ruby-x/rubyx
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
2,394 Commits 2 Branches 0 Tags 5.6 MiB
15337e10be
Commit Graph

1420 Commits

Author SHA1 Message Date
Torsten Ruger
bb1d1495db fix constant propagation through the layers 
so they can end up in the binary
 2018-07-03 10:12:22 +03:00
Torsten Ruger
63dfee0978 move some tests around (and fix them) 2018-07-02 23:20:54 +03:00
Torsten Ruger
55b5884c4e minor test fixes 2018-07-02 23:03:33 +03:00
Torsten Ruger
56f2742e89 fix code listener logic bug 
next's position was only updated initially
previous logic relied on pushing everything long rightly
but it wasn't right, and no harm in setting position twice,
as events only fire if there is real change
 2018-07-02 23:03:00 +03:00
Torsten Ruger
a1197fb70c interpreter tests working again 2018-07-02 17:29:26 +03:00
Torsten Ruger
4b33d1c056 fix many a test especially whole mom 
but some positioning bug still in there preventing
binary to work
 2018-07-02 17:01:07 +03:00
Torsten Ruger
474cd4b348 linker and collector test work again 
had to give space a fixed number of addresses
 2018-07-02 16:19:01 +03:00
Torsten Ruger
4e49c0469c fix compiling support 
which makes all the vool tests pass again
 2018-07-02 15:52:51 +03:00
Torsten Ruger
1132309f6a unify space collection attribute naming 
currently space is still acting as a sort of memory manager.
For proper linking, all objects must be reachable from space, hence the plural versions like messages and addresses (even they are instances, it is the list that is important)
To dish out instance to use, the head must be kept, ie next_XXX for intergers, return addresses and messages
 2018-07-02 15:49:51 +03:00
Torsten Ruger
07a154be70 code listener now get platform explicitly 
used to grab it from global machine.
Now passed in from linker
tests fixed accordingly
 2018-07-02 09:39:07 +03:00
Torsten Ruger
8952b39446 linker goes through assemblers, not space 
used to traverse all methods in space
now "just" use the passed assemblers, which usually should be compiled and builtin (but for testing . . .)
tests fixed accordingly
 2018-07-02 09:37:58 +03:00
Torsten Ruger
f4b1290e58 position tests working 2018-07-02 09:35:57 +03:00
Torsten Ruger
22408b86c6 return the linker from mom_compiler 
linker holds assemblers
assemblers come from method compilation
and so the layers come into focus
 2018-07-01 21:51:48 +03:00
Torsten Ruger
5f2a256608 fix platform derivation and some tests 2018-07-01 21:27:27 +03:00
Torsten Ruger
bb49f1be78 rename ClassCompiler to MomComplier 
ended up not even using the class, it just came from there
It actually compiles methods, and it turns out is the point where builtin comes into the picture as it's boot process also returns method compilers
 2018-07-01 21:26:45 +03:00
Torsten Ruger
c947c27a14 clean up booting 
many machine boot became obsolete
or just neede parfait to boot
actual linker functionality pending
 2018-07-01 14:12:42 +03:00
Torsten Ruger
e10092a4be rename machine to linker 
even work is not quite finished, but helps root out offenders
(needless to say this is wip, deep red wip)
 2018-07-01 13:45:14 +03:00
Torsten Ruger
57fc67ddf6 move address list to object_space 
where i guess it should have been from the start
 2018-07-01 13:43:53 +03:00
Torsten Ruger
35b10c46a3 adds the risc instructions to the compiler 
since they are not in the method anymore
