Main/Console: report installed driver versions to VM release log

Main,FE/VBoxManage: Support exporting machines as appliances which have encrypted disks. Because the OVF standard doesn't support encrypted disks so far we always decrypt exported images which requires the password before starting the export proess

Assorted fixes for FreeBSD hosts, VBox compiles and runs again without further patches (tested on 10.1 amd64 )

Disable building the KEyStore for now because of too old openssl system libraries on some Linux hosts

Main/KeyStore: Simple keystore interface to store single DEKs protected by a password

Missing dependency

testconverter and testconverter2 are outdated.

cleanup

Main: Split the wrapper source generator job into two to speed reduce the time it takes before we can start compiling wrappers. In the process, the xslt code responsible for generating vboxapiwrap.kmk was moved into apiwrap-server-filelist.xsl to prevent more clutter in the original file. Down to 6 seconds for generating wrapper sources, headers and trace points.

apiwrap-server.xsl: Optimizing the parameter code a little, the no-tmp dtrace variant paid off quite a bit (most requently used). Got it down another 5-6 seconds here (ca 18 now).

apiwrap-server.xsl: Use xsl:key + key() to lookup interfaces and enums by @name. Since key works on current document, the super tricky stuff needed to make sure $G_root was select as current, so I took it one step further and make the //interface[@name=$addifname] current as well. Pushing that a step further, and the need for $iface was eliminated. Changes save 5-6 seconds here on a 30 second run.

micro opts.

oops, didn't mean to commit kmk_time.

More micro opts

shut up.

Main/Makefile.kmk: Cleanups and fixed VBoxAPI.d generation.

apiwrap-server.xsl: Fixed a couple of bugs in previous commit and optimize the filelist stuff, keeping it from going unecessarily deep and adding unecessary conditionals.

apiwrap-server.xsl: Generate headers and VBoxAPI.d in separate runs from the source files so kmk can start working on other source files while the wrapper C++ source files are being generated (slow).

build fix

Use the native line ending because kmk_append does and we don't like mixing EOL for a single file (VBoxAPIWrap.kmk).

Main/Makefile.kmk: dtrace fixes.

Main: Added machine state change probe.

Main: Added dtrace probes to the auto generated wrappers (half disabled - VBOX_WITH_DTRACE_R3_MAIN).

fixed resource files

not yet

addded more Windows resource files

Some fixes for windows cross build using wine.

PDM Audio: Branch -> trunk.

Main/HostDnsService: Tcpip\Parameters\NameServer value is not updated when the list of name servers is statically configured, so don't inspect NameServer and DhcpNameServer directly - instead use DnsQueryConfig(DnsConfigDnsServerList, ...) from DnsAPI.dll

DisplayImpl: move legacy guest interfaces to DisplayImplLegacy.cpp.

VBox/Main: #1909: Independent QMTranslator implementation. Some fixes to Utf8Str (added number of characters to be copied when created from CBSTR without terminating zero), added swap method to RTCString for fast exchange between strings. Makefile for VBoxSVC updated in order to compile with QMTranslatorImpl.cpp

Main: build fix.

VBoxRT.dll,VBoxRT-x86.dll,VBoxC.dll: Delay load less frequently used DLLs, by all means try avoid loading user32.dll before it's actually needed.

Main: ISession partially converted to use API wrappers

Main: AutoCaller/VirtualBoxBase refactoring, cleanly splitting out the object state handling, and moving all caller synchronization to one file. Also eliminated a not so vital template (AutoCallerBase) by much simpler inheritance. Theoretically has no visible effects, the behavior should be identical. Done as a preparation for reimplementing the caller synchronization.

DnD: API overhaul; now using IDnDTarget + IDnDSource. Renamed DragAndDrop* enumerations to DnD*. Also rewrote some internal code.

Main,Frontends: IDisplay provides the guest screen bitmap to frontends.

