Commit graph

12210 commits

Author SHA1 Message Date
Bill Currie
24a42dc064 [qfcc] Emit args for ... functions with no other parameters
I missed that the block was < -1, ie at least one real parameters.
2022-02-04 22:00:18 +09:00
Bill Currie
1b40cdbab6 [qfcc] Handle vector scaling by ints
I missed a change when implementing support for the scale instructions.
2022-02-04 21:57:41 +09:00
Bill Currie
2a8fca80a0 [nq,qw] Give the menu and server progs stacks
They're going to need them :P
2022-02-04 21:53:37 +09:00
Bill Currie
7731c469e2 [gamecode] Count calls to builtins in profile
It's a bit disconcerting seeing a builtin in the top 10 when builtins
are counted by call while progs functions are counted by instruction.

Also, show the total profile after the function top-10 list.
2022-02-04 21:49:59 +09:00
Bill Currie
974af36d13 [gamecode] Move profile out of the union
It happens to sit over the builtin data pointer and thus messes up
PR_Profile (any attempt to count calls to builtins).
2022-02-04 21:46:40 +09:00
Bill Currie
b425f449b6 [ruamoko] Separate the two str_mid builtins
pr_argc cannot be used in Ruamoko progs because nothing sets it. This
fixes the parse errors and resulting segfault when trying to parse the
Vulkan pipeline config.
2022-02-04 11:38:36 +09:00
Bill Currie
26fca581fc [qfcc] Make ruamoko the default and update the docs
It's time to do the dog-food.
2022-02-04 10:46:31 +09:00
Bill Currie
f3770cc647 [qfcc] Add --ruamoko command line option and pragma
The command line option works the same way as
--advanced/traditional/extended, as does the pragma. As well, raumoko
(alternative spelling) can be used because both are legitimate and some
people may prefer one spelling over the other.

As always, use of the pragma is at one's own risk: its intended use is
forcing the target in the unit tests.
2022-02-04 09:27:07 +09:00
Bill Currie
52a399daeb [qfcc] Update sizes and alignments for dvec4 and friends
dvec4, lvec4 and ulvec4 need to be aligned to 8 words (32 bytes) in
order to avoid hardware exceptions. Rather than dealing with possibly
mixed alignment when a function has 8-word aligned locals but only
4-word aligned parameters, simply keep the stack frame 8-word aligned at
all times.

As for sizes, the temp def recycler was written before the Ruamoko ISA
was even a pipe dream and thus never expected temp def sizes over 4. At
least now any future adjustments can be done in one place.

My quick and dirty test program works :)

    dvec4 xy = {1d, 2d, 0d, 0.5};
    void printf(string fmt, ...) = #0;
    int main()
    {
	dvec4 u = {3, 4, 3.14};
	dvec4 v = {3, 4, 0, 1};
	dvec4 w = v * xy + u;
	printf ("[%g, %g, %g, %g]\n", w[0], w[1], w[2], w[3]);
	return 0;
    }
2022-02-04 08:46:58 +09:00
Bill Currie
1487fa6b50 [qfcc] Implement some basics for the vector types
They're now properly part of the type system and can be used for
declaring variables, initialized (using {} block initializers), operated
on (=, *, + tested) though much work needs to be done on binary
expressions, and indexed. So far, only ivec2 has been tested.
2022-02-04 00:25:31 +09:00
Bill Currie
ca818eaff8 [qfcc] Ensure ops on globals occur before return
This fixes the return-postop test, and covers calls, too.
2022-02-03 16:40:53 +09:00
Bill Currie
bbac02de24 [qfcc] Add failing test for return of postop
Commit 76b3bedb72 broke more than just the
swap test, but at least I know I need to get an edge in the dag.
Currently, the following code is generated: return and add are reversed.

