RTMpOnPair implementation for linux.

HostDrivers/Support: fix the Linux modules

SUPDrv: Split out the GIP related code into SUPDrvGip.cpp.

HostDrivers/linux: symbol fix.

SUPDrv must be compiled with -DSUPDRV_WITH_RELEASE_LOGGER on linux too.

supdrv,iprt: Added VBOX_WITH_TEXT_MODMEM_HACK for getting VMMR0.r0 address in the linux kernel stack crawl.

SUPDrv: Made dtrace support build against UEKR3.

Linux host modules: added -Wno-declaration-after-statement (introduced with gcc-3.4.0) to prevent warnings coming from re-using C++ files with C

IPRT: Kernel thread-context hooks, linux implementation. Extended tstR0ThreadPreemption testcase to test the thread-context hooks.

Linux r0drv burn fixes

BUGZ:6602 Reverted r83596.

Linux r0drv: implement DO_NOT_DISABLE_PROVE_LOCKING option.

SUPDrv: An DTrace on UEK experiments.

SUPDrv: More exports.

SUPDrv: More exports.

SUPDrv: tracing infastructure coding...

Linux modules: force disabling of CONFIG_MODULE_SIG when building the kernel modules. We don't have any private key available so it does not make much sense to try signing

IRPT: Adding RTErrVarsSave/Restore/++ to preserve errno, h_error/WSAGetLastError and GetLastError accross assertions.

Trunk: forward ported r70873: Installer/linux: no "make kernelrelease" anymore when compiling the Linux host kernel modules because that does not work with Linux 2.6.38 anymore

log rotation review and adjustments: Don't delete any excess files if log roation is disabled - we don't know what these files might be. Moved RTLOGGERFILE into log.c. Keep RTLogCreate simple, anyone needing rotation can use RTLogCreateEx[V]. Made RTLogGetDestinations produce the log rotation bits.

Linux additions / host drivers: fixed compilation with DKMS

GIP,++: Lots of CPUs (disabled).

IPRT: linux R0 threads

IPRT,Drivers: Committed a modified version of the diff_linux_guest_host patch. This mangles the IPRT symbols in kernel space on linux and later other platforms.

IPRT: Implemented RTSemEventWaitEx[Debug] for ring-0 linux.

IPRT,SUPDrv,VBoxGuest: Reimplemented RTSemEventWait* so that it can make use of high resoultion timers when present (2.6.28).

VMM,SUPDrv,IPRT: More changes for related to the priodic preemption timer. (still disabled)

Linux host drivers: another fix for the clean rule (the kernel directory is not necessarily available)

SUPDrv,IPRT,VMM,DevAPIC: Added RTTimerCanDoHighResolution and exposed the RTTimer* API to the ring-0 modules. Fixed two regression from r65858 (TM + APIC).

Linux host kernel modules: some minor Makefile fixes

Linux Installer/Additions: attempt to properly register the host/guest kernel modules at DKMS and use the setup function of the service script to compile the modules instead of doing this separately in the installer scripts

IPRT,SUP: Moved the strformat*.cpp completely out of RuntimeR0 on all platforms so that %RT[pgmpage] will work in ring-0. Minor SUPDrv IOC interface version increment.

HostDrivers/*/linux/Makefile: Disabled IPRT_DEBUG_SEMS.

IPRT/vboxdrv: We have RTTimeSystemNanoTS in ring-0 now, so use it.

Automated rebranding to Oracle copyright/license strings via filemuncher

IPRT,SUPDrv: Basic testcase for ring-0 RTSemMutex*.

Linux HostDrivers/Additions: fixed missing symbol ''RTMemAllocZVar()''; rebranding

VBoxNetFlt-linux: GSO to host and wire experiment, the former is a success the latter needs more magic.

*: The rest of the %V* format specifiers are history.

iprt/assert.h: Added RTAssertMsg2Add[Weak][V] for dumping the lock deadlock chain and stacks. Added the missing internal/assert.h file to the linux and freebsd kernel source trees.

files_*,Makefile: s/RTAssertMsg2V/RTAssertMsg2/

IPRT,SUPDrv,++: AssertMsg[12] -> AssertMsg1Weak, AssertMsg1, AssertMsg2Weak, AssertMsg2, AssertMsg2WeakV and AssertMsg2V. Doing more of the assertion machinery in common/misc/assert.cpp to avoid code duplication (ring-0). Major SUPDrv version bump.