Main: more trivial FIPS preparations

Java COM bindings: fix much bit rot, accidental breakage of the code due to a change meant for WSDL, use an exception class which is available, remove optimization not done in the other API variants, eliminate java warning and fix Makefile

Main: don't build docs.Main if VBOX_WITH_ALL_DOXYGEN_TARGETS is disabled

6813 src-all/ProgressImp.cpp + some formatting/line length sorting

pr6022. Support X509 certification during import OVF appliance.

src/VBox/Devices/Audio, src/VBox/Main/src-client, include/VBox/vmm: src/VBox/Devices/Audio: part of restructuring of audio code. Devices files correspondin to Hda, AC97 and SB16 audio. The structure of files have been modifed as per PDM specs. The modified code is under #ifdef VBOX_WITH_PDM_AUDIO_DRIVER src/VBox/Main/src-client: Driver for the VRDE that interacts with DrvAudio. Enhancement of the CFGM tree for audio. Config.kmk : addition of one configuration parameter that will control whether new audio code is disabled or enabled. "VBOX_WITH_PDM_AUDIO_DRIVER" pdmaudioifs.h: common header file between Device , Intermediate audio driver and Backends specific to audio.

Main: typo

6813 - src-client/GuestDirectoryImpl.cpp

6813 - src-client/GuestDirectoryImpl.cpp - changes backed out due to error on winrel clbr 92400

6813 - src-client/GuestDirectoryImpl.cpp

6813 - src-client/GuestFsObjInfoImpl.cpp wrapper issues sorted

6813 - r92358 changes backed out due to errors on win-dbg clbr and win-rel clbr

6813 - GuestFsObjInfoImpl.cpp 2nd attempt

DnD: Update.

crOpenGL: video recording working

Main/HostDnsService: - Generic part could be used with threading on or off: OS which can monitor changes on DNS structures might need this (e.g. Windows, Darwin, Linux) - ResolvConf based (Linux, Solaris, FreeBSD ans Os/2) doesn't use scanf based resovl.conf parsing new Slirp's parser used instead. It's configured to provide strings instead of addresses. (to resolve xtracker/7034c2) - Darwin monitoring code has been refactored. - linux changed to handle cases described in xtracker/7034c3 - Windows monitors changes on register changes.

Enabled the tarvfs.cpp read code in main. tar.cpp is only used for writing now.

Main: Working on reducing RTTar to a write-only interface. Not there yet, though.

Main: use different file names for IID file, otherwise the SDK build gets confused as it handles both COM and XPCOM on one system

Main: finally handle the IID and CLSID stuff on XPCOM the same way as on MSCOM, those hacks won't be missed,

Devices/Graphics: VMware SVGA II compatible graphics emulation (2D only), including the associated small API and VBoxManage changes, contributed by trivirt AG.

32-bit on 64-bit proxy dll (still disabled), made comregister.cmd register the proxy stuff when present.

Extended the test, nothing exciting.

Main/*.kmk: Moved a couple of global variables into Config.kmk so the testcase/Makefile.kmk can use them. Using variables from a parent makefile in a sub-makefile is absolutely illegal! (If it affects compilation in anyway.)

Quick midl proxy dll experiment.

Main: Some simple COM/XPCOM performance measurements. The idea was getting a baseline before experimenting with proxy/stub stuff in midl. There is some VirtualBox.idl preps related to that. Unforuntately, the proxy/stub generator doesn't support SAFEARRAY parameters with interface pointers other than IUnknown. So, we're probably up against something difficult here.

depend build fix, 2nd try.

Another non-kBuild target property.

depend build fix

Main: tstCollector need to require the wrapper generated header files to exist in order to avoid it being built before they are generated. This necessitated some changes that lead to some more VBoxApiWrap related cleanups.

Main/Makefile.kmk: kmk clean fix

Main/Makefile.kmk: Please, DO NOT invent your own kBuild target properties without at least prefixing them with VBOX_ to indicate that kBuild doesn't have any magic spells for any of them. It's _very_ confusing for everyone reading the makefile.

Main: build fixes.

