities.h | 4 + src/qemu/qemu_driver.c | 216 +++++- src/qemu/qemu_monitor.c | 165 ++++- src/qemu/qemu_monitor.h | 29 +- src/qemu/qemu_monitor_json.c | 223 +++++-- src/qemu/qemu_monitor_json.h | 10 +- src/qemu/qemu_process.c | 345 ++++++++++ src/qemu/qemu_process.h | 32 + tests/cputest.c | 11 +- .../caps_2.10.0.s390x.xml | 60 +- .../caps_2.11.0.s390x.xml | 58 +- .../caps_2.12.0.s390x.xml | 56 +- .../qemucapabilitiesdata/caps_2.8.0.s390x.xml | 32 +- .../qemucapabilitiesdata/caps_2.9.0.s390x.xml | 34 +- .../qemucapabilitiesdata/caps_3.0.0.s390x.xml | 64 +- tests/qemucapabilitiestest.c | 7 + 17 files changed, 1375 insertions(+), 602 deletions(-) -- 2.17.1

Qemu process code in qemu_capabilities.c is moved to qemu_process.c in order to make the code usable outside the original capabilities usecases. The moved code activates and manages Qemu processes without establishing a guest domain. ** This patch is a straight cut/paste move between files. ** Then, subsequent patches modify the process code to make function prefixes and variable names match qemu_process, and make the code usable for more than the capabilities usecase. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 218 +---------------------------------- src/qemu/qemu_process.c | 199 ++++++++++++++++++++++++++++++++ src/qemu/qemu_process.h | 30 +++++ 3 files changed, 230 insertions(+), 217 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index f504db7d05..acd88d990e 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -45,6 +45,7 @@ #define LIBVIRT_QEMU_CAPSPRIV_H_ALLOW #include "qemu_capspriv.h" #include "qemu_qapi.h" +#include "qemu_process.h" #include <fcntl.h> #include <sys/stat.h> @@ -4056,18 +4057,6 @@ virQEMUCapsIsValid(void *data, } -static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED, - virDomainObjPtr vm ATTRIBUTE_UNUSED, - void *opaque ATTRIBUTE_UNUSED) -{ -} - -static qemuMonitorCallbacks callbacks = { - .eofNotify = virQEMUCapsMonitorNotify, - .errorNotify = virQEMUCapsMonitorNotify, -}; - - /** * virQEMUCapsInitQMPArch: * @qemuCaps: QEMU capabilities @@ -4363,211 +4352,6 @@ virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED, } -typedef struct _virQEMUCapsInitQMPCommand virQEMUCapsInitQMPCommand; -typedef virQEMUCapsInitQMPCommand *virQEMUCapsInitQMPCommandPtr; -struct _virQEMUCapsInitQMPCommand { - char *binary; - uid_t runUid; - gid_t runGid; - char **qmperr; - char *monarg; - char *monpath; - char *pidfile; - virCommandPtr cmd; - qemuMonitorPtr mon; - virDomainChrSourceDef config; - pid_t pid; - virDomainObjPtr vm; -}; - - -static void -virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd) -{ - if (cmd->mon) - virObjectUnlock(cmd->mon); - qemuMonitorClose(cmd->mon); - cmd->mon = NULL; - - virCommandAbort(cmd->cmd); - virCommandFree(cmd->cmd); - cmd->cmd = NULL; - - if (cmd->monpath) - unlink(cmd->monpath); - - virDomainObjEndAPI(&cmd->vm); - - if (cmd->pid != 0) { - char ebuf[1024]; - - VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid); - if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH) - VIR_ERROR(_("Failed to kill process %lld: %s"), - (long long)cmd->pid, - virStrerror(errno, ebuf, sizeof(ebuf))); - - VIR_FREE(*cmd->qmperr); - } - if (cmd->pidfile) - unlink(cmd->pidfile); - cmd->pid = 0; -} - - -static void -virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd) -{ - if (!cmd) - return; - - virQEMUCapsInitQMPCommandAbort(cmd); - VIR_FREE(cmd->binary); - VIR_FREE(cmd->monpath); - VIR_FREE(cmd->monarg); - VIR_FREE(cmd->pidfile); - VIR_FREE(cmd); -} - - -static virQEMUCapsInitQMPCommandPtr -virQEMUCapsInitQMPCommandNew(char *binary, - const char *libDir, - uid_t runUid, - gid_t runGid, - char **qmperr) -{ - virQEMUCapsInitQMPCommandPtr cmd = NULL; - - if (VIR_ALLOC(cmd) < 0) - goto error; - - if (VIR_STRDUP(cmd->binary, binary) < 0) - goto error; - - cmd->runUid = runUid; - cmd->runGid = runGid; - cmd->qmperr = qmperr; - - /* the ".sock" sufix is important to avoid a possible clash with a qemu - * domain called "capabilities" - */ - if (virAsprintf(&cmd->monpath, "%s/%s", libDir, - "capabilities.monitor.sock") < 0) - goto error; - if (virAsprintf(&cmd->monarg, "unix:%s,server,nowait", cmd->monpath) < 0) - goto error; - - /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash - * with a qemu domain called "capabilities" - * Normally we'd use runDir for pid files, but because we're using - * -daemonize we need QEMU to be allowed to create them, rather - * than libvirtd. So we're using libDir which QEMU can write to - */ - if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0) - goto error; - - virPidFileForceCleanupPath(cmd->pidfile); - - cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; - cmd->config.data.nix.path = cmd->monpath; - cmd->config.data.nix.listen = false; - - return cmd; - - error: - virQEMUCapsInitQMPCommandFree(cmd); - return NULL; -} - - -/* Returns -1 on fatal error, - * 0 on success, - * 1 when probing QEMU failed - */ -static int -virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, - bool forceTCG) -{ - virDomainXMLOptionPtr xmlopt = NULL; - const char *machine; - int status = 0; - int ret = -1; - - if (forceTCG) - machine = "none,accel=tcg"; - else - machine = "none,accel=kvm:tcg"; - - VIR_DEBUG("Try to probe capabilities of '%s' via QMP, machine %s", - cmd->binary, machine); - - /* - * We explicitly need to use -daemonize here, rather than - * virCommandDaemonize, because we need to synchronize - * with QEMU creating its monitor socket API. Using - * daemonize guarantees control won't return to libvirt - * until the socket is present. - */ - cmd->cmd = virCommandNewArgList(cmd->binary, - "-S", - "-no-user-config", - "-nodefaults", - "-nographic", - "-machine", machine, - "-qmp", cmd->monarg, - "-pidfile", cmd->pidfile, - "-daemonize", - NULL); - virCommandAddEnvPassCommon(cmd->cmd); - virCommandClearCaps(cmd->cmd); - virCommandSetGID(cmd->cmd, cmd->runGid); - virCommandSetUID(cmd->cmd, cmd->runUid); - - virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr); - - /* Log, but otherwise ignore, non-zero status. */ - if (virCommandRun(cmd->cmd, &status) < 0) - goto cleanup; - - if (status != 0) { - VIR_DEBUG("QEMU %s exited with status %d: %s", - cmd->binary, status, *cmd->qmperr); - goto ignore; - } - - if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) { - VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile); - goto ignore; - } - - if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || - !(cmd->vm = virDomainObjNew(xmlopt))) - goto cleanup; - - cmd->vm->pid = cmd->pid; - - if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true, - 0, &callbacks, NULL))) - goto ignore; - - virObjectLock(cmd->mon); - - ret = 0; - - cleanup: - if (!cmd->mon) - virQEMUCapsInitQMPCommandAbort(cmd); - virObjectUnref(xmlopt); - - return ret; - - ignore: - ret = 1; - goto cleanup; -} - - static int virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, const char *libDir, diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index aad6c12552..9698f5b9a1 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8189,3 +8189,202 @@ qemuProcessReconnectAll(virQEMUDriverPtr driver) struct qemuProcessReconnectData data = {.driver = driver}; virDomainObjListForEach(driver->domains, qemuProcessReconnectHelper, &data); } + + +static void virQEMUCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED, + virDomainObjPtr vm ATTRIBUTE_UNUSED, + void *opaque ATTRIBUTE_UNUSED) +{ +} + +static qemuMonitorCallbacks callbacks = { + .eofNotify = virQEMUCapsMonitorNotify, + .errorNotify = virQEMUCapsMonitorNotify, +}; + + +void +virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd) +{ + if (!cmd) + return; + + virQEMUCapsInitQMPCommandAbort(cmd); + VIR_FREE(cmd->binary); + VIR_FREE(cmd->monpath); + VIR_FREE(cmd->monarg); + VIR_FREE(cmd->pidfile); + VIR_FREE(cmd); +} + + +virQEMUCapsInitQMPCommandPtr +virQEMUCapsInitQMPCommandNew(char *binary, + const char *libDir, + uid_t runUid, + gid_t runGid, + char **qmperr) +{ + virQEMUCapsInitQMPCommandPtr cmd = NULL; + + if (VIR_ALLOC(cmd) < 0) + goto error; + + if (VIR_STRDUP(cmd->binary, binary) < 0) + goto error; + + cmd->runUid = runUid; + cmd->runGid = runGid; + cmd->qmperr = qmperr; + + /* the ".sock" sufix is important to avoid a possible clash with a qemu + * domain called "capabilities" + */ + if (virAsprintf(&cmd->monpath, "%s/%s", libDir, + "capabilities.monitor.sock") < 0) + goto error; + if (virAsprintf(&cmd->monarg, "unix:%s,server,nowait", cmd->monpath) < 0) + goto error; + + /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash + * with a qemu domain called "capabilities" + * Normally we'd use runDir for pid files, but because we're using + * -daemonize we need QEMU to be allowed to create them, rather + * than libvirtd. So we're using libDir which QEMU can write to + */ + if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0) + goto error; + + virPidFileForceCleanupPath(cmd->pidfile); + + cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; + cmd->config.data.nix.path = cmd->monpath; + cmd->config.data.nix.listen = false; + + return cmd; + + error: + virQEMUCapsInitQMPCommandFree(cmd); + return NULL; +} + + +/* Returns -1 on fatal error, + * 0 on success, + * 1 when probing QEMU failed + */ +int +virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, + bool forceTCG) +{ + virDomainXMLOptionPtr xmlopt = NULL; + const char *machine; + int status = 0; + int ret = -1; + + if (forceTCG) + machine = "none,accel=tcg"; + else + machine = "none,accel=kvm:tcg"; + + VIR_DEBUG("Try to probe capabilities of '%s' via QMP, machine %s", + cmd->binary, machine); + + /* + * We explicitly need to use -daemonize here, rather than + * virCommandDaemonize, because we need to synchronize + * with QEMU creating its monitor socket API. Using + * daemonize guarantees control won't return to libvirt + * until the socket is present. + */ + cmd->cmd = virCommandNewArgList(cmd->binary, + "-S", + "-no-user-config", + "-nodefaults", + "-nographic", + "-machine", machine, + "-qmp", cmd->monarg, + "-pidfile", cmd->pidfile, + "-daemonize", + NULL); + virCommandAddEnvPassCommon(cmd->cmd); + virCommandClearCaps(cmd->cmd); + virCommandSetGID(cmd->cmd, cmd->runGid); + virCommandSetUID(cmd->cmd, cmd->runUid); + + virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr); + + /* Log, but otherwise ignore, non-zero status. */ + if (virCommandRun(cmd->cmd, &status) < 0) + goto cleanup; + + if (status != 0) { + VIR_DEBUG("QEMU %s exited with status %d: %s", + cmd->binary, status, *cmd->qmperr); + goto ignore; + } + + if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) { + VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile); + goto ignore; + } + + if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || + !(cmd->vm = virDomainObjNew(xmlopt))) + goto cleanup; + + cmd->vm->pid = cmd->pid; + + if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true, + 0, &callbacks, NULL))) + goto ignore; + + virObjectLock(cmd->mon); + + ret = 0; + + cleanup: + if (!cmd->mon) + virQEMUCapsInitQMPCommandAbort(cmd); + virObjectUnref(xmlopt); + + return ret; + + ignore: + ret = 1; + goto cleanup; +} + + +void +virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd) +{ + if (cmd->mon) + virObjectUnlock(cmd->mon); + qemuMonitorClose(cmd->mon); + cmd->mon = NULL; + + virCommandAbort(cmd->cmd); + virCommandFree(cmd->cmd); + cmd->cmd = NULL; + + if (cmd->monpath) + unlink(cmd->monpath); + + virDomainObjEndAPI(&cmd->vm); + + if (cmd->pid != 0) { + char ebuf[1024]; + + VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid); + if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH) + VIR_ERROR(_("Failed to kill process %lld: %s"), + (long long)cmd->pid, + virStrerror(errno, ebuf, sizeof(ebuf))); + + VIR_FREE(*cmd->qmperr); + } + if (cmd->pidfile) + unlink(cmd->pidfile); + cmd->pid = 0; +} diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 67a4d7139f..f76613a7f7 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -214,4 +214,34 @@ int qemuProcessStartManagedPRDaemon(virDomainObjPtr vm); void qemuProcessKillManagedPRDaemon(virDomainObjPtr vm); +typedef struct _virQEMUCapsInitQMPCommand virQEMUCapsInitQMPCommand; +typedef virQEMUCapsInitQMPCommand *virQEMUCapsInitQMPCommandPtr; +struct _virQEMUCapsInitQMPCommand { + char *binary; + uid_t runUid; + gid_t runGid; + char **qmperr; + char *monarg; + char *monpath; + char *pidfile; + virCommandPtr cmd; + qemuMonitorPtr mon; + virDomainChrSourceDef config; + pid_t pid; + virDomainObjPtr vm; +}; + +virQEMUCapsInitQMPCommandPtr virQEMUCapsInitQMPCommandNew(char *binary, + const char *libDir, + uid_t runUid, + gid_t runGid, + char **qmperr); + +void virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd); + +int virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, + bool forceTCG); + +void virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd); + #endif /* LIBVIRT_QEMU_PROCESS_H */ -- 2.17.1

s/virQEMUCapsInitQMPCommand/qemuProcessQMP/ No functionality change. Use file appropriate prefix in moved code. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 14 +++++++------- src/qemu/qemu_process.c | 28 ++++++++++++++-------------- src/qemu/qemu_process.h | 24 ++++++++++++------------ 3 files changed, 33 insertions(+), 33 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index acd88d990e..60ad8ad68d 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4359,15 +4359,15 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, gid_t runGid, char **qmperr) { - virQEMUCapsInitQMPCommandPtr cmd = NULL; + qemuProcessQMPPtr cmd = NULL; int ret = -1; int rc; - if (!(cmd = virQEMUCapsInitQMPCommandNew(qemuCaps->binary, libDir, - runUid, runGid, qmperr))) + if (!(cmd = qemuProcessQMPNew(qemuCaps->binary, libDir, + runUid, runGid, qmperr))) goto cleanup; - if ((rc = virQEMUCapsInitQMPCommandRun(cmd, false)) != 0) { + if ((rc = qemuProcessQMPRun(cmd, false)) != 0) { if (rc == 1) ret = 0; goto cleanup; @@ -4377,8 +4377,8 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, goto cleanup; if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) { - virQEMUCapsInitQMPCommandAbort(cmd); - if ((rc = virQEMUCapsInitQMPCommandRun(cmd, true)) != 0) { + qemuProcessQMPAbort(cmd); + if ((rc = qemuProcessQMPRun(cmd, true)) != 0) { if (rc == 1) ret = 0; goto cleanup; @@ -4391,7 +4391,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, ret = 0; cleanup: - virQEMUCapsInitQMPCommandFree(cmd); + qemuProcessQMPFree(cmd); return ret; } diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 9698f5b9a1..dd3bd441e5 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8204,12 +8204,12 @@ static qemuMonitorCallbacks callbacks = { void -virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd) +qemuProcessQMPFree(qemuProcessQMPPtr cmd) { if (!cmd) return; - virQEMUCapsInitQMPCommandAbort(cmd); + qemuProcessQMPAbort(cmd); VIR_FREE(cmd->binary); VIR_FREE(cmd->monpath); VIR_FREE(cmd->monarg); @@ -8218,14 +8218,14 @@ virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd) } -virQEMUCapsInitQMPCommandPtr -virQEMUCapsInitQMPCommandNew(char *binary, - const char *libDir, - uid_t runUid, - gid_t runGid, - char **qmperr) +qemuProcessQMPPtr +qemuProcessQMPNew(char *binary, + const char *libDir, + uid_t runUid, + gid_t runGid, + char **qmperr) { - virQEMUCapsInitQMPCommandPtr cmd = NULL; + qemuProcessQMPPtr cmd = NULL; if (VIR_ALLOC(cmd) < 0) goto error; @@ -8264,7 +8264,7 @@ virQEMUCapsInitQMPCommandNew(char *binary, return cmd; error: - virQEMUCapsInitQMPCommandFree(cmd); + qemuProcessQMPFree(cmd); return NULL; } @@ -8274,8 +8274,8 @@ virQEMUCapsInitQMPCommandNew(char *binary, * 1 when probing QEMU failed */ int -virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, - bool forceTCG) +qemuProcessQMPRun(qemuProcessQMPPtr cmd, + bool forceTCG) { virDomainXMLOptionPtr xmlopt = NULL; const char *machine; @@ -8345,7 +8345,7 @@ virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, cleanup: if (!cmd->mon) - virQEMUCapsInitQMPCommandAbort(cmd); + qemuProcessQMPAbort(cmd); virObjectUnref(xmlopt); return ret; @@ -8357,7 +8357,7 @@ virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, void -virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd) +qemuProcessQMPAbort(qemuProcessQMPPtr cmd) { if (cmd->mon) virObjectUnlock(cmd->mon); diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index f76613a7f7..4843210f76 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -214,9 +214,9 @@ int qemuProcessStartManagedPRDaemon(virDomainObjPtr vm); void qemuProcessKillManagedPRDaemon(virDomainObjPtr vm); -typedef struct _virQEMUCapsInitQMPCommand virQEMUCapsInitQMPCommand; -typedef virQEMUCapsInitQMPCommand *virQEMUCapsInitQMPCommandPtr; -struct _virQEMUCapsInitQMPCommand { +typedef struct _qemuProcessQMP qemuProcessQMP; +typedef qemuProcessQMP *qemuProcessQMPPtr; +struct _qemuProcessQMP { char *binary; uid_t runUid; gid_t runGid; @@ -231,17 +231,17 @@ struct _virQEMUCapsInitQMPCommand { virDomainObjPtr vm; }; -virQEMUCapsInitQMPCommandPtr virQEMUCapsInitQMPCommandNew(char *binary, - const char *libDir, - uid_t runUid, - gid_t runGid, - char **qmperr); +qemuProcessQMPPtr qemuProcessQMPNew(char *binary, + const char *libDir, + uid_t runUid, + gid_t runGid, + char **qmperr); -void virQEMUCapsInitQMPCommandFree(virQEMUCapsInitQMPCommandPtr cmd); +void qemuProcessQMPFree(qemuProcessQMPPtr cmd); -int virQEMUCapsInitQMPCommandRun(virQEMUCapsInitQMPCommandPtr cmd, - bool forceTCG); +int qemuProcessQMPRun(qemuProcessQMPPtr cmd, + bool forceTCG); -void virQEMUCapsInitQMPCommandAbort(virQEMUCapsInitQMPCommandPtr cmd); +void qemuProcessQMPAbort(qemuProcessQMPPtr cmd); #endif /* LIBVIRT_QEMU_PROCESS_H */ -- 2.17.1

