/[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 40 - (hide annotations)
Mon Oct 8 16:22:11 2007 UTC (16 years, 5 months ago) by dpavlin
File size: 23308 byte(s)
++ trunk/HISTORY	(local)
$Id: HISTORY,v 1.1539 2007/05/01 04:03:51 debug Exp $
20070415	Landisk PCLOCK should be 33.33 MHz, not 50 MHz. (This makes
		the clock run at correct speed.)
		FINALLY found and fixed the bug which caused OpenBSD/landisk
		to randomly bug out: an &-sign was missing in the special case
		handling of FPSCR in the 'LDS.L @Rm+,FPSCR' instruction.
		Adding similar special case handling for 'LDC.L @Rm+,SR'
		(calling sh_update_sr() instead of just loading).
		Implementing the 'FCNVSD FPUL,DRn' and 'FCNVDS DRm,FPUL'
		SuperH instructions.
		The 'LDC Rm,SR' instruction now immediately breaks out of the
		dyntrans loop if an interrupt is to be triggered.
20070416	In memory_rw.c, if mapping a page as writable, make sure to
		invalidate code translations even if the data access was a
		read.
		Minor SuperH updates.
20070418	Removing the dummy M68K emulation mode.
		Minor SH update (turning unnecessary sts_mach_rn, sts_macl_rn,
		and sts_pr_rn instruction handlers into mov_rm_rn).
20070419	Beginning to add a skeleton for an M88K mode: Adding a hack to
		allow OpenBSD/m88k a.out binaries to be loaded, and disassembly
		of a few simple 88K instructions.
		Commenting out the 'LDC Rm,SR' fix from a few days ago, because
		it made Linux/dreamcast bug out.
		Adding a hack to dev_sh4.c (an extra translation cache
		invalidation), which allows OpenBSD/landisk to boot ok after
		an install. Upgrading the Landisk machine mode to stable,
		updating documentation, etc.
20070420	Experimenting with adding a PCI controller (pcic) to dev_sh4.
		Adding a dummy Realtek 8139C+ skeleton device (dev_rtl8139c).
		Implementing the first M88K instructions (br, or[.u] imm), and
		adding disassembly of some more instructions.
20070421	Continuing a little on dev_rtl8139c.
20070422	Implementing the 9346 EEPROM "read" command for dev_rtl8139c.
		Finally found and fixed an old bug in the log n symbol search
		(it sometimes missed symbols). Debug trace (-i, -t etc) should
		now show more symbols. :-)
20070423	Continuing a little on M88K disassembly.
20070428	Fixing a memset arg order bug in src/net/net.c (thanks to
		Nigel Horne for noticing the bug).
		Applying parts of a patch from Carl van Schaik to clear out
		bottom bits of MIPS addresses more correctly, when using large
		page sizes, and doing some other minor cleanup/refactoring.
		Fixing a couple of warnings given by gcc with the -W option (a
		few more warnings than just plain -Wall).
		Reducing SuperH dyntrans physical address space from 64-bit to
		32-bit (since SH5/SH64 isn't imlemented yet anyway).
		Adding address-to-symbol annotation to a few more instructions
		in the SuperH instruction trace output.
		Beginning regression testing for the next release.
		Reverting the value of SCIF_DELAYED_TX_VALUE from 1 to 2,
		because OpenBSD/landisk may otherwise hang randomly.
20070429	The ugly hack/workaround to get OpenBSD/landisk booting without
		crashing does NOT work anymore (with the April 21 snapshot
		of OpenBSD/landisk). Strangely enough, removing the hack
		completely causes OpenBSD/landisk to work (!).
		More regression testing (re-testing everything SuperH-related,
		and some other things).
		Cobalt interrupts were actually broken; fixing by commenting
		out the DEC21143s in the Cobalt machine.
20070430	More regression testing.
20070501	Updating the OpenBSD/landisk install instructions to use
		4.1 instead of the current snapshot.
		GAAAH! OpenBSD/landisk 4.1 _needs_ the ugly hack/workaround;
		reintroducing it again. (The 4.1 kernel is actually from
		2007-03-11.)
		Simplifying the NetBSD/evbarm install instructions a bit.
		More regression testing.