Runtime+Main: the x86 directory is only used for Windows, Solaris doesn't need it right now

Main/VBoxClient-x86: attempt to fix packaging, needs VBoxRT-x86.dll, and has to move to a separate directory to allow having 32 bit msvc*.dll in the same directory

Makefile.kmk: adds HostDnsServiceLinux.cpp to build on Linux host HostDnsService.h: adds declarations of defined functions.

Main/HostVideoInputDeviceImpl: use VBoxHostWebcamList.

move VBoxVolInfo to the correct place and export it

Main: Added KBUILD_TYPE to the release log.

VBoxManage,Main: VideoCaptureDevice -> VideoInputDevice

VBoxManage,Main: IHost::VideoCaptureDevices. Windows host implementation.

IHost::VideoCaptureDevices

Config.kmk,Main/Makefile.kmk: Use quotes instead of escaping the semicolon on dos-systems. The VBOX_SEP is useless outside shell command lines otherwise.

Main: don't build VBoxAPIWrap for SDK

Main/Machine+Session: New generic client session watcher implementation based on token objects, works on all platforms and is used for now on XPCOM. Additionally a better error message when several API clients are racing for a lock, previously it could be quite confusing.

Main,VBoxManage: Implemented IConsole::EmulatedUSB. Removed IMachine::emulatedUSBWebcameraEnabled.

Add src-server/HostDnsService.cpp forgotten in previous.

Main/DNS: adds dns service bits to build.

wine build fix.

Main,Installer: 32-bit on 64-bit COM API support (windows). Currently disabled

Main/Medium: redesign API level medium locking, needed conversions from MediaList to MediumLockLists in several places, forced cleanups elsewhere, too Main/Token: introduced token objects for controlling the unlocking, will be used as a general concept in the future Main/Snapshot: snapshot deletion needed significant cleanups as it was still using many shortcuts, directly calling the API to lock media instead of using lock lists. Now much better, and the online snapshot deletion is also a lot cleaner as it no longer passes unnecessary parameters around which are already known in the machine/snapshot code Main/MediumLock: small improvements, now has a mode which skips locking already locked media, needed by the Snapshot code where we have overlapping lock lists and have to update the original one instead Main/Console+Session+Machine: follow-up changes for the online snapshot merging parameter passing simplification, plus an unrelated lock order violation fix in Machine which happens only for inaccessible machines Main/testcase: update correspondingly doc: update SDK reference

32-bit main API on 64-bit solaris.

Main/idl: Fix small VirtualBox.xidl issues (showing the need to routinely validate the xidl file), bring the DTD up to date, and add a target for validating the xidl file. Unfortunately we don't have the tools everywhere to make this part of the build.

Main/VirtualBox+Machine+Session: separate out the client death detection functionality into separate objects

Main/USB: USB Controller implementation rework. Moved filter handling into a separate interface

Main: re-commit r86967 Main/Network: DHCP server has got the ear in Main, and we able create/describe more complex infrostructures. DHCP server together with Lwip NAT can handle per vm/slot configuration and store them in xml settings. place-holder: Host interface nameserver list, domain name and search strings, I suppose that this functions should be used on initialization stage and then on host configuration change even or directly from event.

backed out r86967,r86968,r86969,r86970,r86971,r86972,r86973,r86975,r86976. will fix build locally.

Main/Network: DHCP server has got the ear in Main, and we able create/describe more complex infrostructures. DHCP server together with Lwip NAT can handle per vm/slot configuration and store them in xml settings. place-holder: Host interface nameserver list, domain name and search strings, I suppose that this functions should be used on initialization stage and then on host configuration change even or directly from event.

Forward ported r85941 and required build fixes (Main: Implemented new event queue to separate system's native event queue and our own. Also, XPCOM is not needed for handling our own events. On Windows this also fixes the system's queue quota limitation).

VBoxDef2LazyLoad: Added tool for generating lazy loading of VBoxVMM.dll to prevent it being loaded into VBoxSVC and the VM selector.