Add the const qualifier on non modified strings (string only copied inside qemuProcessQMPNew) so that const strings can be used directly in calls to qemuProcessQMPNew in future patches. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 2 +- src/qemu/qemu_process.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index dd3bd441e5..7e577f8e7b 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8219,7 +8219,7 @@ qemuProcessQMPFree(qemuProcessQMPPtr cmd) qemuProcessQMPPtr -qemuProcessQMPNew(char *binary, +qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 4843210f76..fe79873967 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -231,7 +231,7 @@ struct _qemuProcessQMP { virDomainObjPtr vm; }; -qemuProcessQMPPtr qemuProcessQMPNew(char *binary, +qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, -- 2.17.1

s/cmd/proc/ in process code imported from qemu_capabilities. No functionality is changed. Just variable renaming. Process code imported from qemu_capabilities was oriented around starting a process to issue a single QMP command. Future usecases (ex. baseline, compare) expect to use a single process to issue multiple different QMP commands. This patch changes the variable naming from cmd to proc to put focus on the process being maintained to issue commands. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 18 ++--- src/qemu/qemu_process.c | 140 +++++++++++++++++------------------ src/qemu/qemu_process.h | 4 +- 3 files changed, 81 insertions(+), 81 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 60ad8ad68d..e062e4daa8 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4359,39 +4359,39 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, gid_t runGid, char **qmperr) { - qemuProcessQMPPtr cmd = NULL; + qemuProcessQMPPtr proc = NULL; int ret = -1; int rc; - if (!(cmd = qemuProcessQMPNew(qemuCaps->binary, libDir, - runUid, runGid, qmperr))) + if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir, + runUid, runGid, qmperr))) goto cleanup; - if ((rc = qemuProcessQMPRun(cmd, false)) != 0) { + if ((rc = qemuProcessQMPRun(proc, false)) != 0) { if (rc == 1) ret = 0; goto cleanup; } - if (virQEMUCapsInitQMPMonitor(qemuCaps, cmd->mon) < 0) + if (virQEMUCapsInitQMPMonitor(qemuCaps, proc->mon) < 0) goto cleanup; if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) { - qemuProcessQMPAbort(cmd); - if ((rc = qemuProcessQMPRun(cmd, true)) != 0) { + qemuProcessQMPAbort(proc); + if ((rc = qemuProcessQMPRun(proc, true)) != 0) { if (rc == 1) ret = 0; goto cleanup; } - if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, cmd->mon) < 0) + if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, proc->mon) < 0) goto cleanup; } ret = 0; cleanup: - qemuProcessQMPFree(cmd); + qemuProcessQMPFree(proc); return ret; } diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 7e577f8e7b..e3f793d1e5 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8204,17 +8204,17 @@ static qemuMonitorCallbacks callbacks = { void -qemuProcessQMPFree(qemuProcessQMPPtr cmd) +qemuProcessQMPFree(qemuProcessQMPPtr proc) { - if (!cmd) + if (!proc) return; - qemuProcessQMPAbort(cmd); - VIR_FREE(cmd->binary); - VIR_FREE(cmd->monpath); - VIR_FREE(cmd->monarg); - VIR_FREE(cmd->pidfile); - VIR_FREE(cmd); + qemuProcessQMPAbort(proc); + VIR_FREE(proc->binary); + VIR_FREE(proc->monpath); + VIR_FREE(proc->monarg); + VIR_FREE(proc->pidfile); + VIR_FREE(proc); } @@ -8225,25 +8225,25 @@ qemuProcessQMPNew(const char *binary, gid_t runGid, char **qmperr) { - qemuProcessQMPPtr cmd = NULL; + qemuProcessQMPPtr proc = NULL; - if (VIR_ALLOC(cmd) < 0) + if (VIR_ALLOC(proc) < 0) goto error; - if (VIR_STRDUP(cmd->binary, binary) < 0) + if (VIR_STRDUP(proc->binary, binary) < 0) goto error; - cmd->runUid = runUid; - cmd->runGid = runGid; - cmd->qmperr = qmperr; + proc->runUid = runUid; + proc->runGid = runGid; + proc->qmperr = qmperr; /* the ".sock" sufix is important to avoid a possible clash with a qemu * domain called "capabilities" */ - if (virAsprintf(&cmd->monpath, "%s/%s", libDir, + if (virAsprintf(&proc->monpath, "%s/%s", libDir, "capabilities.monitor.sock") < 0) goto error; - if (virAsprintf(&cmd->monarg, "unix:%s,server,nowait", cmd->monpath) < 0) + if (virAsprintf(&proc->monarg, "unix:%s,server,nowait", proc->monpath) < 0) goto error; /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash @@ -8252,19 +8252,19 @@ qemuProcessQMPNew(const char *binary, * -daemonize we need QEMU to be allowed to create them, rather * than libvirtd. So we're using libDir which QEMU can write to */ - if (virAsprintf(&cmd->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0) + if (virAsprintf(&proc->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0) goto error; - virPidFileForceCleanupPath(cmd->pidfile); + virPidFileForceCleanupPath(proc->pidfile); - cmd->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; - cmd->config.data.nix.path = cmd->monpath; - cmd->config.data.nix.listen = false; + proc->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; + proc->config.data.nix.path = proc->monpath; + proc->config.data.nix.listen = false; - return cmd; + return proc; error: - qemuProcessQMPFree(cmd); + qemuProcessQMPFree(proc); return NULL; } @@ -8274,7 +8274,7 @@ qemuProcessQMPNew(const char *binary, * 1 when probing QEMU failed */ int -qemuProcessQMPRun(qemuProcessQMPPtr cmd, +qemuProcessQMPRun(qemuProcessQMPPtr proc, bool forceTCG) { virDomainXMLOptionPtr xmlopt = NULL; @@ -8288,7 +8288,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr cmd, machine = "none,accel=kvm:tcg"; VIR_DEBUG("Try to probe capabilities of '%s' via QMP, machine %s", - cmd->binary, machine); + proc->binary, machine); /* * We explicitly need to use -daemonize here, rather than @@ -8297,55 +8297,55 @@ qemuProcessQMPRun(qemuProcessQMPPtr cmd, * daemonize guarantees control won't return to libvirt * until the socket is present. */ - cmd->cmd = virCommandNewArgList(cmd->binary, - "-S", - "-no-user-config", - "-nodefaults", - "-nographic", - "-machine", machine, - "-qmp", cmd->monarg, - "-pidfile", cmd->pidfile, - "-daemonize", - NULL); - virCommandAddEnvPassCommon(cmd->cmd); - virCommandClearCaps(cmd->cmd); - virCommandSetGID(cmd->cmd, cmd->runGid); - virCommandSetUID(cmd->cmd, cmd->runUid); + proc->cmd = virCommandNewArgList(proc->binary, + "-S", + "-no-user-config", + "-nodefaults", + "-nographic", + "-machine", machine, + "-qmp", proc->monarg, + "-pidfile", proc->pidfile, + "-daemonize", + NULL); + virCommandAddEnvPassCommon(proc->cmd); + virCommandClearCaps(proc->cmd); + virCommandSetGID(proc->cmd, proc->runGid); + virCommandSetUID(proc->cmd, proc->runUid); - virCommandSetErrorBuffer(cmd->cmd, cmd->qmperr); + virCommandSetErrorBuffer(proc->cmd, proc->qmperr); /* Log, but otherwise ignore, non-zero status. */ - if (virCommandRun(cmd->cmd, &status) < 0) + if (virCommandRun(proc->cmd, &status) < 0) goto cleanup; if (status != 0) { VIR_DEBUG("QEMU %s exited with status %d: %s", - cmd->binary, status, *cmd->qmperr); + proc->binary, status, *proc->qmperr); goto ignore; } - if (virPidFileReadPath(cmd->pidfile, &cmd->pid) < 0) { - VIR_DEBUG("Failed to read pidfile %s", cmd->pidfile); + if (virPidFileReadPath(proc->pidfile, &proc->pid) < 0) { + VIR_DEBUG("Failed to read pidfile %s", proc->pidfile); goto ignore; } if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || - !(cmd->vm = virDomainObjNew(xmlopt))) + !(proc->vm = virDomainObjNew(xmlopt))) goto cleanup; - cmd->vm->pid = cmd->pid; + proc->vm->pid = proc->pid; - if (!(cmd->mon = qemuMonitorOpen(cmd->vm, &cmd->config, true, true, - 0, &callbacks, NULL))) + if (!(proc->mon = qemuMonitorOpen(proc->vm, &proc->config, true, true, + 0, &callbacks, NULL))) goto ignore; - virObjectLock(cmd->mon); + virObjectLock(proc->mon); ret = 0; cleanup: - if (!cmd->mon) - qemuProcessQMPAbort(cmd); + if (!proc->mon) + qemuProcessQMPAbort(proc); virObjectUnref(xmlopt); return ret; @@ -8357,34 +8357,34 @@ qemuProcessQMPRun(qemuProcessQMPPtr cmd, void -qemuProcessQMPAbort(qemuProcessQMPPtr cmd) +qemuProcessQMPAbort(qemuProcessQMPPtr proc) { - if (cmd->mon) - virObjectUnlock(cmd->mon); - qemuMonitorClose(cmd->mon); - cmd->mon = NULL; + if (proc->mon) + virObjectUnlock(proc->mon); + qemuMonitorClose(proc->mon); + proc->mon = NULL; - virCommandAbort(cmd->cmd); - virCommandFree(cmd->cmd); - cmd->cmd = NULL; + virCommandAbort(proc->cmd); + virCommandFree(proc->cmd); + proc->cmd = NULL; - if (cmd->monpath) - unlink(cmd->monpath); + if (proc->monpath) + unlink(proc->monpath); - virDomainObjEndAPI(&cmd->vm); + virDomainObjEndAPI(&proc->vm); - if (cmd->pid != 0) { + if (proc->pid != 0) { char ebuf[1024]; - VIR_DEBUG("Killing QMP caps process %lld", (long long)cmd->pid); - if (virProcessKill(cmd->pid, SIGKILL) < 0 && errno != ESRCH) + VIR_DEBUG("Killing QMP caps process %lld", (long long)proc->pid); + if (virProcessKill(proc->pid, SIGKILL) < 0 && errno != ESRCH) VIR_ERROR(_("Failed to kill process %lld: %s"), - (long long)cmd->pid, + (long long)proc->pid, virStrerror(errno, ebuf, sizeof(ebuf))); - VIR_FREE(*cmd->qmperr); + VIR_FREE(*proc->qmperr); } - if (cmd->pidfile) - unlink(cmd->pidfile); - cmd->pid = 0; + if (proc->pidfile) + unlink(proc->pidfile); + proc->pid = 0; } diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index fe79873967..05bcf4f12b 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -237,11 +237,11 @@ qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, gid_t runGid, char **qmperr); -void qemuProcessQMPFree(qemuProcessQMPPtr cmd); +void qemuProcessQMPFree(qemuProcessQMPPtr proc); int qemuProcessQMPRun(qemuProcessQMPPtr cmd, bool forceTCG); -void qemuProcessQMPAbort(qemuProcessQMPPtr cmd); +void qemuProcessQMPAbort(qemuProcessQMPPtr proc); #endif /* LIBVIRT_QEMU_PROCESS_H */ -- 2.17.1

s/qemuProcessQMPAbort/qemuProcessQMPStop/ applied to change function name used to stop QEMU processes in process code moved from qemu_capabilities. No functionality change. The new name, qemuProcessQMPStop, is consistent with the existing function qemuProcessStop used to stop Domain processes. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 2 +- src/qemu/qemu_process.c | 6 +++--- src/qemu/qemu_process.h | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index e062e4daa8..2bacd5e40a 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4377,7 +4377,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, goto cleanup; if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) { - qemuProcessQMPAbort(proc); + qemuProcessQMPStop(proc); if ((rc = qemuProcessQMPRun(proc, true)) != 0) { if (rc == 1) ret = 0; diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index e3f793d1e5..2ab1319ab9 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8209,7 +8209,7 @@ qemuProcessQMPFree(qemuProcessQMPPtr proc) if (!proc) return; - qemuProcessQMPAbort(proc); + qemuProcessQMPStop(proc); VIR_FREE(proc->binary); VIR_FREE(proc->monpath); VIR_FREE(proc->monarg); @@ -8345,7 +8345,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc, cleanup: if (!proc->mon) - qemuProcessQMPAbort(proc); + qemuProcessQMPStop(proc); virObjectUnref(xmlopt); return ret; @@ -8357,7 +8357,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc, void -qemuProcessQMPAbort(qemuProcessQMPPtr proc) +qemuProcessQMPStop(qemuProcessQMPPtr proc) { if (proc->mon) virObjectUnlock(proc->mon); diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 05bcf4f12b..28186a256f 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -242,6 +242,6 @@ void qemuProcessQMPFree(qemuProcessQMPPtr proc); int qemuProcessQMPRun(qemuProcessQMPPtr cmd, bool forceTCG); -void qemuProcessQMPAbort(qemuProcessQMPPtr proc); +void qemuProcessQMPStop(qemuProcessQMPPtr proc); #endif /* LIBVIRT_QEMU_PROCESS_H */ -- 2.17.1

virQEMUCapsInitQMP now stops QEMU process in all execution paths, before freeing the process structure. The qemuProcessQMPStop function can be called multiple times without problems... Won't attempt to stop processes and free resources multiple times. Follow the convention established in qemu_process of 1) alloc process structure 2) start process 3) use process 4) stop process 5) free process data structure The process data structure persists after the process activation fails or the process dies or is killed so stderr strings can be retrieved until the process data structure is freed. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 1 + src/qemu/qemu_process.c | 22 +++++++++++++++------- 2 files changed, 16 insertions(+), 7 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 2bacd5e40a..73380fdc2b 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4391,6 +4391,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, ret = 0; cleanup: + qemuProcessQMPStop(proc); qemuProcessQMPFree(proc); return ret; } diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 2ab1319ab9..2208ff72b2 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8359,14 +8359,20 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc, void qemuProcessQMPStop(qemuProcessQMPPtr proc) { - if (proc->mon) + if (!proc) + return; + + if (proc->mon) { virObjectUnlock(proc->mon); - qemuMonitorClose(proc->mon); - proc->mon = NULL; + qemuMonitorClose(proc->mon); + proc->mon = NULL; + } - virCommandAbort(proc->cmd); - virCommandFree(proc->cmd); - proc->cmd = NULL; + if (proc->cmd) { + virCommandAbort(proc->cmd); + virCommandFree(proc->cmd); + proc->cmd = NULL; + } if (proc->monpath) unlink(proc->monpath); @@ -8383,8 +8389,10 @@ qemuProcessQMPStop(qemuProcessQMPPtr proc) virStrerror(errno, ebuf, sizeof(ebuf))); VIR_FREE(*proc->qmperr); + + proc->pid = 0; } + if (proc->pidfile) unlink(proc->pidfile); - proc->pid = 0; } -- 2.17.1

In new process code, move from model where qemuProcessQMP struct can be used to activate a series of Qemu processes to model where one qemuProcessQMP struct is used for one and only one Qemu process. By allowing only one process activation per qemuProcessQMP struct, the struct can safely store process outputs like status and stderr, without being overwritten, until qemuProcessQMPFree is called. By doing this, process outputs like status and stderr can remain stored in the qemuProcessQMP struct without being overwritten by subsequent process activations. The forceTCG parameter (use / don't use KVM) will be passed when the qemuProcessQMP struct is initialized since the qemuProcessQMP struct won't be reused. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 19 +++++++++++++++---- src/qemu/qemu_process.c | 9 +++++---- src/qemu/qemu_process.h | 7 ++++--- 3 files changed, 24 insertions(+), 11 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 73380fdc2b..35aac798d9 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4360,14 +4360,15 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, char **qmperr) { qemuProcessQMPPtr proc = NULL; + qemuProcessQMPPtr procTCG = NULL; int ret = -1; int rc; if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir, - runUid, runGid, qmperr))) + runUid, runGid, qmperr, false))) goto cleanup; - if ((rc = qemuProcessQMPRun(proc, false)) != 0) { + if ((rc = qemuProcessQMPRun(proc)) != 0) { if (rc == 1) ret = 0; goto cleanup; @@ -4377,14 +4378,22 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, goto cleanup; if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) { + /* The second QEMU process probes for TCG capabilities + * in case the first process reported KVM as enabled + * (otherwise the first one already reported TCG capabilities). */ + qemuProcessQMPStop(proc); - if ((rc = qemuProcessQMPRun(proc, true)) != 0) { + + procTCG = qemuProcessQMPNew(qemuCaps->binary, libDir, + runUid, runGid, NULL, true); + + if ((rc = qemuProcessQMPRun(procTCG)) != 0) { if (rc == 1) ret = 0; goto cleanup; } - if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, proc->mon) < 0) + if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, procTCG->mon) < 0) goto cleanup; } @@ -4392,7 +4401,9 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, cleanup: qemuProcessQMPStop(proc); + qemuProcessQMPStop(procTCG); qemuProcessQMPFree(proc); + qemuProcessQMPFree(procTCG); return ret; } diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 2208ff72b2..297a542e3d 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8223,7 +8223,8 @@ qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, - char **qmperr) + char **qmperr, + bool forceTCG) { qemuProcessQMPPtr proc = NULL; @@ -8236,6 +8237,7 @@ qemuProcessQMPNew(const char *binary, proc->runUid = runUid; proc->runGid = runGid; proc->qmperr = qmperr; + proc->forceTCG = forceTCG; /* the ".sock" sufix is important to avoid a possible clash with a qemu * domain called "capabilities" @@ -8274,15 +8276,14 @@ qemuProcessQMPNew(const char *binary, * 1 when probing QEMU failed */ int -qemuProcessQMPRun(qemuProcessQMPPtr proc, - bool forceTCG) +qemuProcessQMPRun(qemuProcessQMPPtr proc) { virDomainXMLOptionPtr xmlopt = NULL; const char *machine; int status = 0; int ret = -1; - if (forceTCG) + if (proc->forceTCG) machine = "none,accel=tcg"; else machine = "none,accel=kvm:tcg"; diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 28186a256f..ccbdb33679 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -229,18 +229,19 @@ struct _qemuProcessQMP { virDomainChrSourceDef config; pid_t pid; virDomainObjPtr vm; + bool forceTCG; }; qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, - char **qmperr); + char **qmperr, + bool forceTCG); void qemuProcessQMPFree(qemuProcessQMPPtr proc); -int qemuProcessQMPRun(qemuProcessQMPPtr cmd, - bool forceTCG); +int qemuProcessQMPRun(qemuProcessQMPPtr cmd); void qemuProcessQMPStop(qemuProcessQMPPtr proc); -- 2.17.1