 2018-07-01 11:59:52 +03:00
Torsten Ruger
22b1fea587 use rubyc_compiler as instance 
not as before class methods
so we can carry the state around
 2018-07-01 11:59:07 +03:00
Torsten Ruger
1505df894a remove translation 
from machine, which is quickly becoming a linker
 2018-07-01 11:58:06 +03:00
Torsten Ruger
05669065ca back to method_compiler 
it is what it is
 2018-06-30 23:26:28 +03:00
Torsten Ruger
91a99b1239 let boot_functions return the compilers 
methods are still added, but this is a good step to removing the risc/cpu level from the methods
 2018-06-30 23:16:17 +03:00
Torsten Ruger
208b98d709 start to move translate code 2018-06-30 22:53:32 +03:00
Torsten Ruger
daf1b56062 start on class compiler 
idea is to get cleaner layer seperation
reduce machine and rework builtin boot
 2018-06-30 19:20:17 +03:00
Torsten Ruger
4a7cc72732 always list of methods in class 
improve vool test dirs
 2018-06-30 10:08:18 +03:00
Torsten Ruger
46d8f5002f hoist parfait boot out of the machine boot 2018-06-30 00:39:23 +03:00
Torsten Ruger
8d7a2fe4d6 fix mom tests 
and always have a list inside a class
 2018-06-29 23:29:10 +03:00
Torsten Ruger
5b87133df2 split rubyx tests 
in vool and mom level
 2018-06-29 23:04:50 +03:00
Torsten Ruger
7006c8e3aa fix remaining tests 
rename ripples
 2018-06-29 22:56:49 +03:00
Torsten Ruger
c8451d0048 create rubyx dir and move previous vool_compiler there 2018-06-29 22:46:39 +03:00
Torsten Ruger
7377522417 fix test ripples and minor parfait naming 
Parfait ruby and boot names desynced, fixed that
 2018-06-29 21:03:06 +03:00
Torsten Ruger
ae35fed0ab fix list to expand 
prevously fixed max length list
now expanding on demand, using next
quite like binary_code, a pattern is emerging
 2018-06-29 20:58:59 +03:00
Torsten Ruger
5036dd68df remove to_mom from ruby_to_vool 
must have slipped in
affects mosty tests
 2018-06-29 14:57:48 +03:00
Torsten Ruger
6bd01fd55f rename method_compiler 
in line with other compiler XX_Compiler being the compiler for that layer
remove type from compiler as it is in method available
 2018-06-29 14:48:52 +03:00
Torsten Ruger
e6f01748ff move adapter stuff around 2018-06-29 14:26:25 +03:00
Torsten Ruger
86b1edb40c add seperate builders 
remove if with polymorphism for different builders
(easier to understand by naming)
 2018-06-29 13:27:57 +03:00
Torsten Ruger
3dffebed3f rename a bit 2018-06-29 13:03:19 +03:00
Torsten Ruger
d50893bb0f rename risc_value to register_value 
almost to register, but it still carries that value
 2018-06-29 11:39:07 +03:00
Torsten Ruger
606c7bf906 remove booted flag 
just boot if needed
 2018-06-29 11:36:14 +03:00
Torsten Ruger
18994d2b4b start on yield statement 2018-06-28 20:15:24 +03:00
Torsten Ruger
f0ba863721 remove to_mom / create_objects dichotomy 
wsa supposed to be clearer, but even to me seems confusing now.
 2018-06-26 20:46:58 +03:00
Torsten Ruger
c6a903073a start on blocks 2018-06-26 20:28:27 +03:00
Torsten Ruger
4103da7490 small test fixes 2018-06-25 20:21:15 +03:00
Torsten Ruger
67a6ef9f67 add rewriting of operator assignment 
foo += 1 becomes foo = foo + 1 in vool
 2018-06-25 16:32:20 +03:00
Torsten Ruger
70d7e654c4 only check object file creation in efl 
remote execution only per request
 2018-06-25 00:39:13 +03:00
Torsten Ruger
b804be5f70 fix dynamic call 
which had the method in the regsiter, not the binary.