==============  RELEASE 0.4.5.1  ==============


1 dpavlin 40 $Id: TODO,v 1.489 2007/05/01 04:05:06 debug Exp $
2 dpavlin 2
3 dpavlin 40 --------------------------------------------
4    
5     Testing for the next release (0.4.5.1):
6    
7     TEST DISK OVERLAY IMAGES
8     TEST LANDISK/SUPERH EMULATION MODES
9     REGRESSION TESTS FOR ALL OTHER SUPPORTED GUEST OSES
10    
11     # NetBSD/pmax 3.1 or 1.6.2 OK
12     # NetBSD/arc 1.6.2 OK
13     # NetBSD/hpcmips 3.1 OK
14     # NetBSD/cobalt 3.1 OK
15     # NetBSD/evbmips 3.1 OK
16     # NetBSD/algor 3.1 OK
17     # NetBSD/sgimips 3.1 OK
18     # NetBSD/cats 3.1 OK
19     # NetBSD/evbarm 2.1 OK
20     # NetBSD/netwinder 3.1 OK
21     # NetBSD/prep 2.1 OK
22     # NetBSD/macppc 3.1 OK
23     # NetBSD/dreamcast 3.1 MD OK
24     # NetBSD/dreamcast 3.1 LiveCD OK
25     # Linux/dreamcast Live CD OK
26     # OpenBSD/pmax 2.8-BETA not tested because of lack of time
27     # OpenBSD/cats 4.0 OK
28     # OpenBSD/landisk 4.1 OK
29     # Ultrix/RISC 4.5 OK
30     # Sprite for DECstation OK
31     # Debian GNU/Linux for pmax not tested because of lack of time
32    
33     Optional:
34     # OpenBSD/sgi FAILED to boot after setup (as expected)
35    
36     --------------------------------------------
37    
38 dpavlin 38 Some things, in totally random order, that I'd like to fix:
39 dpavlin 40 (Some items in this list are possibly out-of-date by now.)
40 dpavlin 20
41 dpavlin 32 Dyntrans:
42 dpavlin 28 x) Instruction combination collisions? How to avoid easily...
43 dpavlin 30 x) Think about how to do both SHmedia and SHcompact in a reasonable
44 dpavlin 32 way! (Or AMD64 long/protected/real, for that matter.)
45     x) 68K emulation; think about how to do variable instruction
46     lengths across page boundaries.
47 dpavlin 28 x) Dyntrans with valgrind-inspired memory checker. (In memory_rw,
48     it would be reasonably simple to add; in each individual fast
49     load/store routine = a lot more work, and it would become
50     kludgy very fast.)
51 dpavlin 24 x) Dyntrans with SMP... lots of work to be done here.
52     x) Dyntrans with cache emulation... lots of work here as well.
53 dpavlin 36 x) Remove the concept of base RAM completely; it would be more
54     generic to allow RAM devices to be used "anywhere".
55 dpavlin 32 o) dev_mp doesn't work well with dyntrans yet
56     o) In general, IPIs, CAS, LL/SC etc must be made to work with dyntrans
57     x) Redesign/rethink the delay slot mechanism used for e.g. MIPS,
58     so that it caches a translation (that is, an instruction
59     word and the instr_call it was translated to the last
60     time), so that it doesn't need to do slow
61     to_be_translated for each end of page?
62     x) Program Counter statistics:
63     Per machine? What about SMP? All data to the same file?
64     A debugger command should be possible to use to enable/
65     disable statistics gathering.
66     Configuration file option!
67     x) Breakpoints:
68     o) Physical vs virtual addresses!
69     o) 32-bit vs 64-bit sign extension for MIPS, and others?
70     x) INVALIDATION should cause translations in _all_ cpus to be
71     invalidated, e.g. on a write to a write-protected page
72     (containing code)
73     x) 16-bit encodings? (MIPS16, ARM Thumb, 32-bit SH on SH64)
74     x) Lots of other stuff: see src/cpus/README_DYNTRANS
75 dpavlin 38 x) Native code generation backends:
76     o) think carefully about this.
77     o) simple syntax for emitting opcodes; backend implementation
78     must be optional, so I don't have to write more code
79     than necessary. after all, the non-native (C) code should
80     always work.
