Commit Graph

2593 Commits

Author SHA1 Message Date
Stanislav Shwartsman
e6adebfe2d Added MMX opcodes to x86-64 mode
Fixed problem with fetching extra byte in ESCx opcodes if FPU is disabled
2002-09-27 09:56:40 +00:00
Kevin Lawton
47f2e7c404 Got rid of the KPL64Hacks macro. The fixes below eliminated it.
Created 64-bit versions of some branch instructions and
  changed fetchdecode64.cc to use them instead.  This keeps the
  #ifdef pollution down for 32-bit code and made fixing them
  easier.  They needed to clear the upper bits of RIP for
  16-bit operand sizes.  They also should not have had a protection
  limit check in them, especially since that field is still
  32-bit in cpu.h, so there's no way to set nominal 64-bit values.
  The 32-bit versions were also not honoring the upper 32-bits
  of RIP.

  LOOPNE64_Jb
  LOOPE64_Jb
  LOOP64_Jb
  JCXZ64_Jb

Changed all occurances of JCC_Jw/JCC_Jd in fetchdecode64.cc to
  use JCC_Jq, which was coded already.  Both JMP_Jq and JCC_Jq are
  now fixed w.r.t. 16-bit opsizes and upper RIP bit clearing.
2002-09-27 07:01:02 +00:00
Peter Tattam
5bfd2c669e SYSRET patch. need to do invalidate_prefetch() for correctness 2002-09-27 03:33:13 +00:00
Bryce Denney
4098cf607e - I screwed up the previous rev by somehow removing a line "typedef struct {".
You can imagine how well it worked without that.
2002-09-27 03:18:44 +00:00
Kevin Lawton
109111202c Modified my fix to JMP_Jq to clear all the upper bits,
63..16 when a 16-bit operand size JMP is executed.  Previous
  fix cleared only 63..32.  I since realized, this is the case
  which does parallel the 32-bit semantics.
2002-09-26 22:34:06 +00:00
Kevin Lawton
6d74a334d6 64-bit bug#1: Instructions such as MOV_ALOq were always
fetching 64-bit address opcode info, which was incorrect.

  Fixed.  Got rid of BxImmediate_Oq.  fetchdecode64.cc now
  uses BxImmediateO, like the fetch routine does.  Addresses which
  are embedded in the opcode, have a size which depends on
  the current addressing size.  For long-mode, this is
  either 64 (default) or 32 (AddrSize over-ride).  BxImmediate_O
  now conditionally fetches based on AddrSize.

64-bit bug#2: In JMP_Jq(), when the current operand size is
  16-bits, the upper dword of RIP was not being cleared.  The
  semantics with this case are weird - one would think the
  top 48 bits would be cleared, but apparently only the top
  32 bits are.  Anyways, I fixed this.

Replaced some of the messy immediate fetching (byte-by-byte) in
  fetchdecode64.cc with ReadHost{Q,D}WordFromLittleEndian() calls
  for cleanliness.  Should do this for all the cases, plus
  the 32-bit stuff.
2002-09-26 21:32:26 +00:00
Bryce Denney
79cd346575 - fix previous rev. It would probably help to put the #include before
the first time that this function is used!
2002-09-26 19:34:44 +00:00
Bryce Denney
5c6d434929 - the raw keycodes code for wxMSW refers to HIWORD, so I made it
conditionally include <windows.h>.  This may seem like a drastic step
  for just one little type, but I expect before long we may want to use
  other symbols like VK_F12 which are also in windows.h.  In a cygwin
  compile this is required.
2002-09-26 19:24:53 +00:00
Stanislav Shwartsman
f987ad036e Changed BxError to UndefinedOpcode function for UD2 opcode (oF 0B) 2002-09-26 18:58:50 +00:00
Bryce Denney
ce8030585f - if cross configuring, don't insist on finding pthreads for the rfb and
wxWindows guis.
- if cross configuring, don't insist on finding curses library.
- on normal configures, when the target platform is win32 (windows, cygwin,
  mingw), don't insist on finding pthread either.

