Set fd non-blocking to avoid common use cases (like reading from a
named pipe) from hanging the agent. This was missed in the original
code.
The patch introduces qemu_set_handle_nonoblocking, the local analog
of qemu_set_nonblock for HANDLES.
The usage of handles in qemu_set_non/block is impossible, because for
win32 there is a difference between file discriptors and file handles,
and all file ops are made via Win32 api.
Signed-off-by: Olga Krishtal <okrishtal@parallels.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
CC: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
CloseHandle use HANDLE as an argument, but not *HANDLE
Signed-off-by: Olga Krishtal <okrishtal@parallels.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
qobject_to_qdict() crashes on null, which is a trap for the unwary.
Return null instead, and simplify a few callers.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1444918537-18107-4-git-send-email-armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
Implemented with base64-encoded strings in qga json protocol.
Glib portable GIOChannel is used for data I/O.
Optinal stdin parameter of guest-exec command is now used as
stdin content for spawned subprocess.
If capture-output bool flag is specified, guest-exec redirects out/err
file descriptiors internally to pipes and collects subprocess
output.
Guest-exe-status is modified to return this collected data to requestor
in base64 encoding.
Signed-off-by: Yuri Pudgorodskiy <yur@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
* switch from 'struct GuestIOExecData' to 'GuestIOExecData'
* s/TRUE/true/g, s/FALSE/false/g for gboolean return values
* s/inp_data/input_data/
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
glib may return G_IO_STATUS_AGAIN which is actually not an error.
Also fixed a bug when on incomplete write buf pointer was not adjusted.
Signed-off-by: Yuri Pudgorodskiy <yur@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
qemu-ga should not exit on guest-file-write to pipe without read end
but proper error code should be returned. The behavior of the
spawned process should be default thus SIGPIPE processing should be
reset to default after fork() but before exec().
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Yuri Pudgorodskiy <yur@virtuozzo.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Guest-exec rewritten in platform-independent style with glib spawn.
Child process is spawn asynchronously and exit status can later
be picked up by guest-exec-status command.
stdin/stdout/stderr of the child now is redirected to /dev/null
Later we will add ability to specify stdin in guest-exec command
and to get collected stdout/stderr with guest-exec-status.
Signed-off-by: Yuri Pudgorodskiy <yur@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
* use g_new0 in place of g_malloc for GuestExec struct
* commit msg spelling fixes
* s/inp-data/input-data
* document capture-input mode as false by default
* use GetProcessId() for pids on w32 instead of casting HANDLE
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
This just makes code shorter and better.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Yuri Pudgorodskiy <yur@virtuozzo.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Some guests don't expose memory blocks via sysfs at all. This
shouldn't be a failure, instead just return an empty list. For
other access failures we still report an error.
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Move the default verbosity settings before loading the configuration
file, or it will overwrite it. Found thanks to writing qga tests :)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Having a environment variable allows to override default configuration
path, useful for testing. Note that this can't easily be an argument,
since loading config is done before parsing the arguments.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
g_new(T, n) is neater than g_malloc(sizeof(T) * n). It's also safer,
for two reasons. One, it catches multiplication overflowing size_t.
Two, it returns T * rather than void *, which lets the compiler catch
more type errors.
This commit only touches allocations with size arguments of the form
sizeof(T). Same Coccinelle semantic patch as in commit b45c03f.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
For consistency with the rest of the comment blocks.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
A number of source files have statements accidentally
terminated by a double semicolon - eg 'foo = bar;;'.
This is harmless but a mistake none the less.
The tcg/ia64/tcg-target.c file is whitelisted because
it has valid use of ';;' in a comment containing assembly
code.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
This is particularly useful when we abort in error_propagate(),
because there the stack backtrace doesn't lead to where the error was
created. Looks like this:
Unexpected error in parse_block_error_action() at .../qemu/blockdev.c:322:
qemu-system-x86_64: -drive if=none,werror=foo: 'foo' invalid write error action
Aborted (core dumped)
Note: to get this example output, I monkey-patched drive_new() to pass
&error_abort to blockdev_init().
To keep the error handling boiler plate from growing even more, all
error_setFOO() become macros expanding into error_setFOO_internal()
with additional __FILE__, __LINE__, __func__ arguments. Not exactly
pretty, but it works.
The macro trickery breaks down when you take the address of an
error_setFOO(). Fortunately, we do that in just one place: qemu-ga's
Windows VSS provider and requester DLL wants to call
error_setg_win32() through a function pointer "to avoid linking glib
to the DLL". Use error_setg_win32_internal() there. The use of the
function pointer is already wrapped in a macro, so the churn isn't
bad.