Main: Removed IGuestErrorInfo, added new attribute resultDetail to IVirtualBoxErrorInfo for (optionally) providing more details on the error happened.

several fixes for video recording

Main: Code generator for (xp)com API implementations, including logging and parameter conversion, so far only used by MediumFormat. Next try, needed significant tweaks to work with xpcom (safearray handling fixes in the parameter conversion helpers), different STL implementation (which doesn't support declaring template type parameters as const), missing build dependencies (which didn't show on the dual core system used for writing the code), and finally the duplicate XPCOM classinfo and AddRef/Release/QueryInterface method definitions needed to be removed.

back out API wrapper generator, too broken with XPCOM

Main/Makefile.kmk: delete debugging leftover

Main: Code generator for (xp)com implementations, including logging and parameter conversion, so far only used by MediumFormat.

GuestCtrl: Implemented using (public) VirtualBox events instead of own callback mechanisms. Bugfixes for testcases (still work in progress).

Ring-1 compression patches, courtesy of trivirt AG: - main: diff to remove the hwvirt requirement for QNX - rem: diff for dealing with raw ring 0/1 selectors and general changes to allowed guest execution states - vmm: changes for using the guest's TSS selector index as our hypervisor TSS selector (makes str safe) (VBOX_WITH_SAFE_STR ) - vmm: changes for dealing with guest ring 1 code (VBOX_WITH_RAW_RING1) - vmm: change to emulate smsw in RC/R0 (QNX uses this old style instruction a lot so going to qemu for emulation is very expensive) - vmm: change (hack) to kick out patm virtual handlers in case they conflict with guest GDT/TSS write monitors; we should allow multiple handlers per page, but that change would be rather invasive

Main/NATNetworks: API+XML serialization for NATNetworks.

reverted: r84411. (need to solve issues with java first.)

Main/NATNetwork API (xTracker/5894).

Main: make VBOX_ONLY_SDK=1 build on Windows hosts

Main/SystemProperties+Machine: new config setting for default VM frontend. Frontend/VirtualBox+VBoxManage: changes to use the default VM frontend when starting a VM, other minor cleanups Main/xml/*.xsd: attempt to bring the XML schema close to reality doc/manual: document the new possibilities, and fix a few long standing inaccuracies

Network-service for lwip-driven proxy. additionally enables SrvNat trunk type for Nat services. Note: all changes disabled by default.

header (C) fixes

VRDP: cleanup: removed obsolete VBOX_WITH_VRDP_VIDEO_CHANNEL.

Main/Metrics: Disk and filesystem metrics for Solaris (#6345)

EFI: permanent NVRAM storage.

*.kmk: Assume VBOX_WITH_COCOA_QT is always set.

Main/Guest control: Removed old API.

Main/EncodeAndVideoRecording Module and API implementation: Integrating an independent encoding and video recording module that will serve all the frontends. Introducing settings settings and API implementation for accessing and modifying video recording parameters: ->target video capture file ->video capture width ->video capture height ->enable video capturing

Guest Control 2.0: Update.

Main: big API naming cleanup, use all caps acronyms everywhere, including SDK docs Frontends/VBoxManage: implement guestcontrol execute for new API, disabled by default

Only find a suitable midl.exe once (Main and VBoxNetFlt was doing it).

case insentitivity paranoia

whatever...

build fix

SDK build fix.

Jettisoned the old compilers.

enabled shared clipboard support for Linux hosts (guest=>host only)

Guest Control 2.0: Update.

VS2010 preps.

Guest Control 2.0: Hacking in progress.

More autostart service updates, work in progress

**/Makefile.kmk: Cosmetic changes.

Main: Delay import VBoxVMM.dll from VBoxC.dll on windows. This is an attempt at avoiding loading VBoxVMM.dll into VBoxSVC.exe.

