Temporarily disable fifo_reset() in recompute_video_memory(). FIFO handling must
be observed and a better solution must be found.
Fixed hang after pressing power button: destructor modification was required since
the swapbuffer command can depend on vertical retrace.
Removed disabled BX_DEBUG code and VGA_TRACE_FEATURE symbol.
Some other small fixes and cleanups.
TODO #1: review MAP13 addressing in case start address > 0.
TODO #2: review chain4 code in medm_write().
- Fixes in MAP13 disabled mode (VGA mode 6): half dot clock, color support
- Modified method get_vga_pixel(): now using row start address as an argument
- Added CMOS image options to runtime menu (except RTC init).
- Fixed default size of Bochs CMOS RAM (s.max_reg).
- Fixed bochsrc parse to make 'rtc_init' parameter work.
- In textconfig now using SERIES_ASK for CMOS image options.
- Show message box while restoring hardware state.
- Set modeless message box handle to NULL after destroying it.
- Reset all config options to default before restoring config.
- Fixed CMOS RTC value after restore if in 12-hour-mode.
- Fixed possible segfault in the plugins code caused by Voodoo plugin in VGA mode
(found at restore code testing).
by Michele Giacomone
Detailed description:
-Observed issues
Due to some limitations only dates between 1980 and 2038 can be
used in a reliable way.
Also, bochs incorrectly assumes a linear correspondence between
the data returned by the <time.h> functions localtime() and
mktime(), and isn't setting the latter properly.
Bochs keeps its internal time value dependent to these functions
after setup, assuming that their internal settings won't change
on the go - which is not the case.
In my OS, and in my timezone, this leads to incorrect startup values
for 5 months each year and unreliable values if the simulation is
kept going for a long time. (a feedback between localtime() and
mktime() is created which keeps shifting back the time)
Also, the RTC simulation is not realistic since the clock fixes
itself across DST changes, without updating any DST related flag,
a behavior that no guest OS expects.
-Proposed fix
This is implemented in such way that no bochs' previous behavior
is changed, a part from the broken ones, with legacy in mind
== the user can keep using bochs exactly as before knowing nothing
of this patch
+Make the internal s.timeval variable a Bit64s, so it can fit all
values that the cmos can correctly represent, reported below:
MIN setting -62167219200 => 0000/01/01 SAT 0:00:00
MAX BCD setting 253402300799 => 9999/12/31 FRI 23:59:59
MAX BIN setting 745690751999 => 25599/12/31 FRI 23:59:59
And then fix each reference to these so it can handle such values
And make bochs correctly wrap around for under/overflows, so that
only the most significant bits of the century are lost.
+Do the same thing to the bochs time0 parameter, so all the above
values can be chosen at startup (despite being now legal values,
1 and 2 will still be treated as "local" and "utc"). Note that
normally only BCD settings are valid since bochs' CMOS defaults
to such operating mode - the only way to use the binary range
is by loading a cmos memory map.
+Make the internal s.timeval variable independent from external
factors. This means providing a small set of time handling
functions, contained in "iodev/utctime.h", which must work in
any environment in which bochs compiles, accessing no external
resource. This also means that after startup, s.timeval will only
be changed internally, and no call to the OS time functions will
be made.
+Make the internal s.timeval variable timezone independent, to
have a linear correlation between its values and valid CMOS
settings. To make it easier, s.timeval is gonna be treated as
if the current timezone was UTC: so,
- if the user selects UTC as time0, s.timeval will become current
time(NULL)
- if the user selects localtime, s.timeval will be computed as
the value which will display the same broken down time as
localtime(&now)
- if the user inputs a time formatted string the proper s.timeval
to displayed will be easily calculated,
- if the user inputs a starting time value, s.timeval will be
computed as the value which will display the same broken down
time as localtime(&user_input) to ensure the same operation as
before.
A "tz=utc" is displayed when bochs prints out the current time
value, to warn users about the difference in meaning between the
internally kept time value and the value they can set through
the "time0=" parameter. This might be changed to communicate
instead the time value they can input to get the same setting,
but performing such calculation (except for the startup time)
suffers from all the mktime()/localtime() problems listed above
so I did not do it.
The range of "time0" is automatically adjusted so all users in
all time zones can set any legal value despite "time0=" having a
local meaning.
A thorough explanation of what I did and why can be found in the
"iodev/utctime.h" library header.
---------
Co-authored-by: Stanislav Shwartsman <sshwarts@users.sourceforge.net>
Co-authored-by: Volker Ruppert <Volker.Ruppert@t-online.de>
The WIN32 configuration for the cdrom was not closing the image file on
eject. This left the handle open while Bochs was running, not allowing
the image file to be modified.