Code size increases by some 35KiB for me (0.7%). Tolerable. Could be
less if we passed relative rather than absolute source file names to
the compiler, or forwent reporting __func__.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
requester.cpp uses this pattern to receive an error and pass it on to
the caller (err_is_set() macro peeled off for clarity):
... code that may set errset->errp ...
if (errset->errp && *errset->errp) {
... handle error ...
}
This breaks when errset->errp is null. As far as I can tell, it
currently isn't, so this is merely fragile, not actually broken.
The robust way to do this is to receive the error in a local variable,
then propagate it up, like this:
Error *err = NULL;
... code that may set err ...
if (err)
... handle error ...
error_propagate(errset->errp, err);
}
See also commit 5e54769, 0f230bf, a903f40.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
qga_vss_fsfreeze() casts error_set_win32() from
void (*)(Error **, int, ErrorClass, const char *, ...)
to
void (*)(void **, int, int, const char *, ...)
The result is later called. Since the two types are not compatible,
the call is undefined behavior. It works in practice anyway.
However, there's no real need for trickery here. Clean it up as
follows:
* Declare struct Error, and fix the first parameter.
* Switch to error_setg_win32(). This gets rid of the troublesome
ErrorClass parameter. Requires converting error_setg_win32() from
macro to function, but that's trivially easy, because this is the
only user of error_set_win32().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Use NetUserSetInfo() to set the user password.
This function is notoriously known to be problematic for users with EFS
encrypted files. But the alternative, NetUserChangePassword() requires
the old password. Nevertheless, The EFS file should be recovered by
changing back to the old password.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
This new option allows to review the agent configuration,
and ease the task of writing a configuration file.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
* removed unecessary keyfile != NULL prior to free
* documented --dump-conf is qemu-ga --help output
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Learn to configure the agent with a system configuration.
This may simplify command-line handling, especially when the blacklist
is long.
Among the other benefits, this may standardize the configuration of an
init service (instead of distro-specific init keys/files)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
* removed unecessary keyfile != NULL prior to free
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Now that main() has a single exit point, we can free a few
more allocations.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Once the options are populated, move the running state to
a run_agent() function.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
* fixed up an s/ga_state/s/ artifact causing segfault
* replaced g_list_free_full with g_list_foreach to maintain glib
2.22 compatibility
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Fill all default options during main(). This is a preparation patch
to allow to dump the configuration.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Move option parsing out of giant main().
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Following patch will return allocated strings, so we must correctly
initialize alloc & free them. The nice side effect is that we no longer
have to check for "fixed_state_dir" to call ga_install_service() with a
NULL state dir. The default values are set after parsing the command
line options.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
'path' is already a global function, rename the variable since it's
going to be in global scope in a later patch.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
In order to avoid any confusion, let's allocate new strings when
splitting.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The function is going to be reused in a later patch.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The option parsing is going to be moved to a separate function,
use exit() consistently.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Previously, if building out-of-tree, the MSI build would fail since
it wasn't able to find the needed files.
Signed-off-by: Leonid Bloch <leonid@daynix.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
* fixed up commit msg formating
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Previously, running the .msi would unregister the QEMU GA VSS service if QEMU GA was already installed on the machine, and then register it only if QEMU GA was NOT previously installed. This behavior caused the service to be registered only after the INITIAL installation, and any subsequent run of the .msi (to redo, repair, or upgrade the installation) ended in the service being unregistered.
Now, the VSS service is still unregistered if QEMU GA is already installed (so that a fix or an update could be performed) but then it is registered again (if the GA is not being uninstalled) thus finishing the repair/upgrade correctly. Additionally, downgrading is now prevented. If a user would like to downgrade a version, he/she must uninstall the newer version first.
Signed-off-by: Leonid Bloch <leonid@daynix.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
This is done to follow the recommendations given here: https://msdn.microsoft.com/en-us/library/aa368269%28VS.85%29.aspx
Signed-off-by: Leonid Bloch <leonid@daynix.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
For compatibility, all the letters in GUID should be capital.
Signed-off-by: Leonid Bloch <leonid@daynix.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Olga Krishtal <okrishtal@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
*added semi-colon to better delineate 2.2 vs. 2.4 versioning
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
PCIAddress inforfation is obtained via SetupApi, which provides the
information about address, bus, etc. We look throught entire device tree
in the system and try to find device object for given volume. For this PDO
SetupDiGetDeviceRegistryProperty is called, which reads PCI configuration
for a given devicei if it is possible.
This is the most convinient way for a userspace service. The lookup is
performed for every volume available. However, this information is
not mandatory for vss-provider.