In the past capabilities could be determined in other ways if QMP messaging didn't succeed so a non-fatal error case was included in the capabilities and QMP Process code. For a while now, QMP capabilities failure has been a fatal case. This patch makes QMP process failures return as a fatal error in all cases consistent with 1) all failures actually being fatal in QMP capabilities code and 2) the QMP process code being made generic. The process changes impact the capabilities code because non-fatal return codes are no longer returned. The rest of the QMP associated capabilities code is updated to make all errors fatal for consistency. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 14 ++------------ src/qemu/qemu_process.c | 15 +++------------ 2 files changed, 5 insertions(+), 24 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 35aac798d9..502bac5e43 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4155,7 +4155,6 @@ virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps, if (qemuMonitorSetCapabilities(mon) < 0) { VIR_DEBUG("Failed to set monitor capabilities %s", virGetLastErrorMessage()); - ret = 0; goto cleanup; } @@ -4164,7 +4163,6 @@ virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps, &package) < 0) { VIR_DEBUG("Failed to query monitor version %s", virGetLastErrorMessage()); - ret = 0; goto cleanup; } @@ -4336,7 +4334,6 @@ virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED, if (qemuMonitorSetCapabilities(mon) < 0) { VIR_DEBUG("Failed to set monitor capabilities %s", virGetLastErrorMessage()); - ret = 0; goto cleanup; } @@ -4362,17 +4359,13 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, qemuProcessQMPPtr proc = NULL; qemuProcessQMPPtr procTCG = NULL; int ret = -1; - int rc; if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir, runUid, runGid, qmperr, false))) goto cleanup; - if ((rc = qemuProcessQMPRun(proc)) != 0) { - if (rc == 1) - ret = 0; + if (qemuProcessQMPRun(proc) < 0) goto cleanup; - } if (virQEMUCapsInitQMPMonitor(qemuCaps, proc->mon) < 0) goto cleanup; @@ -4387,11 +4380,8 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, procTCG = qemuProcessQMPNew(qemuCaps->binary, libDir, runUid, runGid, NULL, true); - if ((rc = qemuProcessQMPRun(procTCG)) != 0) { - if (rc == 1) - ret = 0; + if (qemuProcessQMPRun(procTCG) < 0) goto cleanup; - } if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, procTCG->mon) < 0) goto cleanup; diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 297a542e3d..f662b7d2ce 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8271,10 +8271,6 @@ qemuProcessQMPNew(const char *binary, } -/* Returns -1 on fatal error, - * 0 on success, - * 1 when probing QEMU failed - */ int qemuProcessQMPRun(qemuProcessQMPPtr proc) { @@ -8315,19 +8311,18 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) virCommandSetErrorBuffer(proc->cmd, proc->qmperr); - /* Log, but otherwise ignore, non-zero status. */ if (virCommandRun(proc->cmd, &status) < 0) goto cleanup; if (status != 0) { VIR_DEBUG("QEMU %s exited with status %d: %s", proc->binary, status, *proc->qmperr); - goto ignore; + goto cleanup; } if (virPidFileReadPath(proc->pidfile, &proc->pid) < 0) { VIR_DEBUG("Failed to read pidfile %s", proc->pidfile); - goto ignore; + goto cleanup; } if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || @@ -8338,7 +8333,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) if (!(proc->mon = qemuMonitorOpen(proc->vm, &proc->config, true, true, 0, &callbacks, NULL))) - goto ignore; + goto cleanup; virObjectLock(proc->mon); @@ -8350,10 +8345,6 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) virObjectUnref(xmlopt); return ret; - - ignore: - ret = 1; - goto cleanup; } -- 2.17.1

Expose the process exit status code so the status code can be used to determine if stderr (qmperr) should be logged. In the case of virQEMUCapsInitQmp, we only log stderr (qmperr) if the -first- query process fails with a non zero status code. Logging functions are pulled inside virQEMUCapsInitQmp where we can detect that the -first- (not subsequent) QEMU process activation failed with nonzero status. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 73 ++++++++++++++++++------------------ src/qemu/qemu_process.c | 9 +++-- src/qemu/qemu_process.h | 1 + 3 files changed, 42 insertions(+), 41 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 502bac5e43..12c4ed2eb5 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4349,23 +4349,49 @@ virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED, } +#define MESSAGE_ID_CAPS_PROBE_FAILURE "8ae2f3fb-2dbe-498e-8fbd-012d40afa361" + +static void +virQEMUCapsLogProbeFailure(const char *binary) +{ + virLogMetadata meta[] = { + { .key = "MESSAGE_ID", .s = MESSAGE_ID_CAPS_PROBE_FAILURE, .iv = 0 }, + { .key = "LIBVIRT_QEMU_BINARY", .s = binary, .iv = 0 }, + { .key = NULL }, + }; + + virLogMessage(&virLogSelf, + VIR_LOG_WARN, + __FILE__, __LINE__, __func__, + meta, + _("Failed to probe capabilities for %s: %s"), + binary, virGetLastErrorMessage()); +} + + static int virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, const char *libDir, uid_t runUid, - gid_t runGid, - char **qmperr) + gid_t runGid) { qemuProcessQMPPtr proc = NULL; qemuProcessQMPPtr procTCG = NULL; + char *qmperr = NULL; int ret = -1; if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir, - runUid, runGid, qmperr, false))) + runUid, runGid, &qmperr, false))) goto cleanup; - if (qemuProcessQMPRun(proc) < 0) + if (qemuProcessQMPRun(proc) < 0) { + if (proc->status != 0) + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Failed to probe QEMU binary with QMP: %s"), + qmperr ? qmperr : _("uknown error")); + goto cleanup; + } if (virQEMUCapsInitQMPMonitor(qemuCaps, proc->mon) < 0) goto cleanup; @@ -4390,34 +4416,19 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, ret = 0; cleanup: + if (ret < 0) + virQEMUCapsLogProbeFailure(qemuCaps->binary); + qemuProcessQMPStop(proc); qemuProcessQMPStop(procTCG); qemuProcessQMPFree(proc); qemuProcessQMPFree(procTCG); + VIR_FREE(qmperr); + return ret; } -#define MESSAGE_ID_CAPS_PROBE_FAILURE "8ae2f3fb-2dbe-498e-8fbd-012d40afa361" - -static void -virQEMUCapsLogProbeFailure(const char *binary) -{ - virLogMetadata meta[] = { - { .key = "MESSAGE_ID", .s = MESSAGE_ID_CAPS_PROBE_FAILURE, .iv = 0 }, - { .key = "LIBVIRT_QEMU_BINARY", .s = binary, .iv = 0 }, - { .key = NULL }, - }; - - virLogMessage(&virLogSelf, - VIR_LOG_WARN, - __FILE__, __LINE__, __func__, - meta, - _("Failed to probe capabilities for %s: %s"), - binary, virGetLastErrorMessage()); -} - - virQEMUCapsPtr virQEMUCapsNewForBinaryInternal(virArch hostArch, const char *binary, @@ -4429,7 +4440,6 @@ virQEMUCapsNewForBinaryInternal(virArch hostArch, { virQEMUCapsPtr qemuCaps; struct stat sb; - char *qmperr = NULL; if (!(qemuCaps = virQEMUCapsNew())) goto error; @@ -4456,18 +4466,8 @@ virQEMUCapsNewForBinaryInternal(virArch hostArch, goto error; } - if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid, &qmperr) < 0) { - virQEMUCapsLogProbeFailure(binary); + if (virQEMUCapsInitQMP(qemuCaps, libDir, runUid, runGid) < 0) goto error; - } - - if (!qemuCaps->usedQMP) { - virReportError(VIR_ERR_INTERNAL_ERROR, - _("Failed to probe QEMU binary with QMP: %s"), - qmperr ? qmperr : _("unknown error")); - virQEMUCapsLogProbeFailure(binary); - goto error; - } qemuCaps->libvirtCtime = virGetSelfLastChanged(); qemuCaps->libvirtVersion = LIBVIR_VERSION_NUMBER; @@ -4485,7 +4485,6 @@ virQEMUCapsNewForBinaryInternal(virArch hostArch, } cleanup: - VIR_FREE(qmperr); return qemuCaps; error: diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index f662b7d2ce..10e7bc5f11 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8276,7 +8276,6 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) { virDomainXMLOptionPtr xmlopt = NULL; const char *machine; - int status = 0; int ret = -1; if (proc->forceTCG) @@ -8311,12 +8310,14 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) virCommandSetErrorBuffer(proc->cmd, proc->qmperr); - if (virCommandRun(proc->cmd, &status) < 0) + proc->status = 0; + + if (virCommandRun(proc->cmd, &proc->status) < 0) goto cleanup; - if (status != 0) { + if (proc->status != 0) { VIR_DEBUG("QEMU %s exited with status %d: %s", - proc->binary, status, *proc->qmperr); + proc->binary, proc->status, *proc->qmperr); goto cleanup; } diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index ccbdb33679..149829df52 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -220,6 +220,7 @@ struct _qemuProcessQMP { char *binary; uid_t runUid; gid_t runGid; + int status; char **qmperr; char *monarg; char *monpath; -- 2.17.1

A qemuProcessQMP struct tracks the entire lifespan of a single QEMU Process including storing error output when the process terminates or activation fails. Error output remains available until qemuProcessQMPFree is called. The qmperr variable is renamed stderr (captures stderr from process.) The stderr buffer no longer needs to be maintained outside of the qemuProcessQMP structure because the structure is used for a single QEMU process and the structures can be maintained as long as required to retrieve the process error info. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 8 +++----- src/qemu/qemu_process.c | 9 +++------ src/qemu/qemu_process.h | 3 +-- 3 files changed, 7 insertions(+), 13 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 12c4ed2eb5..b833e0d11e 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4377,18 +4377,17 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, { qemuProcessQMPPtr proc = NULL; qemuProcessQMPPtr procTCG = NULL; - char *qmperr = NULL; int ret = -1; if (!(proc = qemuProcessQMPNew(qemuCaps->binary, libDir, - runUid, runGid, &qmperr, false))) + runUid, runGid, false))) goto cleanup; if (qemuProcessQMPRun(proc) < 0) { if (proc->status != 0) virReportError(VIR_ERR_INTERNAL_ERROR, _("Failed to probe QEMU binary with QMP: %s"), - qmperr ? qmperr : _("uknown error")); + proc->stderr ? proc->stderr : _("uknown error")); goto cleanup; } @@ -4404,7 +4403,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, qemuProcessQMPStop(proc); procTCG = qemuProcessQMPNew(qemuCaps->binary, libDir, - runUid, runGid, NULL, true); + runUid, runGid, true); if (qemuProcessQMPRun(procTCG) < 0) goto cleanup; @@ -4423,7 +4422,6 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, qemuProcessQMPStop(procTCG); qemuProcessQMPFree(proc); qemuProcessQMPFree(procTCG); - VIR_FREE(qmperr); return ret; } diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 10e7bc5f11..71c7fb58ad 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8214,6 +8214,7 @@ qemuProcessQMPFree(qemuProcessQMPPtr proc) VIR_FREE(proc->monpath); VIR_FREE(proc->monarg); VIR_FREE(proc->pidfile); + VIR_FREE(proc->stderr); VIR_FREE(proc); } @@ -8223,7 +8224,6 @@ qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, - char **qmperr, bool forceTCG) { qemuProcessQMPPtr proc = NULL; @@ -8236,7 +8236,6 @@ qemuProcessQMPNew(const char *binary, proc->runUid = runUid; proc->runGid = runGid; - proc->qmperr = qmperr; proc->forceTCG = forceTCG; /* the ".sock" sufix is important to avoid a possible clash with a qemu @@ -8308,7 +8307,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) virCommandSetGID(proc->cmd, proc->runGid); virCommandSetUID(proc->cmd, proc->runUid); - virCommandSetErrorBuffer(proc->cmd, proc->qmperr); + virCommandSetErrorBuffer(proc->cmd, &(proc->stderr)); proc->status = 0; @@ -8317,7 +8316,7 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) if (proc->status != 0) { VIR_DEBUG("QEMU %s exited with status %d: %s", - proc->binary, proc->status, *proc->qmperr); + proc->binary, proc->status, proc->stderr); goto cleanup; } @@ -8381,8 +8380,6 @@ qemuProcessQMPStop(qemuProcessQMPPtr proc) (long long)proc->pid, virStrerror(errno, ebuf, sizeof(ebuf))); - VIR_FREE(*proc->qmperr); - proc->pid = 0; } diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 149829df52..b0b58f9424 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -221,7 +221,7 @@ struct _qemuProcessQMP { uid_t runUid; gid_t runGid; int status; - char **qmperr; + char *stderr; char *monarg; char *monpath; char *pidfile; @@ -237,7 +237,6 @@ qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, const char *libDir, uid_t runUid, gid_t runGid, - char **qmperr, bool forceTCG); void qemuProcessQMPFree(qemuProcessQMPPtr proc); -- 2.17.1

Move a step closer to the function structure used elsewhere in qemu_process where qemuProcessStart and qemuProcessStop are the exposed functions. qemuProcessQMPStart mirrors qemuProcessStart in calling sub functions to initialize, launch the process and connect the monitor to the QEMU process. static functions qemuProcessQMPInit, qemuProcessQMPLaunch and qemuProcessQMPConnectMonitor are introduced. qemuProcessQMPLaunch is just renamed from qemuProcessQMPRun and encapsulates all of the original code. qemuProcessQMPInit and qemuProcessQMPMonitor are nearly empty functions acting as placeholders for later patches where blocks of semi-complicated code are cut/pasted into these functions without modification (hopefully making review easier.) Looking forward, the patch series ultimately moves the code into this partitioning: - qemuProcessQMPInit Becomes the location of ~25 lines of code to create storage directory, in thread safe way, and initialize paths for monpath, monarg and pidfile. - qemuProcessQMPLaunch Becomes the location of ~48 lines of code used to create and run the QEMU command. - qemuProcessQMPConnectMonitor Becomes the final location of ~58 lines of code used to open and initialize the monitor connection between libvirt and qemu. Three smaller, purpose-identifying, functions of ~60 lines or less seem better than a single large process "start" function of > 130 lines. Being able to compare and contrast between the domain and non-domain versions of process code is useful too. There is some significant overlap between what the non-domain and domain functions do. There is also significant additional functionality in the domain functions that might be useful in the non-domain functions in the future. Possibly there could be sharing between non-domain and domain process code in the future but common code would have to be carefully extracted from the domain process code (not trivial.) Mirroring the domain process code has some value, but partitioning the code into logical chunks of < 60 lines is the main reason for the static functions. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 4 +- src/qemu/qemu_process.c | 93 +++++++++++++++++++++++++++++++++++- src/qemu/qemu_process.h | 2 +- 3 files changed, 94 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index b833e0d11e..36d28ba5ea 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4383,7 +4383,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, runUid, runGid, false))) goto cleanup; - if (qemuProcessQMPRun(proc) < 0) { + if (qemuProcessQMPStart(proc) < 0) { if (proc->status != 0) virReportError(VIR_ERR_INTERNAL_ERROR, _("Failed to probe QEMU binary with QMP: %s"), @@ -4405,7 +4405,7 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, procTCG = qemuProcessQMPNew(qemuCaps->binary, libDir, runUid, runGid, true); - if (qemuProcessQMPRun(procTCG) < 0) + if (qemuProcessQMPStart(procTCG) < 0) goto cleanup; if (virQEMUCapsInitQMPMonitorTCG(qemuCaps, procTCG->mon) < 0) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 71c7fb58ad..3b785d64e5 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8270,8 +8270,27 @@ qemuProcessQMPNew(const char *binary, } -int -qemuProcessQMPRun(qemuProcessQMPPtr proc) +/* Initialize configuration and paths prior to starting QEMU + */ +static int +qemuProcessQMPInit(qemuProcessQMPPtr proc) +{ + int ret = -1; + + VIR_DEBUG("proc=%p, emulator=%s", + proc, proc->binary); + + ret = 0; + + VIR_DEBUG("ret=%i", ret); + return ret; +} + + +/* Launch QEMU Process + */ +static int +qemuProcessQMPLaunch(qemuProcessQMPPtr proc) { virDomainXMLOptionPtr xmlopt = NULL; const char *machine; @@ -8348,6 +8367,76 @@ qemuProcessQMPRun(qemuProcessQMPPtr proc) } +/* Connect Monitor to QEMU Process + */ +static int +qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) +{ + int ret = -1; + + VIR_DEBUG("proc=%p, emulator=%s, proc->pid=%lld", + proc, proc->binary, (long long)proc->pid); + + ret = 0; + + VIR_DEBUG("ret=%i", ret); + return ret; +} + + +/** + * qemuProcessQMPStart: + * @proc: Stores process and connection state + * + * Start and connect to QEMU binary so QMP queries can be made. + * + * Usage: + * proc = qemuProcessQMPNew(binary, libDir, runUid, runGid, forceTCG); + * qemuProcessQMPStart(proc); + * ** Send QMP Queries to QEMU using monitor (proc->mon) ** + * qemuProcessQMPStop(proc); + * qemuProcessQMPFree(proc); + * + * Check monitor is not NULL before using. + * + * QEMU probing failure results in monitor being NULL but is not considered + * an error case and does not result in a negative return value. + * + * Both qemuProcessQMPStop and qemuProcessQMPFree must be called to cleanup and + * free resources, even in activation failure cases. + * + * Process error output (proc->stderr) remains available in qemuProcessQMP + * struct until qemuProcessQMPFree is called. + */ +int +qemuProcessQMPStart(qemuProcessQMPPtr proc) +{ + int ret = -1; + + VIR_DEBUG("proc=%p, emulator=%s", + proc, proc->binary); + + if (qemuProcessQMPInit(proc) < 0) + goto cleanup; + + if (qemuProcessQMPLaunch(proc) < 0) + goto stop; + + if (qemuProcessQMPConnectMonitor(proc) < 0) + goto stop; + + ret = 0; + + cleanup: + VIR_DEBUG("ret=%i", ret); + return ret; + + stop: + qemuProcessQMPStop(proc); + goto cleanup; +} + + void qemuProcessQMPStop(qemuProcessQMPPtr proc) { diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index b0b58f9424..9938dd0a69 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -241,7 +241,7 @@ qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, void qemuProcessQMPFree(qemuProcessQMPPtr proc); -int qemuProcessQMPRun(qemuProcessQMPPtr cmd); +int qemuProcessQMPStart(qemuProcessQMPPtr proc); void qemuProcessQMPStop(qemuProcessQMPPtr proc); -- 2.17.1

qemuMonitor code lives in qemuProcessQMPConnectMonitor rather than in qemuProcessQMPNew and qemuProcessQMPLaunch. This is consistent with existing structure in qemu_process.c where qemuConnectMonitor function contains monitor code for domain process activation. Simple code moves in this patch. Improvements in later patch. Only intended functional change in this patch is we don't move (include) code to initiate process stop on failure to create monitor. As comments in qemuProcessQMPStart say... Client must always call qemuProcessStop and qemuProcessQMPFree, even in error cases. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 48 ++++++++++++++++++++--------------------- 1 file changed, 23 insertions(+), 25 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 3b785d64e5..5492ff5a19 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8258,10 +8258,6 @@ qemuProcessQMPNew(const char *binary, virPidFileForceCleanupPath(proc->pidfile); - proc->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; - proc->config.data.nix.path = proc->monpath; - proc->config.data.nix.listen = false; - return proc; error: @@ -8292,7 +8288,6 @@ qemuProcessQMPInit(qemuProcessQMPPtr proc) static int qemuProcessQMPLaunch(qemuProcessQMPPtr proc) { - virDomainXMLOptionPtr xmlopt = NULL; const char *machine; int ret = -1; @@ -8344,6 +8339,28 @@ qemuProcessQMPLaunch(qemuProcessQMPPtr proc) goto cleanup; } + ret = 0; + + cleanup: + return ret; +} + + +/* Connect Monitor to QEMU Process + */ +static int +qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) +{ + int ret = -1; + virDomainXMLOptionPtr xmlopt = NULL; + + VIR_DEBUG("proc=%p, emulator=%s, proc->pid=%lld", + proc, proc->binary, (long long)proc->pid); + + proc->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; + proc->config.data.nix.path = proc->monpath; + proc->config.data.nix.listen = false; + if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || !(proc->vm = virDomainObjNew(xmlopt))) goto cleanup; @@ -8359,27 +8376,8 @@ qemuProcessQMPLaunch(qemuProcessQMPPtr proc) ret = 0; cleanup: - if (!proc->mon) - qemuProcessQMPStop(proc); + VIR_DEBUG("ret=%i", ret); virObjectUnref(xmlopt); - - return ret; -} - - -/* Connect Monitor to QEMU Process - */ -static int -qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) -{ - int ret = -1; - - VIR_DEBUG("proc=%p, emulator=%s, proc->pid=%lld", - proc, proc->binary, (long long)proc->pid); - - ret = 0; - - VIR_DEBUG("ret=%i", ret); return ret; } -- 2.17.1

