upstream/0.4.0/TODO

$Id: TODO,v 1.292 2006/06/23 09:13:34 debug Exp $

Hm. This file is in random order, and not all parts of it are up-to-date.

--------------

Possible release schedule:

0.4.0:
        x)  Quick release, even though performance for non-R3000 MIPS dyntrans
            is really poor. (Assuming everything mentioned in the documentation
            works as expected.)

0.4.1:
        x)  FIX THE NON-R3000 TRANSLATION CACHE INVALIDATION BOTTLENECKS!
        x)  Fix the interrupt problems with Ultrix!
        x)  Find/fix bug which is triggered when building the emulator inside
            NetBSD/pmax 3.0 inside the emulator!

0.4.2 ...?
        x)  Clean-up!
        x)  Clock framework? Go through all clock devices, make sure they
            return correct data, and run at correct speeds!
        x)  Optimizations, continuing on 64-bit issues etc with dyntrans
        x)  Dyntrans with SMP... lots of work to be done here.
        x)  Dyntrans with cache emulation... lots of work here as well.
        x)  Actually use the settings object, better debugger stuff, etc.
        x)  Wait for new releases of NetBSD, and test with those.

--------------

SMP:
        o)  dev_mp doesn't work well with dyntrans yet
        o)  In general, IPIs, CAS, LL/SC etc must be made to work with dyntrans

MIPS:
        o)  Fix invalidate_asid so it works well for non-R3000 too!
        x)  [Re]add an interrupt-asserted bit for MIPS, to speed up
            interrupt handling slightly?
        +)  Print a warning on the first reserved instruction.
        +)  Some more work on opcodes.
                x) MIPS64 revision 2.
                x) _MAYBE_ TX79 and R5900 actually differ in their
                   opcodes? Check this carefully!
        o)  Dyntrans: Count register updates are probably not 100% correct yet.
        o)  Dyntrans: SMP correctness
        o)  Refactor code for performance and readability/maintainability.
        o)  Instruction combinations? Possible candidates (but profile first!):
                o)  multiple loads/stores in a row
                o)  strlen, memset loops etc
                o)  compare + branch
        o)  DROTR32 and similar MIPS64 rev 2 instructions, which have
            a rotation bit which differs from previous ISAs.
        o)  EI and DI instructions for MIPS64/32 rev 2. NOTE: These are
            _NOT_ the same as for R5900!
        o)  R4000 and others:
                x)  watchhi/watchlo exceptions, and other exception
                    handling details
        o)  R10000 and others:  (R12000, R14000 ?)
                x)  memory space, exceptions, ...
                x)  use cop0 framemask for tlb lookups
                    (http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi/hdwr/bks/SGI_Developer/books/R10K_UM/sgi_html/t5.Ver.2.0.book_284.html)

Dyntrans:
        x)  Move the mips_init_64bit_dummy_tables() etc calls into
            src/cpu.c, for all 64-bit cpus?
        x)  64-bit "phystranslation" lookup as in 32-bit mode? Would probably
                help performance a bit.
        x)  Common fatal_abort() function, which drops into the debugger
                without continuing.
        x)  INVALIDATION should cause translations in _all_ cpus to be
            invalidated, e.g. on a write to a write-protected page
            (containing code)
        x)  better (formally defined) instr call statistics (-s command
            line option?), multiple different types? (virtual pc, physical pc)
        x)  Call/return hints?
        x)  16-bit encodings? (MIPS16, ARM Thumb, SH3, ...)
        x)  H8?
        x)  Lots of other stuff: see src/cpus/README_DYNTRANS
        x)  true recompilation backend? think carefully about this,
            experiment in a separate project (not in GXemul)
        x)  Remove the dyntrans_alignment_check functionality; although
            it gives slightly higher peformance sometimes, it increases
            the complexity of the code too much!

Alpha:
        o)  Virtual memory (tlbs etc)
        o)  Get {NetBSD,OpenBSD,Linux}/alpha booting. :)

SPARC:
        o)  Add all registers (floating point, control regs etc)
        o)  Save/restore register windows etc!
        o)  Disassemly of some more instructions.
        o)  Are sll etc 32-bit sign-extending or zero-extending?
        o)  Finish the cmp (subcc) flag computation code.
        o)  Finish the GDB register stuff.