81 dpavlin 34 o) convert into native code only after an entire
82     block has been translated? probably best.
83 dpavlin 38 o) the "almost native" opcodes may be rearranged,
84     "peep-hole optimized", etc. and then as a separate step
85     this list of almost native opcodes is written out
86     as native code.
87     o) think about delay slots at the end of a block!
88 dpavlin 34 o) x86/amd64 code generator can be very similar... perhaps
89 dpavlin 38 o) NOTE that generation is per _ABI_, not per host arch!
90     the configure script must detect ABI!!!
91 dpavlin 34 o) branches to already translated code blocks can
92     link the blocks together
93 dpavlin 38 o) load/store are the most important to optimize
94 dpavlin 24
95 dpavlin 28 Simple Valgrind-like checks?
96     o) Mark every address with bits which tell whether or not the address
97     has been written to.
98     o) What should happen when programs are loaded? Text/data, bss (zero
99     filled). But stack space and heap is uninitialized.
100     o) Uninitialized local variables:
101     A load from a place on the stack which has not previously
102     been stored to => warning. Increasing the stack pointer using
103     any available means should reset the memory to uninitialized.
104     o) If calls to malloc() and free() can be intercepted:
105     o) Access to a memory area after free() => warning.
106     o) Memory returned by malloc() is marked as not-initialized.
107     o) Non-passive, but good to have: Change the argument
108     given to malloc, to return a slightly larger memory
109     area, i.e. margin_before + size + margin_after,
110     and return the pointer + margin_before.
111     Any access to the margin_before or _after space results
112     in warnings. (free() must be modified to free the
113     actually allocated address.)
114    
115 dpavlin 24 MIPS:
116 dpavlin 32 o) Nicer MIPS status bits in register dumps.
117     o) Alignment exceptions.
118     o) Floating point exception correctness.
119     o) Fix this? Triggered by NetBSD/sgimips? Hm:
120     to_be_translated(): TODO: unimplemented instruction:
121     000000000065102c: 00200800 (d) rot_00 at,zr,0
122     o) Some more work on opcodes.
123 dpavlin 24 x) MIPS64 revision 2.
124 dpavlin 28 o) Find out which actual CPUs implement the rev2 ISA!
125 dpavlin 32 o) DROTR32 and similar MIPS64 rev 2 instructions,
126     which have a rotation bit which differs from
127     previous ISAs.
128     o) EI and DI instructions for MIPS64/32 rev 2.
129     NOTE: These are _NOT_ the same as for R5900!
130 dpavlin 24 x) _MAYBE_ TX79 and R5900 actually differ in their
131     opcodes? Check this carefully!
132     o) Dyntrans: Count register updates are probably not 100% correct yet.
133     o) Refactor code for performance and readability/maintainability.
134 dpavlin 28 o) (Re)implement 128-bit loads/stores for R5900.
135 dpavlin 24 o) R4000 and others:
136     x) watchhi/watchlo exceptions, and other exception
137     handling details
138     o) R10000 and others: (R12000, R14000 ?)
139 dpavlin 34 x) The code before the line
140     /* reg[COP0_PAGEMASK] = cpu->cd.mips.coproc[0]->tlbs[0].mask & PAGEMASK_MASK; */
141     in cpu_mips.c is not correct for R10000 according to
142     Lemote's Godson patches for GXemul. TODO: Go through all
143     register definitions according to http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi/hdwr/bks/SGI_Developer/books/R10K_UM/sgi_html/t5.Ver.2.0.book_263.html#HEADING334
144     and make sure everything works with R10000.
145     Then test with OpenBSD/sgi?
146 dpavlin 36 x) Entry LO mask (as above).
147 dpavlin 24 x) memory space, exceptions, ...
148     x) use cop0 framemask for tlb lookups
149     (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)
150    
151 dpavlin 32 SuperH:
152 dpavlin 40 x) SH4 performance is VERY low when running user-space instructions,
153     because I actually simulate the 4-entry ITLB as being separate
154     from the 64-entry DTLB. This is correct, but VERY slow. I need
155     to experiment with _not_ simulating it in too much detail.
