DevAHCI/DevATA: Typo to make some compilers happy. No change in generated code. Thank you Jung-uk Kim.

AHCI: Fix possible crash when suspending a VM (very rare)

AHCI: Fix race condition which results in rare hangs during suspend

Careful...

Storage/AHCI: Free the I/O memory of cached request structures before unmounting the CD/DVD medium. Fixes a crash when a medium was removed from a drive without inserting a new one before the AHCI controller is destroyed

Storage/AHCI: Correct fix for r96392 (AHCI: Update HBA interrupt status register before executing the write)

AHCI: temporarily disabled r96292 to check if it cures the performance regression

AHCI: Fail the request with an overflow error if allocating I/O memory fails

AHCI: Update HBA interrupt status register before executing the write

Storage/AHCI: Switch ordering of register updates when completing the port reset. Update the status register last to make sure the BIOS acts on fully updated registers. Fixes a hang in the BIOS when detecting storage devices on startup

Storage/AHCI: Fix tight race when the guest resets a port leading to an inconsistent active request counter and a release assertion later on. The release assertion must be removed though because it might trigger by mistake now if the request counter is 0 due to thread preemption

AHCI: argh

AHCI: unused variable

Storage/AHCI: Free cached tasks on detach and before ahciR3Destruct is called because drivers are not available at that point anymore

Storage/AHCI: Switch to PDMIBLOCK I/O buffer allocator callbacks

AHCI: Fix request cancelling logic if the log page containing details about the failed request is read

AHCI: Small addendum to r93658, the device emulation needs to update the SCTL all times even if there is no device attached. Fixes error messages with Linux guest if there are ports without devices

Storage/AHCI: Get rid of ASMAtomic*Size and replace with the *U32 variants to get rid of some compiler warnings

Removed scary-looking messages about unimplemented MSI support.

Storage/AHCI: Fix the SSM version check for the legacy port reset handling

Devices/DevAHCI: Fix possible crash if there is no track list for the current medium

Main,DrvVD: Interface to pass the keys to the disk encryption module from Main

Storage/AHCI: Revert to the legacy port reset method if we loaded a saved state

AHCI: Another try to fix The AssertRelease issue during a reset. Move the second part of the port reset to the I/O thread to avoid any races. Needs minor tweaks in the BIOS driver to make it work with the changed behavior

AHCI: Make sure fWrkThreadSleeping is true if there is nothing attached to a port, fixes hangs during boot

AHCI: Fix wrong boolean operator causing the while loop to exit too early

AHCI: Fix race condition during port reset causing a release assertion when canceling outstanding requests

AHCI: Make sure the async notification handler is called from the async I/O worker thread too when a request completes, minor cleanup (ahciTransferComplete doesn't return a status code but a boolean)

AHCI: Fixes for hotplugging, should fix problems with Windows 7 guests

Devices/DevAHCI,DevATA: fixed typo

Storage/AHCI: check the correct return code when unmounting a SATA CDROM; coding style

DevAHCI: Do not let the guest trash PxCMD.CCS bits. Fixes XP SATA CD-ROMs.

DevAHCI: Properly convert MMIO -> PIO return code.

AHCI: Fix possible problem with suspending a VM if there was a non fatal error like a full disk

AHCI: Fix race which can cause a hang during suspend

Storage/AHCI: Make sure all threads are idling before finishing the controller reset, don't process garbage FIS structures

AHCI: Fix compatibility with saved states from previous versions

build fix

AHCI: Additional assertion

AHCI: The port is hot-pluggable by default

Storage/AHCI: Fix detection of CD/DVD media when switching from an empty to a host drive

Storage/AHCI: Add per port hotplug configuration flag

Devices/Storage: Support different sector sizes, booting from images with a sector size != 512 is not supported

Storage/ATA,AHCI: Fix newer Linux kernels not detecting a disk as SSD. The ATA major version must be at least 7

DevAHCI: Ported READ DVD STRUCTURE command from DevATA.

DevAHCI.cpp: Logging nits.

Devices/Storage: Preparations for non 512 byte sector sizes. Add a new getter in the block and media interfaces and remove uses of 512 constants and use the reported sector size from the underlying medium

AHCI: Fix comment

AHCI: Unused variables, typo.