Debugger:
        o)  How does SMP debugging work? Does it simply use "threads"?
                What if the guest OS (running on an emulated SMP machine)
                has a usertask running, with userland threads?
        o)  Try to make the debugger more modular and, if possible, reentrant!
        o)  Remove the emul command? (But show network info if showing
                machines?)
        o)  Generalize the expression evaluator. (debugger_expr.c?)
                settable variables      ("show nr of instructions on average")
                emul[x]                 defaults to current emul
                machine[x]              defaults to current machine
                cpu[x]                  defaults to currently focused cpu
                registers               cpu arch dependent (#-prefix)
                symbols                 @-prefix
                numeric constants       decimal, hex, and octal ($-prefix)
                boolean                 yes,no, true,false
                operators (+ - * / % & | ^ !)
                parentheses for grouping subexpressions
                NOTE: the change from % to # for register prefix!
                examples:
                                emul[0].machine[2].cpu[0].pc
                                machine[test2].cpu[1].ra = main
                                settings.show_trace_tree = yes

                Settings:
                        o)  Remove a setting.
                        o)  Read/write a setting given a name. (Read as
                            string and/or int64_t simultaneously?)

                Help command should have subsections! One for "expressions",
                mirrored in the documentation, but the internal help should
                be the one that should be considered correct.
        o)  see src/debugger.c for more

POWER/PowerPC:
        x)  PPC optimizations; instr combs
        x)  64-bit stuff
        x)  find and fix the bug which causes NetBSD/macppc to fail after
            an install!
        x)  macppc: adb controller; keyboard (for framebuffer mode)
        x)  make OpenBSD/macppc work (PCI controller stuff)

Algor:
        PCI and ISA and LOCAL interrupts! --> wdc could start working
        Add interrupt controller in dev_algor.c.

ARM:
        o)  try to get netbsd/evbarm 3.x running (iq80321)
        o)  make the xscale counter registers (ccnt) work
        o)  make the ata controller usable for FreeBSD!
        o)  zaurus for openbsd...
        o)  debian/cats crashes because of unimplemented coproc stuff.
            fix this?

Cache simulation:
        o)  Separate from the CPU concept, so that multi-core CPUs sharing
            e.g. a L2 cache can be simulated (?)
        o)  Instruction cache emulation is easiest (if separate from the
            data cache); similar hack as the S;I; hack in cpu_dyntrans.c.
            NOTE: if the architecture has a delay slot, then an instruction
            slot can actually be executed as 2 instructions.
        o)  Data cache emulation = harder; each arch's load/store routines
            must include support? running one instruction at a time and
            having a cpu-dependant lookup function for each instruction
            is another option (easier to implement, but very very slow).

Documentation:
        o)  machines, cpus, devices.
        o)  Automagic documentation generation:
                x)  REMEMBER that several machines/devices can be in
                        the same source file!
        o)  Try to rewrite the install instructions for those machines
            that use 3MAX into using CATS? (To remove the need to a raw
            ffs partition using up all of the disk image.)

More generic out_of_memory error reporting, and check everywhere!
        Causes: OpenBSD has low default limits for normal users.
                Host is 32-bit? (32-bit hosts are limited to 4 GB or less
                of userspace memory.)
                You are actually low on RAM. (As trivial as this might sound,
                Unix systems usually allow processes to allocate virtual
                memory beyond the amount of RAM in the machine.)

Breakpoints: 32-bit vs 64-bit sign extension for MIPS, warnings, etc.
        Use the debugger's symbolic name stuff. (which will have to be
        extended soon to support stuff like  "2*x + symbol + y" etc. cool
        stuff)

Sprite (guest OS for DECstation emulation)
        x)  Timing problems during bootup?

The Device subsystem:
        x)  allow devices to be moved and/or changed in size (down to a
            minimum size, etc, or up to a max size)
        x)  keep track of interrupts and busses? actually, allowing any device
            to be a bus might be a nice idea.
        x)  turn interrupt controllers into devices? :-)
        x)  refactor various clocks/nvram/cmos into one device?