156 dpavlin 32 x) SH4 interrupt controller:
157 dpavlin 38 x) MASKING should be possible!
158 dpavlin 34 x) SH4 DMA (0xffa00000)
159     x) SH4 UBC (0xff200000)
160     x) Store queues can copy 32 bytes at a time, there's no need to
161     copy individual 32-bit words. (Performance improvement.)
162 dpavlin 32 x) SH4 BSC (Bus State Controller)
163 dpavlin 34 x) Instruction tracing should include symbols for branch targets,
164     and so on, to make the output more human readable.
165 dpavlin 38 x) SH3-specific devices: Pretty much everything!
166     x) NetBSD/evbsh3, mmeye, hpcsh! Linux?
167 dpavlin 32 x) Replace pc-relative loads with immediate load, if within the
168     same page. (Similar to the same optimization for ARM.)
169 dpavlin 34 x) Floating point speed!
170 dpavlin 32 x) Floating point exception correctness.
171 dpavlin 40 x) NetBSD HEAD (as of April 2007) hangs during bootup, because it
172     turns on/off interrupts in an unfortunately synchronized way
173     with dyntrans. This needs to be fixed.
174     x) Exceptions for unaligned load/stores. OpenBSD/landisk uses
175     this mechanism for its reboot code (machine_reset).
176 dpavlin 32 x) Think carefully about how to implement SH5/SH64 (for evbsh5).
177 dpavlin 18
178 dpavlin 38 Landisk SH4:
179 dpavlin 40 x) When NetBSD/landisk 4.0 has been released, make sure it works
180     in the emulator. (Update documentation, etc.)
181 dpavlin 38
182 dpavlin 32 Dreamcast:
183 dpavlin 34 x) G2 DMA
184     x) LAN adapter (dev_mb8696x.c). NetBSD root-on-nfs.
185 dpavlin 32 x) PVR: Lots of stuff. See dev_pvr.c.
186 dpavlin 38 x) Better GDROM support
187 dpavlin 34 x) Modem
188     x) PCI bridge/bus?
189 dpavlin 32 x) Maple bus:
190     x) Correct controller input
191     x) Mouse input
192 dpavlin 34 x) Software emulation of BIOS calls:
193     x) GD-ROM emulation: Use the GDROM device.
194     x) Use the VGA font as a fake ROM font. (Better than
195     nothing.)
196     x) Make as many as possible of the KOS examples run!
197 dpavlin 32 x) More homebrew demos/games.
198 dpavlin 34 x) SPU: Sound emulation (ARM cpu).
199     x) VME processor emulation? "(Sanyo LC8670 "Potato")" according to
200     Wikipedia, LC86K87 according to Comstedt's page. See
201     http://www.maushammer.com/vmu.html for a good description of
202     the differences between LC86104C and the one used in the VME.
203 dpavlin 32
204 dpavlin 24 Alpha:
205 dpavlin 32 x) OSF1 PALcode, Virtual memory support.
206     x) PALcode replacement! PAL1E etc opcodes...?
207     x) Interrupt/exception/trap handling.
208     x) Floating point exception correctness.
209     x) More work on bootup memory and register contents.
210     x) More Alpha machine types, so it could work with
211     OpenBSD, FreeBSD, and Linux too?
212 dpavlin 24
213 dpavlin 38 SPARC (both the ISA and the machines):
214 dpavlin 32 o) Implement Adress space identifiers; load/stores etc.
215 dpavlin 38 o) Exception/trap/interrupt handling.
216 dpavlin 34 o) Save/restore register windows etc! Both v9 and pre-v9!
217 dpavlin 30 o) Finish the subcc and addcc flag computation code.
218     o) Add more registers (floating point, control regs etc)
219 dpavlin 28 o) Disassemly of some more instructions?
220 dpavlin 24 o) Are sll etc 32-bit sign-extending or zero-extending?
221 dpavlin 38 o) Floating point exception correctness.
222 dpavlin 28 o) SPARC v8, v7 etc?
223 dpavlin 38 o) More machine modes and devices.
224 dpavlin 24
225     Debugger:
226     o) How does SMP debugging work? Does it simply use "threads"?
227     What if the guest OS (running on an emulated SMP machine)
228     has a usertask running, with userland threads?