Single SlotToReg added (and some tests)
Which amazingly fixed all broken binary tests
 2018-06-25 00:19:43 +03:00
Torsten Ruger
068bda492c implement DynamicJump for Arm 
which is in fact the same as return (moving address into pc)
 2018-06-24 19:13:55 +03:00
Torsten Ruger
f83f52faa0 reusing ssh instance for arm 
some speedup , but hoped for more
 2018-06-24 17:44:02 +03:00
Torsten Ruger
c8a903cd83 adds arm testing of same files 2018-06-24 17:10:39 +03:00
Torsten Ruger
ae976c8fb9 generate mains test methods according to file name 2018-06-24 16:48:23 +03:00
Torsten Ruger
8d369dbfa1 automate mains test more 
move sources and include exit code and stdout in filename
 2018-06-24 13:37:36 +03:00
Torsten Ruger
563ed4647a extract the ruby from mains tests 
move up in directory
in preparation for arm tests
 2018-06-24 11:56:10 +03:00
Torsten Ruger
87be6bf9d5 add automated arm tests 2018-06-23 15:58:43 +03:00
Torsten Ruger
ce79617875 fix the exit sequence 
was returning wrong register (1, not 0)
also saving the message for test, which changes order as the return value destroys the message
 2018-06-19 19:52:06 +03:00
Torsten Ruger
5f7683efcf pass return integer back out through exit 
for testing of binaries later (and off course general correctness)
Some tests were using the fact that the interpreter was used, changed those to return ints rather than strings
 2018-06-19 18:55:47 +03:00
Torsten Ruger
db459fcd3d fix label loading in arm 
was still loading the integer address
instead of the return address's address
Hello World working
 2018-06-19 17:35:00 +03:00
Torsten Ruger
956c2ebe54 make arm use branch_lsteners 2018-06-19 10:51:03 +03:00
Torsten Ruger
046617f8dc add branch listener functionaliy 
have to store the branches and loop again as labels
dont neccessarily have positions yet
 2018-06-17 22:25:38 +03:00
Torsten Ruger
3298651238 split create_binary into two phases 
Which gives instructions a chance to check everything
and in Arms case check the constant loads/ instruction adding
So that during assembly no more change happens (and we don't have to reassemble)
 2018-06-17 13:53:17 +03:00
Torsten Ruger
c94f6eaa78 small maintanance 2018-06-16 21:01:15 +03:00
Torsten Ruger
7543236f4f objects didn't get positions 
positions.empty? was wrong check
 2018-06-16 10:58:54 +03:00
Torsten Ruger
698c845297 seperate position create and register 
many test fixes
 2018-06-15 22:00:49 +03:00
Torsten Ruger
9c93b38b8f crete positions while collecting objects 
clear two phase lifecycle for positions
always created invalid and set later
new does not auto add to cache
but create and get_or_create do
 2018-06-15 21:54:21 +03:00
Torsten Ruger
2c765c8f14 rename the objects accessor 
and move to object keys, not object_id
 2018-06-15 09:18:39 +03:00
Torsten Ruger
3cc9175efa start BranchListener 
but on hold, since it needs positions before we have them
Must create them during collection phase
 2018-06-14 21:29:34 +03:00
Torsten Ruger
eaaf61c117 fix position test 2018-06-14 21:19:22 +03:00
Torsten Ruger
5ec235d049 fixes outstanding misc, All green 
Well. it's been a while, positioning is fixed (again)
apart from the upcoming BranchListener
 2018-06-10 13:19:55 +03:00
Torsten Ruger
e1683938ed fix label positioning bug 
where a label at the end of a binary code resulted in the
label actually taking up a position, instead of sharing it with it's next
 2018-06-10 13:18:56 +03:00
Torsten Ruger
0513cd504f update return address in label 
so return jumps go to the right address
 2018-06-10 09:15:57 +03:00
Torsten Ruger
c7ad1d98ca introduce the LabeListener to move instructions along when first code position changes 2018-06-09 22:13:43 +03:00
Torsten Ruger
ec1d38f5a6 reimplement instruction listeners with indexes 
easier to understand, as indexes map to Binarycode indexes, and can thus later be used to assemble directoy into the  code
Almost no change in tests (now only multiples of 4 addresses allowed)
 2018-06-09 17:41:39 +03:00
Torsten Ruger
cf94227b2c fix code init bug 
where initial pos wasnt propagated
 2018-06-09 09:16:23 +03:00
Torsten Ruger
ad3040a846 add position_chaning to event interface 
by reacting to the change _before it happens, we can move any BinaryCode out of the way

So when Instruction are inserted and code gets inserted, we don't need to set up the correct listener explicitly (which is tricky across mathods and changing chains), but instead just move anything that is in the way along