Clocks:
        x)  General framework for automagic clock adjustment for _all_
            kinds of clocks and timers. (Which should be possible to turn
            off, of course, like the way DECstation emulation works now.)

PCI:
        x)  last write was ffffffff ==> fix this, it should be used
            together with a mask to get the correct bits. also, not ALL
            bits are size bits! (lowest 4 vs lowest 2?)
        x)  add support for address fixups
        x)  generalize the interrupt routing stuff (lines etc). this should
            be per machine? or per bus, that's better
        x)  add a "pcn" NIC (AMD PCnet32 Lance 79c970 (PCI 1022:2000)),
            could be useful for several machine modes (Malta, Algor, evbarm,
            hp700?, macppc, etc.)

Network layer:
        o)  DHCP (for Debian and BSD installers :-)
        o)  increase performance
        o)  don't rely on NetBSD-ish usage
        o)  Multiple networks per emulation, and let different
            NICs in machines connect to different networks.
        o)  many other issues: see src/net.c

Busses:
        o)  Redesign the entire "mainbus" concept!
        o)  Busses should be placed in a hierarchical tree!
        o)  Easily configurable interrupt routing in SMP systems.
        o)  Specific clock/bus speeds, cpu speeds etc.
        o)  Synchronization over network? or at least in dyntrans within
            one emulated machine
        o)  dev->bus: TurboChannel, PCMCIA, ADB?

Config file parser:
        o)  Rewrite it from scratch!
        o)  Usage of any expression available through the debugger
        o)  Support for running debugger commands (like the -c
            command line option)

Floating point layer:
        o)  make it common enough to be used by _all_ emulation modes
        o)  implement more stuff
        o)  non-IEEE modes (i.e. x86)?

Userland emulation:
        x)  Lots of stuff; freebsd and netbsd (and linux?) syscalls.
        x)  Dynamic linking? Hm.

Sound:
        x)  generic sound framework
        x)  add one or more sound cards as devices

ASC SCSI controller:
        x)  NetBSD/arc 2.0 uses the ASC controller in a way which GXemul
            cannot yet handle. (NetBSD 1.6.2 works ok.) (Possibly a problem
            in NetBSD itself, http://mail-index.netbsd.org/source-changes/
            2005/11/06/0024.html suggests that.)

Caches / memory hierarchies: (this is mostly MIPS-specific)
        o)  MIPS coproc.c: bits in config registers should reflect
            correct cache sizes for _all_ CPU types. (currently only
            implemented for R4000, R1x000, and a few others)
        o)  src/memory*.c: Implement correct cache emulation for
            all CPU types. (currently only R2000/R3000 is implemented)
            (per CPU, multiple levels should be possible,
            associativity etc!)
        o)  R2000/R3000 isn't _100%_ correct, just almost correct :)
        o)  Move the -S (fill mem with random) functionality into the
            memory.c subsystem, not machine.c or wherever it is now
        o)  ECC stuff, simulation of memory errors?  (Machine dependent)
        o)  More than 4GB of emulated RAM, when run on a 32-bit host?
            (using manual swap-out of blocks to disk, ugly)
        o)  A global command line option should be used to turn
            cache emulation on or off. When off, caches should be
            faked like they are right now. When on, caches and
            memory latencies should be emulated as correctly as
            possible.

File/disk/symbol handling:
        o)  Better handling of tape files
        o)  Read function argument count and types from binaries? (ELF?)
        o)  Better demangling of C++ names. Note: GNU's C++ differs from e.g.
            Microsoft's C++, so multiple schemes must be possible. See
            URL at top of src/symbol_demangle.c for more info.

Userland ABI emulation:
        o)  see src/useremul.c

Terminal/console:
        o)  allow emulated serial ports to be connected to the outside
            world in a more generic way, or even to other emulated
            machines(?)

Save state of the whole emulated machine, to be able to load it back
        in later?  (Memory, all device's states, all registers and
        so on.  Like taking a snapshot. (SimOS seems to do this,
        according to its website.))