This improves and enhances the floppy emulation.
This adds the ability to specify a controller type.
This adds all but one, known, controller command.
This fixes the format command. Issue #123
This fixes the error with older Windows guests.
When the user presses the "Reset" button on the gui ribbon, we need to
clear the active status of all statusbar items or they remain set after
the reset occurs.
This is mentioned in issue https://github.com/bochs-emu/Bochs/issues/136
This allows you to change the CD-ROM via the Bochs Ribbon, as well as
properly "ejecting" a CD from the guest.
Before the CD path would change, but the guest (tested with WinXP and
DOS as guests) would not see the change.
This fix simulates an open tray and then a closed tray for all CD
changes.
avoid using ull modifyer for consts, use BX_CONST64() instead
move definitions from header file to cpp file when possible
add 'const' modifyer for obviosly const methods and pointers
This adds support for the UFI_SEND_DIAGNOSTIC command in USB_FLOPPY.CC
and
This changes the list of defines to enum to match the current changes
made to recent files.
This fixes the CD-ROM hardware by making sure the `drive_ready` flag is
zero after an `ATAPI_DEVICE_RESET` command. As far as I know, no other
Guest needed this fix except for the TEAC-CDI driver referred to below.
This also adds an assumption fix so that the TEAC_CDI driver will work.
The TEAC_CDI driver relies on the fact that the TEAC drive it is
designed for, sets the `interrupt reason` register (`sector count`
register) after the `ATAPI_IDENTIFY` command is issued. It uses this
`i_o` bit of this register to determine the direction of transfer for
the 0xA1 command, even though this is a read-only direction command.
Through other tests, I see no other Guests effected by this addition.
I also added a few minor syntax modifications, tab spacing, and
comments. The DEBUG out string will display the ATAPI command as well as
the 0xA0 command.
This PR adds the ability to choose which companion controller the EHCI
uses.
`#usb_ehci: enabled=1, companion=uhci`
`#usb_ehci: enabled=1, companion=ohci`
UHCI is enabled by default:
`#usb_ehci: enabled=1`
This breaks up `usb_ohci.cc/h` into `usb_ohci.cc/h` and `ohci_core.cc/h`
to be like the existing `usb_uhci.cc/h` and `uhci_core.cc/h`.
This adds the ability to define a Port Routing string in the
`HcspPortRoute` register, and is optional via the `EHCI_PORT_ROUTE`
define.
Currently `EHCI_N_CC` and `EHCI_N_PCC` must remain as currently set (3
and 2 respectively), however with a little more work, I think these can
be modified and have up to 7 companion controllers with more ports per
companion, as long as the combined total is <= 15 total ports.
Please verify the `Makefile.in` changes. I do not use GCC and can not
verify it is correct.
Made all devices pass USB 2.0 Compliance on WinXP (old version of the
compliance test).
USB Hub still needs a little work to be in complete compliance.
Fixed potential bug in EHCI to UHCI hand-off.
Fixed compilation error with Floppy CB/CBI emulation.
Minor syntax fixes (tabulation, old irrelevant comments)
MSD serial numbers must be 12 chars.
Added to CHANGES file
This changes a BX_INFO to a BX_DEBUG. It does not needed to be a
BX_INFO.
Second, this adds (very) minimal support for SCSI command 0xAC.
When emulating a .iso image via USB CD-ROM on Win10, without this
command, the emulation freezes when trying to access the CD-ROM. I don't
know if it is Bochs or Win10. I think Win10 expects the command to work,
or Bochs is not failing correctly.
This PR simply adds code to acknowledge the command and returns a zero
length report. This SCSI (MMC) specification states that a zero length
report is allowed.
The 8 byte header is returned, but the header indicates a zero byte
return: No report segments returned.
This seems to keep Win10 from freezing at USB CD-ROM device access time.
The code also BX_DEBUG's the command sent to the "controller", for
future use when adding the actual support of the command.
This PR allows larger VHD image files. The size in question doesn't
necessarily mean the size of the VHD file on disk. The size is the total
size of the image emulated. This total size is in question.
Without this patch, the total size allowed is 65535 * 16 * 255, or
roughly 32gig.
With this patch, the total size is calculated by who the creator of the
image is, what the CHS values are, and possibly a total size of up to a
limit of 2TB.
Since the original code was ported from QEMU, I ported an updated QEMU
code snippet.
https://gitlab.com/qemu-project/qemu/-/blob/master/block/vpc.c?ref_type=heads#L305
This PR now allows VHD image files with a total size emulated less than
or equal to 2TB.
