- Added support for setting memory write handler to NULL (ROM case).
- Added new PCI device method after_restore_pci_state(). It currently handles
the PCI ROM case only (could be extended).
- Moved VGA timer handler to the VGA core and handle the non-VGA case there.
It calls the specific update() code and finally bx_gui->flush().
- Moved the VGA update frequency parameter handler and the refresh_disply()
method to the VGA core.
- Store the VGA extension parameter in the VGA core.
- Voodoo: Fixed graphics snapshot.
- Fixed probing the CMDFIFO with Voodoo output disabled (send FIFO wakeup
directly, otherwise driven by vertical retrace).
- Use the CMDFIFO member 'enabled' instead of reading the fbiInit7 register bit.
- TODO: The Voodoo2 freeze when testing a Voodoo1 demo is still not fixed.
- voodoo_init() must be called after initializing FIFO mutex.
- The 'op_pending' counter must be reset after resetting FIFOs.
- Fixed an MSVC warning and removed obsolete comment.
- FIXME: There are still some conditions that can freeze the Voodoo display.
it's 64 entries is still not present, since it makes the emulation a few
percent slower due to the overhead of the FIFO code. The memory FIFO has 64k
entries and makes a speedup of around 75 % possible.
- usb_msd.cc: The Device Qualifier is valid for high speed devices only.
- uhci_core.cc: Don't report BX_DEBUG message if the current TD belongs to an
async operation that is still not complete.
- Now using float type for the 'clk0_freq' variable.
- DAC register 6 can switch to half pixel clock (ported from PCem).
- Writing value 0xf8 to PLL register 0x0e completes clock #0 setup.
non-WIN32 platforms instead of BX_MSLEEP(). The donut demo is once again a
little bit slower, but on other tests the IPS values are still okay. Cleaned
up the code a little bit and added save/restore stuff for the new code.
packets and then always sleep for 1 millisecond.
TODO: The first tests with pthread_cond_* stuff showed a slowdown of approx.
10 % compared with the IPS value of the current code.
command and execute it when there is enough data present (code partly ported
from voodoo.h.txt).
TODO: Use "wait for event / condition" functions instead of msleep() if there
is a performance improvement with it.
- Added DNS service support for the server "vnet" and connected clients.
- Added command line option "-loglev=" to control whether or not there is
BX_ERROR, BX_INFO or BX_DEBUG output. Default value is 1 (BX_ERROR only).
variables again to let the threads end normally. Re-implenented the method
closewaveoutput() for some modules, move the destructor code to it and call it
after leaving the mixer thread. This fixes a deadlock in the "soundwin" module
if Bochs is compiled with plugin support.
- usb_common: init_device() now splits device name and primary option before
detecting device type.
- usb_hub: now completely handle port option in constructor.
- plugin.h: removed obsolete defines.
- Removed "pseudo device plugin" containing common code and all connectable
USB devices.
- Moved USB device base class and module handling to the Bochs core.
Now loading USB device plugins in usb_common.cc.
- Created separate plugins for each USB device implementation.
- Modified Bochs plugin system to support new plugin type PLUGTYPE_USBDEV.
- Added entries for the non-plugin case similar to the network and sound stuff.
- Modified win32 thread macros to use the handle instead of the id.
- Added new macro BX_THREAD_KILL to terminate threads.
Voodoo thread changes:
- Let the cmdfifo_thread sleep while the CMDFIFO is disabled.
- Kill the cmdfifo_thread in destructor and removed old thread control code.
- Added capability to save/restore float type values in paramtree / siminterface.
- Added Voodoo2 triangle generator data to save/restore list.
- Implemented bitBLT function "SGRAM fill".
- Minor other changes and cleanups.
- TODO: remaining bitBLT functions, thread handling improvements.
voodoo.h.txt which was a part of the original voodoo patch for DOSBox.
Some Voodoo2 specific parts were not integrated in the main patch file. I guess
they have been removed since Voodoo2 support was incomplete (2D stuff missing).