Storage/AHCI: Fix crash when guest uses READ(10) with a 2352 byte sector size

AHCI: Consistent logging prefix

AHCI: Fix assertion during hotplug (event semaphore not created)

Storage/AHCI: Clear interrupt line on reset

Eliminated leftover variable.

Devices/AHCI: Offload work from EMT to a dedicated thread when using the async interface and make use of the SUPDrv event semaphore to get rid of certain R0/R3 context switches

Update PDMDEVREG initialization comment so they refer to pfnMemSetup instead of pfnIOCtl.

Relevant PDMDevHlpPhysWrite changed to PDMDevHlpPCIPhysWrite. If this breaks anything, add PDM_DO_NOT_RESPECT_PCI_BM_BIT to VMM_COMMON_DEFS to disable the feature.

DevAHCI: Let IOM do the complicated reads. Let IOM filter out unaligned and odd sized writes as well, though we still need to split QWORD writes our selves.

Fixed hungarian typos in MSI registration code. Some E1000 prefix adjustments.

DevAHCI.cpp: Used RT_SUCCESS instead of RT_FAILURE. Duh!

DevAHCI.cpp: Replace the automatic per device lock with a NOP. Some construction/destruction cleanups.

AHCI: DTrace fix

AHCI: Add dtrace probes

AHCI: Clear SActive register when the command list engine is stopped. Might fix some HSM violation error messages during error recovery with Linux guests

AHCI: New CFGM keyto configure the amount of logical sectors per physical sector

AHCI: Use new code for ATAPI passthrough

Add new flag to notify devices during a reset first to make sure there is no I/O pending (see #6434)

AHCI: Fix release assertion when cancelling requests and disabled async I/O

AHCI: A few more statistics and release logging if a request is active for too long

AHCI/Passthrough: Apply fixes for audio CD passthrough

AHCI: Fix ATAPI with EFI

AHCI: Fix EFI boot

AHCI: Move MSI cap to its proper place.

AHCI: Post a PIO setup FIS for PIO data commands

DevAHCI: Saved state restore fix (untested). Made ahciR3LoadLegacyEmulationState static (everything shall be static unless it *REALLY* needs to be global).

Storage/DevAHCI: cosmetic fixes.

Storage/DevAHCI: Remvove complete dependency on ATAController.

Storage/DevAHCI: skipping reading of IOBuffer

Device/DevAHCI: indentation fix.

Devices/DevAHCI: indentation fixes.

Devices/DevAHCI: fix for the failure to load old saved states.

AHCI Controller: DevAHCI.cpp: restoration of the excessive code removed. Saved state breakage in r79045 fix.

AHCI Controller: DevAHCI.cpp removed legacy ATA emulation support.

bugref..

AHCI: Fix possible guest memory corruption, it is possible that timed out requests were not canceled properly because they were not added to the task cache yet

*: gcc-4.7: ~0 => ~0U in initializers (''warning: narrowing conversion of `-1' from `int' to `unsigned int' inside { } is ill-formed in C++11 [-Wnarrowing]'')

Corrected a bunch of HC and GC uses in status codes.

AHCI: Temporary release assertions which hopefully catch inconsistent controller states causing hangs during VM poweroff

AHCI: Fix memory leak after recent optimization

AHCI: Don't always allocate new memory for the data buffers but reuse it when possible. Also use RTMemPageAlloc for page aligned memory to avoid bounce buffers during I/O

AHCI: Long overdue code cleanup, removes over 800 lines of unused or duplicated code and unifies the code path between async and non aasync I/O where possible

Log PRDT entries.

AHCI: Set overflow bit if the guest tries to transfer more data than is available in the buffer

More MMIO refactoring and some other cleanups.

Changed PDMDevHlpMMIORegister to take flags and drop pfnFill. Added PDMDevHlpMMIORegisterEx for the one user of pfnFill.

AHCI: More release logging

AHCI: Log HBA and port resets to the release log

PDM,Devices,Drivers: Add async discard API and make us of it

VMM,++: Try fix the async reset, suspend and power-off problems in PDM wrt conflicting VMM requests. Split them into priority requests and normal requests. The priority requests can safely be processed when PDM is doing async state change waits, the normal ones cannot. (The problem I bumped into was a unmap-chunk request from PGM being processed during PDMR3Reset, causing a recursive VMMR3EmtRendezvous deadlock.)

oops

AHCI: Log if TRIM is enabled

AHCI+DrvBlock+DrvVD: Add support for the TRIM command and connect with the VD discard support.

