- 03 5月, 2012 4 次提交
-
-
由 Li Zhang 提交于
qemuDomainNetsRestart indents with 3 spaces. This patch is to correct it. Signed-off-by: NLi Zhang <zhlcindy@linux.vnet.ibm.com>
-
由 Marc-André Lureau 提交于
The code uses long long for pid_t now. It fails on mingw64 without this change.
-
由 Marc-André Lureau 提交于
Fix a few more places where pid_t is printed with wrong type
-
由 Eugen Feller 提交于
-
- 02 5月, 2012 3 次提交
-
-
由 Eric Blake 提交于
The recent push to use correct scaling terms (kB for 1000, KiB for 1024 - such as commit 9dfdeadc) missed some places in virsh. * tools/virsh.c (prettyCapacity, cmdDominfo, cmdFreecell) (cmdNodeinfo, cmdNodeMemStats, cmdMigrateSetMaxSpeed) (cmdBlockCopy, cmdBlockPull, cmdBlockJob): Use KiB, not kB, when referring to multiples of 1024. * tests/virshtest.c: Update expected output to match.
-
由 Eric Blake 提交于
https://bugzilla.redhat.com/show_bug.cgi?id=817244 mentions that unlike most other tools, where --help or --version prevent all further parsing of all later options, virsh was strange in that --version stopped parsing but --help tried to plow on to the end. There was no rationale for this original implementation (since 2005!), so I think we can safely conform to common usage patterns. * tools/virsh.c (main): Drop useless 'help' variable.
-
由 Laine Stump 提交于
The ATTRIBUTE_NONNULL(m) macro normally resolves to the gcc builtin __attribute__((__nonnull__(m))). The effect of this in gcc is unfortunately only to make gcc believe that "m" can never possibly be NULL, *not* to add in any checks to guarantee that it isn't ever NULL (i.e. it is an optimization aid, *not* something to verify code correctness.) - see the following gcc bug report for more details: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17308 Static source analyzers such as clang and coverity apparently can use ATTRIBUTE_NONNULL(), though, to detect dead code (in the case that the arg really is guaranteed non-NULL), as well as situations where an obviously NULL arg is given to the function. https://bugzilla.redhat.com/show_bug.cgi?id=815270 is a good example of a bug caused by erroneous application of ATTRIBUTE_NONNULL(). Several people spent a long time staring at this code and not finding the problem, because the problem wasn't in the function itself, but in the prototype that specified ATTRIBUTE_NONNULL() for an arg that actually *wasn't* always non-NULL, and caused a segv when dereferenced (even though the code that dereferenced the pointer was inside an if() that checked for a NULL pointer, that code was optimized out by gcc). There may be some very small gain to be had from the optimizations that can be inferred from ATTRIBUTE_NONNULL(), but it seems safer to err on the side of generating code that behaves as expected, while turning on the attribute for static analyzers.
-
- 01 5月, 2012 9 次提交
-
-
由 Daniel P. Berrange 提交于
Once lxcContainerSetStdio is invoked, logging will not work as expected in libvirt_lxc. So make sure this is the last thing to be called, in particular after setting the security process label
-
由 Daniel P. Berrange 提交于
The virLogSetFromEnv call was done too late in startup to catch many log messages (eg from security driver initialization). To assist debugging also explicitly log the security details at startup
-
由 Daniel P. Berrange 提交于
The driver->securityDriverName field may be NULL, if automatic probing is used to determine security driver. This meant that unless selinux was explicitly requested in lxc.conf, it was not being sent to the libvirt_lxc process. The driver->securityManager field is guaranteed non-NULL, since there will always be the 'none' security driver present if nothing else exists. So use that to set the driver name for libvirt_lxc Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>
-
由 Daniel P. Berrange 提交于
Currently the libvirt_lxc process uses VIR_DOMAIN_XML_INACTIVE when loading the XML for the container. This means it loses any dynamic data such as the, just allocated, SELinux label. Further there is an inconsistency in the libvirt LXC driver whereby it saves the live config XML and then later overwrites the file with the live status XML instead. Add a comment about this for future reference. * src/lxc/lxc_controller.c: Remove VIR_DOMAIN_XML_INACTIVE when loading XML * src/lxc/lxc_driver.c: Add comment about inconsistent config file formats Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>
-
由 Eric Blake 提交于
Otherwise, a string such as _("Don't use \"" VAR "\".") would complain about unmarked diagnostics. * cfg.mk (sc_libvirt_unmarked_diagnostics): Handle \" in message.