TODO: Voodoo2 2D support (bitblt engine), more testing, code cleanup...
- Added CMDFIFO thread similar to the sound output thread.
- split register write function in 2 functions: register_w_common() handles
the CMDFIFO writes and valid registers in CMDFIFO mode. It forwards write
accesses to other registers to register_w() (Voodoo1 and non-CMDFIFO mode).
The CMDFIFO thread also calls register_w() when processing data.
- The CMDFIFO thread processes all data, but it does not yet execute type 3
requests (extended triangle generator missing).
- TODO #1: implement Voodoo2 specific triangle functions (for CMDFIFO).
- TODO #2: implement 2D bitblt engine.
- Zero out voodoo device state in init().
- Added the CMDFIFO state to save/restore list.
- Added CMDFIFO register handling in read/write functions.
- Added new function cmdfifo_put().
- TODO: CMDFIFO thread.
- Fixed normal register access in CMDFIFO mode.
- Implemented cmdFifoBaseAddr register behaviour.
- Write CMDFIFO data to FBI memory (has no effect yet).
- Changed some log messages to BX_DEBUG.
- Report unimplemented Voodoo2 registers.
- TODO: Add FIFO thread to process the CMDFIFO data.
valid. The hRetrace value can be set now, since Bochs no longer crashes.
TODO: Implement enough of the FIFO stuff to make the Win9x driver for the
Voodoo2 work.
- Voodoo: Use the VGA/Cirrus timers to drive the gui screen update.
The vertical retrace emulation is still driven by a separate timer.
- VGA/Cirrus: Since we are using the virtual timer, we also have to read
time_usec() from it.
- vgacore: removed unused timer methods.
- prepared hRetrace value implementation (cannot be enabled yet, since the
driver passes this test and gets confused in next stage (segfault).
- Detect CMDFIFO access (not present yet).
- Report disabling Voodoo graphics mode.
- TODO #1: Fix Voodoo2 to make it usable.
- TODO #2: Split update code and drive gui screen updates from the VGA timer
(finally remove new realtime option again).
- Since we are using the virtual timer, so we have to read time_usec()
from it, too (TODO: realtime mode is hardcoded, maybe add an option).
- Voodoo1 and Voodoo2 have different timing register widths.
- Fixed vertical retrace value returned by get_retrace().
- Prepared separate hvRetrace register (TODO: hRetrace).
- Store the Voodoo model in structure bx_voodoo_t.
- The PCI initEnable register of the Voodoo2 model has a read-only secondary
revision ID in bits 12..15. Now the Voodoo2 is correctly detected by the
Windows driver, but display is not yet working (blank screen).
- Reset now disables VGA override.
- PCI: Moved PCI device base class methods to the end of file devices.cc and
re-define LOG_THIS to use the correct log module and prefix.
- Enabled Winsock2 usage and define _WIN32 symbol for Cygwin (both 32 and 64 bit).
- Changed include order in some files to resolve conflicts.
- Modified some data types to fix errors and warnings.
- Added workarounds for the incorrectly defined symbols FIONBIO, FIONREAD and
EWOULDBLOCK.
- Verified slirp usability with Cygwin 32-bit, MinGW/MSYS and MSVC builds of
Bochs. Other ports should not be affected.
- socket: Define the FIONBIO workaround for 64-bit Cygwin only.
- slirp: Don't mix usage of WIN32 and _WIN32 definitions.
- TODO: The slirp module still fails at UDP socket creation when Bochs is
compiled in Cygwin64.
- slirp: Increment IP packet ID outside of htons() to avoid undefined behaviour.
- socket: Workaround for a bug in the w32api code that generates a constant
with wrong value. Setting up our own FIONBIO fixes the panic.
- TODO: The slirp module still fails when Bochs is compiled in Cygwin64.
NOTE: networking still fails when compiled with Cygwin ('socket' mode fails
to set up non-blocking i/o / 'slirp' doesn't complain, but fails to connect
the internet).