AHCI: Fixes for CD/DVD hotplugging

AHCI: Fix read access beyond 4GB for CD/DVD drives

Device/AHCI: 32bit alignment fixes

Main/Console+Machine: add notification for guest triggered eject, which right now results in updating the VM config Devices/Storage/ATA+AHCI: trigger the eject notification

AHCI: Hotplug fixes

AHCI: Initialize vendor product data for hotplugged devices

Changed FNIOMMMIOWRITE to take a const buffer pointer.

DevAHCI.cpp: Don't use the virtual sync clock for the command completion coalescing timer. It's apparently not engaged, so this change should be harmless.

VMM,Devices: Automatically use a per-device lock instead of the giant IOM lock. With exception of the PIC, APIC, IOAPIC and PCI buses which are all using the PDM crit sect, there should be no calls between devices. So, this change should be relatively safe.

TM,Devices: Fixed default critical section screwup and adjusted its usage in the devices.

AHCI: Fix return code for index/Data pair access

Storage/ATA+AHCI: support the non-rotational medium flag

AHCI: Fixes for hotplugging

AHCI: Controller part for device hotplug

Devices/AHCI: fixed mask for Fis address

AHCI: Support access through the index/data register pair from the I/O port space (as described in 10.14 of the AHCI 1.2 spec). Required for the BIOS driver

AHCI: Cancel active tasks during a device, port or HBA reset

AHCI+ATAController: Port r70670 + r70674 (Signal no current profile if no medium is loaded and add mandatory features to the GET CONFIGURATION reply)

Storage/DevATA+DevAHCI+ATAController: make assertion less strict, locked media and double unmount are harmless

AHCI: Make the number of available command slots configurable

PDM: introduced fEject parameter to PDMIMOUNT::pfnUnmount which is false if we don't need to eject the medium during unmount

Move the misc files the in src/VBox/Devices/ directory into a build/ subdirectory, changing their names to match the target module.

VMM reorg: Moving the public include files from include/VBox to include/VBox/vmm.

AHCI: Fix wrong Led assignments

AHCI: Fix loading older saved states

AHCI: Fix ATAPI device detection with Solaris guests. The command shouldn't be cleared in the CI register if it failed. The controller should stop processing commands instead to preserve the CCS field in the CMD register and the bits in CI

Storage: Introduce interface to query the location of a medium (device + instance + LUN) in a VM to generate IDs which are constant for saved states

AHCI: Better fix for r67808. Don't ignore the return values for R3 usage and use better status code for rcBusy

AHCI: Fix ahciHbaSetInterrupt from non R3 contexts

Devices/Storage: Introduce Bootable CFGM key to enable BIOS access

src/*: more spelling fixes (logging), thanks Timeless!

*: spelling fixes, thanks Timeless!

one more

SATA: 32-bit alignment issues

PCI: MSI-X work

AHCI: put MSI in place which allows to avoid conflicts

AHCI: Some cleanup Until now there were two different methods to forward new requests from GC to R3. For sync I/O a FIFO and a PDM queue to kick the I/O thread was used. For async I/O every task was passed through the PDM queue and processed by the consumer. This is messy and requires too much hyper heap. The new method is the same for sync and async I/O. Because a port can only have 32 active requests at a time it is enough to use a bitmap which is processed in R3 to get the slots of the new tasks. Requires a PDM queue with much less items then before and saves us almost 20KB of hyper heap on a 64bit host.

AHCI: Implement suspend on error for the async I/O path

AHCI: More verbose assertion

AHCI: checksumming in device identification

Devices: global define for using MSI support in devices

AHCI: Implement ATA_DEVICE_RESET for ATAPI devices (mandatory command for ATAPI)

AHCI: Implement SuspendOnError to give the user a chance to fix certain errors like broken iSCSI connections.

AHCI: MSI support

AHCI: Clear status and error fields of a task for the async I/O case for every new command, fixes booting newer Solaris guests. Don't process the ATA command if the Control bit is set in the FIS.