Store libDir path in the qemuProcessQMP struct in anticipation of moving path construction code into qemuProcessQMPInit function. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 8 +++++--- src/qemu/qemu_process.h | 1 + 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 5492ff5a19..dc2237f0e6 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8211,6 +8211,7 @@ qemuProcessQMPFree(qemuProcessQMPPtr proc) qemuProcessQMPStop(proc); VIR_FREE(proc->binary); + VIR_FREE(proc->libDir); VIR_FREE(proc->monpath); VIR_FREE(proc->monarg); VIR_FREE(proc->pidfile); @@ -8231,7 +8232,8 @@ qemuProcessQMPNew(const char *binary, if (VIR_ALLOC(proc) < 0) goto error; - if (VIR_STRDUP(proc->binary, binary) < 0) + if (VIR_STRDUP(proc->binary, binary) < 0 || + VIR_STRDUP(proc->libDir, libDir) < 0) goto error; proc->runUid = runUid; @@ -8241,7 +8243,7 @@ qemuProcessQMPNew(const char *binary, /* the ".sock" sufix is important to avoid a possible clash with a qemu * domain called "capabilities" */ - if (virAsprintf(&proc->monpath, "%s/%s", libDir, + if (virAsprintf(&proc->monpath, "%s/%s", proc->libDir, "capabilities.monitor.sock") < 0) goto error; if (virAsprintf(&proc->monarg, "unix:%s,server,nowait", proc->monpath) < 0) @@ -8253,7 +8255,7 @@ qemuProcessQMPNew(const char *binary, * -daemonize we need QEMU to be allowed to create them, rather * than libvirtd. So we're using libDir which QEMU can write to */ - if (virAsprintf(&proc->pidfile, "%s/%s", libDir, "capabilities.pidfile") < 0) + if (virAsprintf(&proc->pidfile, "%s/%s", proc->libDir, "capabilities.pidfile") < 0) goto error; virPidFileForceCleanupPath(proc->pidfile); diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 9938dd0a69..e07584c11c 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -218,6 +218,7 @@ typedef struct _qemuProcessQMP qemuProcessQMP; typedef qemuProcessQMP *qemuProcessQMPPtr; struct _qemuProcessQMP { char *binary; + char *libDir; uid_t runUid; gid_t runGid; int status; -- 2.17.1

Move code for setting paths and prepping file system from qemuProcessQMPNew to qemuProcessQMPInit. This keeps qemuProcessQMPNew limited to data structures and path initialization is done in qemuProcessQMPInit. The patch is a non-functional, cut / paste change, however goto is now "cleanup" rather than "error". Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 44 +++++++++++++++++++++-------------------- 1 file changed, 23 insertions(+), 21 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index dc2237f0e6..430136251d 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8240,14 +8240,33 @@ qemuProcessQMPNew(const char *binary, proc->runGid = runGid; proc->forceTCG = forceTCG; + return proc; + + error: + qemuProcessQMPFree(proc); + return NULL; +} + + +/* Initialize configuration and paths prior to starting QEMU + */ +static int +qemuProcessQMPInit(qemuProcessQMPPtr proc) +{ + int ret = -1; + + VIR_DEBUG("proc=%p, emulator=%s", + proc, proc->binary); + /* the ".sock" sufix is important to avoid a possible clash with a qemu * domain called "capabilities" */ if (virAsprintf(&proc->monpath, "%s/%s", proc->libDir, "capabilities.monitor.sock") < 0) - goto error; + goto cleanup; + if (virAsprintf(&proc->monarg, "unix:%s,server,nowait", proc->monpath) < 0) - goto error; + goto cleanup; /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash * with a qemu domain called "capabilities" @@ -8256,30 +8275,13 @@ qemuProcessQMPNew(const char *binary, * than libvirtd. So we're using libDir which QEMU can write to */ if (virAsprintf(&proc->pidfile, "%s/%s", proc->libDir, "capabilities.pidfile") < 0) - goto error; + goto cleanup; virPidFileForceCleanupPath(proc->pidfile); - return proc; - - error: - qemuProcessQMPFree(proc); - return NULL; -} - - -/* Initialize configuration and paths prior to starting QEMU - */ -static int -qemuProcessQMPInit(qemuProcessQMPPtr proc) -{ - int ret = -1; - - VIR_DEBUG("proc=%p, emulator=%s", - proc, proc->binary); - ret = 0; + cleanup: VIR_DEBUG("ret=%i", ret); return ret; } -- 2.17.1

The monitor config data is removed from the qemuProcessQMP struct. The monitor config data can be initialized immediately before call to qemuMonitorOpen and does not need to be maintained after the call because qemuMonitorOpen copies any strings it needs. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 9 +++++---- src/qemu/qemu_process.h | 1 - 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 430136251d..91532c19ce 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8357,13 +8357,14 @@ qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) { int ret = -1; virDomainXMLOptionPtr xmlopt = NULL; + virDomainChrSourceDef monConfig; VIR_DEBUG("proc=%p, emulator=%s, proc->pid=%lld", proc, proc->binary, (long long)proc->pid); - proc->config.type = VIR_DOMAIN_CHR_TYPE_UNIX; - proc->config.data.nix.path = proc->monpath; - proc->config.data.nix.listen = false; + monConfig.type = VIR_DOMAIN_CHR_TYPE_UNIX; + monConfig.data.nix.path = proc->monpath; + monConfig.data.nix.listen = false; if (!(xmlopt = virDomainXMLOptionNew(NULL, NULL, NULL, NULL, NULL)) || !(proc->vm = virDomainObjNew(xmlopt))) @@ -8371,7 +8372,7 @@ qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) proc->vm->pid = proc->pid; - if (!(proc->mon = qemuMonitorOpen(proc->vm, &proc->config, true, true, + if (!(proc->mon = qemuMonitorOpen(proc->vm, &monConfig, true, true, 0, &callbacks, NULL))) goto cleanup; diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index e07584c11c..0bf594ad2e 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -228,7 +228,6 @@ struct _qemuProcessQMP { char *pidfile; virCommandPtr cmd; qemuMonitorPtr mon; - virDomainChrSourceDef config; pid_t pid; virDomainObjPtr vm; bool forceTCG; -- 2.17.1

qemuProcessQMPNew is one of the 4 public functions used to create and manage a qemu process for QMP command exchanges outside of domain operations. Add descriptive comment block, Debug statement and make source consistent with the cleanup / VIR_STEAL_PTR format used elsewhere. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 29 ++++++++++++++++++++++++----- 1 file changed, 24 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 91532c19ce..58842a0f1c 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8220,6 +8220,18 @@ qemuProcessQMPFree(qemuProcessQMPPtr proc) } +/** + * qemuProcessQMPNew: + * @binary: QEMU binary + * @libDir: Directory for process and connection artifacts + * @runUid: UserId for QEMU Process + * @runGid: GroupId for QEMU Process + * @forceTCG: Force TCG mode if true + * + * Allocate and initialize domain structure encapsulating + * QEMU Process state and monitor connection to QEMU + * for completing QMP Queries. + */ qemuProcessQMPPtr qemuProcessQMPNew(const char *binary, const char *libDir, @@ -8227,24 +8239,31 @@ qemuProcessQMPNew(const char *binary, gid_t runGid, bool forceTCG) { + qemuProcessQMPPtr ret = NULL; qemuProcessQMPPtr proc = NULL; + VIR_DEBUG("exec=%s, libDir=%s, runUid=%u, runGid=%u, forceTCG=%d", + binary, libDir, runUid, runGid, forceTCG); + if (VIR_ALLOC(proc) < 0) - goto error; + goto cleanup; if (VIR_STRDUP(proc->binary, binary) < 0 || VIR_STRDUP(proc->libDir, libDir) < 0) - goto error; + goto cleanup; proc->runUid = runUid; proc->runGid = runGid; proc->forceTCG = forceTCG; - return proc; + VIR_STEAL_PTR(ret, proc); - error: + cleanup: qemuProcessQMPFree(proc); - return NULL; + + VIR_DEBUG("ret=%p", ret); + + return ret; } -- 2.17.1

qemuProcessQMPStop is one of the 4 public functions used to create and manage a Qemu process for QMP command exchanges. Add comment header and debug message. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 58842a0f1c..fb14ed35f9 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8459,12 +8459,21 @@ qemuProcessQMPStart(qemuProcessQMPPtr proc) } +/** + * qemuProcessStop: + * @proc: Stores process and connection state + * + * Stop monitor connection and QEMU process + */ void qemuProcessQMPStop(qemuProcessQMPPtr proc) { if (!proc) return; + VIR_DEBUG("proc=%p emulator=%s mon=%p pid=%lld", + proc, proc->binary, proc->mon, (long long)proc->pid); + if (proc->mon) { virObjectUnlock(proc->mon); qemuMonitorClose(proc->mon); -- 2.17.1

Catch execution paths where qemuProcessQMPFree is called before qemuProcessQMPStop then report error and force stop before proceeding. Also added public function header and debug message. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index fb14ed35f9..15a38144f5 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8203,13 +8203,28 @@ static qemuMonitorCallbacks callbacks = { }; +/** + * qemuProcessQMPFree: + * @proc: Stores process and connection state + * + * Free process data structure. + */ void qemuProcessQMPFree(qemuProcessQMPPtr proc) { if (!proc) return; - qemuProcessQMPStop(proc); + VIR_DEBUG("proc=%p, proc->mon=%p", proc, proc->mon); + + /* This should never be non-NULL if we get here, but just in case... */ + if (proc->mon || proc->pid) { + VIR_ERROR(_("Unexpected QEMU still active during process free" + " emulator: %s, pid: %lld, mon: %p"), + proc->binary, (long long)proc->pid, proc->mon); + qemuProcessQMPStop(proc); + } + VIR_FREE(proc->binary); VIR_FREE(proc->libDir); VIR_FREE(proc->monpath); -- 2.17.1

qemuProcessQMPStart starts a QEMU process and monitor connection that can be used by multiple functions possibly for multiple QMP commands. The QMP exchange to exit capabilities negotiation mode and enter command mode can only be performed once after the monitor connection is established. Move responsibility for entering QMP command mode into the qemuProcessQMP code so multiple functions can issue QMP commands in arbitrary orders. This also simplifies the functions using the connection provided by qemuProcessQMPStart to issue QMP commands. Test code now needs to call qemuMonitorSetCapabilities to send the message to switch to command mode because the test code does not use the qemuProcessQMP command that internally calls qemuMonitorSetCapabilities. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 12 ------------ src/qemu/qemu_process.c | 8 ++++++++ tests/qemucapabilitiestest.c | 7 +++++++ 3 files changed, 15 insertions(+), 12 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 36d28ba5ea..bac8487a78 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -4152,12 +4152,6 @@ virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps, /* @mon is supposed to be locked by callee */ - if (qemuMonitorSetCapabilities(mon) < 0) { - VIR_DEBUG("Failed to set monitor capabilities %s", - virGetLastErrorMessage()); - goto cleanup; - } - if (qemuMonitorGetVersion(mon, &major, &minor, µ, &package) < 0) { @@ -4331,12 +4325,6 @@ virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps ATTRIBUTE_UNUSED, { int ret = -1; - if (qemuMonitorSetCapabilities(mon) < 0) { - VIR_DEBUG("Failed to set monitor capabilities %s", - virGetLastErrorMessage()); - goto cleanup; - } - if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, mon, true) < 0) goto cleanup; diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 15a38144f5..d2d067b1a6 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8412,6 +8412,14 @@ qemuProcessQMPConnectMonitor(qemuProcessQMPPtr proc) virObjectLock(proc->mon); + /* Exit capabilities negotiation mode and enter QEMU command mode + * by issuing qmp_capabilities command to QEMU */ + if (qemuMonitorSetCapabilities(proc->mon) < 0) { + VIR_DEBUG("Failed to set monitor capabilities %s", + virGetLastErrorMessage()); + goto cleanup; + } + ret = 0; cleanup: diff --git a/tests/qemucapabilitiestest.c b/tests/qemucapabilitiestest.c index b3a9a4b9fc..785b445672 100644 --- a/tests/qemucapabilitiestest.c +++ b/tests/qemucapabilitiestest.c @@ -58,6 +58,9 @@ testQemuCaps(const void *opaque) if (!(mon = qemuMonitorTestNewFromFileFull(repliesFile, &data->driver, NULL))) goto cleanup; + if (qemuMonitorSetCapabilities(qemuMonitorTestGetMonitor(mon)) < 0) + goto cleanup; + if (!(capsActual = virQEMUCapsNew()) || virQEMUCapsInitQMPMonitor(capsActual, qemuMonitorTestGetMonitor(mon)) < 0) @@ -65,6 +68,10 @@ testQemuCaps(const void *opaque) if (virQEMUCapsGet(capsActual, QEMU_CAPS_KVM)) { qemuMonitorResetCommandID(qemuMonitorTestGetMonitor(mon)); + + if (qemuMonitorSetCapabilities(qemuMonitorTestGetMonitor(mon)) < 0) + goto cleanup; + if (virQEMUCapsInitQMPMonitorTCG(capsActual, qemuMonitorTestGetMonitor(mon)) < 0) goto cleanup; -- 2.17.1

Multiple QEMU processes for QMP commands can operate concurrently. Use a unique directory under libDir for each QEMU processes to avoid pidfile and unix socket collision between processes. The pid file name is changed from "capabilities.pidfile" to "qmp.pid" because we no longer need to avoid a possible clash with a qemu domain called "capabilities" now that the processes artifacts are stored in their own unique temporary directories. "Capabilities" was changed to "qmp" in the pid file name because these processes are no longer specific to the capabilities usecase and are more generic in terms of being used for any general purpose QMP message exchanges with a QEMU process that is not associated with a domain. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_process.c | 25 +++++++++++++++---------- src/qemu/qemu_process.h | 1 + 2 files changed, 16 insertions(+), 10 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index d2d067b1a6..79a914a627 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -8227,6 +8227,7 @@ qemuProcessQMPFree(qemuProcessQMPPtr proc) VIR_FREE(proc->binary); VIR_FREE(proc->libDir); + VIR_FREE(proc->uniqDir); VIR_FREE(proc->monpath); VIR_FREE(proc->monarg); VIR_FREE(proc->pidfile); @@ -8287,32 +8288,33 @@ qemuProcessQMPNew(const char *binary, static int qemuProcessQMPInit(qemuProcessQMPPtr proc) { + char *template = NULL; int ret = -1; VIR_DEBUG("proc=%p, emulator=%s", proc, proc->binary); - /* the ".sock" sufix is important to avoid a possible clash with a qemu - * domain called "capabilities" - */ - if (virAsprintf(&proc->monpath, "%s/%s", proc->libDir, - "capabilities.monitor.sock") < 0) + if (virAsprintf(&template, "%s/qmp-XXXXXX", proc->libDir) < 0) + goto cleanup; + + if (!(proc->uniqDir = mkdtemp(template))) + goto cleanup; + + if (virAsprintf(&proc->monpath, "%s/%s", proc->uniqDir, + "qmp.monitor") < 0) goto cleanup; if (virAsprintf(&proc->monarg, "unix:%s,server,nowait", proc->monpath) < 0) goto cleanup; - /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash - * with a qemu domain called "capabilities" + /* * Normally we'd use runDir for pid files, but because we're using * -daemonize we need QEMU to be allowed to create them, rather * than libvirtd. So we're using libDir which QEMU can write to */ - if (virAsprintf(&proc->pidfile, "%s/%s", proc->libDir, "capabilities.pidfile") < 0) + if (virAsprintf(&proc->pidfile, "%s/%s", proc->uniqDir, "qmp.pid") < 0) goto cleanup; - virPidFileForceCleanupPath(proc->pidfile); - ret = 0; cleanup: @@ -8528,4 +8530,7 @@ qemuProcessQMPStop(qemuProcessQMPPtr proc) if (proc->pidfile) unlink(proc->pidfile); + + if (proc->uniqDir) + rmdir(proc->uniqDir); } diff --git a/src/qemu/qemu_process.h b/src/qemu/qemu_process.h index 0bf594ad2e..d429c76723 100644 --- a/src/qemu/qemu_process.h +++ b/src/qemu/qemu_process.h @@ -226,6 +226,7 @@ struct _qemuProcessQMP { char *monarg; char *monpath; char *pidfile; + char *uniqDir; virCommandPtr cmd; qemuMonitorPtr mon; pid_t pid; -- 2.17.1