Main: remote smartcard

Main+Frontends: removed unused and confusing VirtualBoxErrorInfo implementation, and cleaned up lots of misleading comments and other leftovers about the earlier ErrorInfo mess

Main+Frontends: move release logging setup to the glue code (to eliminate spreading code duplication), various minor fixes and cleanups

NetCfg/win: disable/enable/update HostOnly

Main/Makefile.kmk: correctly handle COM sdk binding files in INSTALL, avoids clumsy manual copying including the related dependency problems

lib vars fixes

yet another err, sigh..

osx burn fix

3D check: direct invocation on the client side

NetShaper: TX thread implementation (#5582)

hash the teleporter token.

Ancient Solaris crossbow/TAP no longer relevant.

spaces.

Main: initial DnD support

Main/glue/tests: clean up sample Java app using the glue code

Main/glue: Add javadoc comments to the java glue code, use them to generate a documentation jar file. Add an OSGi compliant manifest file to the webservice jar.

whitespace fix

GuestCtrl: Update.

GuestCtrl: Update.

Main: bits for debugging cardreader.

build fix

*.kmk,++: Mode fixes to maximize hardlinking potential during staging.

GuestCtrl: Update of copy from guest (work in progress).

*.kmk: Installation/staging adjustments (kBuild 0.2.0 preps).

eliminate use of LOG_ENABLED

ain/Makefile.kmk: Two missing slashes in _INST properties.

fix OSE

OSE build fix

Main/VBoxSVC: add release logging, with conservative log rotation settings enabled by default

Main-CloneVM: factor out the clone VM task

Main,ExtPacks/Puel/UsbWebcam: UsbWebcam intergration.

API+Frontends: Generic network attachment driver support which obsoletes the special case for VDE. Big API cleanup in the same area. Adapt all frontends to these changes (full implementation in VBoxManage, minimum implementation in GUI).

usb rework

Main/Makefile.kmk: Re r70893: TARGET_xxx -> xxx_1_TARGET. The former form is deprecated.

netflt/adp/win: switch to new VBoxDrvCfg lib

Main/src-server/xpcom: Avoid initializing static variables with a constructor. Unsafe c++, and gcc on freebsd choked on it when optimizing. Who knows which gcc version will be next.

Main/VBoxManage: Update on guest facilities, added IAdditionsFacility; now it's possible to iterate over known facilities (see VBoxManage showvminfo --details). Also, facilities now are grouped into several classes (AdditionsFacilityClass).

Main, Device: passthrough conditional

Main, VMM: PCI passthrough work

typo

docs build fix

The DIRDEP is deprecated, use dir/ or $$(dir $$@) instead.

Makefiles: The PATH_<target> variable is deprated, use <target>_0_OUTDIR instead.

moved comregister.cmd.

build fix.

Main: source re-org.

backed out 69217

VMM, Main: PCI passthrough work

Main/Display: removed obsolete code.

Main/Display, VBoxTray: removed obsolete code.

introduced VBOX_WITH_S3

Main: Moved Guest Control code to own file for better separation.

VBoxSVC: solaris should preload VBoxVMM.so so a development build won't have any problems loading the retail extension pack.

VBoxExtPackHelperApp: Authorize root execution the standard way and drop the set-uid-to-root stuff on darwin.

drop the console... though that makes it a pop-under-blinking-task-bar thing. hmm.