- Created framework based on our I/O APIC code and ported HPET core from Qemu
with some required changes for the Bochs timer and IRQ handling.
- Enabled HPET-specific code in the ACPI and rombios32 sources and generated
new ACPI table with iasl.
- The HPET device plugin is now always loaded if the i440FX chipset is selected
(same as ACPI). We have to rethink this when we have implemented a more
modern chipset.
- TODO: Rewrite of the virtual timer code for nanosecond support to make the
realtime synchronization possible with HPET.
directly sent to the optional keyboard (USB keypad). It is now independent
from the legacy keyboard and the check for supported keys is simplified. The
legacy keyboard still receives the unhandled key event, but now sent from the
common devices code.
bx_devmodel_c. The early version of the combined vga / svga_cirrus code
required the now obsolete implementation. Some related cleanups in the
devices code.
- Since the pci_read_handler() method is identical in most devices, move it
to the base class to reduce code duplication. Only the 'pcidev' device has
it's own implementation (NOTE: it is not maintained yet).
- Minor other fixes and cleanups in some PCI devices.
- In "non-plugin" mode loading and unloading network module is not necessary.
Now enabled loading for the plugin mode only and removed unloading function
for that case. The "pseudo plugin" entries of the dummy entry points are
necessary to make sure the self-registering network driver objects are not
optimized out.
- slirp / vnet: Since the plugin init function is not called in non-plugin
mode, revert the instance count init change.
- Fixed a makefile dependecy.
- TODO: Implement self-registering objects in the sound driver code.
- Removed "pseudo device plugin" containing common code and specific drivers.
- Moved network module base classes and module handling to the Bochs core.
Now loading network driver plugins in netmod.cc.
- Moved shared networking code used by the 'vnet' driver and 'bxhub' from
netmod.cc/.h to new files netutil.cc/.h.
- Created separate plugins for each network driver implementation.
- Modified Bochs plugin system to support new plugin type PLUGTYPE_NETWORK.
The format of the 'ethdev' string must be 'host:port' to use this feature.
Using the port number only for connecting 'bxhub' on the same machine is still
supported.
- Small changes to avoid crash after failure.
- Several cleanups in the socket mode.
- Fixed reporting active mode after init is done.
- Reporting sent/received byte with BX_DEBUG.
- The 'vnet' module now uses the 'script' parameter for an alternative log
file name.
- Renamed log file names in some networking modules (replaced 'ne2k' by a name
related to the module name).
- Documentation updates (networking stuff and cmosimage).
- Increase maximum number of Bochs client sessions to 6.
- Define bx_devmodel_c to 'void' for the bxhub case to simplify netmod code.
- Minor cleanups in bxhub and related vnet code.
- According to the specs masked edge-sensitive IRQs are not held pending.
- Print BX_DEBUG message in set_irq_level() only in case of level change to
avoid flooding log file.
- eth_socket.cc: Added missing winsock init.
- bxhub.cc: Added missing CDECL macro to the signal handler.
- Added some definitions for non-Windows platforms similar to the serial and
rfb code.
- bxhub.cc: Put all client data in a structure and added send_packet() function.
- bxhub.cc: Added SVN ID line.
- netmod.cc: Changed some DHCP messages to BX_DEBUG.
- eth_socket.cc: Minor cleanups.
external program 'bxhub' (simulating a two port ethernet hub). It is based on
an old SF patch and it has been updated and extended:
- Windows support (tested with MinGW/MSYS and MSVC nmake).
- Command line arguments for 'bxhub' (base port number, MAC address, TFTP).
- Added support for DHCP, TFTP, ARP and ICMP echo (ping). It uses parts of the
code for the 'vnet' module with some additions from the former "slirp backend".
- TODO: code cleanup to reduce duplication, rewrite of the networking plugins
stuff similar to the work done with the sound drivers.
- Builtin sound driver plugin support should depend on BX_SUPPORT_SOUNDLOW.
- Added some symbols required for the MSVC plugin of sound driver "file".