229     o) Try to make the debugger more modular and, if possible, reentrant!
230     o) Remove the emul command? (But show network info if showing
231     machines?)
232 dpavlin 34 o) Evaluate expressions within []? That would allow stuff like
233     cpu[x] where x is an expression.
234 dpavlin 32 o) Settings:
235     x) Special handlers for Write!
236     +) MIPS coproc regs
237     +) Alpha/MIPS/SPARC zero registers
238     +) x86 64/32/16-bit registers
239     x) Value formatter for resulting output.
240 dpavlin 24 o) see src/debugger.c for more
241    
242     POWER/PowerPC:
243 dpavlin 34 x) Fix DECR timer speed, so it matches the host.
244 dpavlin 32 x) NetBSD/prep 3.x triggers a possible bug in the emulator:
245     <wdc_exec_command(0xd005e514,0xd60cdd30,0,8,..)>
246     <ata_get_xfer(0,0xd60cdd30,0,8,..)>
247     <0x26c550(&ata_xfer_pool,2,0,8,..)>
248     <0x35c71c(0x3f27000,0,52,8,..)>
249     <ata_exec_xfer(0xd005e4c8,0x3f27000,0,13,..)>
250     <atastart(0xd005e4c8,0x3f27000,0,13,..)>
251     <__wdccommand_start(0xd005e4c8,0x3f27000,0,13,..)>
252     <bsw1(&prep_isa_io_space_tag,0x800001f6,0,176,..)>
253     [ wdc: write to SDH: 0xb0 (sectorsize 2, lba=1, drive 1, head 0) ]
254     <wdcwait(0xd005e4c8,72,64,0xbb8,..)>
255     <0x198120(0xd005e4c8,72,64,0xbb8,..)>
256     <bsr1(&prep_isa_io_space_tag,0,0,0xbb8,..)>
257     <delay(100,0,0,0xbb8,..)>
258     Note: <bsr1(&prep_isa_io_space_tag,0,0,0xbb8,..)>
259 dpavlin 24 x) PPC optimizations; instr combs
260 dpavlin 28 x) 64-bit stuff: either Linux on G5, or perhaps some hobbyist
261     version of AIX? (if there exists such a thing)
262 dpavlin 24 x) macppc: adb controller; keyboard (for framebuffer mode)
263     x) make OpenBSD/macppc work (PCI controller stuff)
264 dpavlin 32 x) Floating point exception correctness.
265     x) Alignment exceptions.
266 dpavlin 24
267 dpavlin 34 PReP:
268 dpavlin 38 x) Clock time! ("Bad battery blah blah")
269 dpavlin 34
270 dpavlin 24 Algor:
271 dpavlin 32 o) Other models than the P5064?
272     o) PCI interrupts... needed for stuff like the tlp NIC?
273 dpavlin 24
274 dpavlin 34 BeBox:
275     o) Interrupts. There seems to be a problem with WDC interrupts
276     "after a short while", although a few interrupts get through?
277     o) Perhaps find a copy of BeOS and try it?
278    
279 dpavlin 32 HPCmips:
280     x) Mouse/pad support! :)
281     x) A NIC? (As a PCMCIA device?)
282    
283 dpavlin 40 M88K:
284     o) Everything. :)
285     o) More instruction disassembly!
286     o) Implement more instructions.
287     o) has-delay-slot (for debugging)
288     o) Find manuals!
289     o) MMU stuff
290     o) Exceptions
291     o) FPU
292     o) Control registers
293    
294 dpavlin 32 AVR:
295     o) Everything.
296    
297 dpavlin 24 ARM:
298 dpavlin 32 o) See netwinder_reset() in NetBSD; the current "an internal error
299     occured" message after reboot/halt is too ugly.
300     o) ARM "wait"-like instruction?
301 dpavlin 34 o) try to get netbsd/evbarm 3.x or 4.x running (iq80321)
302 dpavlin 24 o) make the xscale counter registers (ccnt) work
303     o) make the ata controller usable for FreeBSD!
304 dpavlin 36 o) Zaurus emulation:
305     x) OpenBSD/zaurus
306     x) NetBSD/zaurus? See the following URL:
307     http://mail-index.netbsd.org/port-arm/2006/11/19/0000.html
308 dpavlin 34 o) Debian/cats crashes because of unimplemented coproc stuff.