Better framebuffer and X-windows functionality:
        o)  -Yx sometimes causes crashes.
        o)  Simple device access to framebuffer_blockcopyfill() etc,
            and text output (using the built-in fonts), for dev_fb.
        o)  CLEAN UP the ugly event code
        o)  Mouse clicks can be "missed" in the current system; this is
            not good. They should be put on a stack of some kind.
        o)  More 2D and 3D framebuffer acceleration.
        o)  Non-resizable windows?  Or choose scaledown depending
                on size (and center the image, with a black border).
        o)  Different scaledown on different windows?
        o)  Switch scaledown during runtime? (Ala CTRL-ALT-plus/minus)
        o)  Bug reported by Elijah Rutschman on MacOS with weird
            keys (F5 = cursor down?).
        o)  Keyboard and mouse events:
                x)  Do this for more machines than just DECstation
                x)  more X11 cursor keycodes
                x)  Keys like CTRL, ALT, SHIFT do not get through
                    by themselves (these are necessary for example
                    to change the font of an xterm in X in the
                    emulator)
        o)  Generalize the framebuffer stuff by moving _ALL_ X11
                specific code to src/x11.c!

Statistics:  (this could be interesting)
        o)  Save to file and show graphics. It should be possible to
            run gxemul after a simulation to just show the graphics,
            or convert to a .ppm or .tga or similar.
        o)  memory accesses (to measure cache efficiency and
                page coloring efficiency)
        o)  nr of simultaneous ASIDs in use in the TLB, for MIPS
        o)  percentage of time spent in different "states", such as
            running userland code, kernel code, or idling (for CPUs
            that have such an instruction, or whenever the PC is
            inside a specific idle-function (address range)).
            Possible additional state (for example on R3000): caches
            disabled.
        o)  position of read/write on (SCSI) disks