 2018-06-09 08:10:41 +03:00
Torsten Ruger
74c15d45a3 some more mocking worked 
some didnt
 2018-06-08 20:43:36 +03:00
Torsten Ruger
5815d32bde make the instructions jump over the BinaryCode end 2018-06-07 19:27:44 +03:00
Torsten Ruger
2d218bbc48 work on jump insertion continues 
Now registering CodeListener instead of PositionListener
Also instead of on the previous, in itself, which is simpler
and allows to react to insertion at end
 2018-06-07 19:26:02 +03:00
Torsten Ruger
7fa8397b56 insertion pushes, split position tests 2018-06-06 10:19:18 +03:00
Torsten Ruger
c22aff4c4f start on insertion events and handling 2018-06-06 10:00:07 +03:00
Torsten Ruger
4789b63fcb fix the gapping of instructions 
so last binary stays empty and new starts at 8
 2018-06-06 01:16:00 +03:00
Torsten Ruger
4ab6d62acf small arm fixes etc 2018-06-06 00:53:41 +03:00
Torsten Ruger
d7b3368b28 fix some pesty errors 
need to fix instruction overlap beofre interpreter starts again
 2018-06-05 19:05:12 +03:00
Torsten Ruger
8d953a619f propagate instruction positions 
still overlapping onto binaries, but a start
 2018-06-05 18:11:25 +03:00
Torsten Ruger
f35ee6425a renaming and test fixing 2018-06-02 23:48:12 +03:00
Torsten Ruger
1d1c7105b4 introduce some helper methods 2018-06-02 23:02:59 +03:00
Torsten Ruger
c2d450f779 fold position module and object position 
simpler that way, aslo code is moving to listners
 2018-06-02 21:59:41 +03:00
Torsten Ruger
24f6e30b54 start on redoing instruction positions 
using insruction listeners (wip)
 2018-06-02 21:20:15 +03:00
Torsten Ruger
3bc35c2275 rework binary code positioning setup 
dependency chain set up explicitly.
Next have to react to events correctly
 2018-06-02 17:29:38 +03:00
Torsten Ruger
91c7903848 start simple with code listner 2018-06-02 16:34:44 +03:00
Torsten Ruger
aa6707337a positioning wip 2018-06-02 16:12:01 +03:00
Torsten Ruger
d1cc482f8f bit of test refactoring 2018-06-01 19:13:53 +03:00
Torsten Ruger
46fbfb7101 adds the simple object dependency 
where the next object is dependent on the previous one
and just behind it, the padded_length away
 2018-05-31 20:41:04 +03:00
Torsten Ruger
21a9c56ba9 change position event code to use util 
general purpose event fireing
can then be used for several event types
 2018-05-31 19:05:53 +03:00
Torsten Ruger
2544bc5a85 move position module to folder 2018-05-31 19:01:10 +03:00
Torsten Ruger
49849939a1 little event mechanism for changed positions 2018-05-31 17:01:23 +03:00
Torsten Ruger
0e155315aa movs is not allowed into pc 
Illegal instruction it says
Otherwise the status update is nice (for branches) and we’ll keep
 2018-05-31 14:03:25 +03:00
Torsten Ruger
67100a3ef8 write adjusted address 
and rename integer to address in label
1k hurray
 2018-05-31 00:07:58 +03:00
Torsten Ruger
e39e96f646 create return address as own class to hold return addresses 
to distinguish from integer, which does not need adjusting
 2018-05-30 23:49:01 +03:00
Torsten Ruger
e86ca5ae9d integer was wrong padded§ length 
or rather right (2 as should be and will be)
but currently we use next AND value + type means 3, rounded 4
because we can’t swap types at compile time (as is planned at runtime,
we use next OR value)
 2018-05-30 14:55:17 +03:00
Torsten Ruger
53107d3ef8 reducing the label's int for return 
placing the raw address in the register to jump to
puts the extra instruction at the risc level (not arm), thus changing a
lot of (brittle?) tests
 2018-05-30 12:54:40 +03:00
Torsten Ruger
01a2911483 using label int in interpreter 2018-05-30 11:49:29 +03:00
Torsten Ruger
0dc89c772a get the label int to work consistently 
still need to use it in the return
 2018-05-30 10:54:18 +03:00
Torsten Ruger
074ec34659 wip, fixed some label, need more fixing 2018-05-30 10:29:38 +03:00
Torsten Ruger
c55b41afae sort objects by class before positioning 
since we don’t have pages yet, but want consistent layout
 2018-05-29 20:52:58 +03:00
Torsten Ruger
e94c9ea37b add failfast 
tired of waiting for the almost 1000 tests to run through, just to get
an error message
 2018-05-29 20:40:02 +03:00
Torsten Ruger
296617cbaf set label integer to position 2018-05-29 20:30:02 +03:00
Torsten Ruger
8322fca7b3 give labels an integer that will end up being the position at runtime 
Since integers are first class objects, we need to use an integer object
as the return address. The actual address can not be stored in an
instance variable since it is not an object.
