/[gxemul]/trunk/TODO
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/TODO

Parent Directory Parent Directory | Revision Log Revision Log


Revision 28 - (hide annotations)
Mon Oct 8 16:20:26 2007 UTC (13 years ago) by dpavlin
File size: 16846 byte(s)
++ trunk/HISTORY	(local)
$Id: HISTORY,v 1.1298 2006/07/22 11:27:46 debug Exp $
20060626	Continuing on SPARC emulation (beginning on the 'save'
		instruction, register windows, etc).
20060629	Planning statistics gathering (new -s command line option),
		and renaming speed_tricks to allow_instruction_combinations.
20060630	Some minor manual page updates.
		Various cleanups.
		Implementing the -s command line option.
20060701	FINALLY found the bug which prevented Linux and Ultrix from
		running without the ugly hack in the R2000/R3000 cache isol
		code; it was the phystranslation hint array which was buggy.
		Removing the phystranslation hint code completely, for now.
20060702	Minor dyntrans cleanups; invalidation of physpages now only
		invalidate those parts of a page that have actually been
		translated. (32 parts per page.)
		Some MIPS non-R3000 speed fixes.
		Experimenting with MIPS instruction combination for some
		addiu+bne+sw loops, and sw+sw+sw.
		Adding support (again) for larger-than-4KB pages in MIPS tlbw*.
		Continuing on SPARC emulation: adding load/store instructions.
20060704	Fixing a virtual vs physical page shift bug in the new tlbw*
		implementation. Problem noticed by Jakub Jermar. (Many thanks.)
		Moving rfe and eret to cpu_mips_instr.c, since that is the
		only place that uses them nowadays.
20060705	Removing the BSD license from the "testmachine" include files,
		placing them in the public domain instead; this enables the
		testmachine stuff to be used from projects which are
		incompatible with the BSD license for some reason.
20060707	Adding instruction combinations for the R2000/R3000 L1
		I-cache invalidation code used by NetBSD/pmax 3.0, lui+addiu,
		various branches followed by addiu or nop, and jr ra followed
		by addiu. The time it takes to perform a full NetBSD/pmax R3000
		install on the laptop has dropped from 573 seconds to 539. :-)
20060708	Adding a framebuffer controller device (dev_fbctrl), which so
		far can be used to change the fb resolution during runtime, but
		in the future will also be useful for accelerated block fill/
		copy, and possibly also simplified character output.
		Adding an instruction combination for NetBSD/pmax' strlen.
20060709	Minor fixes: reading raw files in src/file.c wasn't memblock
		aligned, removing buggy multi_sw MIPS instruction combination,
		etc.
20060711	Adding a machine_qemu.c, which contains a "qemu_mips" machine.
		(It mimics QEMU's MIPS machine mode, so that a test kernel
		made for QEMU_MIPS also can run in GXemul... at least to some
		extent.)  Adding a short section about how to run this mode to
		doc/guestoses.html.
20060714	Misc. minor code cleanups.
20060715	Applying a patch which adds getchar() to promemul/yamon.c
		(from Oleksandr Tymoshenko).
		Adding yamon.h from NetBSD, and rewriting yamon.c to use it
		(instead of ugly hardcoded numbers) + some cleanup.
20060716	Found and fixed the bug which broke single-stepping of 64-bit
		programs between 0.4.0 and 0.4.0.1 (caused by too quick
		refactoring and no testing). Hopefully this fix will not
		break too many other things.
20060718	Continuing on the 8253 PIT; it now works with Linux/QEMU_MIPS.
		Re-adding the sw+sw+sw instr comb (the problem was that I had
		ignored endian issues); however, it doesn't seem to give any
		big performance gain.
20060720	Adding a dummy Transputer mode (T414, T800 etc) skeleton (only
		the 'j' and 'ldc' instructions are implemented so far). :-}
20060721	Adding gtreg.h from NetBSD, updating dev_gt.c to use it, plus
		misc. other updates to get Linux 2.6 for evbmips/malta working
		(thanks to Alec Voropay for the details).
