* option "cmosimage" must be saved, too
* most important options (config_interface, display_library, megs, romimage,
vgaromimage, boot) moved to the top of the saved file
- add the "journal" option to the ataX-xxx directive, so the user can
set a specific path for the redolog file (applies only for undoable
and volatile disk modes)
- add the enable handler, so the status and journal options of the ataX-xxx
directive can be hidden even if the ataX-xxx parameter is enabled.
- add handlers called when a parameter is enabled or disabled.
This is useful for parameter dependent on another, that may
not wish to be enabled when its father parameter is enabled.
fixes a bug in exception handling in v8086 mode
Here are comments from the author:
Since the bug of bochs-2.0.win32 was found and
corrected, it reports. cpu/exception.cpp of src : Within
an interrupt() function, when present is the V8086 mode,
a bug is in the portion which processes 386 (286)
int/trap gate. From the V8086 mode, this portion is
performed, when it is going to execute an int imm
command. The portion in which push_32 () is called in
the state of VM=1 is still a mistake. Although this
push_32 () tends to write in to the stack of a protected
mode, if it is still VM=1 of EFLAGS, the
write_virtual_dword() function called out of push_32 () will
take out a segment protection exception. After
performing clear_VM() etc., it is necessary to make it
call push_32 () correctly, since EFLAGS is saved locally.
Syslinux 1.48 that is used to boot freesco, did not like the latest
int15 function e801 returning AX=0 and BX=0, although my real
system does, and Ralph Brown list allow it.
Interestingly enough, the symptoms were floppy accesses with a track
number > 100. Funny.
Now int15 function e801 returns both extended and configured memory.
Syntax error fixes:
[x] fixed error with array bounds in virt_timer.cc
[x] fixed error with returning value from void functions
[x] fixed const class member initialization (according C++ standard)
[x] for-loop local variable declaration moved out of loop
(for VC 6.0 compliance)
Removed already committed patch from Dirk Thierbach
Replaced the patch from Vitaly Vorobyov by set of splitted patches.
Now any of his patches could be committed separatelly from others !
[x] fixed bug in int01 (opcode 0xF1) emulation
[x] fixed bug in x86 debugger with dr0-dr3 registers
Committed disassembler bugfix from Dirk Thierbach:
[x] fixed bug in relative addresses in Jmp, Jcc, Call and so on