The address is unique to the label and never changes after positioning
(using the int is next up)
 2018-05-29 20:26:00 +03:00
Torsten Ruger
7847420d49 make ints compile time mutable 
to change the label value and use the allocated stubs
 2018-05-29 17:03:55 +03:00
Torsten Ruger
b81d9565de fix binary code offset when calling 
in other words, remember the arm pipeline being 8
 2018-05-29 17:03:20 +03:00
Torsten Ruger
4b34546c11 fixing the binary code indexes 
AGAIN, that third of all bugs, the off by one
 2018-05-28 18:20:09 +03:00
Torsten Ruger
8ef1a471a4 finally change List to derive from data object 
can’t derive from data16 as some lists are longer
have to get the delegation to work first
 2018-05-28 15:45:29 +03:00
Torsten Ruger
f9a89db10c use fake memory 
fix integer offset bug
(which only didn’t cause errors as fixnums are still an order too big
and the famous +1 error hit the empty space)
 2018-05-28 15:09:59 +03:00
Torsten Ruger
6c06f61ab8 start on fake memory 
get index checks on the array access of parfait
 2018-05-28 14:46:26 +03:00
Torsten Ruger
be1bc63ff2 jump was written off the end of binary code, fixed 2018-05-28 11:45:04 +03:00
Torsten Ruger
1c09d4202f fix all test 
and thus all green, two weeks of side branch positioning  done
(luckily not on arm, but interpreter)
 2018-05-25 20:40:39 +03:00
Torsten Ruger
c0cd1e0740 fix an issue where instruction was in wrong code 
embarrassingly used instance variable where it should have been local
 2018-05-25 19:16:13 +03:00
Torsten Ruger
6f0fad0957 dragging the extra through resets 
as the binary the instruction is in may change when repositioning
 2018-05-25 19:04:48 +03:00
Torsten Ruger
8d8cc4b016 more test fixing 
only one bug to go
 2018-05-24 21:20:56 +03:00
Torsten Ruger
bf23883270 busy fixing tests 2018-05-24 19:38:48 +03:00
Torsten Ruger
183d4152d5 loading label must translate the labels too 
(psst: like arm translator already did. duh)
 2018-05-24 19:20:06 +03:00
Torsten Ruger
dae8e88b7a recurring bug, don't save label positions 
labels are ethereal, no length, so they don’t get to occupy a position
 2018-05-24 16:26:56 +03:00
Torsten Ruger
8d510c2e7e getting instruction repositioning right 
big bug hunt, involving complicated maths (%)
 2018-05-24 14:27:53 +03:00
Torsten Ruger
0293320bb8 use a constant for the binary code offset 
where the instructions start
 2018-05-23 21:35:22 +03:00
Torsten Ruger
8ca70a6835 Steamline objectPosition init 
check for resetting on the same position with different class, which
should not happen
Ie it’s ok for an object to push another object up,
but not for an instruction to land on code
 2018-05-23 21:34:49 +03:00
Torsten Ruger
f5d1090c39 no-ops really 2018-05-23 18:06:55 +03:00
Torsten Ruger
a2e7d7c469 give interpreter a clock and pc 
where the pc, like in cpu’s is the memory position.