1	$Id: TODO,v 1.292 2006/06/23 09:13:34 debug Exp $
2
3	Hm. This file is in random order, and not all parts of it are up-to-date.
4
5	--------------
6
7	Possible release schedule:
8
9	0.4.0:
10	x) Quick release, even though performance for non-R3000 MIPS dyntrans
11	is really poor. (Assuming everything mentioned in the documentation
12	works as expected.)
13
14	0.4.1:
15	x) FIX THE NON-R3000 TRANSLATION CACHE INVALIDATION BOTTLENECKS!
16	x) Fix the interrupt problems with Ultrix!
17	x) Find/fix bug which is triggered when building the emulator inside
18	NetBSD/pmax 3.0 inside the emulator!
19
20	0.4.2 ...?
21	x) Clean-up!
22	x) Clock framework? Go through all clock devices, make sure they
23	return correct data, and run at correct speeds!
24	x) Optimizations, continuing on 64-bit issues etc with dyntrans
25	x) Dyntrans with SMP... lots of work to be done here.
26	x) Dyntrans with cache emulation... lots of work here as well.
27	x) Actually use the settings object, better debugger stuff, etc.
28	x) Wait for new releases of NetBSD, and test with those.
29
30	--------------
31
32	SMP:
33	o) dev_mp doesn't work well with dyntrans yet
34	o) In general, IPIs, CAS, LL/SC etc must be made to work with dyntrans
35
36	MIPS:
37	o) Fix invalidate_asid so it works well for non-R3000 too!
38	x) [Re]add an interrupt-asserted bit for MIPS, to speed up
39	interrupt handling slightly?
40	+) Print a warning on the first reserved instruction.
41	+) Some more work on opcodes.
42	x) MIPS64 revision 2.
43	x) _MAYBE_ TX79 and R5900 actually differ in their
44	opcodes? Check this carefully!
45	o) Dyntrans: Count register updates are probably not 100% correct yet.
46	o) Dyntrans: SMP correctness
47	o) Refactor code for performance and readability/maintainability.
48	o) Instruction combinations? Possible candidates (but profile first!):
49	o) multiple loads/stores in a row
50	o) strlen, memset loops etc
51	o) compare + branch
52	o) DROTR32 and similar MIPS64 rev 2 instructions, which have
53	a rotation bit which differs from previous ISAs.
54	o) EI and DI instructions for MIPS64/32 rev 2. NOTE: These are
55	_NOT_ the same as for R5900!
56	o) R4000 and others:
57	x) watchhi/watchlo exceptions, and other exception
58	handling details
59	o) R10000 and others: (R12000, R14000 ?)
60	x) memory space, exceptions, ...
61	x) use cop0 framemask for tlb lookups
62	(http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi/hdwr/bks/SGI_Developer/books/R10K_UM/sgi_html/t5.Ver.2.0.book_284.html)
63
64	Dyntrans:
65	x) Move the mips_init_64bit_dummy_tables() etc calls into
66	src/cpu.c, for all 64-bit cpus?
67	x) 64-bit "phystranslation" lookup as in 32-bit mode? Would probably
68	help performance a bit.
69	x) Common fatal_abort() function, which drops into the debugger
70	without continuing.
71	x) INVALIDATION should cause translations in _all_ cpus to be
72	invalidated, e.g. on a write to a write-protected page
73	(containing code)
74	x) better (formally defined) instr call statistics (-s command
75	line option?), multiple different types? (virtual pc, physical pc)
76	x) Call/return hints?
77	x) 16-bit encodings? (MIPS16, ARM Thumb, SH3, ...)
78	x) H8?
79	x) Lots of other stuff: see src/cpus/README_DYNTRANS
80	x) true recompilation backend? think carefully about this,
81	experiment in a separate project (not in GXemul)
82	x) Remove the dyntrans_alignment_check functionality; although
83	it gives slightly higher peformance sometimes, it increases
84	the complexity of the code too much!
85
86	Alpha:
87	o) Virtual memory (tlbs etc)
88	o) Get {NetBSD,OpenBSD,Linux}/alpha booting. :)
89
90	SPARC:
91	o) Add all registers (floating point, control regs etc)
92	o) Save/restore register windows etc!
93	o) Disassemly of some more instructions.
94	o) Are sll etc 32-bit sign-extending or zero-extending?
95	o) Finish the cmp (subcc) flag computation code.
96	o) Finish the GDB register stuff.
97
98	Debugger:
99	o) How does SMP debugging work? Does it simply use "threads"?
100	What if the guest OS (running on an emulated SMP machine)
101	has a usertask running, with userland threads?
102	o) Try to make the debugger more modular and, if possible, reentrant!
103	o) Remove the emul command? (But show network info if showing
104	machines?)
105	o) Generalize the expression evaluator. (debugger_expr.c?)
106	settable variables ("show nr of instructions on average")
107	emul[x] defaults to current emul
108	machine[x] defaults to current machine
109	cpu[x] defaults to currently focused cpu
110	registers cpu arch dependent (#-prefix)
111	symbols @-prefix
112	numeric constants decimal, hex, and octal ($-prefix)
113	boolean yes,no, true,false