-
由 Josh Durgin 提交于
Signed-off-by: NJosh Durgin <josh.durgin@dreamhost.com>
-
由 Josh Durgin 提交于
This works with newer qemu that doesn't allow escaping spaces. It's backwards compatible as well. Signed-off-by: NJosh Durgin <josh.durgin@dreamhost.com>
-
由 Alex Jia 提交于
In fact, the 'tapfd' is always NULL, the function 'virNetDevTapCreate()' hasn't assign 'fd' to 'tapfd', when the function 'virNetDevSetMAC()' is failed then goto 'error' label, finally, the VIR_FORCE_CLOSE() will deref a NULL 'tapfd'. * util/virnetdevtap.c (virNetDevTapCreateInBridgePort): fix a NULL pointer derefing. * How to reproduce? $ cat > /tmp/net.xml <<EOF <network> <name>test</name> <forward mode='nat'/> <bridge name='br1' stp='off' delay='1' /> <mac address='00:00:00:00:00:00'/> <ip address='192.168.100.1' netmask='255.255.255.0'> <dhcp> <range start='192.168.100.2' end='192.168.100.254' /> </dhcp> </ip> </network> EOF $ virsh net-define /tmp/net.xml $ virsh net-start test error: Failed to start network brTest error: End of file while reading data: Input/output error Signed-off-by: NAlex Jia <ajia@redhat.com>
-
由 Eric Blake 提交于
The previous storage patch missed an instance affected by the struct member rename. It also had some botched whitespace detected by 'make check'. * src/storage/storage_backend_iscsi.c (virStorageBackendISCSIFindPoolSources): Adjust to new struct. * src/conf/storage_conf.c (virStoragePoolSourceFormat): Fix indentation.
-
- 30 4月, 2012 3 次提交
-
-
由 Osier Yang 提交于
It doesn't break out the "for" loop even if duplicate pool is found, and thus the "matchpool" could be overriden as NULL again if there is different pool afterwards. To address the problem in libvirt-user list: https://www.redhat.com/archives/libvirt-users/2012-April/msg00150.html
-
由 Wido den Hollander 提交于
The current storage pools for NFS and iSCSI only require one host to connect to. Future storage pools like RBD and Sheepdog will require multiple hosts. This patch allows multiple source hosts and rewrites the current storage drivers. Signed-off-by: NWido den Hollander <wido@widodh.nl>
-
由 Jiri Denemark 提交于
When libvirtd is started, we create "libvirt/qemu" directories under hugetlbfs mount point. Only the "qemu" subdirectory is chowned to qemu user and "libvirt" remains owned by root. If umask was too restrictive when libvirtd started, qemu user may lose access to "qemu" subdirectory. Let's explicitly grant search permissions to "libvirt" directory for all users.
-
- 28 4月, 2012 7 次提交
-
-
由 Michal Privoznik 提交于
Currently, qemu GA is not providing 'desc' field for errors like we are used to from qemu monitor. Therefore, we fall back to this general 'unknown error' string. However, GA is reporting 'class' which is not perfect, but much more helpful than generic error string. Thus we should fall back to class firstly and if even no class is presented, then we can fall back to that generic string. Before this patch: virsh # dompmsuspend --target mem f16 error: Domain f16 could not be suspended error: internal error unable to execute QEMU command 'guest-suspend-ram': unknown QEMU command error After this patch: virsh # dompmsuspend --target mem f16 error: Domain f16 could not be suspended error: internal error unable to execute QEMU command 'guest-suspend-ram': The command has not been found
-
由 Stefan Berger 提交于
More bug extermination in the category of: Error: CHECKED_RETURN: /libvirt/src/conf/network_conf.c:595: check_return: Calling function "virAsprintf" without checking return value (as is done elsewhere 515 out of 543 times). /libvirt/src/qemu/qemu_process.c:2780: unchecked_value: No check of the return value of "virAsprintf(&msg, "was paused (%s)", virDomainPausedReasonTypeToString(reason))". /libvirt/tests/commandtest.c:809: check_return: Calling function "setsid" without checking return value (as is done elsewhere 4 out of 5 times). /libvirt/tests/commandtest.c:830: unchecked_value: No check of the return value of "virTestGetDebug()". /libvirt/tests/commandtest.c:831: check_return: Calling function "virTestGetVerbose" without checking return value (as is done elsewhere 41 out of 42 times). /libvirt/tests/commandtest.c:833: check_return: Calling function "virInitialize" without checking return value (as is done elsewhere 18 out of 21 times). One note about the error in commandtest line 809: setsid() seems to fail when running the test -- could be removed ?