    ../tools/qfcc/test/return-postop.r:8:   return counter++;
    0001 store.i counter, .tmp0
    0002 return .tmp0
    0003 add.i .tmp0, (1), counter

However, I don't want to deal with it right now, so it's marked XFAIL.
2022-02-03 16:40:53 +09:00
Bill Currie
3d8ee5df43 [qfcc] Ensure ops on globals occur before return
This fixes the return-postop test, and covers calls, too.
2022-02-03 16:33:42 +09:00
Bill Currie
a3c37201b2 [qfcc] Emit constant pointers as direct def references
When possible, of course. However, this tightens up struct and constant
index array accesses, and avoids issues with flow analysis losing track
of the def (such trucking is something I want to do, but haven't decided
out to get the information out to the right statements).
2022-02-03 14:41:46 +09:00
Bill Currie
008359862b [qfcc] Avoid pointer alias of address expressions
Since address expressions always product a pointer type, aliasing one to
another pointer type is redundant. Instead, simply return an address
expression with the desired type.
2022-02-03 14:38:26 +09:00
Bill Currie
80c6431544 [qfcc] Clear up a FIXME
The FIXME was there because I couldn't remember why the test was
type_compatible but the internal error complains about the types being
the same size. The compatibility check is to see if the op can be used
directly or whether a temp is required. The offset check is because
types that are the same size (which they must be if they are
compatible) is because it is not possible to create an offset alias def
that escapes the bounds of the real def, which any non-zero offset will
do if the types are the same size.
2022-02-03 14:15:20 +09:00
Bill Currie
6f49b919ec [qfcc] Move constant pointer offset into address expr
This is the intended purpose of the offset field in address expressions,
and will make struct and array accesses more efficient when I sort out
the code generation side.
2022-02-03 10:55:37 +09:00
Bill Currie
b668759b7d [qfcc] Add a very basic attribute system
Ruamoko passes va_list (@args) through the ... parameter (as such), but
IMP uses ... to defeat parameter type and count checking and doesn't
want va_list. While possibly not the best solution, adding a no_va_list
flag to function types and skipping ex_args entirely does take care of
the problem without hard-coding anything specific to IMP.

The system currently just sets some bits in the type specifier (the
attribute list should probably be carried around with the specifier),
but it gets the job done for now, and at least gets things started.
2022-02-02 23:51:37 +09:00
Bill Currie
6fe72b0420 [qfcc] Check load/store operand type before mangling
This fixes the incorrect use of assign64 for quaternions. All tests
except return-postop pass \o/.
2022-02-02 19:14:22 +09:00
Bill Currie
0fa9d0d256 [qfcc] Tweak the printf to make more sense
Though I'm sure I had a good reason at the time, seeing 6.98487e-315
when expecting pi is a bit disconcerting.
2022-02-02 19:04:43 +09:00
Bill Currie
a64f91129f [qfcc] Give lea its own statement type
This makes it much easier to check (and more robust to name changes),
allowing for effectively killing the node to which the variable being
addressed is attached. This fixes the incorrect address being used for
va_list, which is what caused double-alias to fail.
2022-02-02 18:55:01 +09:00
Bill Currie
38550922cc [qfcc] Map 64-bit load/store/assign instructions
In order to not waste instructions, the Ruamoko ISA does not provide 1
and 2 component 64-bit load/store instructions since they can be
implemented using 2 and 4 component 32-bit instructions (load and store
are independent of the interpretation of the data). This fixes the
double test, and technically the double-alias test, but it fails due to
a problem with the optimizer causing lea to use the wrong reference for
the address. It also breaks the quaternion test due to what seems to be
a type error that may have been lurking for a while, further
investigation is needed there.
2022-02-02 16:06:15 +09:00
Bill Currie
0211b6ec5b [gamecode] Add 64-bit load/store instructions
Only widths 3 and 4 have been added because widths 1 and 2 can be
implemented by widths 2 and 4 of the 32-bit load/store instructions.
2022-02-02 14:47:36 +09:00
Bill Currie
93840d9892 [qfcc] Handle Ruamoko's call return destination
Since the call instruction in the Ruamoko ISA specifies the destination
of the return value of the called function, it is much like any
expression type instruction in that the def referenced by its c operand
is both defined and killed by the instruction. However, unlike other
instructions, it really has many pseudo-operands: the arguments placed
on the stack. The problem is that when one of the arguments is also the
destination of the return value, the dags code wants to use the stack
argument as it was the last use of the real argument. Thus, instead of
using the value of the child node for the result, use the value label
attached to the call node (there should be only one such label).