Use a helper function to allocate and initializes CPU Model Info structs. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 8 ++++---- src/qemu/qemu_monitor.c | 27 ++++++++++++++++++++++----- src/qemu/qemu_monitor.h | 2 ++ src/qemu/qemu_monitor_json.c | 5 +---- 4 files changed, 29 insertions(+), 13 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index bac8487a78..02911a2a10 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -3063,16 +3063,16 @@ virQEMUCapsLoadHostCPUModelInfo(virQEMUCapsPtr qemuCaps, goto cleanup; } - if (VIR_ALLOC(hostCPU) < 0) - goto cleanup; - - if (!(hostCPU->name = virXMLPropString(hostCPUNode, "model"))) { + if (!(str = virXMLPropString(hostCPUNode, "model"))) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("missing host CPU model name in QEMU " "capabilities cache")); goto cleanup; } + if (!(hostCPU = qemuMonitorCPUModelInfoNew(str))) + goto cleanup; + if (!(str = virXMLPropString(hostCPUNode, "migratability")) || (val = virTristateBoolTypeFromString(str)) <= 0) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index 7ce5bd668e..815a30c275 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -3686,6 +3686,26 @@ qemuMonitorGetCPUModelExpansion(qemuMonitorPtr mon, } +qemuMonitorCPUModelInfoPtr +qemuMonitorCPUModelInfoNew(const char *name) +{ + qemuMonitorCPUModelInfoPtr ret = NULL; + qemuMonitorCPUModelInfoPtr model; + + if (VIR_ALLOC(model) < 0) + return NULL; + + if (VIR_STRDUP(model->name, name) < 0) + goto cleanup; + + VIR_STEAL_PTR(ret, model); + + cleanup: + qemuMonitorCPUModelInfoFree(model); + return ret; +} + + void qemuMonitorCPUModelInfoFree(qemuMonitorCPUModelInfoPtr model_info) { @@ -3709,18 +3729,15 @@ qemuMonitorCPUModelInfoFree(qemuMonitorCPUModelInfoPtr model_info) qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig) { - qemuMonitorCPUModelInfoPtr copy; + qemuMonitorCPUModelInfoPtr copy = NULL; size_t i; - if (VIR_ALLOC(copy) < 0) + if (!(copy = qemuMonitorCPUModelInfoNew(orig->name))) goto error; if (VIR_ALLOC_N(copy->props, orig->nprops) < 0) goto error; - if (VIR_STRDUP(copy->name, orig->name) < 0) - goto error; - copy->migratability = orig->migratability; copy->nprops = orig->nprops; diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index 8fcac8850f..d17bda75c3 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1067,6 +1067,8 @@ int qemuMonitorGetCPUModelExpansion(qemuMonitorPtr mon, void qemuMonitorCPUModelInfoFree(qemuMonitorCPUModelInfoPtr model_info); +qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoNew(const char *name); + qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig); diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index 1f7eb20a25..dcb468c747 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -5642,10 +5642,7 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, goto cleanup; } - if (VIR_ALLOC(machine_model) < 0) - goto cleanup; - - if (VIR_STRDUP(machine_model->name, cpu_name) < 0) + if (!(machine_model = qemuMonitorCPUModelInfoNew(cpu_name))) goto cleanup; if (VIR_ALLOC_N(machine_model->props, virJSONValueObjectKeysNumber(cpu_props)) < 0) -- 2.17.1

Conversion functions are used convert CPUModelInfo structs into QMP JSON and the reverse. QMP JSON is of form: {"model": {"name": "IvyBridge", "props": {}}} qemuMonitorCPUModelInfoBoolPropAdd is used to add boolean properties to CPUModelInfo struct. qemuMonitorJSONGetCPUModelExpansion makes full use of conversions and propAdd in prep to support input of full cpu model in future. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_monitor.c | 25 ++++++ src/qemu/qemu_monitor.h | 5 ++ src/qemu/qemu_monitor_json.c | 152 +++++++++++++++++++++++++---------- 3 files changed, 140 insertions(+), 42 deletions(-) diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index 815a30c275..ac9fc1cdc2 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -3775,6 +3775,31 @@ qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig) } +int +qemuMonitorCPUModelInfoBoolPropAdd(qemuMonitorCPUModelInfoPtr model, + const char *prop_name, + bool prop_value) +{ + int ret = -1; + qemuMonitorCPUProperty prop = { + .type = QEMU_MONITOR_CPU_PROPERTY_BOOLEAN, + .value.boolean = prop_value + }; + + if (VIR_STRDUP(prop.name, prop_name) < 0) + goto cleanup; + + if (VIR_APPEND_ELEMENT(model->props, model->nprops, prop) < 0) + goto cleanup; + + ret = 0; + + cleanup: + VIR_FREE(prop.name); + return ret; +} + + int qemuMonitorGetCommands(qemuMonitorPtr mon, char ***commands) diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index d17bda75c3..bb74f29cb4 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1072,6 +1072,11 @@ qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoNew(const char *name); qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig); +int qemuMonitorCPUModelInfoBoolPropAdd(qemuMonitorCPUModelInfoPtr model, + const char *prop_name, + bool prop_value) + ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2); + int qemuMonitorGetCommands(qemuMonitorPtr mon, char ***commands); int qemuMonitorGetEvents(qemuMonitorPtr mon, diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index dcb468c747..a3998decf0 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -5541,6 +5541,102 @@ qemuMonitorJSONParseCPUModelProperty(const char *key, return 0; } + +/* model_json: {"name": "z13-base", "props": {}} + */ +static virJSONValuePtr +qemuMonitorJSONBuildCPUModelInfoToJSON(qemuMonitorCPUModelInfoPtr model) +{ + virJSONValuePtr cpu_props = NULL; + virJSONValuePtr model_json = NULL; + size_t i; + + if (!model) + goto cleanup; + + if (model->nprops > 0 && !(cpu_props = virJSONValueNewObject())) + goto cleanup; + + for (i = 0; i < model->nprops; i++) { + qemuMonitorCPUPropertyPtr prop = &(model->props[i]); + + switch (prop->type) { + case QEMU_MONITOR_CPU_PROPERTY_BOOLEAN: + if (virJSONValueObjectAppendBoolean(cpu_props, prop->name, + prop->value.boolean) < 0) + goto cleanup; + break; + + case QEMU_MONITOR_CPU_PROPERTY_STRING: + if (virJSONValueObjectAppendString(cpu_props, prop->name, + prop->value.string) < 0) + goto cleanup; + break; + + case QEMU_MONITOR_CPU_PROPERTY_NUMBER: + if (virJSONValueObjectAppendNumberLong(cpu_props, prop->name, + prop->value.number) < 0) + goto cleanup; + break; + + case QEMU_MONITOR_CPU_PROPERTY_LAST: + default: + virReportEnumRangeError(qemuMonitorCPUPropertyPtr, prop->type); + goto cleanup; + } + } + + ignore_value(virJSONValueObjectCreate(&model_json, "s:name", model->name, + "A:props", &cpu_props, NULL)); + + cleanup: + virJSONValueFree(cpu_props); + return model_json; +} + + +/* model_json: {"name": "IvyBridge", "props": {}} + */ +static qemuMonitorCPUModelInfoPtr +qemuMonitorJSONBuildCPUModelInfoFromJSON(virJSONValuePtr cpu_model) +{ + virJSONValuePtr cpu_props; + qemuMonitorCPUModelInfoPtr model = NULL; + qemuMonitorCPUModelInfoPtr ret = NULL; + char const *cpu_name; + + if (!(cpu_name = virJSONValueObjectGetString(cpu_model, "name"))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Parsed JSON reply missing 'name'")); + goto cleanup; + } + + if (!(model = qemuMonitorCPUModelInfoNew(cpu_name))) + goto cleanup; + + if (!(cpu_props = virJSONValueObjectGetObject(cpu_model, "props"))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("query-cpu-model-expansion reply data was missing 'props'")); + goto cleanup; + } + + if (VIR_ALLOC_N(model->props, + virJSONValueObjectKeysNumber(cpu_props)) < 0) + goto cleanup; + + if (virJSONValueObjectForeachKeyValue(cpu_props, + qemuMonitorJSONParseCPUModelProperty, + model) < 0) + goto cleanup; + + VIR_STEAL_PTR(ret, model); + + cleanup: + qemuMonitorCPUModelInfoFree(model); + + return ret; +} + int qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelExpansionType type, @@ -5549,32 +5645,25 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelInfoPtr *model_info) { int ret = -1; - virJSONValuePtr model = NULL; - virJSONValuePtr props = NULL; + virJSONValuePtr json_model_in = NULL; virJSONValuePtr cmd = NULL; virJSONValuePtr reply = NULL; virJSONValuePtr data; virJSONValuePtr cpu_model; - virJSONValuePtr cpu_props; - qemuMonitorCPUModelInfoPtr machine_model = NULL; - char const *cpu_name; + qemuMonitorCPUModelInfoPtr model_in = NULL; const char *typeStr = ""; *model_info = NULL; - if (!(model = virJSONValueNewObject())) + if (!(model_in = qemuMonitorCPUModelInfoNew(model_name))) goto cleanup; - if (virJSONValueObjectAppendString(model, "name", model_name) < 0) + if (!migratable && + qemuMonitorCPUModelInfoBoolPropAdd(model_in, "migratable", false) < 0) goto cleanup; - if (!migratable) { - if (!(props = virJSONValueNewObject()) || - virJSONValueObjectAppendBoolean(props, "migratable", false) < 0 || - virJSONValueObjectAppend(model, "props", props) < 0) - goto cleanup; - props = NULL; - } + if (!(json_model_in = qemuMonitorJSONBuildCPUModelInfoToJSON(model_in))) + goto cleanup; retry: switch (type) { @@ -5590,7 +5679,7 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, if (!(cmd = qemuMonitorJSONMakeCommand("query-cpu-model-expansion", "s:type", typeStr, - "a:model", &model, + "a:model", &json_model_in, NULL))) goto cleanup; @@ -5621,7 +5710,9 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, * on the result of the initial "static" expansion. */ if (type == QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC_FULL) { - if (!(model = virJSONValueCopy(cpu_model))) + virJSONValueFree(json_model_in); + + if (!(json_model_in = virJSONValueCopy(cpu_model))) goto cleanup; virJSONValueFree(cmd); @@ -5630,39 +5721,16 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, goto retry; } - if (!(cpu_name = virJSONValueObjectGetString(cpu_model, "name"))) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("query-cpu-model-expansion reply data was missing 'name'")); - goto cleanup; - } - - if (!(cpu_props = virJSONValueObjectGetObject(cpu_model, "props"))) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("query-cpu-model-expansion reply data was missing 'props'")); - goto cleanup; - } - - if (!(machine_model = qemuMonitorCPUModelInfoNew(cpu_name))) - goto cleanup; - - if (VIR_ALLOC_N(machine_model->props, virJSONValueObjectKeysNumber(cpu_props)) < 0) - goto cleanup; - - if (virJSONValueObjectForeachKeyValue(cpu_props, - qemuMonitorJSONParseCPUModelProperty, - machine_model) < 0) + if (!(*model_info = qemuMonitorJSONBuildCPUModelInfoFromJSON(cpu_model))) goto cleanup; ret = 0; - *model_info = machine_model; - machine_model = NULL; cleanup: - qemuMonitorCPUModelInfoFree(machine_model); + qemuMonitorCPUModelInfoFree(model_in); virJSONValueFree(cmd); virJSONValueFree(reply); - virJSONValueFree(model); - virJSONValueFree(props); + virJSONValueFree(json_model_in); return ret; } -- 2.17.1

Create an augmented CPUModelInfo identifying which props are/aren't migratable based on a diff between migratable and non-migratable inputs. This patch pulls existing logic out of virQEMUCapsProbeQMPHostCPU and wraps the existing logic in a standalone function hopefully simplifying both functions and making the inputs and outputs clearer. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 130 ++++++++++++++++++++++++----------- 1 file changed, 91 insertions(+), 39 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 02911a2a10..cd8d1439f6 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -2394,14 +2394,91 @@ virQEMUCapsProbeQMPCPUDefinitions(virQEMUCapsPtr qemuCaps, } +/* virQEMUCapsMigratablePropsCalc + * @migratable: input where all migratable props have value true + * and non-migratable and unsupported props have value false + * + * @nonMigratable: input where all migratable & non-migratable props + * have value true and unsupported props have value false + * + * @augmented: output including all props listed in @migratable but + * both migratable & non-migratable props have value true, + * unsupported props have value false, + * and prop->migratable is set to VIR_TRISTATE_BOOL_{YES/NO} + * for each supported prop + * + * Differences in expanded CPUModelInfo inputs @migratable and @nonMigratable are + * used to create output @augmented where individual props have prop->migratable + * set to indicate if prop is or isn't migratable. + */ +static int +virQEMUCapsMigratablePropsCalc(qemuMonitorCPUModelInfoPtr migratable, + qemuMonitorCPUModelInfoPtr nonMigratable, + qemuMonitorCPUModelInfoPtr *augmented) +{ + int ret = -1; + qemuMonitorCPUModelInfoPtr tmp; + qemuMonitorCPUPropertyPtr prop; + qemuMonitorCPUPropertyPtr mProp; + qemuMonitorCPUPropertyPtr nonmProp; + virHashTablePtr hash = NULL; + size_t i; + + *augmented = NULL; + + if (!(tmp = qemuMonitorCPUModelInfoCopy(migratable))) + goto cleanup; + + if (!nonMigratable) + goto done; + + if (!(hash = virHashCreate(0, NULL))) + goto cleanup; + + for (i = 0; i < tmp->nprops; i++) { + prop = tmp->props + i; + + if (virHashAddEntry(hash, prop->name, prop) < 0) + goto cleanup; + } + + for (i = 0; i < nonMigratable->nprops; i++) { + nonmProp = nonMigratable->props + i; + + if (!(mProp = virHashLookup(hash, nonmProp->name)) || + mProp->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN || + mProp->type != nonmProp->type) + continue; /* In non-migratable list but not in migratable list */ + + if (mProp->value.boolean) { + mProp->migratable = VIR_TRISTATE_BOOL_YES; + } else if (nonmProp->value.boolean) { + mProp->value.boolean = true; + mProp->migratable = VIR_TRISTATE_BOOL_NO; + } + } + + tmp->migratability = true; + + done: + VIR_STEAL_PTR(*augmented, tmp); + ret = 0; + + cleanup: + qemuMonitorCPUModelInfoFree(tmp); + virHashFree(hash); + return ret; +} + + static int virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, qemuMonitorPtr mon, bool tcg) { - qemuMonitorCPUModelInfoPtr modelInfo = NULL; + qemuMonitorCPUModelInfoPtr migratable = NULL; qemuMonitorCPUModelInfoPtr nonMigratable = NULL; - virHashTablePtr hash = NULL; + qemuMonitorCPUModelInfoPtr augmented = NULL; const char *model; qemuMonitorCPUModelExpansionType type; virDomainVirtType virtType; @@ -2431,54 +2508,29 @@ virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, else type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC; - if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &modelInfo) < 0) + if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &migratable) < 0) goto cleanup; + if (!migratable) { + ret = 0; /* Qemu can't expand the model name, exit without error */ + goto cleanup; + } + /* Try to check migratability of each feature. */ - if (modelInfo && - qemuMonitorGetCPUModelExpansion(mon, type, model, false, + if (qemuMonitorGetCPUModelExpansion(mon, type, model, false, &nonMigratable) < 0) goto cleanup; - if (nonMigratable) { - qemuMonitorCPUPropertyPtr prop; - qemuMonitorCPUPropertyPtr nmProp; - size_t i; + if (virQEMUCapsMigratablePropsCalc(migratable, nonMigratable, &augmented) < 0) + goto cleanup; - if (!(hash = virHashCreate(0, NULL))) - goto cleanup; - - for (i = 0; i < modelInfo->nprops; i++) { - prop = modelInfo->props + i; - if (virHashAddEntry(hash, prop->name, prop) < 0) - goto cleanup; - } - - for (i = 0; i < nonMigratable->nprops; i++) { - nmProp = nonMigratable->props + i; - if (!(prop = virHashLookup(hash, nmProp->name)) || - prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN || - prop->type != nmProp->type) - continue; - - if (prop->value.boolean) { - prop->migratable = VIR_TRISTATE_BOOL_YES; - } else if (nmProp->value.boolean) { - prop->value.boolean = true; - prop->migratable = VIR_TRISTATE_BOOL_NO; - } - } - - modelInfo->migratability = true; - } - - VIR_STEAL_PTR(cpuData->info, modelInfo); + VIR_STEAL_PTR(cpuData->info, augmented); ret = 0; cleanup: - virHashFree(hash); + qemuMonitorCPUModelInfoFree(migratable); qemuMonitorCPUModelInfoFree(nonMigratable); - qemuMonitorCPUModelInfoFree(modelInfo); + qemuMonitorCPUModelInfoFree(augmented); return ret; } -- 2.17.1