- Clean up threads and audio buffers only when shutting down the primary sound
driver (pcm_callback_id >= 0). Fixes crash on exit with wavemode 3.
- Renamed OSS sound driver files from soundlnx.cc/.h to soundoss.cc/.h.
- Removed "pseudo device plugin" containing common code and specific drivers.
- Moved common sound code, base classes of sound drivers and the dummy driver
to the Bochs core. Now loading sound driver plugins in soundmod.cc.
- Created separate plugins for the platform/library specific drivers and the
"file" driver.
- Modified Bochs plugin system to support now plugin type PLUGTYPE_SOUND.
Prepared "biosdetect" option in the harddrv code.
TODO #1: Add support for a comma-separated list of BIOS options.
TODO #2: Implement support for at least "biosdetect=none" in the Bochs BIOS.
either libsamplerate or the SoX resampler library installed. If not installed,
the data is only copied to the output buffer and sample rate of the output
driver is changed similar to legacy code. Related changes:
- Added check for libsamplerate or SoX resampler header files in configure
script.
- Added functions for converting source format to float (requied by resampler)
and float to output format.
- Added support for float type data in the audio buffer code. Buffer chain #0
receives float data from sendwavepacket() and buffer #1 receives data for
mixing and output in format "16-bit signed little endian stereo".
- ALSA: Disable builtin resampling feature if resampler is present to avoid
doing it twice.
- Create a file with image name plus extension ".lock" if an image is opened
in read/write mode. This file will be deleted after closing the image.
- Check for a lock file before opening an image and fail if it exists.
- Check for a lock file before creating a undoable/volatile redolog file and
fail if it exists.
"fatal". It can be used for all cases when there is no workaroud present to
avoid application crash or incorrect simulation behaviour. As a first step now
using BX_FATAL() for some data structure size and bit field errors.
- The retry timer now calls process_transfer_ring() for enabled slots only.
- On USB3-only ports ignore device speed setup and try to enable super-speed mode.
- Added support for HID devices in high-speed mode.
- Added separate descriptors for MSD devices in high-speed mode.
- NOTE: USB 2.0 values ported from Qemu / tested with USB EHCI.
async mode by default for all USB devices (used by disk, cdrom and floppy).
Whenever we implement a new HC model, it is still possible to set the async mode
to 0 until the async packet completion is working correctly.
- Extended USB callback definition to support different event types sent from
device to HC. Currently we only have the two events "async completion" and
"remote wakeup".
- Implemented event handlers in the UHCI, OHCI and EHCI. For the EHCI version
we currently have no test case.
- The external USB hub now uses this feature at device connect/disconnect.
Now the device change on the hub is correctly detected by the guest OS.
- TODO #1: remote wakeup and async completion for the xHCI.
- TODO #2: event handler for the external hub to make a USB device work on
a chain of hubs.
- TODO #3: the event handler possibly could be used for the packet copy code.
- use get_dwords() and put_dwords() from EHCI sources for reading / writing
xHCI data structures as dword array. This fixes some endianness issues.
- simplified copying data from / to data structures since we can now use the
dword array with the correct byte order directly.
- fixed all remaining "strict-aliasing" GCC warnings.
configuration, get/set status, get descriptor) to the base class to reduce code
duplication. The device calls the common handle_control() method and then it
processes unhandled requests in it's specific code. (similar to Qemu).
- USB xHCI NAK handling fix by Ben. Now openSUSE 13.2 boots without hang with
hub connected to an xHCI port.
- Updated Ben's email address and year in some copyright headers.
- Get rid of device_buffer and use packet init / cleanup functions.
- Read immediate data from memory location instead of possibly byte-swapped
TRB parameter.
- Moved UHCI async packet functions to usb_common.h for using then on both
UHCI and OHCI.
- Modified OHCI packet handling based on the UHCI implementation.
- Renamed EHCI helper function to avoid conflicts.
- TODO: async packet support on xHCI.
- Added missing queue handling code for init, reset and device disconnect.