		FINALLY found and fixed the bug which made tlbw* for non-R3000
		buggy; it was a reference count problem in the dyntrans core.
20060722	Testing stuff; things seem stable enough for a new release.

==============  RELEASE 0.4.1  ==============


1 dpavlin 28 $Id: TODO,v 1.324 2006/07/22 10:23:39 debug Exp $
2 dpavlin 2
3 dpavlin 22 Hm. This file is in random order, and not all parts of it are up-to-date.
4 dpavlin 20
5 dpavlin 28 Code cleanup:
6     x) 64-bit ranges in src/cpus/memory_mips_v2p.c
7     x) Revert the dyntrans page template experiment? Hm.
8     x) Refactor the cpu type detection/initialization/listing.
9 dpavlin 2
10 dpavlin 28 Documentation:
11     x) Rewrite the section about experimental devices, after the
12     framebuffer acceleration has been implemented, and demos
13     written. (Symbolic names instead of numbers; example
14     use cases, etc. Mention demo files that use the various
15     features?)
16     x) "a very simple linear framebuffer device (for graphics output)"
17     under "which machines does gxemul emulate" ==> better
18     description?
19     x) Better description on how to set up a cross compiler?
20     Example for MIPS64.
21    
22     Long-term implementation:
23     x) Testmachine includes:
24     + dev_fb block fill and copy
25     + dev_fb draw characters (from the built-in font)?
26     + dev_fb input device? mouse pointer coordinates and buttons
27     (allow changes in these to cause interrupts as well?)
28     + Redefine the halt() function so that it stops "sometimes
29     soon", i.e. usage in demo code should be:
30     for (;;) {
31     halt();
32     }
33     x) Rewrite the networking stack; make OpenBSD work better as a guest
34     OS, fix the performance problems, make Linux work with DHCP, etc.
35     x) Make the wdc controller work with modern versions of NetBSD!
36     x) Continue on SPARC emulation
37     + Enable it in the configure script as soon as it can
38     run all the demo programs.
39     x) Continue on Alpha emulation (virtual memory, etc). Cleanup.
40     x) Alignment exceptions (MIPS, PPC, ARM?, ...)
41    
42     Long-term design:
43     x) Instruction combination collisions? How to avoid easily...
44     o) Actually use the settings object, better debugger stuff, etc!
45     o) Debugger command for enabling/disabling instruction statistics
46     during runtime. machine.statistics = on|off
47     x) MAINBUS REDESIGN!
48     x) Clock framework! Go through all clock devices, make sure they
49 dpavlin 24 return correct data, and run at correct speeds!
50 dpavlin 28 x) Dyntrans with valgrind-inspired memory checker. (In memory_rw,
51     it would be reasonably simple to add; in each individual fast
52     load/store routine = a lot more work, and it would become
53     kludgy very fast.)
54 dpavlin 24 x) Dyntrans with SMP... lots of work to be done here.
55     x) Dyntrans with cache emulation... lots of work here as well.
56 dpavlin 28 x) Reimplement the config file parser from scratch.
57 dpavlin 24
58 dpavlin 28 Test:
59     x) Test with more than one Sprite instance on an emulated network!
60     x) NetBSD 4.x, once it is out.
61 dpavlin 24
62 dpavlin 28 -------------------------------------------------------------------------------
63    
64     Simple Valgrind-like checks?
65     o) Mark every address with bits which tell whether or not the address
66     has been written to.
67     o) What should happen when programs are loaded? Text/data, bss (zero
68     filled). But stack space and heap is uninitialized.
69     o) Uninitialized local variables:
70     A load from a place on the stack which has not previously
71     been stored to => warning. Increasing the stack pointer using
72     any available means should reset the memory to uninitialized.
73     o) If calls to malloc() and free() can be intercepted:
74     o) Access to a memory area after free() => warning.
75     o) Memory returned by malloc() is marked as not-initialized.