309 dpavlin 24 fix this?
310    
311 dpavlin 32 Test machines:
312 dpavlin 34 o) dev_fb block fill and copy
313     o) dev_fb draw characters (from the built-in font)?
314     o) dev_fb input device? mouse pointer coordinates and buttons
315 dpavlin 32 (allow changes in these to cause interrupts as well?)
316 dpavlin 34 o) Redefine the halt() function so that it stops "sometimes
317     soon", i.e. usage in demo code should be:
318 dpavlin 32 for (;;) {
319     halt();
320     }
321    
322     Better CD Image file support:
323     x) Support CD formats that contain more than 1 track, e.g.
324     CDI files (?). These can then contain a mixture of e.g. sound
325     and data tracks, and booting from an ISO filesystem path
326     would boot from [by default] the first data track.
327     (This would make sense for e.g. Dreamcast CD images, or
328     possibly other live-CD formats.)
329    
330     Networking:
331 dpavlin 38 x) Redesign of the networking subsystem, at least the NAT translation
332     part. The current way of allowing raw ethernet frames to be
333     transfered to/from the emulator via UDP should probably be
334     extended to allow the frames to be transmitted other ways as
335     well.
336     x) Also adding support for connecting ttys (either to xterms, or to
337     pipes/sockets etc, or even to PPP->NAT or SLIP->NAT :-).
338     x) Documentation updates (!) are very important, making it easier to
339     use the (already existing) network emulation features.
340 dpavlin 32 x) Fix performance problems caused by only allowing a
341     single TCP packet to be unacked.
342     x) Don't hardcode offsets into packets!
343     x) Test with lower than 100 max tcp/udp connections,
344     to make sure that reuse works!
345     x) Make OpenBSD work better as a guest OS!
346     x) DHCP? Debian doesn't actually send DHCP packets, even
347     though it claims to? So it is hard to test.
348     x) Multiple networks per emulation, and let different
349     NICs in machines connect to different networks.
350     x) Support VDE (vde.sf.net)? Easiest/cleanest (before a
351     redesign of the network framework has been done) is
352     probably to connect it using the current (udp) solution.
353     x) Allow SLIP connections, possibly PPP, in addition to
354     ethernet?
355    
356 dpavlin 24 Cache simulation:
357 dpavlin 28 o) Command line flags for:
358     o) CPU endianness?
359     o) Cache sizes? (multiple levels)
360 dpavlin 24 o) Separate from the CPU concept, so that multi-core CPUs sharing
361     e.g. a L2 cache can be simulated (?)
362     o) Instruction cache emulation is easiest (if separate from the
363     data cache); similar hack as the S;I; hack in cpu_dyntrans.c.
364     NOTE: if the architecture has a delay slot, then an instruction
365     slot can actually be executed as 2 instructions.
366     o) Data cache emulation = harder; each arch's load/store routines
367     must include support? running one instruction at a time and
368     having a cpu-dependant lookup function for each instruction
369     is another option (easier to implement, but very very slow).
370    
371     Documentation:
372 dpavlin 32 x) Note about sandboxing/security:
373     Not all emulated instructions fail in the way they would
374     do on real hardware (e.g. a userspace program writing to
375     a system register might work in GXemul, but it would
376     fail on real hardware). Sandbox = contain from the
377     host OS. But the emulated programs will run "less
378     securely".
379     x) Try NetBSD/arc 4.x! (It seems to work with disk images!)
380     x) NetBSD/pmax 4 install instructions: xterm instead of vt100!
381 dpavlin 34 x) BETTER DEVICE EXAMPLES!
382     o) Move away from technical.html to somewhere new.
383     o) DEVICE_TICK
384     o) Implement example devices using interrupts, dyntrans
385     memory access, etc.?
386     x) Document the dyntrans core?
387 dpavlin 32 x) Rewrite the section about experimental devices, after the
388     framebuffer acceleration has been implemented, and demos
389     written. (Symbolic names instead of numbers; example
390     use cases, etc. Mention demo files that use the various
391     features?)