Main: Bandwidth groups for disks (and later network) This introduces two new interfaces. The first one named IBandwidthGroup represents one I/O limit and can be assigned to several mediums which share this limit (which works only for harddisk images with the disabled host cache). The second one IBandwdithControl manages the groups and can create new ones and destroy them if not required anymore. VBoxManage: commands to access the bandwidth groups Syntax: VBoxManage storageattach <uuid|vmname> ... --bandwidthgroup <name> --bandwidthgroup assigns the specified device to the given group. VBoxManage bandwidthctl <uuid|vmname> --name <name> --add disk|network --limit <megabytes per second> --delete The --name parameter gives the name of the bandwidth group. --add creates a new group of the given type (only disk is implemented so far) with the given name. --limit sets the limit to the given amount of MB/s Note that limit can be changed while the VM is running. The VM will immediately pick up the new limit for the given group name. --delete deletes the group with the given name if it isn't used anymore. Trying to delete a still used group will result in an error. Example: VBoxManage bandwidthctl "Test VM" --name Limit --add disk --limit 20 Creates a group named Test having a 20 MB/s limit. VBoxManage storageattach "Test VM" --storagectl "SATA Controller" --port 0 --device 0 --type hdd --medium test.vdi --bandwidthgroup Limit Adds a new disk to the SATA controller and assigns the bandwidth group Limit to it. VBoxManage storageattach "Test VM" --storagectl "SATA Controller" --port 0 --device 0 --type hdd --medium test.vdi --bandwidthgroup none Removes the bandwidth limit from the disk. VBoxManage bandwidthctl "Test VM" --name Limit --add disk --limit 10 Changes the limit of bandwidth group Limit to 10 MB/s. If the VM is running the limit will be picked up immediately.

Main/VirtualBoxClient: new client-side singleton interface for convenient monitoring of VBoxSVC liveness, plus a big whitespace cleanup of the xidl file

Main: remove dead and obsolete code

Main: better type checks when firing events

PCI: more 4.0 interfaces

Main: PCI passthrough work

builds on windows, removing hack.

build fix #1

Let's see VBoxExtPackHelperApp this builds on windows now...

space

ExtPack changes, related IPRT changes.

Main: Added ExtPackManager to Console and implemented the Console and VirtualBox hooks.

Main: More ExtPack code.

PCI: part 1 of PCI slots assignment logic in Main

Disabled the building of VBoxExtPackHelperApp.

build fix

ExtPack: More code.

VRDE: renamed VRDPServer.h, .cpp to VRDEServer.

VRDE: removed VBOX_WITH_VRDP from source code, also some obsolete code removed.

ExtPack: Sketches and some code.

Removed VBOX_DYNAMIC_NET_ATTACH, we don't need to keep this optional.

wddm/3d: chromium hgsmi, host part + guest part debugging

Main;OVF/OVA: online calculation of the SHA1 sum; directly stream into the ova (no temporary files anymore); cache writing

Guest Copy/VBoxManage+Main: Added very first bits (not enabled by default yet, untested).

backed out accidental changes

com/string: Windows build fixes

Devices: refactoring, further PCI work

Main: Save guest dimension in save state; add method for queering this info to IMachine.

Main/USBProxyServiceLinux: split out USB enumeration from USBProxyServiceLinux.cpp and HostHardwareLinux.cpp into a separate file

Main/Makefile.kmk: $(QUIET)

*: a few OTHER_CLEAN typos and fixes

quiet

VRDP leak detector: use the same define for VRDP and libjpeg, disabled the detector in VBoxC

Applied a simple memory leak detector to libjpeg and VBoxC (Windows host only), the code completely disabled by default.

Main: zlib needed by PNG

Main: thumbnail PNG API, refactoring

make the VBOX_WITH_NEW_USB_CODE_ON_SOLARIS code mandatory

Main/Solaris: New DVD drive enumeration, works on Solaris 10 & Nevada. Removed now obsolete NSL_RESOLVED stuff on Solaris.

Java: samples rule

Java bridge: started common JAX bridge work

Main: callbacks removal

Main: remove VirtualBoxSupportTranslation template, add translation support to generic base class, clean up COM headers more, remove SupportErrorInfo.cpp|h

burn

Main: autogenerate event implementations

Main: generic events (disabled)

Main: some events work

Main: console COM events, cleanup, locking

Main: COM events work (smth along those lines could be done for generic events mechanisms too)

Main: ProgressProxyImpl.cpp/h initial coding.

Java bridge: MSCOM basics

bits

Java bridge: mscom work

SVC: link lsocket on Solaris too

Main/HostHardwareLinux: use inotify instead of FAM for USB hotplug probing, and disable the hal/dbus method for now to get testing

VRDP video channel configuration API.

Automated rebranding to Oracle copyright/license strings via filemuncher