76     o) Non-passive, but good to have: Change the argument
77     given to malloc, to return a slightly larger memory
78     area, i.e. margin_before + size + margin_after,
79     and return the pointer + margin_before.
80     Any access to the margin_before or _after space results
81     in warnings. (free() must be modified to free the
82     actually allocated address.)
83    
84 dpavlin 24 SMP:
85     o) dev_mp doesn't work well with dyntrans yet
86     o) In general, IPIs, CAS, LL/SC etc must be made to work with dyntrans
87    
88     MIPS:
89     +) Some more work on opcodes.
90     x) MIPS64 revision 2.
91 dpavlin 28 o) Find out which actual CPUs implement the rev2 ISA!
92 dpavlin 24 x) _MAYBE_ TX79 and R5900 actually differ in their
93     opcodes? Check this carefully!
94     o) Dyntrans: Count register updates are probably not 100% correct yet.
95     o) Refactor code for performance and readability/maintainability.
96     o) Instruction combinations? Possible candidates (but profile first!):
97 dpavlin 28 o) R2000/R3000 cache cleaner!
98 dpavlin 24 o) strlen, memset loops etc
99 dpavlin 28 o) multiple loads/stores in a row, e.g. relative to
100     the stack pointer
101     o) lui + or, lui + add, and 64-bit variants
102     o) jr ra + addiu to the v0 register?
103 dpavlin 24 o) compare + branch
104     o) DROTR32 and similar MIPS64 rev 2 instructions, which have
105     a rotation bit which differs from previous ISAs.
106     o) EI and DI instructions for MIPS64/32 rev 2. NOTE: These are
107     _NOT_ the same as for R5900!
108 dpavlin 28 o) (Re)implement 128-bit loads/stores for R5900.
109 dpavlin 24 o) R4000 and others:
110     x) watchhi/watchlo exceptions, and other exception
111     handling details
112     o) R10000 and others: (R12000, R14000 ?)
113     x) memory space, exceptions, ...
114     x) use cop0 framemask for tlb lookups
115     (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)
116    
117 dpavlin 14 Dyntrans:
118 dpavlin 28 x) Redesign/rethink the delay slot mechanism used for e.g. MIPS,
119     so that it caches a translation (that is, an instruction
120     word and the instr_call it was translated to the last
121     time), so that it doesn't need to do slow
122     to_be_translated for each end of page?
123     x) Program Counter statistics:
124     Per machine? What about SMP? All data to the same file?
125     A debugger command should be possible to use to enable/
126     disable statistics gathering.
127     Configuration file option!
128 dpavlin 24 x) Common fatal_abort() function, which drops into the debugger
129     without continuing.
130 dpavlin 22 x) INVALIDATION should cause translations in _all_ cpus to be
131     invalidated, e.g. on a write to a write-protected page
132     (containing code)
133     x) Call/return hints?
134     x) 16-bit encodings? (MIPS16, ARM Thumb, SH3, ...)
135 dpavlin 24 x) H8?
136 dpavlin 18 x) Lots of other stuff: see src/cpus/README_DYNTRANS
137 dpavlin 22 x) true recompilation backend? think carefully about this,
138     experiment in a separate project (not in GXemul)
139 dpavlin 28 o) First test would be to just implement a simple
140     instruction such as MIPS' addiu or lui, on AMD64
141     hosts...
142     x) Idle loop detection? (Depends on target.) Could be turned
143     into usleep(1) or similar on the host... except when doing
144     e.g. SMP emulation. Then it becomes trickier.
145 dpavlin 18
146 dpavlin 24 Alpha:
147     o) Virtual memory (tlbs etc)
148     o) Get {NetBSD,OpenBSD,Linux}/alpha booting. :)
149    
150     SPARC:
151     o) Add all registers (floating point, control regs etc)
152     o) Save/restore register windows etc!
153 dpavlin 28 o) Load/stores!
154     o) Disassemly of some more instructions?