AHCI: few PCI device config notes

Only dump used statistics

AHCI/ATAPI: Make it work with Windows guests

AHCI/ATAPI: port fixes for audio CD over and implement support for saved states

CD/Passthrough: Prevent the host from crashing when trying to access a audio CD with passthrough enabled

CD/Passthrough: READ CD requires the numbers of blocks to read and not the number of bytes

AHCI: CD/DVD fixes

Build fix

AHCI: Support for CD/DVD devices (inclusive passthrough and runtime media changes)

AHCI: Add debugger info callback to inspect controller state

Removed more instances of obsolete PCISetIrqNoWait calls.

SATA: Always raise an interrupt after task completion; delaying this (interrupt coalescing) increases latency and has a significant impact on performance (see #5071)

Devices: three size_t warnings.

iprt/asm.h,*: Revised the ASMAtomic*Ptr functions and macros. The new saves lots of unsafe (void * volatile *) casts as well as adding some type safety when using GCC (typeof rulez).

DevAHCI.cpp: r=bird: Using TMCLOCK_VIRTUAL_SYNC in devices which aren't time sources requires some justificiation. See DevEHCI for instance.

AHCI: Fix crash on reset after a snapshot was taken. The Driver needs to be updated in the ATA emulation

AHCI: Count non queued async tasks too. Fixes rare crash during suspend if an async flush is still active

AHCI: Always use an intermediate buffer for now until we have native S/G operations (fixes OS X read performance)

AHCI: Fix assertion

AHCI: Bug fix for error reporting

AHCI: Clear the I/O thread handle after the thread was destroyed

DevAHCI: Alignment fix?

AHCI: Fix error reporting protocol violation

AHCI: Implement missing error reporting bits

AHCI: Error reporting fix

Automated rebranding to Oracle copyright/license strings via filemuncher

AHCI: Report errors of completed requests to the guest (not tested yet)

Interface changes for error reporting and async I/O

AHCI: Fix race condition where a task could be used even if it is already active

VBoxHDD: async I/O updates. Mostly complete except for bug fixes

Missing parameters

RTMemFree -> RTMemPageFree (DevAHCI.cpp)

AHCI: Use async I/O flush API

AHCI: Fix

Storage: Convert from PDMDATASEG to RTSGSEG to avoid casting between those two in VBoxHDD and more async I/O updates

AHCI: Report I/O errors to the guest

AHCI: Remove unwanted change 2

AHCI: Remove unwanted change

AHCI: Add ATA_SLEEP to the list of supported commands (does nothing actually)

PDM: s/pCfgHandle/pCfg/g - part 2.

PDM: s/pCfgHandle/pCfg/g - part 1.

PDM: s/szDeviceName/szName/g - PDMDEVREG & PDMUSBREG.

PDM: s/pDevReg/pReg/g

PDMDEVHLPR3 cleanup; reduced the number of strict functions.

PDM,*: Redid the PDM structure versions. Check the instance and helper versions in every device and driver constructor.

pdmifs.h: the final batch of refactored interface ID code.

pdmifs.h: the penultimate batch of refactored interface ID code.

pdmifs.h: another batch of _IID changes.

PDMIBASE refactoring; use UUID as interface IDs.

Devices: more -Wshadow

PDMCritSect: Deployed lock ordering. (ring-3 only, only DEBUG_bird atm)

DevAHCI: warning

AHCI: Fix 64bit build

AHCI: Fix STAM alignment

DevAHCI,ATAController: Async hanlding of reset, suspend and power off.

DevAHCI: A little cleanup.

DevAHCI: copy&past comments.

DevAHCI: async suspend, reset and power off handling for the SATA part.

DevAHCI: Ditto.

Devices,VMM: Replaced all VERR_SSM_LOAD_CONFIG_MISMATCH returns with SSMR3SetCfgError calls.

DevAHCI: ACHI -> AHCI.

AHCI: Try to fix the build again

AHCI: Alignment for STAM

AHCI: Try to fix the build

AHCI: Add additional VPD keys

DevAHCI: Try avoid the 100ms timeouts during suspend and shutdown. Ported over some of the ATA changes stuff.

DevAHCI,DevATA: More RTThreadSleep(100) todos.

reformat pg_dev_achi. no real change.

DevAHCI: even more config verification bugfix.

DevAHCI: more config bugfix.

DevAHCI: bugfix.

DevAHCI: save and verify config (saved state).