392     x) "a very simple linear framebuffer device (for graphics output)"
393     under "which machines does gxemul emulate" ==> better
394     description?
395     x) Better description on how to set up a cross compiler?
396     Example for MIPS64.
397     o) Automagic documentation generation?
398     x) machines, cpus, devices.
399 dpavlin 24 x) REMEMBER that several machines/devices can be in
400     the same source file!
401     o) Try to rewrite the install instructions for those machines
402 dpavlin 32 that use 3MAX into using CATS or hpcmips? (To remove the need
403     to use a raw ffs partition, using up all of the disk image.)
404 dpavlin 24
405 dpavlin 22 More generic out_of_memory error reporting, and check everywhere!
406     Causes: OpenBSD has low default limits for normal users.
407     Host is 32-bit? (32-bit hosts are limited to 4 GB or less
408     of userspace memory.)
409     You are actually low on RAM. (As trivial as this might sound,
410     Unix systems usually allow processes to allocate virtual
411     memory beyond the amount of RAM in the machine.)
412    
413     The Device subsystem:
414     x) allow devices to be moved and/or changed in size (down to a
415 dpavlin 32 minimum size, etc, or up to a max size); if there is a collision,
416     return false. It is up to the caller to handle this situation!
417     x) NOTE: Translations must be invalidated, both for
418     registering new devices, and for moving existing ones.
419     cpu->invalidate translation caches, for all CPUs that
420     are connected to a specific memory.
421 dpavlin 22 x) keep track of interrupts and busses? actually, allowing any device
422 dpavlin 24 to be a bus might be a nice idea.
423     x) turn interrupt controllers into devices? :-)
424 dpavlin 22 x) refactor various clocks/nvram/cmos into one device?
425    
426     PCI:
427 dpavlin 38 x) Pretty much everything related to runtime configuration, device
428     slots, interrupts, etc must be redesigned/cleaned up. The current
429     code is very hardcoded and ugly.
430     o) Allow cards to be added/removed during runtime more easily.
431     o) Allow cards to be enabled/disabled (i/o ports, etc, like
432     NetBSD needs for disk controller detection).
433     o) Allow devices to be moved in memory during runtime.
434     o) Interrupts per PCI slot, etc. (A-D).
435     o) PCI interrupt controller logic... very hard to get right,
436     because these differ a lot from one machine to the next.
437 dpavlin 24 x) last write was ffffffff ==> fix this, it should be used
438     together with a mask to get the correct bits. also, not ALL
439     bits are size bits! (lowest 4 vs lowest 2?)
440 dpavlin 22 x) add support for address fixups
441 dpavlin 32 x) generalize the interrupt routing stuff (lines etc)
442 dpavlin 22
443 dpavlin 32 Clocks and timers:
444 dpavlin 38 x) Fix the PowerPC DECR interrupt speed! (MacPPC and PReP speed, etc.)
445 dpavlin 32 x) DON'T HARDCODE 100 HZ IN cpu_mips_coproc.c!
446     x) Test the 8253? Right now it doesn't seem to be used?
447 dpavlin 40 x) NetWinder timeofday is incorrect! It seems to be exactly
448     1 day ahead of actual time?
449 dpavlin 32 x) Cobalt TOD is incorrect!
450     x) Go through all other machines, one by one, and fix them.
451 dpavlin 22
452     Config file parser:
453 dpavlin 24 o) Rewrite it from scratch!
454 dpavlin 22 o) Usage of any expression available through the debugger
455 dpavlin 34 o) Allow interrupt controllers to be added! and interrupts
456     to be used in more ways than before
457 dpavlin 22 o) Support for running debugger commands (like the -c
458     command line option)
459    
460     Floating point layer:
461     o) make it common enough to be used by _all_ emulation modes
462 dpavlin 32 o) implement correct error/exception handling and rounding modes
463     o) implement more helper functions (i.e. add, sub, mul...)
464 dpavlin 22 o) non-IEEE modes (i.e. x86)?
465    
466 dpavlin 14 Userland emulation:
467 dpavlin 38 x) Dynamic linking!
468     x) Lots of stuff; freebsd, netbsd, linux, ... syscalls.
469     x) Initial register/stack contents (environment, command line args).
470     x) Return value (from main).