114	operators (+ - * / % & \| ^ !)
115	parentheses for grouping subexpressions
116	NOTE: the change from % to # for register prefix!
117	examples:
118	emul[0].machine[2].cpu[0].pc
119	machine[test2].cpu[1].ra = main
120	settings.show_trace_tree = yes
121
122	Settings:
123	o) Remove a setting.
124	o) Read/write a setting given a name. (Read as
125	string and/or int64_t simultaneously?)
126
127	Help command should have subsections! One for "expressions",
128	mirrored in the documentation, but the internal help should
129	be the one that should be considered correct.
130	o) see src/debugger.c for more
131
132	POWER/PowerPC:
133	x) PPC optimizations; instr combs
134	x) 64-bit stuff
135	x) find and fix the bug which causes NetBSD/macppc to fail after
136	an install!
137	x) macppc: adb controller; keyboard (for framebuffer mode)
138	x) make OpenBSD/macppc work (PCI controller stuff)
139
140	Algor:
141	PCI and ISA and LOCAL interrupts! --> wdc could start working
142	Add interrupt controller in dev_algor.c.
143
144	ARM:
145	o) try to get netbsd/evbarm 3.x running (iq80321)
146	o) make the xscale counter registers (ccnt) work
147	o) make the ata controller usable for FreeBSD!
148	o) zaurus for openbsd...
149	o) debian/cats crashes because of unimplemented coproc stuff.
150	fix this?
151
152	Cache simulation:
153	o) Separate from the CPU concept, so that multi-core CPUs sharing
154	e.g. a L2 cache can be simulated (?)
155	o) Instruction cache emulation is easiest (if separate from the
156	data cache); similar hack as the S;I; hack in cpu_dyntrans.c.
157	NOTE: if the architecture has a delay slot, then an instruction
158	slot can actually be executed as 2 instructions.
159	o) Data cache emulation = harder; each arch's load/store routines
160	must include support? running one instruction at a time and
161	having a cpu-dependant lookup function for each instruction
162	is another option (easier to implement, but very very slow).
163
164	Documentation:
165	o) machines, cpus, devices.
166	o) Automagic documentation generation:
167	x) REMEMBER that several machines/devices can be in
168	the same source file!
169	o) Try to rewrite the install instructions for those machines
170	that use 3MAX into using CATS? (To remove the need to a raw
171	ffs partition using up all of the disk image.)
172
173	More generic out_of_memory error reporting, and check everywhere!
174	Causes: OpenBSD has low default limits for normal users.
175	Host is 32-bit? (32-bit hosts are limited to 4 GB or less
176	of userspace memory.)
177	You are actually low on RAM. (As trivial as this might sound,
178	Unix systems usually allow processes to allocate virtual
179	memory beyond the amount of RAM in the machine.)
180
181	Breakpoints: 32-bit vs 64-bit sign extension for MIPS, warnings, etc.
182	Use the debugger's symbolic name stuff. (which will have to be
183	extended soon to support stuff like "2*x + symbol + y" etc. cool
184	stuff)
185
186	Sprite (guest OS for DECstation emulation)
187	x) Timing problems during bootup?
188
189	The Device subsystem:
190	x) allow devices to be moved and/or changed in size (down to a
191	minimum size, etc, or up to a max size)
192	x) keep track of interrupts and busses? actually, allowing any device
193	to be a bus might be a nice idea.
194	x) turn interrupt controllers into devices? :-)
195	x) refactor various clocks/nvram/cmos into one device?
196
197	Clocks:
198	x) General framework for automagic clock adjustment for _all_
199	kinds of clocks and timers. (Which should be possible to turn
200	off, of course, like the way DECstation emulation works now.)
201
202	PCI:
203	x) last write was ffffffff ==> fix this, it should be used
204	together with a mask to get the correct bits. also, not ALL
205	bits are size bits! (lowest 4 vs lowest 2?)
206	x) add support for address fixups
207	x) generalize the interrupt routing stuff (lines etc). this should
208	be per machine? or per bus, that's better
209	x) add a "pcn" NIC (AMD PCnet32 Lance 79c970 (PCI 1022:2000)),
210	could be useful for several machine modes (Malta, Algor, evbarm,
211	hp700?, macppc, etc.)
212
213	Network layer:
214	o) DHCP (for Debian and BSD installers :-)
215	o) increase performance
216	o) don't rely on NetBSD-ish usage
217	o) Multiple networks per emulation, and let different
218	NICs in machines connect to different networks.
219	o) many other issues: see src/net.c
220
221	Busses:
222	o) Redesign the entire "mainbus" concept!
223	o) Busses should be placed in a hierarchical tree!
224	o) Easily configurable interrupt routing in SMP systems.
225	o) Specific clock/bus speeds, cpu speeds etc.