This fixes iterfunc, typedef, zerolinker and vkgen when optimizing. Now
all but the double tests and return postop tests pass (and the retun
postop test is not related to the Ruamoko ISA, so fails either way).
2022-02-01 21:46:28 +09:00
Bill Currie
9b81d27f1a [qfcc] Add test for var = func(var)
That is, updating a variable using a function that takes the same
variable, probably very common in iterators, thus the name. It happens
to be the first qfcc test specific to Ruamoko. It's really just the
typedef, zerolinker, and vkgen type encoding loop stripped down for ease
of debugging.

Of course, it fails :)
2022-02-01 21:40:59 +09:00
Bill Currie
4ec3486a4b [qfcc] Make lea generate a pointer operand
I really need to come up with a better way to get the result type into
the flow analyser. However, this fixes the aliasing ICE when optimizing
Ruamoko code that uses struct assignment.
2022-02-01 20:01:13 +09:00
Bill Currie
6514e09e7c [gamecode] Use an explicit size for the null page
It's currently only 4 (or even 3 for v6) words, but this fixes false
positives when checking for null pointers in Ruamoko progs due to
pr_return pointing to the return buffer and thus outside the progs
memory map resulting in an impossible to exceed value.
2022-02-01 16:43:29 +09:00
Bill Currie
c84fb3e6d3 [qfcc] Use a hidden local variable for pascal functions
This gets gcd.pas working nicely with the Ruamoko ISA, and keeps things
reasonably nice vor v6p (it will likely do better with global CSE).
2022-02-01 16:08:58 +09:00
Bill Currie
fc56d1c6e2 [qfcc] Use dynamic addressing for Ruamoko move source
This fixes the incorrect pointer being used in movep instructions in
Ruamoko progs, as well as 3 of the test cases.
2022-02-01 14:57:16 +09:00
Bill Currie
e0c5c475ea [qfcc] Modify the modulo tests to be compatible with Ruamoko ISA
Surprisingly, it passes (I didn't expect it to due to the doubles). I'll
look into to it further later on.
2022-02-01 14:57:16 +09:00
Bill Currie
64c8c02eac [qfcc] Add pragma to control optimization
And force a couple of tests to be built with optimization. I'll probably
add it to most, if not all, but for now I'm clearing up tests as I go.
2022-02-01 14:57:16 +09:00
Bill Currie
f714b6fbea [qwaq] Up qwaq progs heap to 2MB
or 512kW (kilowatts? :P). Barely enough for vkgen to run (it runs out if
auto release is run during scan_types, probably due to fragmentation). I
imagine I need to look into better memory management schemes, especially
since I want to make zone allocations 64-byte aligned (instead of the
current 8). And it doesn't help that 16 words per allocation are
dedicated to the zone management.

Anyway, with this, vgken runs and produces sufficiently correct results
for the rest of QF to build, so long as qfcc is not optimizing.
2022-02-01 14:57:16 +09:00
Bill Currie
bea1155e9e [qwaq] Give qwaq progs a stack of 64kB
16kW should be enough as a default (when I get around to making it
configurable). Stops vkgen from trashing its heap with the stack.
2022-02-01 14:57:16 +09:00
Bill Currie
2fcec6e5cb [zone] Move heap check to Z_TagMalloc
Since Z_Malloc uses Z_TagMalloc to do the work, this ensures the check
is always run.

Also, add the check to Z_Realloc when it needs to adjust an existing
block.
2022-02-01 14:56:47 +09:00
Bill Currie
b8c2b7f856 [ruamoko] Make a common sprintf wrapper function
This takes care of converting from progs varargs to what PR_Sprintf
expects. I got tired of modifying the wrappers when I found a third one.
2022-02-01 09:27:03 +09:00
Bill Currie
5d41e90cc7 [gamecode] Remove pushregs and popregs specs
They have been redundant since the operations were moved into with.
2022-01-31 23:47:47 +09:00
Bill Currie
e7ac637687 [ruamoko] Wrap hash table callbacks with push/pop frame
Builtins that call progs with parameters now must always wrap the call
to PR_ExecuteProgram so that the data stack is properly preserved across
the call.