That is what the interpreter works on.
But for humans, the clock is a simpler way to count where the program
is at, no. of instructions executed
 2018-05-23 18:05:22 +03:00
Torsten Ruger
ef2dc932ad use positions in interpreter 2018-05-20 15:52:13 +03:00
Torsten Ruger
a7a62d53b2 use modified opal logger 
logger was having bizarre format errors
copied and simplified (to get on)
 2018-05-20 14:45:48 +03:00
Torsten Ruger
15e4533a2f fix interpreter to use positions 
tick from one (simulated memory) position to the next
(not from one instruction to the next)
 2018-05-17 20:14:59 +03:00
Torsten Ruger
8966a50a8a getter by position 
extract and expand position testing
never return labels (that have the same position as their target)
 2018-05-17 20:13:33 +03:00
Torsten Ruger
37d62d298e assemble risc to it's position (not itself) 
So in the next step the interpreter can use positions as program counter
and would be much more like the real thing
 2018-05-17 09:49:01 +03:00
Torsten Ruger
49c4d170ce make the interpreter platform 
but still using the risc_instruction stream
 2018-05-17 09:31:36 +03:00
Torsten Ruger
07ed16d765 groundwork for interpreter platform 
especially positioning makes the previous interpreter approach to
fragile (too far off the real thing)
Interpreter should be it’s own platform (still use the risc
instructions),
get a translator and use positions
 2018-05-16 21:00:14 +03:00
Torsten Ruger
9597fc5756 more slot tests, also in separate files 2018-05-16 12:49:46 +03:00
Torsten Ruger
36e59ebdea some slot load tests 2018-05-16 12:35:44 +03:00
Torsten Ruger
473866e76b start with some slot definition tests 2018-05-15 19:29:27 +03:00
Torsten Ruger
ab4bc370ed another +1 bug 2018-05-15 16:25:55 +03:00
Torsten Ruger
e237bc625a remove unused methods 
and a whole lot more index fixes
 2018-05-14 20:50:52 +03:00
Torsten Ruger
5d3c70da89 fix type api 
tests were using methods that were only used in tests.
 2018-05-14 16:13:50 +03:00
Torsten Ruger
3c00239f36 another million index fixes 2018-05-14 15:17:04 +03:00
Torsten Ruger
4a88f342d3 random checkin 
still suffering -1 trauma
 2018-05-14 12:38:44 +03:00
Torsten Ruger
ab01fa3862 switch to 0 based indexing 
the world rocked for a moment (and more fixes to follow)
also the crumbling of idealism was heard
 2018-05-14 11:55:01 +03:00
Torsten Ruger
4856b9891d simplify writing binary 
loop as should be, fixes the problem (one problem at least)
 2018-05-13 18:01:45 +03:00
Torsten Ruger
866467ee5e change binary labels 2018-05-13 17:21:48 +03:00
Torsten Ruger
39902401b9 positioning code by setting first method code 
codes will initial (and on reset) propagate the whole chain
 2018-05-13 15:28:10 +03:00
Torsten Ruger
7ad36380c2 add next_value for dictionary 
also next for list
and add types attributes to space
 2018-05-13 13:03:04 +03:00
Torsten Ruger
2d901bf7b6 not wrapping the cpu initial jump anymore 
also introduce padding after cpu_init (wip)
 2018-05-12 18:36:59 +03:00
Torsten Ruger
232fe67c09 introduce platform to abstract cpu and load address 2018-05-12 18:32:10 +03:00
Torsten Ruger
776a97986d fix instruction positioning 
use at as at and only compute difference when needed (to determine if
there is a jump in binary)
Easier to understand that way
 2018-05-11 18:36:45 +03:00
Torsten Ruger
bc1e29e4f6 seperate different Position classes into own files 
also tests
and have Position module keep all positions
(singletons should be at module, not class level)
 2018-05-10 20:56:12 +03:00