- Added EHCI status access methods for async and periodic state.
- Added a lot of comments (mostly from the original EHCI core code).
- UFI/CBI floppy: bus-powered, no remote wakeup
- HID (mouse, tablet, keypad): bus-powered, remote wakeup
- MSD (disk, cdrom), printer: self-powered, no remote wakeup
TODO: Remote wakeup is not yet implemented. A device should be able to resume
it's upstream port if suspended.
- HcLSThreshold is writable, but it's usually not modified (same as Qemu).
- BX_ERROR should only occur on an attempt to change the number of ports.
- Don't register read-only items for save/restore.
- Changed some device descriptor fields (using values from Qemu).
- Ported device reset code from Qemu.
- Check device speed setting on connect (only low/full speed supported yet).
- TODO: External hub doesn't work properly in some test cases (device change
not detected by guest OS, openSUSE 13.2 hangs with hub on xHCI).
- USB mouse and keypad now return NAK response when idle and there is no change.
- Implemented HID commands for get/set idle duration.
- Fixed NAK behaviour in UHCI and OHCI.
TODO list: USB tablet should also return NAK (doesn't work yet), review xHCI for
the NAK behaviour, accurate implementation of the idle duration
is currently supported per hub at the same time (same as OHCI).
TODO #1: We should support at least one async operation per connected device.
TODO #2: Async packet support for EHCI and xHCI.
- added a hack to make EHCI work in Bochs, but without asynchronous packet
support (tested with "high-speed" USB disk).
- added missing list for connected USB device state.
- fixed some of the endianness issues
- fixed some of the GCC warnings
- changed some BX_INFO to BX_DEBUG
- FIXME: non-bulk devices not yet working on xHCI. Linux guest hangs during boot
when HID or hub device is connected.
instead of the hub's remove_device() method. The external hub removes connected
devices before clearing it's own list. This fixes crash on exit when devices
are present on external hub.
ignores connected high speed device (fixes panic with EHCI on guest OS init or
shutdown). The other HC's now panic and disconnect device if the user decides
to continue.
Asynchronous transfers are not yet working properly (tested with USB disk device
in high speed mode). That's why the support for "speed:high" is still disabled.
Since we have no test case for the isosynchronous mode, I have not yet ported
the main code for it.
- added Android case to the configure script.
- renamed file memory.h to memory-bochs.h to fix conflict with NDK.
- fixed Android issues in some files.
almost implemented, the port owner handling and the UHCI companion controllers
are present. High speed devices are not yet supported, since the EHCI scheduler
code has not been written yet (e.g. porting from Qemu). Low and full speed
devices should be detected correctly and work after port ownership change to
UHCI. 6 ports are available and the bochsrc syntax is similar to other HCs.
- added method to change port ownership.
- fixed UHCI reset and initialized device pointers.
- added symbol UHCI_FULL_DEBUG to reduce debug output if not set.
- renamed connection handling method.
emulation and usb_uhci.cc implements the setup for the PIIX3 chipset and the
runtime device change code. The UHCI core is required for the EHCI emulation,
since there can be multiple "companion" controllers for USB 1.1 device handling.
- vgacore: double font and screen width at half dot clock.
- X gui: improved font bitmap generation to support width > 9.
- win32 gui: adjust values to use existing stretching feature.
- guis rfb, sdl, sdl2, vncsrv and wx: double pixels if font width > 9.
sector transfers (buffer pointer update was missing). Multiple sectors transfers
could for the USB disk emulation and for image conversion with bximage.
- More accurate seek emulation for read commands.
- Moved status indicator code to the SCSI layer.
- Fixed dump size for async packets.
TODO: Seek emulation for disk write commands.
Added latency time for the "implied seek" of read/write/format commands. Both
features can be used when asynchronous packet handling is available. Some
related changes in the non-async read and write code.
- Added missing update of the write protection flag after runtime config.
- Added sense codes for invalid cases of some commands.
- Added buffer pointer save/restore handling and use BXRS macros now.