This improves the detection of the default speed of a device when given
in bochsrc.
It will first get the max speed of the port given. For example, uhci and
ohci will return FULL, ehci will return HIGH.
xHCI will return SUPER if the port is a super-speed port, else it will
return HIGH.
Then it will compare this with the max speed allowed for that device,
returning the highest common speed.
Of course, if the speed option is given, the above is overwritten by the
specified speed.
Both checks are still performed on the result.
1) does device allow the result speed?
2) does the controller allow the result speed?
This adds three additional checks to make sure the Guest is working
correctly.
- checks that the guest allocated the Scratchpad Area. If left zero
(NULL) the controller may access low memory.
- checks that the guest uses correct segment sizes in the interrupter
ring(s)
- checks the burst size value given to be within normal range
This also adds an internal register value for the HcCrcr (Command Ring
Control) register. Since this register reads zero by the Guest, we keep
an internal value so that the emulation can read the value, internally.
This is for features soon to be released.
Minor other syntax/comment changes (misspelled word, etc)
Checked with WinXP, Win7, and Win10
This fixes zero length packet handling.
See the following example that requests 128 bytes:
```
SETUP(8)
IN(64)
IN(64)
STATUS(0)
```
The current code erroneously assumes that there are no more IN packets
after the first two 64-byte packets.
However, what happens with the following:
```
SETUP(8)
IN(64)
IN(64)
IN(64) <--- current code assumes this will be, and expects a STATUS packet
STATUS(0)
```
Currently, the third IN(64) above will result in a coding error because
the code is expecting the STATUS packet, not another IN packet.
The "controller" must allow for more packets than expected, returning a
short packet detect on the third IN(64) packet shown above, actually
returning zero bytes, hence the Short Packet Detect.
This patch was tested on WinXP, Win7, and Win10.
The check for the amount of byte transferred in one frame time was done
before the update of the vertical pointer in the current Queue of the
last executed TD. Therefore, on the next frame time, the last executed
TD was being "executed" again. The check has now been moved to the top
of the loop.
This adds over-current signaling to the USBs four host controllers.
To signal an OC, use the runtime configuration and set the checkbox (GUI) or text config's parameter to 1.
This pull request also adds USB documentation to user.dbk.
The 'd.pcap_image_t pcapture' in usb_common.h called its constructor and set 'fd' to -1. However, in the usb_device_c() constructor, we then cleared all of 'd', resetting pcapture.fd back to zero. Then on Bochs exit, the pcap_image_t destructor tried to close the file since 'fd' was zero instead of -1. To fix this, I removed the pcap_image_t constructor and now call pcap_image_init() within the usb_device_c constructor.
Added examples for new options.
- xhci: model= option
- xhci: allow the user to select number of ports
- usb disk: proto= option (bbb or uasp)
Added xhci primary stream support.
(Secondary stream support is not included)
Added (experimental) usb disk protocol UASP for super- and high-speed usb disk devices.
(high-speed uasp support is not thoroughly tested)
xhci: fixed "change event" function
scsi: fixed bug in command 0x25
scsi: added command 9E/10
Minor:
Check speed indicator in the Transfer Descriptor matching the device speed.
Max packet size check, checking that the amount requested/sent is within limits.
Checking that the first packet sent is the GetDescriptor Request with a length equal to or less than the max packet size.
Checking the command length in the USB SCSI emulation.
Checking command parameters within various emulations (SCSI (BBB), etc.)
Implemented "Boot Protocol" for the (HID) Mouse emulation.
Fixed wheel mouse HID report (along with the other HID reports)
Added the Over Current bits to the UHCI code
Added the monitoring of the toggle bit in the TDs
Major:
Re-wrote the UHCI stack processing so that control and bulk reclamation can be used.
(This made it so Windows 7 now works)
Adding function to the USB SCSI emulation.
Enhanced the xHCI emulation to further check for errors in a driver.
- Added check in config.h whether or not Bochs is configured and compiled for
the same target platform (WIN32 / WIN64).
- Fixed some warnings that occur in the MSVC WIN64 case.
- Added simple clip check method to be called for each destination x / y change.
- Use the new clip check in both host-to-screen methods.
- Some related changes.
- TODO: verify other bitblt methods for clipping correctness.
- Calculate host-to-screen data size for "source stride packing" in color mode
similar to the monochrome mode.
- Set up data swizzle mode for all launch area operations.
- Temporarily draw polygon in normal mode in case the currently unsupported
pattern mode is selected.