In order to use SetupApi we need to notify linker about it. We do not need
to install additional libs, so we do not make separate configuration
option to use libsetupapi.su
SetupApi gives as the same information as kernel driver
with IRP_MN_QUERY_INTERFACE.
https://support.microsoft.com/en-us/kb/253232
Signed-off-by: Olga Krishtal <okrishtal@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
* stub out get_pci_info if !CONFIG_QGA_NTDDSCSI
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
According to Microsoft disk location path can be obtained via
IOCTL_SCSI_GET_ADDRESS. Unfortunately this ioctl can not be used for all
devices. There are certain bus types which could be obtained with this
API. Please, refer to the following link for more details
https://technet.microsoft.com/en-us/library/ee851589(v=ws.10).aspx
Bus type could be obtained using IOCTL_STORAGE_QUERY_PROPERTY. Enum
STORAGE_BUS_TYPE describes all buses supported by OS.
Windows defines more bus types than Linux. Thus some values have been added
to GuestDiskBusType.
Signed-off-by: Olga Krishtal <okrishtal@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
* fixed warning in CreateFile due to use of NULL instead of 0
* only provide disk info when CONFIG_QGA_NTDDSCSI=y
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
We should use GetVolumeXXX api to work with volumes. This will help us to
resolve the situation with volumes without drive letter, i.e. when the
volume is mounted as a folder. Such volume is called mounted folder.
This volume is a regular mounted volume from all other points of view.
The information about non mounted volume is reported as System Reserved.
This volume is not mounted and thus it is not writable.
GuestDiskAddressList API is not used because operations are performed with
volumes but no with disks. This means that spanned disk will
be counted and handled as a single volume. It is worth mentioning
that the information about every disk in the volume can be queried
via IOCTL_VOLUME_GET_VOLUME_DISK_EXTENTS.
Signed-off-by: Olga Krishtal <okrishtal@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
We need qmp_quest_get_fsinfo togather with vss-provider, which works with
volumes. The call to this function is implemented via
FindFirst/NextVolumes. Moreover, volumes in Windows OS are filesystem unit,
so it will be more effective to work with them rather with devices.
Signed-off-by: Olga Krishtal <okrishtal@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
It's possible to set system time with dates after 2070, however, it's
not possible to set the RTC. It has limitation to up to year
2070 (1970+100). In order to keep both clock in sync and before the
kernel complains on invalid values, bail out early.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
By default, IPv4 prefixes will be derived by matching the address
to those returned by GetAdaptersInfo. IPv6 prefixes can not be
matched this way due to the unpredictable order of entries.
In Windows Vista/2008 guests and newer, both IPv4 and IPv6 prefixes
can be retrieved from OnLinkPrefixLength. Setting --extra-cflags
in the build configuration to "-D_WIN32_WINNT=0x600"
or greater makes OnLinkPrefixLength available. Setting --extra-cflags
is not required and if not set, the default approach to get the prefix
will be taken.
Signed-off-by: Kirk Allan <kallan@suse.com>
* drop ws2ipdef.h, it's missing on old mingw, and ws2tcpip.h already
includes it automatically on new builds
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Since we now require GLib 2.22+ (commit f40685c), we don't have to
work around lack of g_strcmp0() anymore.
This reverts commit 8f47747899.
Conflicts:
qemu-ga.c
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The current guest-fstrim support only returns an error if some
mountpoint was unable to be trimmed, skipping any possible additional
mountpoints. The result of the TRIM operation itself is also discarded.
This change returns a per mountpoint result of the TRIM operation. If an
error occurs on some mountpoints that error is returned and the
guest-fstrim continue with any additional mountpoints.
The returned values for errors, minimum and trimmed are dependant on the
filesystem, storage stacks and kernel version.
Signed-off-by: Justin Ossevoort <justin@quarantainenet.nl>
* s/type/struct/ in schema type definitions
* moved version annotation for new guest-fstrim return field to
the field itself rather than applying to the entire command
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The FITRIM ioctl updates the fstrim_range structure it receives. This
way the caller can determine how many bytes were trimmed. The
guest-fstrim logic reuses the same fstrim_range for each filesystem,
effectively limiting each filesystem to trim at most as much as the
previous was able to trim.
If a previous filesystem would have trimmed 0 bytes, than the next
filesystem would report an error 'Invalid argument' because a FITRIM
request with length 0 is not valid.
This change resets the fstrim_range structure for each filesystem.
Signed-off-by: Justin Ossevoort <justin@quarantainenet.nl>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
These macros expand into error class enumeration constant, comma,
string. Unclean. Has been that way since commit 13f59ae.
The error class is always ERROR_CLASS_GENERIC_ERROR since the previous
commit.
Clean up as follows:
* Prepend every use of a QERR_ macro by ERROR_CLASS_GENERIC_ERROR, and
delete it from the QERR_ macro. No change after preprocessing.
* Rewrite error_set(ERROR_CLASS_GENERIC_ERROR, ...) into
error_setg(...). Again, no change after preprocessing.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