226	o) Synchronization over network? or at least in dyntrans within
227	one emulated machine
228	o) dev->bus: TurboChannel, PCMCIA, ADB?
229
230	Config file parser:
231	o) Rewrite it from scratch!
232	o) Usage of any expression available through the debugger
233	o) Support for running debugger commands (like the -c
234	command line option)
235
236	Floating point layer:
237	o) make it common enough to be used by _all_ emulation modes
238	o) implement more stuff
239	o) non-IEEE modes (i.e. x86)?
240
241	Userland emulation:
242	x) Lots of stuff; freebsd and netbsd (and linux?) syscalls.
243	x) Dynamic linking? Hm.
244
245	Sound:
246	x) generic sound framework
247	x) add one or more sound cards as devices
248
249	ASC SCSI controller:
250	x) NetBSD/arc 2.0 uses the ASC controller in a way which GXemul
251	cannot yet handle. (NetBSD 1.6.2 works ok.) (Possibly a problem
252	in NetBSD itself, http://mail-index.netbsd.org/source-changes/
253	2005/11/06/0024.html suggests that.)
254
255	Caches / memory hierarchies: (this is mostly MIPS-specific)
256	o) MIPS coproc.c: bits in config registers should reflect
257	correct cache sizes for _all_ CPU types. (currently only
258	implemented for R4000, R1x000, and a few others)
259	o) src/memory*.c: Implement correct cache emulation for
260	all CPU types. (currently only R2000/R3000 is implemented)
261	(per CPU, multiple levels should be possible,
262	associativity etc!)
263	o) R2000/R3000 isn't _100%_ correct, just almost correct :)
264	o) Move the -S (fill mem with random) functionality into the
265	memory.c subsystem, not machine.c or wherever it is now
266	o) ECC stuff, simulation of memory errors? (Machine dependent)
267	o) More than 4GB of emulated RAM, when run on a 32-bit host?
268	(using manual swap-out of blocks to disk, ugly)
269	o) A global command line option should be used to turn
270	cache emulation on or off. When off, caches should be
271	faked like they are right now. When on, caches and
272	memory latencies should be emulated as correctly as
273	possible.
274
275	File/disk/symbol handling:
276	o) Better handling of tape files
277	o) Read function argument count and types from binaries? (ELF?)
278	o) Better demangling of C++ names. Note: GNU's C++ differs from e.g.
279	Microsoft's C++, so multiple schemes must be possible. See
280	URL at top of src/symbol_demangle.c for more info.
281
282	Userland ABI emulation:
283	o) see src/useremul.c
284
285	Terminal/console:
286	o) allow emulated serial ports to be connected to the outside
287	world in a more generic way, or even to other emulated
288	machines(?)
289
290	Save state of the whole emulated machine, to be able to load it back
291	in later? (Memory, all device's states, all registers and
292	so on. Like taking a snapshot. (SimOS seems to do this,
293	according to its website.))
294
295	Better framebuffer and X-windows functionality:
296	o) -Yx sometimes causes crashes.
297	o) Simple device access to framebuffer_blockcopyfill() etc,
298	and text output (using the built-in fonts), for dev_fb.
299	o) CLEAN UP the ugly event code
300	o) Mouse clicks can be "missed" in the current system; this is
301	not good. They should be put on a stack of some kind.
302	o) More 2D and 3D framebuffer acceleration.
303	o) Non-resizable windows? Or choose scaledown depending
304	on size (and center the image, with a black border).
305	o) Different scaledown on different windows?
306	o) Switch scaledown during runtime? (Ala CTRL-ALT-plus/minus)
307	o) Bug reported by Elijah Rutschman on MacOS with weird
308	keys (F5 = cursor down?).
309	o) Keyboard and mouse events:
310	x) Do this for more machines than just DECstation
311	x) more X11 cursor keycodes
312	x) Keys like CTRL, ALT, SHIFT do not get through
313	by themselves (these are necessary for example
314	to change the font of an xterm in X in the
315	emulator)
316	o) Generalize the framebuffer stuff by moving _ALL_ X11
317	specific code to src/x11.c!
318
319	Statistics: (this could be interesting)
320	o) Save to file and show graphics. It should be possible to
321	run gxemul after a simulation to just show the graphics,
322	or convert to a .ppm or .tga or similar.
323	o) memory accesses (to measure cache efficiency and
324	page coloring efficiency)
325	o) nr of simultaneous ASIDs in use in the TLB, for MIPS
326	o) percentage of time spent in different "states", such as
327	running userland code, kernel code, or idling (for CPUs
328	that have such an instruction, or whenever the PC is
329	inside a specific idle-function (address range)).
330	Possible additional state (for example on R3000): caches
331	disabled.
332	o) position of read/write on (SCSI) disks
333