(I'm starting to wonder if when cross_configure=1 we shouldn't just skip over
ALL of the library and header checks.  When you're going to configure on one
platform and build on another, all that information is useless anyway.)
2002-09-26 18:23:34 +00:00
Mike Lerwill
e0275c644f Silence compiler warnings on VC6.0 2002-09-26 09:00:52 +00:00
Bryce Denney
a50f374a28 - fix typo: "event-type" to "event->type" 2002-09-26 03:01:13 +00:00
Bryce Denney
bbc43ba25b - avoid segfaults by checking if the term gui has already been initialized
or not, before calling curses functions.
2002-09-26 02:53:58 +00:00
Bryce Denney
d7251f8693 - move comment on bx_gui_c::specific_init down so that it's next to the
method it describes
2002-09-26 02:46:39 +00:00
Bryce Denney
73681b3a53 - for term compile, search for a function called "color_set" and compile it
away if color_set doesn't exist on that platform.  On Solaris this was
  the only thing keeping term from compiling.  See this bug
  [ 613393 ] solaris: term compile missing color_set
  http://sourceforge.net/tracker/index.php?func=detail&aid=613393&group_id=12580&atid=112580
2002-09-26 02:36:04 +00:00
Bryce Denney
4be28bc5ed - RFB_LIBS was getting overwritten by pthread libs. arg! 2002-09-26 02:19:14 +00:00
Bryce Denney
ec096b9653 *** empty log message *** 2002-09-26 02:15:56 +00:00
Bryce Denney
2352bd4992 - oops, the substitution variable is RFB_LIBS 2002-09-26 02:15:20 +00:00
Bryce Denney
48e0e9dc7d - update so that it applies clean again 2002-09-26 02:12:00 +00:00
Bryce Denney
45acd5b67a - rfb gui needs the socket function. on solaris, you need to link with
-lsocket to get it.
2002-09-26 01:53:29 +00:00
Bryce Denney
085ebec771 - rfb gui needs the socket function. on solaris, you need to link with
-lsocket to get it.
2002-09-26 01:45:10 +00:00
Bryce Denney
87835fd841 - fix memory leaks in the text mode interface that I introduced while working
on the wxWindows interface.  There are many more changes here than
  absolutely required to fix the memory leaks.  Instead, I've tried to
  clean things up so that it does the right thing, and is easier to
  read and maintain.
- For events that the text mode interface is going to ignore anyway, I #ifdefed
  the event creation code instead of calling new and then delete.
- now all synchronous events in siminterface.cc are created as local variables
  on the stack.  Some of them were allocated with new before, and yes some of
  them leaked.
- now I ignore the result of sim_to_ci_event (&event).  It was always
  returning a pointer to the input event anyway.  This makes the event
  sending code simpler.
- wxmain.cc:
  - in the BxEvent handling functions, now all cases "break" down to common
    code at the end which deletes async events.  This is easier to read than
    having each case handle the delete individually.
  - in OnLogMsg, do not delete the event here because it is now handled
    in the common code of OnSim2CIEvent instead.
- thanks to Christophe for pointing out the location of the worst
  memory leak.
2002-09-25 22:54:23 +00:00
Bryce Denney
56b58e6226 - remove "Create Image" button for cdrom from the sketch 2002-09-25 21:28:49 +00:00
Christophe Bothamy
83235c10ec - malloc'ed memory was one byte too short 2002-09-25 19:24:26 +00:00
Bryce Denney
e6e2c08809 - make parse_bochsrc return -1 if any parse errors occur.
- make bx_init_main return -1 if any parse errors occur.
- wxWindows: if bx_init_main returns -1, don't even show the application
  window.  You'll get an error dialog and then it exits.  So far this
  only happens if you run with -q and the parse fails.
- non-wxWindows: if bx_init_main returns -1, just exit.  So far this
  only happens if you run with -q and the parse fails.
- with these changes, handling of bochsrc parse errors seems to work
  as you would expect.  And it certainly doesn't go into an infinite
  recursive loop, as it used to!
- a little more testing and I can close
  bug 614175: infinite panic loop if bochsrc buggy
- modified: main.cc gui/siminterface.h gui/wxmain.cc
2002-09-25 19:05:01 +00:00
Bryce Denney
2b4fb2b0ed - until init_done is set, don't dump CPU registers. It's frustrating to
get a screen full of cpu registers when you actually have a bochsrc
  parse error.
2002-09-25 18:49:35 +00:00
Bryce Denney
a4f0c5b53e - move read configuration and save configuration behavior out of wx.cc
and into wxmain.cc, like other actions.
- set a default siminterface callback for the whole application, which is used
  whenever the simulator is not running.  This is important when the wx code
  calls simulator or param code and triggers a BX_PANIC or something.
  The default callback is responsible for displaying error messages which
  appear while reading the bochsrc, for example.
- move the implementation of BX_SYNC_EVT_LOG_ASK and BX_ASYNC_EVT_LOG_MSG
  into a separate function OnLogMsg().  In the future, OnLogMsg() may be called
  from the application default callback on errors.
- modified: gui/wx.cc gui/wxmain.cc gui/wxmain.h
2002-09-25 18:40:15 +00:00
Bryce Denney
e3760bedc8 - change default action on panic to ASK, for wx only. Once I get the
wx behavior worked out, I'll take out the ifdef and settle on something.
- use new enums BX_LOG_ASK_CHOICE_CONTINUE, BX_LOG_ASK_CHOICE_DIE, etc.
- do not use BX_PANIC inside fatal function! it is very likely to reenter,
  causing infinite recursion.  Use fprintf instead.
2002-09-25 18:35:06 +00:00
Bryce Denney
fd2566023a - add enums for the LOG_ASK choices: continue, die, etc. 2002-09-25 18:32:36 +00:00
Bryce Denney
13de366861 - now parse_line_formatted and parse_line_unformatted return a signed int
(0=success, -1=failure).  If parse_line_unformatted returns failure, I
  stop parsing the bochsrc.
- implement all parsing BX_PANICs with new PARSE_ERR macro, so that we can
  easily change the behavior of all parse errors at once.  For now, I want
  it to BX_PANIC and then return -1 (case panic continues).  Eventually I
  might turn it into BX_ERROR and return -1.
2002-09-25 18:31:38 +00:00
Bryce Denney
2ee888f3bb - create the quit_context as early as possible, in case fatal() gets called
from bx_init_main, bx_do_text_config_interface, or bx_config_interface.
  This should fix bug #614175: infinite panic loop if bochsrc buggy
2002-09-25 16:26:29 +00:00
Bryce Denney
ba6afa5e97 minor cleanups:
- moved cpu_online_map into the BX_CPU_C structure as a static member
  (there is only one per bochs, not one per CPU)
- reduced the diffs in several places to make it more clear what had changed
- removed lots of whitespace diffs
2002-09-25 14:52:45 +00:00
Peter Tattam
67082a5b50 Implemented SWAPGS instruction.
Note that it is unusual to decode (see SGDT instruction)
2002-09-25 14:09:08 +00:00
Bryce Denney
4570f1ba77 - add highmem patch from Zwane Mwaikambo 2002-09-25 13:29:11 +00:00
Bryce Denney
8f9bec3919 - remove unused, and incorrect MSR fields 2002-09-25 13:26:04 +00:00
Bryce Denney
b77654f907 - add APIC patch from Zwane Mwaikambo 2002-09-25 13:16:04 +00:00
Bryce Denney
610d7f00d0 - make the parser a little more informative when things go wrong. 2002-09-25 13:01:30 +00:00
Peter Tattam
a0d90e9b39 Implemented SYSCALL and SYSRET as part of x86-64 emulation.
Since the SYSCALL replaces the LOADALL instruction, it is incompatible with
earlier CPU types.