-
由 Eric Blake 提交于
With RHEL 6.2, virDomainBlockPull(dom, dev, bandwidth, 0) has a race with non-zero bandwidth: there is a window between the block_stream and block_job_set_speed monitor commands where an unlimited amount of data was let through, defeating the point of a throttle. This race was first identified in commit a9d3495e, and libvirt was able to reduce the size of the window for that race. In the meantime, the qemu developers decided to fix things properly; per this message: https://lists.gnu.org/archive/html/qemu-devel/2012-04/msg03793.html the fix will be in qemu 1.1, and changes block-job-set-speed to use a different parameter name, as well as adding a new optional parameter to block-stream, which eliminates the race altogether. Since our documentation already mentioned that we can refuse a non-zero bandwidth for some hypervisors, I think the best solution is to do just that for RHEL 6.2 qemu, so that the race is obvious to the user (anyone using stock RHEL 6.2 binaries won't have this patch, and anyone building their own libvirt with this patch for RHEL can also rebuild qemu to get the modern semantics, so it is no real loss in behavior). Meanwhile the code must be fixed to honor actual qemu 1.1 naming. Rename the parameter to 'modern', since the naming difference now covers more than just 'async' block-job-cancel. And while at it, fix an unchecked integer overflow. * src/qemu/qemu_monitor.h (enum BLOCK_JOB_CMD): Drop unused value, rename enum to match conventions. * src/qemu/qemu_monitor.c (qemuMonitorBlockJob): Reflect enum rename. * src/qemu_qemu_monitor_json.h (qemuMonitorJSONBlockJob): Likewise. * src/qemu/qemu_monitor_json.c (qemuMonitorJSONBlockJob): Likewise, and support difference between RHEL 6.2 and qemu 1.1 block pull. * src/qemu/qemu_driver.c (qemuDomainBlockJobImpl): Reject bandwidth during pull with too-old qemu. * src/libvirt.c (virDomainBlockPull, virDomainBlockRebase): Document this.
-
由 Stefan Berger 提交于
Error: UNINIT: /libvirt/src/lxc/lxc_driver.c:1412: var_decl: Declaring variable "fd" without initializer. /libvirt/src/lxc/lxc_driver.c:1460: uninit_use_in_call: Using uninitialized value "fd" when calling "virFileClose". /libvirt/src/util/virfile.c:50: read_parm: Reading a parameter value. Error: DEADCODE: /libvirt/src/lxc/lxc_controller.c:960: dead_error_condition: On this path, the condition "ret == 4" cannot be true. /libvirt/src/lxc/lxc_controller.c:959: at_most: After this line, the value of "ret" is at most -1. /libvirt/src/lxc/lxc_controller.c:959: new_values: Noticing condition "ret < 0". /libvirt/src/lxc/lxc_controller.c:961: dead_error_line: Execution cannot reach this statement "continue;". Error: UNINIT: /libvirt/src/lxc/lxc_controller.c:1104: var_decl: Declaring variable "consoles" without initializer. /libvirt/src/lxc/lxc_controller.c:1237: uninit_use: Using uninitialized value "consoles".
-
由 Cole Robinson 提交于
We were using the libvirt release version (like 0.9.11) and not the configure version (which for stable releases is 0.9.11.X) Most other places got this right so hopefully that's all the fallout from the version format change :) Signed-off-by: NCole Robinson <crobinso@redhat.com>
-
由 Cole Robinson 提交于
Use a witness file approach like we do for python/generator.py, as suggested by Eric. Fixes the build issue reported here: https://www.redhat.com/archives/libvir-list/2012-April/msg01435.htmlSigned-off-by: NCole Robinson <crobinso@redhat.com>
-
由 Cole Robinson 提交于
Since for stable releases, some test files were over the 99 char limit for traditional tar filenames. Suggested by Osier here: https://www.redhat.com/archives/libvir-list/2012-April/msg01435.htmlSigned-off-by: NCole Robinson <crobinso@redhat.com>
-
- 27 4月, 2012 6 次提交
-
-
由 Jiri Denemark 提交于
QEMU binary is called several times when we probe different kinds of capabilities the binary supports. This patch introduces new common helper so that all probes use a consistent way of invoking qemu.