A Full CPUModelInfo structure with props is sent to QEMU for expansion. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 8 +++++--- src/qemu/qemu_monitor.c | 20 +++++++++++++++----- src/qemu/qemu_monitor.h | 5 +++-- src/qemu/qemu_monitor_json.c | 12 +++++++----- src/qemu/qemu_monitor_json.h | 4 ++-- tests/cputest.c | 11 ++++++++--- 6 files changed, 40 insertions(+), 20 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index cd8d1439f6..463c803416 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -2476,6 +2476,7 @@ virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, qemuMonitorPtr mon, bool tcg) { + qemuMonitorCPUModelInfoPtr input; qemuMonitorCPUModelInfoPtr migratable = NULL; qemuMonitorCPUModelInfoPtr nonMigratable = NULL; qemuMonitorCPUModelInfoPtr augmented = NULL; @@ -2508,7 +2509,8 @@ virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, else type = QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC; - if (qemuMonitorGetCPUModelExpansion(mon, type, model, true, &migratable) < 0) + if (!(input = qemuMonitorCPUModelInfoNew(model)) || + qemuMonitorGetCPUModelExpansion(mon, type, input, true, &migratable) < 0) goto cleanup; if (!migratable) { @@ -2517,8 +2519,7 @@ virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, } /* Try to check migratability of each feature. */ - if (qemuMonitorGetCPUModelExpansion(mon, type, model, false, - &nonMigratable) < 0) + if (qemuMonitorGetCPUModelExpansion(mon, type, input, false, &nonMigratable) < 0) goto cleanup; if (virQEMUCapsMigratablePropsCalc(migratable, nonMigratable, &augmented) < 0) @@ -2528,6 +2529,7 @@ virQEMUCapsProbeQMPHostCPU(virQEMUCapsPtr qemuCaps, ret = 0; cleanup: + qemuMonitorCPUModelInfoFree(input); qemuMonitorCPUModelInfoFree(migratable); qemuMonitorCPUModelInfoFree(nonMigratable); qemuMonitorCPUModelInfoFree(augmented); diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index ac9fc1cdc2..d6bb2488d1 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -3669,20 +3669,30 @@ qemuMonitorCPUDefInfoFree(qemuMonitorCPUDefInfoPtr cpu) } +/** + * qemuMonitorGetCPUModelExpansion: + * @mon: + * @type: qemuMonitorCPUModelExpansionType + * @input: Input model + * @migratable: Prompt QEMU to include non-migratable props for X86 models if false + * @expansion: Expanded output model (or NULL if QEMU rejects model or request) + * + * Re-represent @input CPU props using a new constructed CPUModelInfo +*/ int qemuMonitorGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelExpansionType type, - const char *model_name, + qemuMonitorCPUModelInfoPtr input, bool migratable, - qemuMonitorCPUModelInfoPtr *model_info) + qemuMonitorCPUModelInfoPtr *expansion + ) { VIR_DEBUG("type=%d model_name=%s migratable=%d", - type, model_name, migratable); + type, input->name, migratable); QEMU_CHECK_MONITOR(mon); - return qemuMonitorJSONGetCPUModelExpansion(mon, type, model_name, - migratable, model_info); + return qemuMonitorJSONGetCPUModelExpansion(mon, type, input, migratable, expansion); } diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index bb74f29cb4..4440e2f3d6 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1061,9 +1061,10 @@ typedef enum { int qemuMonitorGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelExpansionType type, - const char *model_name, + qemuMonitorCPUModelInfoPtr input, bool migratable, - qemuMonitorCPUModelInfoPtr *model_info); + qemuMonitorCPUModelInfoPtr *expansion) + ATTRIBUTE_NONNULL(3) ATTRIBUTE_NONNULL(5); void qemuMonitorCPUModelInfoFree(qemuMonitorCPUModelInfoPtr model_info); diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index a3998decf0..e5e56747a7 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -5637,12 +5637,13 @@ qemuMonitorJSONBuildCPUModelInfoFromJSON(virJSONValuePtr cpu_model) return ret; } + int qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelExpansionType type, - const char *model_name, + qemuMonitorCPUModelInfoPtr input, bool migratable, - qemuMonitorCPUModelInfoPtr *model_info) + qemuMonitorCPUModelInfoPtr *expansion) { int ret = -1; virJSONValuePtr json_model_in = NULL; @@ -5653,9 +5654,9 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelInfoPtr model_in = NULL; const char *typeStr = ""; - *model_info = NULL; + *expansion = NULL; - if (!(model_in = qemuMonitorCPUModelInfoNew(model_name))) + if (!(model_in = qemuMonitorCPUModelInfoCopy(input))) goto cleanup; if (!migratable && @@ -5721,13 +5722,14 @@ qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, goto retry; } - if (!(*model_info = qemuMonitorJSONBuildCPUModelInfoFromJSON(cpu_model))) + if (!(*expansion = qemuMonitorJSONBuildCPUModelInfoFromJSON(cpu_model))) goto cleanup; ret = 0; cleanup: qemuMonitorCPUModelInfoFree(model_in); + virJSONValueFree(cmd); virJSONValueFree(reply); virJSONValueFree(json_model_in); diff --git a/src/qemu/qemu_monitor_json.h b/src/qemu/qemu_monitor_json.h index 62772228fe..731f1ec010 100644 --- a/src/qemu/qemu_monitor_json.h +++ b/src/qemu/qemu_monitor_json.h @@ -364,9 +364,9 @@ int qemuMonitorJSONGetCPUDefinitions(qemuMonitorPtr mon, int qemuMonitorJSONGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelExpansionType type, - const char *model_name, + qemuMonitorCPUModelInfoPtr input, bool migratable, - qemuMonitorCPUModelInfoPtr *model_info) + qemuMonitorCPUModelInfoPtr *expansion) ATTRIBUTE_NONNULL(3) ATTRIBUTE_NONNULL(5); int qemuMonitorJSONGetCommands(qemuMonitorPtr mon, diff --git a/tests/cputest.c b/tests/cputest.c index b75d864d8e..228f737824 100644 --- a/tests/cputest.c +++ b/tests/cputest.c @@ -480,6 +480,7 @@ cpuTestMakeQEMUCaps(const struct data *data) virQEMUCapsPtr qemuCaps = NULL; qemuMonitorTestPtr testMon = NULL; qemuMonitorCPUModelInfoPtr model = NULL; + qemuMonitorCPUModelInfoPtr expansion = NULL; char *json = NULL; if (virAsprintf(&json, "%s/cputestdata/%s-cpuid-%s.json", @@ -489,9 +490,12 @@ cpuTestMakeQEMUCaps(const struct data *data) if (!(testMon = qemuMonitorTestNewFromFile(json, driver.xmlopt, true))) goto error; + if (!(model = qemuMonitorCPUModelInfoNew("host"))) + goto cleanup; + if (qemuMonitorGetCPUModelExpansion(qemuMonitorTestGetMonitor(testMon), QEMU_MONITOR_CPU_MODEL_EXPANSION_STATIC, - "host", true, &model) < 0) + model, true, &expansion) < 0) goto error; if (!(qemuCaps = virQEMUCapsNew())) @@ -503,8 +507,8 @@ cpuTestMakeQEMUCaps(const struct data *data) virQEMUCapsSet(qemuCaps, QEMU_CAPS_QUERY_CPU_DEFINITIONS); virQEMUCapsSetArch(qemuCaps, data->arch); - virQEMUCapsSetCPUModelInfo(qemuCaps, VIR_DOMAIN_VIRT_KVM, model); - model = NULL; + virQEMUCapsSetCPUModelInfo(qemuCaps, VIR_DOMAIN_VIRT_KVM, expansion); + expansion = NULL; if (virQEMUCapsProbeQMPCPUDefinitions(qemuCaps, qemuMonitorTestGetMonitor(testMon), @@ -513,6 +517,7 @@ cpuTestMakeQEMUCaps(const struct data *data) cleanup: qemuMonitorCPUModelInfoFree(model); + qemuMonitorCPUModelInfoFree(expansion); qemuMonitorTestFree(testMon); VIR_FREE(json); -- 2.17.1

Move existing code to convert between cpu model info structures (qemuMonitorCPUModelInfoPtr into virCPUDef) into a reusable function. The new function is used in this and future patches. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 88 ++++++++++++++++++++++++++---------- src/qemu/qemu_capabilities.h | 3 ++ 2 files changed, 68 insertions(+), 23 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 463c803416..2bb56763b0 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -2834,7 +2834,8 @@ virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps, virCPUDefPtr cpu, bool migratable) { - size_t i; + virCPUDefPtr tmp = NULL; + int ret = -1; if (!modelInfo) { if (type == VIR_DOMAIN_VIRT_KVM) { @@ -2847,32 +2848,18 @@ virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps, return 2; } - if (VIR_STRDUP(cpu->model, modelInfo->name) < 0 || - VIR_ALLOC_N(cpu->features, modelInfo->nprops) < 0) - return -1; + if (!(tmp = virQEMUCapsCPUModelInfoToCPUDef(modelInfo, migratable))) + goto cleanup; - cpu->nfeatures_max = modelInfo->nprops; - cpu->nfeatures = 0; + /* Free original then copy over model, vendor, vendor_id and features */ + virCPUDefStealModel(cpu, tmp, true); - for (i = 0; i < modelInfo->nprops; i++) { - virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures; - qemuMonitorCPUPropertyPtr prop = modelInfo->props + i; + ret = 0; - if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN) - continue; + cleanup: + virCPUDefFree(tmp); - if (VIR_STRDUP(feature->name, prop->name) < 0) - return -1; - - if (!prop->value.boolean || - (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO)) - feature->policy = VIR_CPU_FEATURE_DISABLE; - else - feature->policy = VIR_CPU_FEATURE_REQUIRE; - cpu->nfeatures++; - } - - return 0; + return ret; } @@ -3677,6 +3664,61 @@ virQEMUCapsLoadCache(virArch hostArch, } +/** + * virQEMUCapsCPUModelInfoToCPUDef: + * @model: input model + * @migratable: mark non-migratable features as disabled if true else allow all + * + * qemuMonitorCPUModelInfo name => virCPUDef model + * qemuMonitorCPUModelInfo boolean properties => virCPUDef features + */ +virCPUDefPtr +virQEMUCapsCPUModelInfoToCPUDef(qemuMonitorCPUModelInfoPtr model, + bool migratable) +{ + virCPUDefPtr cpu = NULL; + virCPUDefPtr ret = NULL; + size_t i; + + VIR_DEBUG("model= %p, model->name= %s", model, NULLSTR(model ? model->name : NULL)); + + if (!model || VIR_ALLOC(cpu) < 0) + goto cleanup; + + if (VIR_STRDUP(cpu->model, model->name) < 0 || + VIR_ALLOC_N(cpu->features, model->nprops) < 0) + goto cleanup; + + cpu->nfeatures_max = model->nprops; + cpu->nfeatures = 0; + + for (i = 0; i < model->nprops; i++) { + virCPUFeatureDefPtr feature = cpu->features + cpu->nfeatures; + qemuMonitorCPUPropertyPtr prop = model->props + i; + + if (prop->type != QEMU_MONITOR_CPU_PROPERTY_BOOLEAN) + continue; + + if (VIR_STRDUP(feature->name, prop->name) < 0) + goto cleanup; + + if (!prop->value.boolean || + (migratable && prop->migratable == VIR_TRISTATE_BOOL_NO)) + feature->policy = VIR_CPU_FEATURE_DISABLE; + else + feature->policy = VIR_CPU_FEATURE_REQUIRE; + + cpu->nfeatures++; + } + + VIR_STEAL_PTR(ret, cpu); + + cleanup: + virCPUDefFree(cpu); + return ret; +} + + static void virQEMUCapsFormatHostCPUModelInfo(virQEMUCapsPtr qemuCaps, virBufferPtr buf, diff --git a/src/qemu/qemu_capabilities.h b/src/qemu/qemu_capabilities.h index 6d5ed8a3cc..834844c1be 100644 --- a/src/qemu/qemu_capabilities.h +++ b/src/qemu/qemu_capabilities.h @@ -583,6 +583,9 @@ int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps, void virQEMUCapsFilterByMachineType(virQEMUCapsPtr qemuCaps, const char *machineType); +virCPUDefPtr virQEMUCapsCPUModelInfoToCPUDef(qemuMonitorCPUModelInfoPtr model, + bool migratable); + virFileCachePtr virQEMUCapsCacheNew(const char *libDir, const char *cacheDir, uid_t uid, -- 2.17.1

Create public function to convert virCPUDef data structure into qemuMonitorCPUModelInfoPtr data structure. There was no existing code to reuse to create this function so this new virQEMUCapsCPUModelInfoFromCPUDef function was based on reversing the action of the existing virQEMUCapsCPUModelInfoToCPUDef function. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_capabilities.c | 52 ++++++++++++++++++++++++++++++++++++ src/qemu/qemu_capabilities.h | 1 + 2 files changed, 53 insertions(+) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 2bb56763b0..d0a2a42c97 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -3664,6 +3664,58 @@ virQEMUCapsLoadCache(virArch hostArch, } +/** + * virQEMUCapsCPUModelInfoFromCPUDef: + * @cpuDef: input model + * + * virCPUDef model => qemuMonitorCPUModelInfo name + * virCPUDef features => qemuMonitorCPUModelInfo boolean properties + * + * property have true value if feature policy is undefined (-1), FORCE or REQUIRE + * These semantics may need to be modified for other cases. + */ +qemuMonitorCPUModelInfoPtr +virQEMUCapsCPUModelInfoFromCPUDef(const virCPUDef *cpuDef) +{ + size_t i; + qemuMonitorCPUModelInfoPtr cpuModel = NULL; + qemuMonitorCPUModelInfoPtr ret = NULL; + + VIR_DEBUG("cpuDef = %p, cpuDef->model = %s", cpuDef, NULLSTR(cpuDef ? cpuDef->model : NULL)); + + if (!cpuDef || (VIR_ALLOC(cpuModel) < 0)) + goto cleanup; + + if (VIR_STRDUP(cpuModel->name, cpuDef->model) < 0 || + VIR_ALLOC_N(cpuModel->props, cpuDef->nfeatures) < 0) + goto cleanup; + + cpuModel->nprops = 0; + + for (i = 0; i < cpuDef->nfeatures; i++) { + qemuMonitorCPUPropertyPtr prop = &(cpuModel->props[cpuModel->nprops]); + virCPUFeatureDefPtr feature = &(cpuDef->features[i]); + + if (VIR_STRDUP(prop->name, feature->name) < 0) + goto cleanup; + + prop->type = QEMU_MONITOR_CPU_PROPERTY_BOOLEAN; + + prop->value.boolean = feature->policy == -1 || /* policy undefined */ + feature->policy == VIR_CPU_FEATURE_FORCE || + feature->policy == VIR_CPU_FEATURE_REQUIRE; + + cpuModel->nprops++; + } + + VIR_STEAL_PTR(ret, cpuModel); + + cleanup: + qemuMonitorCPUModelInfoFree(cpuModel); + return ret; +} + + /** * virQEMUCapsCPUModelInfoToCPUDef: * @model: input model diff --git a/src/qemu/qemu_capabilities.h b/src/qemu/qemu_capabilities.h index 834844c1be..e05596ae06 100644 --- a/src/qemu/qemu_capabilities.h +++ b/src/qemu/qemu_capabilities.h @@ -583,6 +583,7 @@ int virQEMUCapsGetMachineTypesCaps(virQEMUCapsPtr qemuCaps, void virQEMUCapsFilterByMachineType(virQEMUCapsPtr qemuCaps, const char *machineType); +qemuMonitorCPUModelInfoPtr virQEMUCapsCPUModelInfoFromCPUDef(const virCPUDef *cpuDef); virCPUDefPtr virQEMUCapsCPUModelInfoToCPUDef(qemuMonitorCPUModelInfoPtr model, bool migratable); -- 2.17.1

Introduce monitor functions to use QEMU to compute baseline cpu from an input of two cpu models. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_monitor.c | 12 ++++++++ src/qemu/qemu_monitor.h | 6 ++++ src/qemu/qemu_monitor_json.c | 60 ++++++++++++++++++++++++++++++++++++ src/qemu/qemu_monitor_json.h | 6 ++++ 4 files changed, 84 insertions(+) diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index d6bb2488d1..7d82f25fc2 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -4526,3 +4526,15 @@ qemuMonitorGetPRManagerInfo(qemuMonitorPtr mon, virHashFree(info); return ret; } + + +int +qemuMonitorGetCPUModelBaseline(qemuMonitorPtr mon, + qemuMonitorCPUModelInfoPtr model_a, + qemuMonitorCPUModelInfoPtr model_b, + qemuMonitorCPUModelInfoPtr *model_baseline) +{ + QEMU_CHECK_MONITOR(mon); + + return qemuMonitorJSONGetCPUModelBaseline(mon, model_a, model_b, model_baseline); +} diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index 4440e2f3d6..2b3ea6ab8e 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1230,4 +1230,10 @@ struct _qemuMonitorPRManagerInfo { int qemuMonitorGetPRManagerInfo(qemuMonitorPtr mon, virHashTablePtr *retinfo); +int qemuMonitorGetCPUModelBaseline(qemuMonitorPtr mon, + qemuMonitorCPUModelInfoPtr model_a, + qemuMonitorCPUModelInfoPtr model_b, + qemuMonitorCPUModelInfoPtr *model_baseline) + ATTRIBUTE_NONNULL(2) ATTRIBUTE_NONNULL(3) ATTRIBUTE_NONNULL(4); + #endif /* LIBVIRT_QEMU_MONITOR_H */ diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index e5e56747a7..792c3a1e0a 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -8547,3 +8547,63 @@ qemuMonitorJSONGetPRManagerInfo(qemuMonitorPtr mon, return ret; } + + +int +qemuMonitorJSONGetCPUModelBaseline(qemuMonitorPtr mon, + qemuMonitorCPUModelInfoPtr model_a, + qemuMonitorCPUModelInfoPtr model_b, + qemuMonitorCPUModelInfoPtr *model_baseline) +{ + int ret = -1; + virJSONValuePtr cmd = NULL; + virJSONValuePtr reply = NULL; + virJSONValuePtr data = NULL; + virJSONValuePtr modela = NULL; + virJSONValuePtr modelb = NULL; + virJSONValuePtr cpu_model = NULL; + + *model_baseline = NULL; + + if (!(modela = qemuMonitorJSONBuildCPUModelInfoToJSON(model_a)) || + !(modelb = qemuMonitorJSONBuildCPUModelInfoToJSON(model_b))) + goto cleanup; + + if (!(cmd = qemuMonitorJSONMakeCommand("query-cpu-model-baseline", + "a:modela", &modela, + "a:modelb", &modelb, + NULL))) + goto cleanup; + + if (qemuMonitorJSONCommand(mon, cmd, &reply) < 0) + goto cleanup; + + if (qemuMonitorJSONCheckReply(cmd, reply, VIR_JSON_TYPE_OBJECT) < 0) { + virReportError(VIR_ERR_INVALID_ARG, + _("QEMU doesn't support baseline or recognize model %s or %s"), + model_a->name, + model_b->name); + goto cleanup; + } + + data = virJSONValueObjectGetObject(reply, "return"); + + if (!(cpu_model = virJSONValueObjectGetObject(data, "model"))) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("query-cpu-model-baseline reply data was missing 'model'")); + goto cleanup; + } + + if (!(*model_baseline = qemuMonitorJSONBuildCPUModelInfoFromJSON(cpu_model))) + goto cleanup; + + ret = 0; + + cleanup: + virJSONValueFree(cmd); + virJSONValueFree(reply); + virJSONValueFree(modela); + virJSONValueFree(modelb); + + return ret; +} diff --git a/src/qemu/qemu_monitor_json.h b/src/qemu/qemu_monitor_json.h index 731f1ec010..df2759ca11 100644 --- a/src/qemu/qemu_monitor_json.h +++ b/src/qemu/qemu_monitor_json.h @@ -578,4 +578,10 @@ int qemuMonitorJSONGetPRManagerInfo(qemuMonitorPtr mon, virHashTablePtr info) ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2); +int qemuMonitorJSONGetCPUModelBaseline(qemuMonitorPtr mon, + qemuMonitorCPUModelInfoPtr model_a, + qemuMonitorCPUModelInfoPtr model_b, + qemuMonitorCPUModelInfoPtr *model_baseline) + ATTRIBUTE_NONNULL(2) ATTRIBUTE_NONNULL(3) ATTRIBUTE_NONNULL(4); + #endif /* LIBVIRT_QEMU_MONITOR_JSON_H */ -- 2.17.1

Simple cut/paste operations within function qemuConnectBaselineHypervisorCPU to group together code specific to computing baseline using only libvirt utility functions. This is done in anticipation of creating new utility functions for 1) baseline using libvirt utilities (this code) 2) baseline using qemu qmp command (future) Future patches are easier to follow if the code for using libvirt is consolidated. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_driver.c | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 1d961707cc..161b82d229 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -13711,7 +13711,6 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, virQEMUCapsPtr qemuCaps = NULL; virArch arch; virDomainVirtType virttype; - virDomainCapsCPUModelsPtr cpuModels; bool migratable; virCPUDefPtr cpu = NULL; char *cpustr = NULL; @@ -13723,8 +13722,6 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, if (virConnectBaselineHypervisorCPUEnsureACL(conn) < 0) goto cleanup; - migratable = !!(flags & VIR_CONNECT_BASELINE_CPU_MIGRATABLE); - if (!(cpus = virCPUDefListParse(xmlCPUs, ncpus, VIR_CPU_TYPE_AUTO))) goto cleanup; @@ -13737,17 +13734,21 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, if (!qemuCaps) goto cleanup; - if (!(cpuModels = virQEMUCapsGetCPUDefinitions(qemuCaps, virttype)) || - cpuModels->nmodels == 0) { - virReportError(VIR_ERR_OPERATION_UNSUPPORTED, - _("QEMU '%s' does not support any CPU models for " - "virttype '%s'"), - virQEMUCapsGetBinary(qemuCaps), - virDomainVirtTypeToString(virttype)); - goto cleanup; - } - if (ARCH_IS_X86(arch)) { + migratable = !!(flags & VIR_CONNECT_BASELINE_CPU_MIGRATABLE); + + virDomainCapsCPUModelsPtr cpuModels; + + if (!(cpuModels = virQEMUCapsGetCPUDefinitions(qemuCaps, virttype)) || + cpuModels->nmodels == 0) { + virReportError(VIR_ERR_OPERATION_UNSUPPORTED, + _("QEMU '%s' does not support any CPU models for " + "virttype '%s'"), + virQEMUCapsGetBinary(qemuCaps), + virDomainVirtTypeToString(virttype)); + goto cleanup; + } + int rc = virQEMUCapsGetCPUFeatures(qemuCaps, virttype, migratable, &features); if (rc < 0) -- 2.17.1

