--- trunk/TODO 2007/10/08 16:20:18 27 +++ trunk/TODO 2007/10/08 16:20:26 28 @@ -1,49 +1,111 @@ -$Id: TODO,v 1.298 2006/06/25 11:08:04 debug Exp $ +$Id: TODO,v 1.324 2006/07/22 10:23:39 debug Exp $ Hm. This file is in random order, and not all parts of it are up-to-date. --------------- +Code cleanup: + x) 64-bit ranges in src/cpus/memory_mips_v2p.c + x) Revert the dyntrans page template experiment? Hm. + x) Refactor the cpu type detection/initialization/listing. - x) FIX THE NON-R3000 TRANSLATION CACHE INVALIDATION BOTTLENECKS! - x) Find a way to get rid of the cpu_create_or_reset_tc in the - R2000/R3000 cache isolation code. (NetBSD works without it, - but not Ultrix and Linux yet.) - x) Formalize the statistics gathering stuff for dyntrans... - x) ... and use it to optimize MIPS dyntrans stuff. - x) Clock framework? Go through all clock devices, make sure they +Documentation: + x) Rewrite the section about experimental devices, after the + framebuffer acceleration has been implemented, and demos + written. (Symbolic names instead of numbers; example + use cases, etc. Mention demo files that use the various + features?) + x) "a very simple linear framebuffer device (for graphics output)" + under "which machines does gxemul emulate" ==> better + description? + x) Better description on how to set up a cross compiler? + Example for MIPS64. + +Long-term implementation: + x) Testmachine includes: + + dev_fb block fill and copy + + dev_fb draw characters (from the built-in font)? + + dev_fb input device? mouse pointer coordinates and buttons + (allow changes in these to cause interrupts as well?) + + Redefine the halt() function so that it stops "sometimes + soon", i.e. usage in demo code should be: + for (;;) { + halt(); + } + x) Rewrite the networking stack; make OpenBSD work better as a guest + OS, fix the performance problems, make Linux work with DHCP, etc. + x) Make the wdc controller work with modern versions of NetBSD! + x) Continue on SPARC emulation + + Enable it in the configure script as soon as it can + run all the demo programs. + x) Continue on Alpha emulation (virtual memory, etc). Cleanup. + x) Alignment exceptions (MIPS, PPC, ARM?, ...) + +Long-term design: + x) Instruction combination collisions? How to avoid easily... + o) Actually use the settings object, better debugger stuff, etc! + o) Debugger command for enabling/disabling instruction statistics + during runtime. machine.statistics = on|off + x) MAINBUS REDESIGN! + 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 valgrind-inspired memory checker. (In memory_rw, + it would be reasonably simple to add; in each individual fast + load/store routine = a lot more work, and it would become + kludgy very fast.) 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. + x) Reimplement the config file parser from scratch. --------------- +Test: + x) Test with more than one Sprite instance on an emulated network! + x) NetBSD 4.x, once it is out. + +------------------------------------------------------------------------------- + +Simple Valgrind-like checks? + o) Mark every address with bits which tell whether or not the address + has been written to. + o) What should happen when programs are loaded? Text/data, bss (zero + filled). But stack space and heap is uninitialized. + o) Uninitialized local variables: + A load from a place on the stack which has not previously + been stored to => warning. Increasing the stack pointer using + any available means should reset the memory to uninitialized. + o) If calls to malloc() and free() can be intercepted: + o) Access to a memory area after free() => warning. + o) Memory returned by malloc() is marked as not-initialized. + o) Non-passive, but good to have: Change the argument + given to malloc, to return a slightly larger memory + area, i.e. margin_before + size + margin_after, + and return the pointer + margin_before. + Any access to the margin_before or _after space results + in warnings. (free() must be modified to free the + actually allocated address.) 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. + o) Find out which actual CPUs implement the rev2 ISA! 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) R2000/R3000 cache cleaner! o) strlen, memset loops etc + o) multiple loads/stores in a row, e.g. relative to + the stack pointer + o) lui + or, lui + add, and 64-bit variants + o) jr ra + addiu to the v0 register? 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) (Re)implement 128-bit loads/stores for R5900. o) R4000 and others: x) watchhi/watchlo exceptions, and other exception handling details @@ -53,26 +115,33 @@ (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) Redesign/rethink the delay slot mechanism used for e.g. MIPS, + so that it caches a translation (that is, an instruction + word and the instr_call it was translated to the last + time), so that it doesn't need to do slow + to_be_translated for each end of page? + x) Program Counter statistics: + Per machine? What about SMP? All data to the same file? + A debugger command should be possible to use to enable/ + disable statistics gathering. + Configuration file option! 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! + o) First test would be to just implement a simple + instruction such as MIPS' addiu or lui, on AMD64 + hosts... + x) Idle loop detection? (Depends on target.) Could be turned + into usleep(1) or similar on the host... except when doing + e.g. SMP emulation. Then it becomes trickier. Alpha: o) Virtual memory (tlbs etc) @@ -81,10 +150,12 @@ SPARC: o) Add all registers (floating point, control regs etc) o) Save/restore register windows etc! - o) Disassemly of some more instructions. + o) Load/stores! + 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. + o) SPARC v8, v7 etc? Debugger: o) How does SMP debugging work? Does it simply use "threads"? @@ -114,6 +185,13 @@ o) Remove a setting. o) Read/write a setting given a name. (Read as string and/or int64_t simultaneously?) + o) Warnings when exiting the emulator, if the + settings have not been removed exactly in + the same way as they were added? This would + improve code cleanliness in the long term. + (I.e. require a corresponding _destroy() + function for all _new functions... machine_ + cpu_ etc.) Help command should have subsections! One for "expressions", mirrored in the documentation, but the internal help should @@ -122,7 +200,8 @@ POWER/PowerPC: x) PPC optimizations; instr combs - x) 64-bit stuff + x) 64-bit stuff: either Linux on G5, or perhaps some hobbyist + version of AIX? (if there exists such a thing) x) find and fix the bug which causes NetBSD/macppc to fail after an install! x) macppc: adb controller; keyboard (for framebuffer mode) @@ -141,6 +220,9 @@ fix this? Cache simulation: + o) Command line flags for: + o) CPU endianness? + o) Cache sizes? (multiple levels) 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 @@ -174,9 +256,6 @@ 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) @@ -244,13 +323,9 @@ 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!) + (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 @@ -264,6 +339,8 @@ possible. File/disk/symbol handling: + o) Remove some of the complexity in file format guessing, for + Ultrix kernels that are actually disk images? 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.