155 dpavlin 24 o) Are sll etc 32-bit sign-extending or zero-extending?
156     o) Finish the cmp (subcc) flag computation code.
157     o) Finish the GDB register stuff.
158 dpavlin 28 o) SPARC v8, v7 etc?
159 dpavlin 24
160     Debugger:
161     o) How does SMP debugging work? Does it simply use "threads"?
162     What if the guest OS (running on an emulated SMP machine)
163     has a usertask running, with userland threads?
164     o) Try to make the debugger more modular and, if possible, reentrant!
165     o) Remove the emul command? (But show network info if showing
166     machines?)
167     o) Generalize the expression evaluator. (debugger_expr.c?)
168     settable variables ("show nr of instructions on average")
169     emul[x] defaults to current emul
170     machine[x] defaults to current machine
171     cpu[x] defaults to currently focused cpu
172     registers cpu arch dependent (#-prefix)
173     symbols @-prefix
174     numeric constants decimal, hex, and octal ($-prefix)
175     boolean yes,no, true,false
176     operators (+ - * / % & | ^ !)
177     parentheses for grouping subexpressions
178     NOTE: the change from % to # for register prefix!
179     examples:
180     emul[0].machine[2].cpu[0].pc
181     machine[test2].cpu[1].ra = main
182     settings.show_trace_tree = yes
183    
184     Settings:
185     o) Remove a setting.
186     o) Read/write a setting given a name. (Read as
187     string and/or int64_t simultaneously?)
188 dpavlin 28 o) Warnings when exiting the emulator, if the
189     settings have not been removed exactly in
190     the same way as they were added? This would
191     improve code cleanliness in the long term.
192     (I.e. require a corresponding _destroy()
193     function for all _new functions... machine_
194     cpu_ etc.)
195 dpavlin 24
196     Help command should have subsections! One for "expressions",
197     mirrored in the documentation, but the internal help should
198     be the one that should be considered correct.
199     o) see src/debugger.c for more
200    
201     POWER/PowerPC:
202     x) PPC optimizations; instr combs
203 dpavlin 28 x) 64-bit stuff: either Linux on G5, or perhaps some hobbyist
204     version of AIX? (if there exists such a thing)
205 dpavlin 24 x) find and fix the bug which causes NetBSD/macppc to fail after
206     an install!
207     x) macppc: adb controller; keyboard (for framebuffer mode)
208     x) make OpenBSD/macppc work (PCI controller stuff)
209    
210     Algor:
211     PCI and ISA and LOCAL interrupts! --> wdc could start working
212     Add interrupt controller in dev_algor.c.
213    
214     ARM:
215     o) try to get netbsd/evbarm 3.x running (iq80321)
216     o) make the xscale counter registers (ccnt) work
217     o) make the ata controller usable for FreeBSD!
218     o) zaurus for openbsd...
219     o) debian/cats crashes because of unimplemented coproc stuff.
220     fix this?
221    
222     Cache simulation:
223 dpavlin 28 o) Command line flags for:
224     o) CPU endianness?
225     o) Cache sizes? (multiple levels)
226 dpavlin 24 o) Separate from the CPU concept, so that multi-core CPUs sharing
227     e.g. a L2 cache can be simulated (?)
228     o) Instruction cache emulation is easiest (if separate from the
229     data cache); similar hack as the S;I; hack in cpu_dyntrans.c.
230     NOTE: if the architecture has a delay slot, then an instruction
231     slot can actually be executed as 2 instructions.
232     o) Data cache emulation = harder; each arch's load/store routines
233     must include support? running one instruction at a time and
234     having a cpu-dependant lookup function for each instruction
235     is another option (easier to implement, but very very slow).
236    
237     Documentation:
238     o) machines, cpus, devices.
239     o) Automagic documentation generation:
240     x) REMEMBER that several machines/devices can be in
241     the same source file!
242     o) Try to rewrite the install instructions for those machines
243     that use 3MAX into using CATS? (To remove the need to a raw
244     ffs partition using up all of the disk image.)