SUPDrv,++: Made SUPDrv.c platform agnostic and renamed SUPDrvAgnostic.c to SUPDrvSem.c. dprintf and dprintf2 are no more (Log/Log2), while OSDBGPRINT simply calls SUPR0Printf.

Linux installer: ''depmod -ae'' => ''depmod -a'' (the latter is sufficient and prevents a warning on recent Linux distributions)

SUPDrv: Started splitting out platform agonstic code to avoid clashes with kernel headers (linux + silly ES #define).

iprt/log.h: Use the spinning mutexes.

HstDrv: Made vboxnetadp and vboxnetflt use IPRT from vboxdrv.

SUPDrv: Export the RTR0MemUser/Kernel APIs. (Needed for the tstRTR0MemUserKernel testcase.)

HostDrivers/*/linux/Makefile: Added --no-print-directory to the $(MAKE) kernelrelease invocation to prevent it from being messed up when doing "make -C out/linux.amd64/debug/bin/src". MAKEFLAGS+MFLAGS passes -m which seems to overrule the -s.

SUP: Expose ring-0 event semaphores to ring-3, part 1.

linux kernel modules: prevent some superflous output during build

leftover

enable ''vm_insert_page()'' for Linux 2.6.23+

Additions/linux: fixed kernel detection

Linux host kernel modules: removed 2.4 cruft; fixed VBOX_WITH_64_BITS_GUESTS

deb/rpm: fixed vboxnetflt module compilation

fix linux builds

clean one more file

disabled VBOX_USE_INSERT_PAGE by default - somewhat unclear what goes on in #3280

set VBOX_USE_INSERT_PAGE = 1 as default linux host kernel module flag

Linux module fixes

small fix

Created assert-r0drv-linux.c (finally).

IPRT,++: some assert.h cleanup, making a suitable place to call panic() in ring-0.

Implemented kernel driver build flag to allow better debuggability of VBox on Linux (VBOX_USE_INSERT_PAGE when compiling vboxdrv.ko on kernels after 2.6.22)

iprt: splitting out some log stuff from log.cpp so we can avoid some APIs that includes the ellipsis between VBoxDrv and the dynamically loaded code (VMMR0.r0 and friends).

#3076: Merged in the branch with the alternate driver authentication method. (34468:HEAD)

Linux support: define VBOX_WITH_64_BITS_GUESTS on AMD64 hosts; added -g to CFLAGS when compiling vboxdrv in debug mode

left a test change in (26 instead of 24).

lose the space in ifeq.

EXPORT_SYMTAB on 2.4

SUPDRVShared.c -> SUPDrv.c

Cleaned out the last bits of the USE_NEW_OS_* stuff.

is needed for some older kernels/distros

fixed + re-enabled USE_NEW_OS_INTERFACE_FOR_GIP code for Linux

disabled the new GIP timer interface

fixed vboxdrv compilation

__udivdi3

Generic timer code for async GIP mode.

The Big Sun Rebranding Header Change

clean rule

Linking fix.

Compilation fix.

RTMpOn stubs for linux.

Added RTStrFormatTypeRegister/SetUser/Deregister for runtime registration of custom format types. Example: RTStrFormat(,,"%R[pgmpage]", pPage);

Linux kmod: VBOX_REDHAT_KABI

Ring-0 semaphore debugging aid (linux).

Linux modules: unbreak compilation against Linux 2.4

Linux kernel modules: added override

Linux kernel modules: allow to build from kBuild (KERNELRELEASE set), cosmetical fixes

Support driver (Linux): added gcc (u)div/mod support

Made it possible to make debug versions of vboxdrv on linux. This will check for BUILD_TYPE=debug, so for those who doesn't want ring-0 assertions in vboxdrv make sure you built it with BUILD_TYPE <> debug.

Linux kernel module Makefiles: use KDIR if set

The Giant CDDL Dual-License Header Change.

improve kernel version check

semaphore-r0drv-lnx.c split.

disable IDT_PATCHING code in the Linux vboxdrv module

inverted VBOX_WITHOUT_IDT_PATCHING.

2.6.10-gentoo-r6 needed __divdi3 for some reason.

Use the RTR0MemObj API on Linux. (seems to work - just disable it in the Makefile if it does not).

Biggest check-in ever. New source code headers for all (C) innotek files.