-
由 Eric Blake 提交于
https://bugzilla.redhat.com/show_bug.cgi?id=816662 pointed out that attempting 'virsh blockpull' on an offline domain gave a misleading error message about qemu lacking support for the operation, even when qemu was specifically updated to support it. The real problem is that we have several capabilities that are only determined when starting a domain, and therefore are still clear when first working with an inactive domain (namely, any capability set by qemuMonitorJSONCheckCommands). While this patch was able to hoist an existing check in one of the three culprits, it had to add redundant checks in the other two places (because you always have to check for an active domain after obtaining a VM job lock, but the capability bits were being checked prior to obtaining the job lock). Someday it would be nice to patch libvirt to cache the set of capabilities per qemu binary (as determined by inode and timestamp), rather than re-probing the binary every time a domain is started, and to teach the cache how to query the monitor during the one time the probe is made rather than having to wait until a guest is started; then, a capability probe would succeed even for offline guests because it just refers to the cache, and the single check for an active domain after grabbing the job lock would be sufficient. But since that will involve a lot more coding, I'm happy to go with this simpler solution for an immediate solution. * src/qemu/qemu_driver.c (qemuDomainPMSuspendForDuration) (qemuDomainSnapshotCreateXML, qemuDomainBlockJobImpl): Check for offline state before checking an online-only cap.
-
由 Stefan Berger 提交于
Below patch fixes the following coverity findings Error: OVERRUN_STATIC: /libvirt/src/qemu/qemu_command.c:152: overrun-buffer-val: Overrunning static array "net->mac" of size 6 bytes by passing it as an argument to a function which indexes it at byte position 15. /libvirt/src/util/virnetdevmacvlan.c:948: access_dbuff_const: Calling "virNetDevMacVLanVPortProfileRegisterCallback" indexes array "macaddress" at byte position 15. /libvirt/src/util/virnetdevmacvlan.c:773: access_dbuff_const: Calling "memcpy" indexes array "macaddress" with index "16UL" at byte position 15. Error: OVERRUN_STATIC: /libvirt/src/qemu/qemu_migration.c:2744: overrun-buffer-val: Overrunning static array "net->mac" of size 6 bytes by passing it as an argument to a function which indexes it at byte position 15. /libvirt/src/util/virnetdevmacvlan.c:773: access_dbuff_const: Calling "memcpy" indexes array "macaddress" with index "16UL" at byte position 15. Error: OVERRUN_STATIC: /libvirt/src/qemu/qemu_driver.c:435: overrun-buffer-val: Overrunning static array "net->mac" of size 6 bytes by passing it as an argument to a function which indexes it at byte position 15. /libvirt/src/util/virnetdevmacvlan.c:1036: access_dbuff_const: Calling "virNetDevMacVLanVPortProfileRegisterCallback" indexes array "macaddress" at byte position 15. /libvirt/src/util/virnetdevmacvlan.c:773: access_dbuff_const: Calling "memcpy" indexes array "macaddress" with index "16UL" at byte position 15.
-
由 Stefan Berger 提交于
This patch addresses the following coverity findings: /libvirt/src/conf/nwfilter_params.c:390: var_assigned: Assigning: "varValue" = null return value from "virHashLookup". /libvirt/src/conf/nwfilter_params.c:392: dereference: Dereferencing a pointer that might be null "varValue" when calling "virNWFilterVarValueGetNthValue". /libvirt/src/conf/nwfilter_params.c:399: dereference: Dereferencing a pointer that might be null "tmp" when calling "virNWFilterVarValueGetNthValue".
-
由 Stefan Berger 提交于
This patch addresses the following coverity findings: /libvirt/src/conf/nwfilter_params.c:157: deref_parm: Directly dereferencing parameter "val". /libvirt/src/conf/nwfilter_params.c:473: negative_returns: Using variable "iterIndex" as an index to array "res->iter". /libvirt/src/nwfilter/nwfilter_ebiptables_driver.c:2891: unchecked_value: No check of the return value of "virAsprintf(&protostr, "-d 01:80:c2:00:00:00 ")". /libvirt/src/nwfilter/nwfilter_ebiptables_driver.c:2894: unchecked_value: No check of the return value of "virAsprintf(&protostr, "-p 0x%04x ", l3_protocols[protoidx].attr)". /libvirt/src/nwfilter/nwfilter_ebiptables_driver.c:3590: var_deref_op: Dereferencing null variable "inst".
-
由 Laine Stump 提交于
Some of the error messages in this function should have been virReportSystemError (since they have an errno they want to log), but were mistakenly written as netlinkError, which expects a libvirt error code instead. The result was that when one of the errors was encountered, "No error message provided" would be printed instead of something meaningful (see https://bugzilla.redhat.com/show_bug.cgi?id=816465 for an example).