At moment, the SYSCALL is only enabled by x86-64 emulation, but the code
can be incorporated in IA32 only emulations.

Instructions added:

0F 05		SYSCALL		(replaces LOADALL)
0F 07		SYSRET		(new)

TODO:  restructure #if ... so that it can be used by non x86-64 emulations.
2002-09-25 12:54:41 +00:00
Bryce Denney
6ba1fd1b06 - VC++ doesn't like declaring a var inside a for loop, and then later
declaring another.  I guess it doesn't think the var went out of
  scope at end of first loop.  Oh well, it's easy to fix.
2002-09-25 08:04:07 +00:00
Bryce Denney
3889b9339c - add missing right alt keycode XK_Alt_R to all keymaps 2002-09-25 08:00:25 +00:00
Bryce Denney
a66f903731 - add a few more missing keys: left&right windows keys, pause
- clean up remainder of the old scancode-fixing hack
- change the BX_INFO on each keypress to a BX_DEBUG
2002-09-25 07:24:41 +00:00
Bryce Denney
ae1ac3e266 - add cases in switch statement for keys: Right CTRL, Right ALT, Scroll Lock,
Print Screen, Pause.
2002-09-25 07:21:38 +00:00
Peter Tattam
0de63e8492 Fix to stop external debugger trapping for all exceptions. 2002-09-25 06:36:42 +00:00
Bryce Denney
2107fb3af9 - in the keycode conversion switch statement, rename bx_key to key_event so
that it is as close as possible to the x.cc version.  I've made some
  improvements in x.cc and I want to apply them to wxWindows as well.
2002-09-25 05:35:36 +00:00
Bryce Denney
b99b6a157e - change the way x86-64 default options work to hopefully make them
smarter.  If you use --enable-x86-64, it has the following side
  effects on other options:
  - cpu level defaults to 6 instead of 5
  - ignore bad msr defaults to on (Peter T. requested this)
- if cpu level is 6 or greater, these options are enabled by default.
  - 4meg pages (PSE)
  - physical address extensions (PAE)
  - global pages (PGE)
  - apic support
2002-09-25 04:24:46 +00:00
Bryce Denney
6ccc57320f - add sanity checks for x86-64 compiles to ensure that the required
features are enabled.  Currently checks for cpu level > 5, support
  for PAE, PGE, and PSE.
2002-09-25 03:58:40 +00:00
Bryce Denney
e3539d7a4a - from Peter Tattam: add missing instruction MOVNTI to decode table 2002-09-25 03:32:12 +00:00
Bryce Denney
62066dbd4a - I modified Zwane's patch. see my comments inside 2002-09-25 03:17:32 +00:00
Bryce Denney
8e103b3ffb - fix compile warning about checking that an unsigned int was >= 0 2002-09-25 01:50:14 +00:00
Bryce Denney
70a7056b23 - add tsc patch from Zwane Mwaikambo 2002-09-25 00:00:11 +00:00