245    
246 dpavlin 22 More generic out_of_memory error reporting, and check everywhere!
247     Causes: OpenBSD has low default limits for normal users.
248     Host is 32-bit? (32-bit hosts are limited to 4 GB or less
249     of userspace memory.)
250     You are actually low on RAM. (As trivial as this might sound,
251     Unix systems usually allow processes to allocate virtual
252     memory beyond the amount of RAM in the machine.)
253    
254     Breakpoints: 32-bit vs 64-bit sign extension for MIPS, warnings, etc.
255     Use the debugger's symbolic name stuff. (which will have to be
256     extended soon to support stuff like "2*x + symbol + y" etc. cool
257     stuff)
258    
259     The Device subsystem:
260     x) allow devices to be moved and/or changed in size (down to a
261     minimum size, etc, or up to a max size)
262     x) keep track of interrupts and busses? actually, allowing any device
263 dpavlin 24 to be a bus might be a nice idea.
264     x) turn interrupt controllers into devices? :-)
265 dpavlin 22 x) refactor various clocks/nvram/cmos into one device?
266    
267 dpavlin 24 Clocks:
268     x) General framework for automagic clock adjustment for _all_
269     kinds of clocks and timers. (Which should be possible to turn
270     off, of course, like the way DECstation emulation works now.)
271 dpavlin 22
272     PCI:
273 dpavlin 24 x) last write was ffffffff ==> fix this, it should be used
274     together with a mask to get the correct bits. also, not ALL
275     bits are size bits! (lowest 4 vs lowest 2?)
276 dpavlin 22 x) add support for address fixups
277     x) generalize the interrupt routing stuff (lines etc). this should
278     be per machine? or per bus, that's better
279 dpavlin 24 x) add a "pcn" NIC (AMD PCnet32 Lance 79c970 (PCI 1022:2000)),
280     could be useful for several machine modes (Malta, Algor, evbarm,
281     hp700?, macppc, etc.)
282 dpavlin 22
283     Network layer:
284     o) DHCP (for Debian and BSD installers :-)
285     o) increase performance
286     o) don't rely on NetBSD-ish usage
287     o) Multiple networks per emulation, and let different
288     NICs in machines connect to different networks.
289     o) many other issues: see src/net.c
290    
291     Busses:
292     o) Redesign the entire "mainbus" concept!
293     o) Busses should be placed in a hierarchical tree!
294     o) Easily configurable interrupt routing in SMP systems.
295     o) Specific clock/bus speeds, cpu speeds etc.
296     o) Synchronization over network? or at least in dyntrans within
297     one emulated machine
298     o) dev->bus: TurboChannel, PCMCIA, ADB?
299    
300     Config file parser:
301 dpavlin 24 o) Rewrite it from scratch!
302 dpavlin 22 o) Usage of any expression available through the debugger
303     o) Support for running debugger commands (like the -c
304     command line option)
305    
306     Floating point layer:
307     o) make it common enough to be used by _all_ emulation modes
308 dpavlin 24 o) implement more stuff
309 dpavlin 22 o) non-IEEE modes (i.e. x86)?
310    
311 dpavlin 14 Userland emulation:
312 dpavlin 22 x) Lots of stuff; freebsd and netbsd (and linux?) syscalls.
313 dpavlin 14 x) Dynamic linking? Hm.
314 dpavlin 12
315 dpavlin 22 Sound:
316     x) generic sound framework
317     x) add one or more sound cards as devices
318 dpavlin 12
319 dpavlin 24 ASC SCSI controller:
320     x) NetBSD/arc 2.0 uses the ASC controller in a way which GXemul
321     cannot yet handle. (NetBSD 1.6.2 works ok.) (Possibly a problem
322     in NetBSD itself, http://mail-index.netbsd.org/source-changes/
323     2005/11/06/0024.html suggests that.)
