/[gxemul]/trunk/RELEASE

This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!

Diff of /trunk/RELEASE

Parent Directory | Revision Log | View Patch Patch

-revision 28 by dpavlin,
Mon Oct  8 16:20:26 2007 UTC
+revision 32 by dpavlin,
Mon Oct  8 16:20:58 2007 UTC
 Line 1
- Release notes for Gavare's eXperimental Emulator (GXemul), 0.4.1
+ Release notes for Gavare's eXperimental Emulator (GXemul), 0.4.3
  ================================================================
  Copyright (C) 2003-2006  Anders Gavare.
 Line 14 
 be regarded as "working" in GXemul. The
  systems are probably NetBSD/pmax, NetBSD/cats, and OpenBSD/cats.
- Changes between release 0.4.0.1 and 0.4.1 include, among other things:
+ Changes between release 0.4.2 and 0.4.3 include, among other things:
-     o)  Some bugs in the dynamic translation core have been fixed, making
+     o)  SuperH (SH4) emulation is now stable enough to let a NetBSD/dreamcast
-         all emulation modes (especially the MIPS mode) more stable.
+         GENERIC_MD (ramdisk) kernel reach userland.
-     o)  In the refactoring effort between 0.4.0 and 0.4.0.1, single-stepping
+     o)  There is now a simple framework for letting emulated clocks, as seen
-         (and instruction tracing) of 64-bit programs was accidentally broken.
+         by guest operating systems, run at the same speed as the host clock.
-         Single-stepping in 64-bit mode could result in strange exceptions.
-         This has been fixed in 0.4.1.
-     o)  MIPS emulation performance has been improved somewhat:
+         So far, the DECstation, MobilePro (hpcmips), NetWinder, CATS, Malta
+         (evbmips), Cobalt, Algor, Dreamcast, and testmips machine modes
+         use the new clock/timer framework.
-         R3000: After removing some buggy code (hints for physical page
+     o)  Some changes to the way expressions are evaluated in the built-in
-         translations), it was possible to remove the workaround for R3000
+         debugger, and some changes in command behaviour:
-         caches which was needed in 0.4.0.1 to make Linux and Ultrix run.
-         This gives an overall speedup for R2000/R3000.
-         For non-R3000, there have been some speedups as well. After fixing
+           x)  Expressions (including assignments) can now be arbitrarily
-         reference count bugs for 64-bit addressing in the dyntrans system,
+               complex, using parentheses, and the following operators:
-         workarounds/hacks in the tlbwr/tlbwi instructions and in the ASID
-         change helper function could be removed.
-     o)  A new -s command line option is now available, for dumping
+                   + - * /    % (modulo)  ^ (xor)  & (and)  | (or)
-         raw runtime data/statistics on every instruction to a file.
-         Currently, the following kinds of data can be dumped:
+           x)  Some internal emulator variables can now be read/written using
+               normal expressions. Examples of commands that did not work
+               earlier, but should work now:
-. the program counter (virtual address)
+                   print verbose
-. the physical address representation of the program counter
+                   r5 = sp - arch_pagesize * 4
-. the internal dyntrans instruction call pointer, useful when
+                   machine[0].statistics_enabled = 1
-            developing/optimizing the emulator
-         This data can then be analyzed by external tools.
+           x)  To force a name to be interpreted as a setting/register name,
+               a hash sign (#) is now used instead of the percentage sign (%).
+               (In the new expression evaluator, % means arithmetic modulo.)
-         (There was a -s command line option in previous releases of
+           x)  The 'focus' command now also selects a cpu, in addition to
-         the emulator, but it did not work as intended, and was not
+               selecting machine and emul.
-         usable from dyntrans emulation modes.)
+           x)  The 'reg' command only prints registers for one cpu now, not
+               all cpus in the currently focused machine.
+     o)  The wdc (standard IDE controller) had a bug which prevented disk
+         images larger than 2 GB to work correctly. This has been fixed.
+     o)  For MIPS emulation, some combinations of emulated processor + guest
+         operating system should now work better when idling (i.e. the host
+         should not run at 100% CPU):
+           x)  For MIPS32/MIPS64 and RM5200, the 'wait' instruction should
+               now work more or less as expected.
+           x)  For VR41xx (e.g. MobilePro) emulation, the standby instruction
+               should work like the 'wait' instruction.
+           x)  For R3000 emulation, where there is no hardware wait instruction,
+               I've implemented "instruction combination" hacks for both
+               NetBSD/pmax and Debian/pmax, so that their cpu idle loops are
+               detected and treated almost as a wait instruction.
+     o)  MIPS 64-bit address translation (X=1) was not fully working before;
+         TLB exception handling for xkseg and larger-than-2GB-userland should
+         now actually work. (Thanks to Juli Mallett and Carl van Schaik for
+         noticing these problems.)
+     o)  The mouse cursor update routines in DECstation (LK201) emulation
+         previously used the fact that guest OSes set the _hardware_
+         cursor position. In order to support X Windows when emulating
+         modern versions of NetBSD/pmax, which don't set the hardware
+         position anymore, a workaround has been implemented which only
+         sends relative coordinates to the guest OS. This has two drawbacks:
+. Ultrix emulation with dual- and tripple-head emulation will
+            most likely feel very strange. It will still work, though.
+. Cursor movement feels "accelerated", because the emulator
+            sends unaccelerated movements to the guest OS, which then
+            accelerates them. This can however be compensated to some
+            degree by running 'xset m 1 0' in the guest OS.
+         Having weird accelerated mouse movement is better than having no
+         mouse support at all, so this change was necessary.
  Please read the HISTORY files for more details.
-Line 149 
 them like this is in order:
+Line 190 
 them like this is in order:
      This product includes software developed by Ichiro FUKUHARA.
+     This product includes software developed by Marcus Comstedt.
  Also, src/include/alpha_rpb.h requires the following:
      Copyright (c) 1994, 1995, 1996 Carnegie-Mellon University.
-Line 166 
 See individual files for license details
+Line 209 
 See individual files for license details
  or reuse code.
- Thanks to (in no specific order) Joachim Buss, Olivier Houchard, Juli Mallett,
- Juan Romero Pardines, Alec Voropay, G�ran Weinholt, Alexander Yurchenko, and
- everyone else who has provided me with feedback.
  If you have found GXemul useful in some way, or feel like sending me comments
  or feedback in general, then mail me at anders(at)gavare.se.

 Legend:



Removed from v.28
 


changed lines


 
Added in v.32
 Legend:



Removed from v.28
 


changed lines


 
Added in v.32
-Removed from v.28
+Added in v.32

	ViewVC Help
Powered by ViewVC 1.1.26