Create utility function encapsulating code to calculate hypervisor baseline cpu using the local libvirt utility functions. Similar function encapsulating code to calculating hypervisor baseline using QEMU QMP messages will be introduced in later commit. Patch is a cut and paste of existing code into a utility function wrapper. s/cpu/*baseline/ (change output variable name ) and initialize variable "rc" are the only code changes. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_driver.c | 78 ++++++++++++++++++++++++++++-------------- 1 file changed, 52 insertions(+), 26 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 161b82d229..2a030bed2f 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -13696,6 +13696,55 @@ qemuConnectBaselineCPU(virConnectPtr conn ATTRIBUTE_UNUSED, } +static int +qemuConnectBaselineHypervisorCPUViaLibvirt( + virQEMUCapsPtr qemuCaps, + bool migratable, + virDomainVirtType virttype, + virArch arch, + virCPUDefPtr *cpus, + unsigned int ncpus, + virCPUDefPtr *baseline) +{ + char **features = NULL; + int ret = -1; + int rc = -1; + virDomainCapsCPUModelsPtr cpuModels; + + *baseline = NULL; + + if (!(cpuModels = virQEMUCapsGetCPUDefinitions(qemuCaps, virttype)) || + cpuModels->nmodels == 0) { + virReportError(VIR_ERR_OPERATION_UNSUPPORTED, + _("QEMU '%s' does not support any CPU models for " + "virttype '%s'"), + virQEMUCapsGetBinary(qemuCaps), + virDomainVirtTypeToString(virttype)); + goto cleanup; + } + + rc = virQEMUCapsGetCPUFeatures(qemuCaps, virttype, + migratable, &features); + if (rc < 0) + goto cleanup; + if (features && rc == 0) { + /* We got only migratable features from QEMU if we asked for them, + * no further filtering in virCPUBaseline is desired. */ + migratable = false; + } + + if (!(*baseline = virCPUBaseline(arch, cpus, ncpus, cpuModels, + (const char **)features, migratable))) + goto cleanup; + + ret = 0; + + cleanup: + virStringListFree(features); + return ret; +} + + static char * qemuConnectBaselineHypervisorCPU(virConnectPtr conn, const char *emulator, @@ -13714,7 +13763,6 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, bool migratable; virCPUDefPtr cpu = NULL; char *cpustr = NULL; - char **features = NULL; virCheckFlags(VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES | VIR_CONNECT_BASELINE_CPU_MIGRATABLE, NULL); @@ -13737,30 +13785,9 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, if (ARCH_IS_X86(arch)) { migratable = !!(flags & VIR_CONNECT_BASELINE_CPU_MIGRATABLE); - virDomainCapsCPUModelsPtr cpuModels; - - if (!(cpuModels = virQEMUCapsGetCPUDefinitions(qemuCaps, virttype)) || - cpuModels->nmodels == 0) { - virReportError(VIR_ERR_OPERATION_UNSUPPORTED, - _("QEMU '%s' does not support any CPU models for " - "virttype '%s'"), - virQEMUCapsGetBinary(qemuCaps), - virDomainVirtTypeToString(virttype)); - goto cleanup; - } - - int rc = virQEMUCapsGetCPUFeatures(qemuCaps, virttype, - migratable, &features); - if (rc < 0) - goto cleanup; - if (features && rc == 0) { - /* We got only migratable features from QEMU if we asked for them, - * no further filtering in virCPUBaseline is desired. */ - migratable = false; - } - - if (!(cpu = virCPUBaseline(arch, cpus, ncpus, cpuModels, - (const char **)features, migratable))) + if (qemuConnectBaselineHypervisorCPUViaLibvirt(qemuCaps, migratable, + virttype, arch, + cpus, ncpus, &cpu) < 0) goto cleanup; } else { virReportError(VIR_ERR_OPERATION_UNSUPPORTED, @@ -13781,7 +13808,6 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, virCPUDefListFree(cpus); virCPUDefFree(cpu); virObjectUnref(qemuCaps); - virStringListFree(features); return cpustr; } -- 2.17.1

Hypervisor baseline cpu can be computed locally using libvirt utility functions or remotely using QEMU QMP commands. Likewise, cpu feature expansion can be computed locally using libvirt utility functions or remotely using QEMU QMP commands. This patch identifies using libvirt as a distinct case in the hypervisor baseline logic and triggers that case when the X86 architecture is being baselined. There is one functionality change introduced by this patch. Local libvirt functions are only used for feature expansion when the local utility functions are used for CPU baseline and an error is generated when no method is available to expand cpu featues. The useQEMU option will be introduced in a future patch for using QEMU to compute baseline rather than using libvirt. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_driver.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 2a030bed2f..e9f5686dbe 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -13763,6 +13763,7 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, bool migratable; virCPUDefPtr cpu = NULL; char *cpustr = NULL; + bool useLibvirt = false; virCheckFlags(VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES | VIR_CONNECT_BASELINE_CPU_MIGRATABLE, NULL); @@ -13782,7 +13783,9 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, if (!qemuCaps) goto cleanup; - if (ARCH_IS_X86(arch)) { + useLibvirt = ARCH_IS_X86(arch); + + if (useLibvirt) { migratable = !!(flags & VIR_CONNECT_BASELINE_CPU_MIGRATABLE); if (qemuConnectBaselineHypervisorCPUViaLibvirt(qemuCaps, migratable, @@ -13798,9 +13801,17 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, cpu->fallback = VIR_CPU_FALLBACK_FORBID; - if ((flags & VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES) && - virCPUExpandFeatures(arch, cpu) < 0) - goto cleanup; + if (flags & VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES) { + if (useLibvirt && virCPUExpandFeatures(arch, cpu) < 0) { + goto cleanup; + } else { + virReportError(VIR_ERR_OPERATION_UNSUPPORTED, + _("expand features while " + "computing baseline hypervisor CPU is not supported " + "for arch %s"), virArchToString(arch)); + goto cleanup; + } + } cpustr = virCPUDefFormat(cpu, NULL); -- 2.17.1

Add capability to calculate hypervisor baseline using QMP message exchanges with QEMU in addition to existing capability to calculate baseline using libvirt utility functions. A new utility function encapsulates the core logic for interacting with QEMU using QMP baseline messages. The QMP messages only allow two cpu models to be evaluated at a time so a sequence of QMP baseline messages must be used to include all the models in the baseline when more than 2 cpu models are input. A QEMU process must be started prior to sending the series of QEMU messages to compute baseline. The QEMU process is started and maintained in the main hypervisor baseline function and only a pointer to the QEMU process monitor is passed to the baseline utility function. The QEMU process is maintained in the main hypervisor baseline function because the process will also be used for feature expansion (via QEMU) in a later patch. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_driver.c | 97 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 97 insertions(+) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index e9f5686dbe..5068805f51 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -13696,6 +13696,78 @@ qemuConnectBaselineCPU(virConnectPtr conn ATTRIBUTE_UNUSED, } +/* in: + * cpus[0]->model = "z14"; + * cpus[0]->features[0].name = "xxx"; + * cpus[0]->features[1].name = "yyy"; + * *** + * cpus[n]->model = "z13"; + * cpus[n]->features[0].name = "xxx"; + * cpus[n]->features[1].name = "yyy"; + * + * out: + * *baseline->model = "z13-base"; + * *baseline->features[0].name = "yyy"; + */ +static int +qemuConnectBaselineHypervisorCPUViaQEMU(qemuMonitorPtr mon, + virCPUDefPtr *cpus, + virCPUDefPtr *baseline) +{ + qemuMonitorCPUModelInfoPtr model_baseline = NULL; + qemuMonitorCPUModelInfoPtr new_model_baseline = NULL; + qemuMonitorCPUModelInfoPtr next_model = NULL; + bool migratable = true; + int ret = -1; + size_t i; + + *baseline = NULL; + + if (!cpus || !cpus[0]) { + virReportError(VIR_ERR_INVALID_ARG, "%s", _("no cpus")); + goto cleanup; + } + + for (i = 0; cpus[i]; i++) { /* cpus terminated by NULL element */ + virCPUDefPtr cpu = cpus[i]; + + VIR_DEBUG("cpu[%lu]->model = %s", i, NULLSTR(cpu->model)); + + if (!(next_model = virQEMUCapsCPUModelInfoFromCPUDef(cpu))) + goto cleanup; + + if (i == 0) { + model_baseline = next_model; + continue; + } + + if (qemuMonitorGetCPUModelBaseline(mon, model_baseline, + next_model, &new_model_baseline) < 0) + goto cleanup; + + qemuMonitorCPUModelInfoFree(model_baseline); + qemuMonitorCPUModelInfoFree(next_model); + + next_model = NULL; + + model_baseline = new_model_baseline; + } + + if (!(*baseline = virQEMUCapsCPUModelInfoToCPUDef(model_baseline, migratable))) + goto cleanup; + + VIR_DEBUG("baseline->model = %s", (*baseline)->model); + + ret = 0; + + cleanup: + qemuMonitorCPUModelInfoFree(model_baseline); + qemuMonitorCPUModelInfoFree(next_model); + + return ret; +} + + static int qemuConnectBaselineHypervisorCPUViaLibvirt( virQEMUCapsPtr qemuCaps, @@ -13764,6 +13836,11 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, virCPUDefPtr cpu = NULL; char *cpustr = NULL; bool useLibvirt = false; + bool useQemu = false; + bool forceTCG = false; + qemuMonitorCPUModelInfoPtr modelInfo = NULL; + qemuMonitorCPUModelInfoPtr expansion = NULL; + qemuProcessQMPPtr proc = NULL; virCheckFlags(VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES | VIR_CONNECT_BASELINE_CPU_MIGRATABLE, NULL); @@ -13784,6 +13861,7 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, goto cleanup; useLibvirt = ARCH_IS_X86(arch); + useQemu = ARCH_IS_S390(arch); if (useLibvirt) { migratable = !!(flags & VIR_CONNECT_BASELINE_CPU_MIGRATABLE); @@ -13792,6 +13870,21 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, virttype, arch, cpus, ncpus, &cpu) < 0) goto cleanup; + + } else if (useQemu) { + const char *binary = virQEMUCapsGetBinary(qemuCaps); + virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver); + + if (!(proc = qemuProcessQMPNew(binary, cfg->libDir, + cfg->user, cfg->group, forceTCG))) + goto cleanup; + + if (qemuProcessQMPStart(proc) < 0) + goto cleanup; + + if ((qemuConnectBaselineHypervisorCPUViaQEMU(proc->mon, cpus, &cpu) < 0) || !cpu) + goto cleanup; + } else { virReportError(VIR_ERR_OPERATION_UNSUPPORTED, _("computing baseline hypervisor CPU is not supported " @@ -13819,6 +13912,10 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, virCPUDefListFree(cpus); virCPUDefFree(cpu); virObjectUnref(qemuCaps); + qemuMonitorCPUModelInfoFree(modelInfo); + qemuMonitorCPUModelInfoFree(expansion); + qemuProcessQMPStop(proc); + qemuProcessQMPFree(proc); return cpustr; } -- 2.17.1

Support using QEMU to do feature expansion when also using QEMU to compute hypervisor baseline. A QEMU process is already created to send the QMP messages to baseline using QEMU. The same QEMU process is used for the CPU feature expansion. QEMU only returns migratable features when expanding CPU model in architectures where QEMU is used for baseline so no attempt is made to ask for non-migratable features in expansions when using QEMU for baseline. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_driver.c | 25 +++++++++++++++++++++++++ src/qemu/qemu_monitor.c | 30 ++++++++++++++++++++++++++++++ src/qemu/qemu_monitor.h | 4 ++++ 3 files changed, 59 insertions(+) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 5068805f51..b57123b585 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -13897,6 +13897,31 @@ qemuConnectBaselineHypervisorCPU(virConnectPtr conn, if (flags & VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES) { if (useLibvirt && virCPUExpandFeatures(arch, cpu) < 0) { goto cleanup; + } else if (useQemu && + virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION)) { + + if (!(modelInfo = virQEMUCapsCPUModelInfoFromCPUDef(cpu))) + goto cleanup; + + virCPUDefFree(cpu); + cpu = NULL; + + /* QEMU can only include migratable features + for all archs that use QEMU for baseline calculation */ + migratable = true; + + if (qemuMonitorGetCPUModelExpansion(proc->mon, + QEMU_MONITOR_CPU_MODEL_EXPANSION_FULL, + modelInfo, migratable, &expansion) < 0) + goto cleanup; + + /* Expansion enumerates all features + * Baselines output enumerates only in-model (true) features */ + qemuMonitorCPUModelInfoRemovePropByBoolValue(expansion, false); + + if (!(cpu = virQEMUCapsCPUModelInfoToCPUDef(expansion, migratable))) + goto cleanup; + } else { virReportError(VIR_ERR_OPERATION_UNSUPPORTED, _("expand features while " diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index 7d82f25fc2..56a3bbcea1 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -3785,6 +3785,36 @@ qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig) } +/* Squash CPU Model Info property list + * removing props of type boolean matching value */ +void +qemuMonitorCPUModelInfoRemovePropByBoolValue(qemuMonitorCPUModelInfoPtr model, + bool value) +{ + qemuMonitorCPUPropertyPtr src; + qemuMonitorCPUPropertyPtr dst; + size_t i; + size_t dst_nprops = 0; + + for (i = 0; i < model->nprops; i++) { + src = &(model->props[i]); + dst = &(model->props[dst_nprops]); + + if (src->type == QEMU_MONITOR_CPU_PROPERTY_BOOLEAN && + src->value.boolean == value) + continue; + + *dst = *src; + + dst_nprops++; + } + + model->nprops = dst_nprops; + + ignore_value(VIR_REALLOC_N_QUIET(model->props, dst_nprops)); +} + + int qemuMonitorCPUModelInfoBoolPropAdd(qemuMonitorCPUModelInfoPtr model, const char *prop_name, diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index 2b3ea6ab8e..c54f311632 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1078,6 +1078,10 @@ int qemuMonitorCPUModelInfoBoolPropAdd(qemuMonitorCPUModelInfoPtr model, bool prop_value) ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2); +void qemuMonitorCPUModelInfoRemovePropByBoolValue(qemuMonitorCPUModelInfoPtr model, + bool value) + ATTRIBUTE_NONNULL(1); + int qemuMonitorGetCommands(qemuMonitorPtr mon, char ***commands); int qemuMonitorGetEvents(qemuMonitorPtr mon, -- 2.17.1