324    
325 dpavlin 22 Caches / memory hierarchies: (this is mostly MIPS-specific)
326     o) src/memory*.c: Implement correct cache emulation for
327     all CPU types. (currently only R2000/R3000 is implemented)
328 dpavlin 28 (per CPU, multiple levels should be possible, associativity etc!)
329 dpavlin 22 o) R2000/R3000 isn't _100%_ correct, just almost correct :)
330     o) Move the -S (fill mem with random) functionality into the
331     memory.c subsystem, not machine.c or wherever it is now
332     o) ECC stuff, simulation of memory errors? (Machine dependent)
333     o) More than 4GB of emulated RAM, when run on a 32-bit host?
334     (using manual swap-out of blocks to disk, ugly)
335     o) A global command line option should be used to turn
336     cache emulation on or off. When off, caches should be
337     faked like they are right now. When on, caches and
338     memory latencies should be emulated as correctly as
339     possible.
340 dpavlin 2
341 dpavlin 22 File/disk/symbol handling:
342 dpavlin 28 o) Remove some of the complexity in file format guessing, for
343     Ultrix kernels that are actually disk images?
344 dpavlin 22 o) Better handling of tape files
345     o) Read function argument count and types from binaries? (ELF?)
346 dpavlin 24 o) Better demangling of C++ names. Note: GNU's C++ differs from e.g.
347     Microsoft's C++, so multiple schemes must be possible. See
348     URL at top of src/symbol_demangle.c for more info.
349 dpavlin 2
350 dpavlin 22 Userland ABI emulation:
351     o) see src/useremul.c
352 dpavlin 12
353 dpavlin 22 Terminal/console:
354     o) allow emulated serial ports to be connected to the outside
355     world in a more generic way, or even to other emulated
356     machines(?)
357 dpavlin 2
358 dpavlin 22 Save state of the whole emulated machine, to be able to load it back
359     in later? (Memory, all device's states, all registers and
360     so on. Like taking a snapshot. (SimOS seems to do this,
361     according to its website.))
362 dpavlin 6
363 dpavlin 22 Better framebuffer and X-windows functionality:
364     o) -Yx sometimes causes crashes.
365     o) Simple device access to framebuffer_blockcopyfill() etc,
366     and text output (using the built-in fonts), for dev_fb.
367     o) CLEAN UP the ugly event code
368     o) Mouse clicks can be "missed" in the current system; this is
369     not good. They should be put on a stack of some kind.
370     o) More 2D and 3D framebuffer acceleration.
371     o) Non-resizable windows? Or choose scaledown depending
372     on size (and center the image, with a black border).
373     o) Different scaledown on different windows?
374     o) Switch scaledown during runtime? (Ala CTRL-ALT-plus/minus)
375 dpavlin 24 o) Bug reported by Elijah Rutschman on MacOS with weird
376     keys (F5 = cursor down?).
377 dpavlin 22 o) Keyboard and mouse events:
378     x) Do this for more machines than just DECstation
379     x) more X11 cursor keycodes
380     x) Keys like CTRL, ALT, SHIFT do not get through
381     by themselves (these are necessary for example
382     to change the font of an xterm in X in the
383     emulator)
384     o) Generalize the framebuffer stuff by moving _ALL_ X11
385     specific code to src/x11.c!
386 dpavlin 2
387 dpavlin 22 Statistics: (this could be interesting)
388     o) Save to file and show graphics. It should be possible to
389     run gxemul after a simulation to just show the graphics,
390     or convert to a .ppm or .tga or similar.
391     o) memory accesses (to measure cache efficiency and
392     page coloring efficiency)
393     o) nr of simultaneous ASIDs in use in the TLB, for MIPS
394     o) percentage of time spent in different "states", such as
395     running userland code, kernel code, or idling (for CPUs
396     that have such an instruction, or whenever the PC is
397     inside a specific idle-function (address range)).
398     Possible additional state (for example on R3000): caches
399     disabled.
400     o) position of read/write on (SCSI) disks
401 dpavlin 2

  ViewVC Help
Powered by ViewVC 1.1.26