-
- 26 4月, 2012 7 次提交
-
-
由 Jiri Denemark 提交于
Once qemu monitor reports migration has completed, we just closed our end of the pipe and let migration tunnel die. This generated bogus error in case we did so before the thread saw EOF on the pipe and migration was aborted even though it was in fact successful. With this patch we first wake up the tunnel thread and once it has read all data from the pipe and finished the stream we close the filedescriptor. A small additional bonus of this patch is that real errors reported inside qemuMigrationIOFunc are not overwritten by virStreamAbort any more.
-
由 Jiri Denemark 提交于
When QEMU reported failed or canceled migration, we correctly detected it but didn't really consider it as an error condition and migration protocol just went on. Luckily, some of the subsequent steps eventually failed end we reported an (unrelated and mostly random) error back to the caller.
-
由 Jiri Denemark 提交于
Currently, non-blocking calls are either sent immediately or discarded in case sending would block. This was implemented based on the assumption that the non-blocking keepalive call is not needed as there are other calls in the queue which would keep the connection alive. However, if those calls are no-reply calls (such as those carrying stream data), the remote party knows the connection is alive but since we don't get any reply from it, we think the connection is dead. This is most visible in tunnelled migration. If it happens to be longer than keepalive timeout (30s by default), it may be unexpectedly aborted because the connection is considered to be dead. With this patch, we only discard non-blocking calls when the last call with a thread is completed and thus there is no thread left to keep sending the remaining non-blocking calls.
-
由 Jiri Denemark 提交于
In some cases (spotted with broken connection during tunneled migration) we were overwriting the original error with worse or even misleading errors generated when we were cleaning up after failed migration.
-
由 Peter Krempa 提交于
The docs for virConnectSetKeepAlive() advertise that this function should be able to disable keepalives on negative or zero interval time. This patch removes the check that prohibited this and adds code to disable keepalives on negative/zero interval. * src/libvirt.c: virConnectSetKeepAlive(): - remove check for negative values * src/rpc/virnetclient.c * src/rpc/virnetclient.h: - add virNetClientKeepAliveStop() to disable keepalive messages * src/remote/remote_driver.c: remoteSetKeepAlive(): -add ability to disable keepalives
-
由 Laine Stump 提交于
This patch resolves https://bugzilla.redhat.com/show_bug.cgi?id=815270 The function virNetDevMacVLanVPortProfileRegisterCallback() takes an arg "virtPortProfile", and was checking it for non-NULL before using it. However, the prototype for virNetDevMacVLanPortProfileRegisterCallback had marked that arg with ATTRIBUTE_NONNULL(). Contrary to what one may think, ATTRIBUTE_NONNULL() does not provide any guarantee that an arg marked as such really is always non-null; the only effect to the code generated by gcc, is that gcc *assumes* it is non-NULL; this results in, for example, the check for a non-NULL value being optimized out. (Unfortunately, this code removal only occurs when optimization is enabled, and I am in the habit of doing local builds with optimization off to ease debugging, so the bug didn't show up in my earlier local testing). In general, virPortProfile might always be NULL, so it shouldn't be marked as ATTRIBUTE_NONNULL. One other function prototype made this same error, so this patch fixes it as well.
-
由 Eric Blake 提交于
Commit 8fe455fd tried to work around a regression introduced in upstream gnulib that requires gettext 0.18 or newer on all projects using bootstrap, by making libvirt require gettext 0.18. But this fails on RHEL 6.2, which still ships gettext 0.17. Revert that change, and instead, import the latest round of gnulib updates that fix that problem properly. If you have already built in the window where libvirt required 0.18, be aware that incremental updates may run into problems: this is because 'autopoint --force' will not downgrade m4/po.m4 back to an older version, but it must be downgraded back to 0.17 levels to work with this patch. You may either manually remove that file then rerun bootstrap, or it may prove easier to just clean up all non-git files to start from a clean slate. * bootstrap.conf: Revert minimum gettext back to 0.17. * configure.ac: Likewise. * .gnulib: Update to latest, for bootstrap fixes. * bootstrap: Resync from gnulib.
-
- 25 4月, 2012 1 次提交
-
-
由 Stefan Berger 提交于
Add 2 new functions to the virSocketAddr 'class': - virSocketAddrEqual: tests whether two IP addresses and their ports are equal - virSocketaddSetIPv4Addr: set a virSocketAddr given a 32 bit int
-