QEMU only returns migratable props when expanding model unless explicitly told to also include non-migratable props. Props will be marked migratable when we are certain QEMU returned only migratable props resulting in consistent information and expansion output for s390 that is consistent with x86. After this change, immediately default prop->migratable = _YES for all props when we know QEMU only included migratable props in CPU Model. Set model->migratability = true when we have set prop->migratable. Signed-off-by: Chris Venteicher <cventeic@redhat.com> --- src/qemu/qemu_monitor.c | 55 +++++++++++++++- src/qemu/qemu_monitor.h | 7 +- .../caps_2.10.0.s390x.xml | 60 ++++++++--------- .../caps_2.11.0.s390x.xml | 58 ++++++++--------- .../caps_2.12.0.s390x.xml | 56 ++++++++-------- .../qemucapabilitiesdata/caps_2.8.0.s390x.xml | 32 +++++----- .../qemucapabilitiesdata/caps_2.9.0.s390x.xml | 34 +++++----- .../qemucapabilitiesdata/caps_3.0.0.s390x.xml | 64 +++++++++---------- 8 files changed, 211 insertions(+), 155 deletions(-) diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index 56a3bbcea1..d9171dbf83 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -3687,12 +3687,31 @@ qemuMonitorGetCPUModelExpansion(qemuMonitorPtr mon, qemuMonitorCPUModelInfoPtr *expansion ) { + int ret = -1; + qemuMonitorCPUModelInfoPtr tmp; + VIR_DEBUG("type=%d model_name=%s migratable=%d", type, input->name, migratable); + *expansion = NULL; + QEMU_CHECK_MONITOR(mon); - return qemuMonitorJSONGetCPUModelExpansion(mon, type, input, migratable, expansion); + if ((ret = qemuMonitorJSONGetCPUModelExpansion(mon, type, input, migratable, &tmp)) < 0) + goto cleanup; + + if (migratable) { + /* Only migratable props were included in expanded CPU model */ + *expansion = qemuMonitorCPUModelInfoCopyDefaultMigratable(tmp); + } else { + VIR_STEAL_PTR(*expansion, tmp); + } + + ret = 0; + + cleanup: + qemuMonitorCPUModelInfoFree(tmp); + return ret; } @@ -3742,7 +3761,7 @@ qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig) qemuMonitorCPUModelInfoPtr copy = NULL; size_t i; - if (!(copy = qemuMonitorCPUModelInfoNew(orig->name))) + if (!orig || !(copy = qemuMonitorCPUModelInfoNew(orig->name))) goto error; if (VIR_ALLOC_N(copy->props, orig->nprops) < 0) @@ -3785,6 +3804,38 @@ qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig) } +qemuMonitorCPUModelInfoPtr +qemuMonitorCPUModelInfoCopyDefaultMigratable(const qemuMonitorCPUModelInfo *orig) +{ + qemuMonitorCPUModelInfoPtr ret = NULL; + qemuMonitorCPUModelInfoPtr tmp = NULL; + qemuMonitorCPUPropertyPtr prop = NULL; + size_t i; + + if (!(tmp = qemuMonitorCPUModelInfoCopy(orig))) + goto cleanup; + + for (i = 0; i < tmp->nprops; i++) { + prop = tmp->props + i; + + /* Default prop thats in cpu model (true) to migratable (_YES) + * unless prop already explicitly set not migratable (_NO) + */ + if (prop->type == QEMU_MONITOR_CPU_PROPERTY_BOOLEAN && + prop->value.boolean && + prop->migratable != VIR_TRISTATE_BOOL_NO) + prop->migratable = VIR_TRISTATE_BOOL_YES; + } + + tmp->migratability = true; /* prop->migratable = YES/NO for all CPU props */ + + VIR_STEAL_PTR(ret, tmp); + + cleanup: + return ret; +} + + /* Squash CPU Model Info property list * removing props of type boolean matching value */ void diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index c54f311632..0487cc2ca3 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -1050,7 +1050,7 @@ struct _qemuMonitorCPUModelInfo { char *name; size_t nprops; qemuMonitorCPUPropertyPtr props; - bool migratability; + bool migratability; /* true if prop->migratable is YES/NO for all CPU props */ }; typedef enum { @@ -1073,6 +1073,11 @@ qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoNew(const char *name); qemuMonitorCPUModelInfoPtr qemuMonitorCPUModelInfoCopy(const qemuMonitorCPUModelInfo *orig); +qemuMonitorCPUModelInfoPtr +qemuMonitorCPUModelInfoCopyDefaultMigratable(const qemuMonitorCPUModelInfo *orig) + ATTRIBUTE_NONNULL(1); + + int qemuMonitorCPUModelInfoBoolPropAdd(qemuMonitorCPUModelInfoPtr model, const char *prop_name, bool prop_value) diff --git a/tests/qemucapabilitiesdata/caps_2.10.0.s390x.xml b/tests/qemucapabilitiesdata/caps_2.10.0.s390x.xml index 180a688ba2..07f4da579b 100644 --- a/tests/qemucapabilitiesdata/caps_2.10.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_2.10.0.s390x.xml @@ -120,36 +120,36 @@ <microcodeVersion>306247</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='z14-base' migratability='no'> - <property name='aen' type='boolean' value='true'/> - <property name='cmmnt' type='boolean' value='true'/> - <property name='aefsi' type='boolean' value='true'/> - <property name='mepoch' type='boolean' value='true'/> - <property name='msa8' type='boolean' value='true'/> - <property name='msa7' type='boolean' value='true'/> - <property name='msa6' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='vx' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='vxeh' type='boolean' value='true'/> - <property name='vxpd' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='iep' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='ais' type='boolean' value='true'/> - <property name='gs' type='boolean' value='true'/> - <property name='zpci' type='boolean' value='true'/> - <property name='sea_esop2' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='z14-base' migratability='yes'> + <property name='aen' type='boolean' value='true' migratable='yes'/> + <property name='cmmnt' type='boolean' value='true' migratable='yes'/> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='mepoch' type='boolean' value='true' migratable='yes'/> + <property name='msa8' type='boolean' value='true' migratable='yes'/> + <property name='msa7' type='boolean' value='true' migratable='yes'/> + <property name='msa6' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='vx' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='vxeh' type='boolean' value='true' migratable='yes'/> + <property name='vxpd' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='iep' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='ais' type='boolean' value='true' migratable='yes'/> + <property name='gs' type='boolean' value='true' migratable='yes'/> + <property name='zpci' type='boolean' value='true' migratable='yes'/> + <property name='sea_esop2' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z10EC-base' usable='yes'/> <cpu type='kvm' name='z9EC-base' usable='yes'/> diff --git a/tests/qemucapabilitiesdata/caps_2.11.0.s390x.xml b/tests/qemucapabilitiesdata/caps_2.11.0.s390x.xml index 7dc643d9a3..c2f222bd41 100644 --- a/tests/qemucapabilitiesdata/caps_2.11.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_2.11.0.s390x.xml @@ -127,35 +127,35 @@ <microcodeVersion>345099</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='z14-base' migratability='no'> - <property name='aen' type='boolean' value='true'/> - <property name='cmmnt' type='boolean' value='true'/> - <property name='aefsi' type='boolean' value='true'/> - <property name='mepoch' type='boolean' value='true'/> - <property name='msa8' type='boolean' value='true'/> - <property name='msa7' type='boolean' value='true'/> - <property name='msa6' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='vx' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='vxeh' type='boolean' value='true'/> - <property name='vxpd' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='iep' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='gs' type='boolean' value='true'/> - <property name='zpci' type='boolean' value='true'/> - <property name='sea_esop2' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='z14-base' migratability='yes'> + <property name='aen' type='boolean' value='true' migratable='yes'/> + <property name='cmmnt' type='boolean' value='true' migratable='yes'/> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='mepoch' type='boolean' value='true' migratable='yes'/> + <property name='msa8' type='boolean' value='true' migratable='yes'/> + <property name='msa7' type='boolean' value='true' migratable='yes'/> + <property name='msa6' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='vx' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='vxeh' type='boolean' value='true' migratable='yes'/> + <property name='vxpd' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='iep' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='gs' type='boolean' value='true' migratable='yes'/> + <property name='zpci' type='boolean' value='true' migratable='yes'/> + <property name='sea_esop2' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z890.2' usable='yes'/> <cpu type='kvm' name='z990.4' usable='yes'/> diff --git a/tests/qemucapabilitiesdata/caps_2.12.0.s390x.xml b/tests/qemucapabilitiesdata/caps_2.12.0.s390x.xml index 49249f5adf..e2a001d30f 100644 --- a/tests/qemucapabilitiesdata/caps_2.12.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_2.12.0.s390x.xml @@ -138,34 +138,34 @@ <microcodeVersion>375102</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='z14-base' migratability='no'> - <property name='aen' type='boolean' value='true'/> - <property name='aefsi' type='boolean' value='true'/> - <property name='msa8' type='boolean' value='true'/> - <property name='msa7' type='boolean' value='true'/> - <property name='msa6' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='vx' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='vxeh' type='boolean' value='true'/> - <property name='vxpd' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='iep' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='gs' type='boolean' value='true'/> - <property name='ppa15' type='boolean' value='true'/> - <property name='zpci' type='boolean' value='true'/> - <property name='sea_esop2' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='z14-base' migratability='yes'> + <property name='aen' type='boolean' value='true' migratable='yes'/> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='msa8' type='boolean' value='true' migratable='yes'/> + <property name='msa7' type='boolean' value='true' migratable='yes'/> + <property name='msa6' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='vx' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='vxeh' type='boolean' value='true' migratable='yes'/> + <property name='vxpd' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='iep' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='gs' type='boolean' value='true' migratable='yes'/> + <property name='ppa15' type='boolean' value='true' migratable='yes'/> + <property name='zpci' type='boolean' value='true' migratable='yes'/> + <property name='sea_esop2' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z890.2' usable='yes'/> <cpu type='kvm' name='z990.4' usable='yes'/> diff --git a/tests/qemucapabilitiesdata/caps_2.8.0.s390x.xml b/tests/qemucapabilitiesdata/caps_2.8.0.s390x.xml index f3a32ad376..8fa95c1f90 100644 --- a/tests/qemucapabilitiesdata/caps_2.8.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_2.8.0.s390x.xml @@ -109,22 +109,22 @@ <microcodeVersion>244554</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='zEC12.2-base' migratability='no'> - <property name='aefsi' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='zEC12.2-base' migratability='yes'> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z10EC-base'/> <cpu type='kvm' name='z9EC-base'/> diff --git a/tests/qemucapabilitiesdata/caps_2.9.0.s390x.xml b/tests/qemucapabilitiesdata/caps_2.9.0.s390x.xml index 4bb61f09f0..93a7c72f3a 100644 --- a/tests/qemucapabilitiesdata/caps_2.9.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_2.9.0.s390x.xml @@ -114,23 +114,23 @@ <microcodeVersion>267973</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='z13.2-base' migratability='no'> - <property name='aefsi' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='vx' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='z13.2-base' migratability='yes'> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='vx' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z10EC-base'/> <cpu type='kvm' name='z9EC-base'/> diff --git a/tests/qemucapabilitiesdata/caps_3.0.0.s390x.xml b/tests/qemucapabilitiesdata/caps_3.0.0.s390x.xml index 8569de6538..fa56597c6f 100644 --- a/tests/qemucapabilitiesdata/caps_3.0.0.s390x.xml +++ b/tests/qemucapabilitiesdata/caps_3.0.0.s390x.xml @@ -140,38 +140,38 @@ <microcodeVersion>388416</microcodeVersion> <package></package> <arch>s390x</arch> - <hostCPU type='kvm' model='z14-base' migratability='no'> - <property name='aen' type='boolean' value='true'/> - <property name='cmmnt' type='boolean' value='true'/> - <property name='aefsi' type='boolean' value='true'/> - <property name='mepoch' type='boolean' value='true'/> - <property name='msa8' type='boolean' value='true'/> - <property name='msa7' type='boolean' value='true'/> - <property name='msa6' type='boolean' value='true'/> - <property name='msa5' type='boolean' value='true'/> - <property name='msa4' type='boolean' value='true'/> - <property name='msa3' type='boolean' value='true'/> - <property name='msa2' type='boolean' value='true'/> - <property name='msa1' type='boolean' value='true'/> - <property name='sthyi' type='boolean' value='true'/> - <property name='edat' type='boolean' value='true'/> - <property name='ri' type='boolean' value='true'/> - <property name='edat2' type='boolean' value='true'/> - <property name='vx' type='boolean' value='true'/> - <property name='ipter' type='boolean' value='true'/> - <property name='mepochptff' type='boolean' value='true'/> - <property name='vxeh' type='boolean' value='true'/> - <property name='vxpd' type='boolean' value='true'/> - <property name='esop' type='boolean' value='true'/> - <property name='iep' type='boolean' value='true'/> - <property name='cte' type='boolean' value='true'/> - <property name='bpb' type='boolean' value='true'/> - <property name='gs' type='boolean' value='true'/> - <property name='ppa15' type='boolean' value='true'/> - <property name='zpci' type='boolean' value='true'/> - <property name='sea_esop2' type='boolean' value='true'/> - <property name='te' type='boolean' value='true'/> - <property name='cmm' type='boolean' value='true'/> + <hostCPU type='kvm' model='z14-base' migratability='yes'> + <property name='aen' type='boolean' value='true' migratable='yes'/> + <property name='cmmnt' type='boolean' value='true' migratable='yes'/> + <property name='aefsi' type='boolean' value='true' migratable='yes'/> + <property name='mepoch' type='boolean' value='true' migratable='yes'/> + <property name='msa8' type='boolean' value='true' migratable='yes'/> + <property name='msa7' type='boolean' value='true' migratable='yes'/> + <property name='msa6' type='boolean' value='true' migratable='yes'/> + <property name='msa5' type='boolean' value='true' migratable='yes'/> + <property name='msa4' type='boolean' value='true' migratable='yes'/> + <property name='msa3' type='boolean' value='true' migratable='yes'/> + <property name='msa2' type='boolean' value='true' migratable='yes'/> + <property name='msa1' type='boolean' value='true' migratable='yes'/> + <property name='sthyi' type='boolean' value='true' migratable='yes'/> + <property name='edat' type='boolean' value='true' migratable='yes'/> + <property name='ri' type='boolean' value='true' migratable='yes'/> + <property name='edat2' type='boolean' value='true' migratable='yes'/> + <property name='vx' type='boolean' value='true' migratable='yes'/> + <property name='ipter' type='boolean' value='true' migratable='yes'/> + <property name='mepochptff' type='boolean' value='true' migratable='yes'/> + <property name='vxeh' type='boolean' value='true' migratable='yes'/> + <property name='vxpd' type='boolean' value='true' migratable='yes'/> + <property name='esop' type='boolean' value='true' migratable='yes'/> + <property name='iep' type='boolean' value='true' migratable='yes'/> + <property name='cte' type='boolean' value='true' migratable='yes'/> + <property name='bpb' type='boolean' value='true' migratable='yes'/> + <property name='gs' type='boolean' value='true' migratable='yes'/> + <property name='ppa15' type='boolean' value='true' migratable='yes'/> + <property name='zpci' type='boolean' value='true' migratable='yes'/> + <property name='sea_esop2' type='boolean' value='true' migratable='yes'/> + <property name='te' type='boolean' value='true' migratable='yes'/> + <property name='cmm' type='boolean' value='true' migratable='yes'/> </hostCPU> <cpu type='kvm' name='z890.2' usable='yes'/> <cpu type='kvm' name='z990.4' usable='yes'/> -- 2.17.1

May I ask what the current status of this item is? Is this still being worked on or has it been shelved? On 1/12/19 7:49 PM, Chris Venteicher wrote:
Some architectures (S390) depend on QEMU to compute baseline CPU model and expand a models feature set.
Interacting with QEMU requires starting the QEMU process and completing one or more query-cpu-model-baseline QMP exchanges with QEMU in addition to a query-cpu-model-expansion QMP exchange to expand all features in the model.
See "s390x CPU models: exposing features" patch set on Qemu-devel for discussion of QEMU aspects.
This is part of resolution of: https://bugzilla.redhat.com/show_bug.cgi?id=1511999
----- v6 addresses these issues from v5:
qemu_process: Move process code from qemu_capabilities to qemu_process [x] Remove exception statement from commit message
qemu_process: Use qemuProcessQmp prefix [x] s/ProcessQmp/ProcessQMP/ in this and all subsequent patches
qemu_process: Refer to proc not cmd in process code [x] Back port indentation fix in qemuProcessQMPRun
qemu_capabilities: Stop QEMU process before freeing [x] Back port !proc check in qemuProcessQmpStop
qemu_process: Use qemuProcessQmp struct for a single process [x] Remove extra empty lines
qemu_process: All ProcessQMP errors are fatal [x] Split into multiple patches
qemu_process: Introduce qemuProcessQmpStart [x] Log function parameters separately in the first function debug message [x] Removed NULLSTR check on proc-binary (would fail previously on proc activate) [x] Back port referring to stderr rather than qmperr in comment [ ] Combine qemuProcessQmpStop and qemuProcessQmpFree (See reply to v5 email)
emu_process: Collect monitor code in single function [x] Back port qemuMonitorOpen indent fix to earlier patch
qemu_process: Don't open monitor if process failed [x] Drop patch
qemu_process: Cleanup qemuProcessQmp alloc function [x] s/Qemu/QEMU/ in comments for qemuProcessQMPNew [x] Remove unneeded use of NULLSTR and NULL check before calling qemuProcessQmpFree
qemu_process: Cleanup qemuProcessQmpStop function [x] Remove lines from from commit message [x] Remove capitalization from comment [x] Fix func def to match coding style [x] Back port !proc check to earlier patch
qemu_process: Catch process free before process stop [x] Remove capitalization from comment [x] Don't do proc NULL check in VIR_DEBUG [ ] Combine qemuProcessQmpStop and qemuProcessQmpFree (See reply to v5 email)
qemu_monitor: Make monitor callbacks optional [x] Drop patch
qemu_process: Enter QMP command mode when starting QEMU Process [ ] Don't call qemuMonitorSetCapabilities in qemuProcessQMPConnectMonitor (See reply to v5 email)
qemu_process: Use unique directories for QMP processes [x] s/qemu./qmp-/ for QMP process directory name prefix [x] Catch mkdtemp returns NULL
qemu_process: Stop locking QMP process monitor immediately [x] Drop patch
qemu_monitor: Introduce qemuMonitorCPUModelInfoNew [x] Use qemuMonitorCPUModelInfoNew to set model name [x] Drop variable initialization [x] Use qemuMonitorCPUModelInfoNew in qemuMonitorJSONGetCPUModelExpansion (Pull forward) [ ] Remove orig NULL check (Unit tests fail if I do this)
qemu_monitor: Introduce qemuMonitorCPUModelInfo / JSON conversion [x] Make variable initialization easier to read in PropAdd [x] Require at least one "prop" (don't allow zero in this commit)
qemu_capabilities: Introduce virQEMuCapsMigratablePropsDiff [x] Remove template txt in commit message [x] Change function name [x] Distinguish nmProp from mProp [x] Don't set cpuData->info to NULL [ ] Don't use tmp (defer to next patch set... out of time)
qemu_monitor: qemuMonitorGetCPUModelExpansion inputs and outputs CPUModelInfo [x] Remove extra line from commit message [x] Change parameter order [x] Don't use VIR_STEAL_PTR
qemu_capabilities: Introduce CPUModelInfo to virCPUDef function [x] Change comment [x] Change parameter order [x] Separate lines [x] Move debug message before alloc [x] Back port function spacing fix from next patch
qemu_capabilities: Introduce virCPUDef to CPUModelInfo function [x] Change comment [x] Move debug message before alloc [x] Remove NULL check on feature->name [x] Document prop value semantics
Reviewed in v5 but no changes required: qemu_process: Limit qemuProcessQmpNew to const input strings qemu_process: Use consistent name for stop process function qemu_process: Persist stderr in qemuProcessQmp struct qemu_process: Store libDir in qemuProcessQmp struct qemu_process: Setup paths within qemuProcessQmpInit qemu_process: Stop retaining Monitor config in qemuProcessQmp
Not reviewed in v5: qemu_monitor: Support query-cpu-model-baseline QMP command qemu_driver: Consolidate code to baseline using libvirt qemu_driver: Decouple code for baseline using libvirt qemu_driver: Identify using libvirt as a distinct way to compute baseline qemu_driver: Support baseline calculation using QEMU qemu_driver: Support feature expansion via QEMU when baselining cpu qemu_driver: Remove unsupported props in expanded hypervisor baseline output qemu_monitor: Default props to migratable when expanding cpu model
----- v5:
Fixes all process issues identified here: https://www.redhat.com/archives/libvir-list/2018-November/msg00349.html
- Make the process code generic (not capabilities specific) for use by BaselineHypervisorCPU
- Many of the process patches are simple code moves with implementation changes in other distinct patches
- A thread safe library function creates a unique directory under libDir for each QEMU process (for QMP messaging) to decouple processes in terms of sockets and file system footprint.
The remaining (non-process) patches in v4 address all issues in v1-v4 of 'BaselineHypervisorCPU using QEMU QMP exchanges'
Thanks, Chris
*** BLURB HERE ***
Chris Venteicher (33): qEmu_process: Move process code from qemu_capabilities to qemu_process qemu_process: Use qemuProcessQMP prefix qemu_process: Limit qemuProcessQMPNew to const input strings qemu_process: Refer to proc not cmd in process code qemu_process: Use consistent name for stop process function qemu_capabilities: Stop QEMU process before freeing qemu_process: Use qemuProcessQMP struct for a single process qemu_process: All ProcessQMP errors are fatal qemu_process: Expose process exit status code qemu_process: Persist stderr in qemuProcessQMP struct qemu_process: Introduce qemuProcessQMPStart qemu_process: Collect monitor code in single function qemu_process: Store libDir in qemuProcessQMP struct qemu_process: Setup paths within qemuProcessQMPInit qemu_process: Stop retaining Monitor config in qemuProcessQMP qemu_process: Cleanup qemuProcessQMP alloc function qemu_process: Cleanup qemuProcessQMPStop function qemu_process: Catch process free before process stop qemu_process: Enter QMP command mode when starting QEMU Process qemu_process: Use unique directories for QMP processes qemu_monitor: Introduce qemuMonitorCPUModelInfoNew qemu_monitor: Introduce qemuMonitorCPUModelInfo / JSON conversion qemu_capabilities: Introduce virQEMuCapsMigratablePropsCalc qemu_monitor: qemuMonitorGetCPUModelExpansion inputs and outputs CPUModelInfo qemu_capabilities: Introduce CPUModelInfo to virCPUDef function qemu_capabilities: Introduce virCPUDef to CPUModelInfo function qemu_monitor: Support query-cpu-model-baseline QMP command qemu_driver: Consolidate code to baseline using libvirt qemu_driver: Decouple code for baseline using libvirt qemu_driver: Identify using libvirt as a distinct way to compute baseline qemu_driver: Support baseline calculation using QEMU qemu_driver: Support feature expansion via QEMU when baselining cpu qemu_monitor: Default props to migratable when expanding cpu model
src/qemu/qemu_capabilities.c | 631 ++++++++---------- src/qemu/qemu_capabilities.h | 4 + src/qemu/qemu_driver.c | 216 +++++- src/qemu/qemu_monitor.c | 165 ++++- src/qemu/qemu_monitor.h | 29 +- src/qemu/qemu_monitor_json.c | 223 +++++-- src/qemu/qemu_monitor_json.h | 10 +- src/qemu/qemu_process.c | 345 ++++++++++ src/qemu/qemu_process.h | 32 + tests/cputest.c | 11 +- .../caps_2.10.0.s390x.xml | 60 +- .../caps_2.11.0.s390x.xml | 58 +- .../caps_2.12.0.s390x.xml | 56 +- .../qemucapabilitiesdata/caps_2.8.0.s390x.xml | 32 +- .../qemucapabilitiesdata/caps_2.9.0.s390x.xml | 34 +- .../qemucapabilitiesdata/caps_3.0.0.s390x.xml | 64 +- tests/qemucapabilitiestest.c | 7 + 17 files changed, 1375 insertions(+), 602 deletions(-)
participants (2)
-
Chris Venteicher
-
Collin Walling