- TODO #1: Rewrite of the clipping code at least for host-to-screen functions.
- TODO #2: Drawing polygon with pattern if requested.
monochrome source and "source stride" packing. Now adding the source stride
converted to pixels to the 5-bit srcX value.
TODO #1: color source needs to be handled in a similar way (no test case yet).
TODO #2: clipping needs to be rewritten at least for host-to-screen blts.
mode sets both banks to the same value. Reading VBE bank register always
returns the write aperture setting. This feature is designed for the scrolling
function of the VBE 'write character' support. VBE code should use compatibility
mode for now.
TODO: update VGABIOS to use this feature.
- VBE enable register returns 32k granularity flag if enabled.
- VBE bank register returns 32k granularity flag if VBE_DISPI_GETCAPS is set.
- Defining new DISPI ID not necessary.
- TODO: update VGABIOS to use this feature.
- Limit VBE bank size to 64k as expected.
- Enabling VBE mode now clears the whole video memory for all bpp modes unless
"noclear" bit is set.
- Prepared 32k bank granularity mode required for text output functions. Added
new VBE_DISPI_ID6 to indicate this capability. The default is still 64k.
- Added new files with small changes, output can be viewed with Wireshark.
- FIXME: Windows ports crash with original Win98 guest and USB disk connected.
- TODO: mention new feature in documentation.
- vgacore: method get_crtc_params() now returns selected video clock value.
- banshee: now using selected video clock instead of hardcoded VCLK #3.
- cirrus: added vertical frequency to mode switch BX_INFO.
- Moved setup of default i/o register values to reset().
- Set up strapInfo register bits in init().
- Don't overwrite readonly bits in miscInit1 register.
- Set up subsystem ID in VGABIOS ROM in case it's > 32k (doesn't exist yet).
- Copy subsystem ID from ROM to PCI config registers in reset().
cannot unload itself, BX_UNREGISTER_DEVICE_DEVMODEL() call will fail. Setting
plugin control flag to 0 like other devices do. When starting simulation, all
unused plugins are removed.
NOTE: The pcidev stuff is unmaintained, functionality not tested.
- Don't include paramtree.h in siminterface.h to get rid of duplicates
in a lot of Makefile dependency lines.
- Regenerated Makefile dependencies in iodev and affected subdirectories.
- FIXME #1: including debug.h causes duplicate entries of config.h and osdep.h.
- FIXME #2: cpudb.h appears in most of the iodev* dependencies, but only needed
by config.cc and the cpu class.
- Moved serial and parallel config parameter list creation to the corresponding
device code.
- Unregister serial and parallel bochsrc options in PLUGIN_FINI mode.
- Moved USB config and runtime parameter list creation to bx_init_usb_options().
- textconfig.cc: Small fixes in the optional plugin control.
- Changed variable type of the plugin_t "type" member from enum to Bit16u.
- Added support for returning device flags with the new mode PLUGIN_FLAGS in
the plugin entry functions. It is currently only used for devices that can
be connected to a PCI slot.
- Code cleanup in core device plugins: checking type no longer necessary.
- The "non-plugin" mode now also uses the "loadtype" member of plugin_t.
TODO: Change PCI slot options to bx_param_enum_c and build the choices list
using the new capabilities of the plugin API.
- Disable port "options" parameter after successful connect. While connected,
a device may install runtime options to control it's behaviour.
- Added enable handlers for the "options" parameter to ensure an attempt to
change the device while connected doesn't enable it.
- Some related code cleanups.
device handling. Added some code to handle plugins that provide more than one
device type (usb_hid, usb_msd) or with different name (usb_cbi). The device
type member is now only used by multi-function devices.
- USB hub: disable device "options" parameter if device is set to "none".
- USB hub: handle device disconnect in runtime_config().
- Removed handling of USB device type in host controllers and external hub.
- Fixed USB devices list by adding a terminating NULL entry.
- Disable USB device options parameter if device is set to "none".
- Related documentation updates.
- Moved main option from the portX parameter to the optionsX parameter. Added
new option "path" for the disk, cdrom and floppy devices, "ports" for the
USB hub and "file" for the USB printer. The legacy syntax is still supported.
- Register USB device state after init() to ensure all options that may have
effect on the save/restore list are applied.
- Report available devices on top of log file similar to hdimage, network and
sound modules.
- Changed USB device type variable typein host controllers to int.
- Prepared improved usbdev_ctl init / exit implementation.
- TODO: Create list of USB devices dynamicly from available plugins.
- Fixed loading of all plugins with one type using wildcard.
- Reduced maximum length of line when listing available modules for hdimage,
network and sound.