471     x) mmap emulation layer
472     x) errno emulation layer
473     x) struct conversions for may syscalls
474 dpavlin 12
475 dpavlin 22 Sound:
476     x) generic sound framework
477 dpavlin 32 x) add one or more sound cards as devices; add a testmachine
478     sound card first?
479 dpavlin 34 x) Dreamcast sound? Generic PCI sound cards?
480 dpavlin 12
481 dpavlin 24 ASC SCSI controller:
482     x) NetBSD/arc 2.0 uses the ASC controller in a way which GXemul
483     cannot yet handle. (NetBSD 1.6.2 works ok.) (Possibly a problem
484     in NetBSD itself, http://mail-index.netbsd.org/source-changes/
485     2005/11/06/0024.html suggests that.)
486 dpavlin 32 NetBSD 4.x seems to work? :)
487 dpavlin 24
488 dpavlin 22 Caches / memory hierarchies: (this is mostly MIPS-specific)
489     o) src/memory*.c: Implement correct cache emulation for
490     all CPU types. (currently only R2000/R3000 is implemented)
491 dpavlin 28 (per CPU, multiple levels should be possible, associativity etc!)
492 dpavlin 22 o) R2000/R3000 isn't _100%_ correct, just almost correct :)
493     o) Move the -S (fill mem with random) functionality into the
494     memory.c subsystem, not machine.c or wherever it is now
495     o) ECC stuff, simulation of memory errors? (Machine dependent)
496     o) More than 4GB of emulated RAM, when run on a 32-bit host?
497     (using manual swap-out of blocks to disk, ugly)
498     o) A global command line option should be used to turn
499     cache emulation on or off. When off, caches should be
500     faked like they are right now. When on, caches and
501     memory latencies should be emulated as correctly as
502     possible.
503 dpavlin 2
504 dpavlin 22 File/disk/symbol handling:
505 dpavlin 34 o) Make sure that disks can be added/removed during runtime!
506     (Perhaps this needs a reasonably large re-write.)
507 dpavlin 28 o) Remove some of the complexity in file format guessing, for
508     Ultrix kernels that are actually disk images?
509 dpavlin 34 o) Better handling of tape files
510 dpavlin 22 o) Read function argument count and types from binaries? (ELF?)
511 dpavlin 24 o) Better demangling of C++ names. Note: GNU's C++ differs from e.g.
512     Microsoft's C++, so multiple schemes must be possible. See
513     URL at top of src/symbol_demangle.c for more info.
514 dpavlin 2
515 dpavlin 22 Userland ABI emulation:
516     o) see src/useremul.c
517 dpavlin 12
518 dpavlin 22 Better framebuffer and X-windows functionality:
519 dpavlin 32 o) Generalize the update_x1y1x2y2 stuff to an extend-region()
520     function...
521 dpavlin 22 o) -Yx sometimes causes crashes.
522     o) Simple device access to framebuffer_blockcopyfill() etc,
523     and text output (using the built-in fonts), for dev_fb.
524     o) CLEAN UP the ugly event code
525     o) Mouse clicks can be "missed" in the current system; this is
526     not good. They should be put on a stack of some kind.
527     o) More 2D and 3D framebuffer acceleration.
528     o) Non-resizable windows? Or choose scaledown depending
529     on size (and center the image, with a black border).
530     o) Different scaledown on different windows?
531 dpavlin 32 o) Non-integral scale-up? (E.g. 640x480 -> 1024x768)
532 dpavlin 22 o) Switch scaledown during runtime? (Ala CTRL-ALT-plus/minus)
533 dpavlin 24 o) Bug reported by Elijah Rutschman on MacOS with weird
534     keys (F5 = cursor down?).
535 dpavlin 22 o) Keyboard and mouse events:
536     x) Do this for more machines than just DECstation
537     x) more X11 cursor keycodes
538     x) Keys like CTRL, ALT, SHIFT do not get through
539     by themselves (these are necessary for example
540     to change the font of an xterm in X in the
541     emulator)
542     o) Generalize the framebuffer stuff by moving _ALL_ X11
543     specific code to src/x11.c!
544 dpavlin 2

  ViewVC Help
Powered by ViewVC 1.1.26