I need to do an audit of all the calls to PR_ExecuteProgram.
2022-01-31 23:47:02 +09:00
Bill Currie
175cc408d8 [qwaq] Update bi_printf to work with Ruamoko's va_list
Same deal as for test_bi in qfcc's test harness.  I really need to
consolidate all these little functions.
2022-01-31 23:44:09 +09:00
Bill Currie
e3f88b2b9c [qwaq] Align progs memory to 64 bytes
It doesn't do much good for dynamic progs memory because zone currently
aligns to 8 bytes (oops, forgot to fix that), but at least the stack and
globals are properly aligned.
2022-01-31 23:40:21 +09:00
Bill Currie
98215b46ad [ruamoko] Cast autoreleaseIMP to avoid ...
The ABI for the Ruamoko ISA set currently puts va_list into the
parameter stream whenever a varargs function is called. Unfortunately,
IMP is declared with ... and thus cannot be used directly unless all
methods become varargs, but I think that would cause even more
headaches.
2022-01-31 23:37:03 +09:00
Bill Currie
712d800491 [gamecode] Save return ptr for chained calls
It turns out the return pointer still needs to be saved even when a
builtin sets up a chain call to progs, but rather than the pointer being
simply restored, it needs to be saved in the call stack exactly as if
the function was called directly by progs. This fixes the invalid self
issue quite thoroughly: parameter state seems to be correct across all
calls now.

I should set up an automated test now that I know and understand the
situation.
2022-01-31 23:35:56 +09:00
Bill Currie
92711e778c [ruamoko] Set params in obj_msgSend_super only for v6p progs
In Ruamoko ISA progs, the param pointers point to the stack and
generally must most be manipulated by builtins, and there is no need
anyway as Ruamoko doesn't have RCALL. Fixes the mangling of .super.
2022-01-31 19:04:18 +09:00
Bill Currie
cabb53e693 [gamecode] Skip return ptr restore if depth changed
When calling a builtin, normally the return pointer needs to be
restored, but if the builtin changes the call depth (usually by
effecting "return foo()" as in support for objects, but possibly
setjmp/longjmp when they are implemented), then the return pointer must
not be restored. This gets vkgen past object allocation, but it dies
when trying to send messages to super. This appears to be a compiler
bug.
2022-01-31 16:51:46 +09:00
Bill Currie
c5ae1ae13c [ruamoko] Check self is within progs memory
Changes a segfault to a runtime error, which beats the risk of self
pointing somewhere that doesn't segrault.
2022-01-31 14:03:36 +09:00
Bill Currie
edf7a781fd [qfcc] Map uint to int for some intructions
Many math instructions don't care about the difference between signed
and unsigned operands and are thus specified using int, but need to be
usable with uint. div is NOT mapped because there is a difference:
0x8000 / 2 (16-bit) is 0x4000 unsigned but 0xc000 signed, and 0x8000 /
0xfffe is 0 unsigned and 0x4000 signed. This means I'll need to add some
more instructions. Not sure what to do about % and %% though as that's a
lot of instructions (12).
2022-01-30 22:47:57 +09:00
Bill Currie
8dc4a0ea80 [qfcc] Change v6p's jumpb opname to jump
More ease of searching, since the operand types help greatly.
2022-01-30 22:39:21 +09:00
Bill Currie
7971bcd91c [gamecode] Sort out shr's opname for easier searching
Since the operand types sort out the difference between asr and shr, no
need to give them different opnames. Means qfcc doesn't need to worry
about which one it's searching for.
2022-01-30 22:37:03 +09:00
Bill Currie
766bf758ab [gamecode] Redesign jump's B addressing
Yet another redundant addressing mode (since ptr + 0 can be used), so
replace it with a variable-indexed array (same as in v6p). Was forced
into noticing the problem when trying to compile Machine.r.
2022-01-30 22:34:40 +09:00
Bill Currie
b8df11b2cb [qfcc] Really get vecaddr.r working on both ISAs
Deliberately defeating an optimiser is not so easy (but I really needed
that variable to be set).
2022-01-30 20:01:32 +09:00
Bill Currie
e1a0c31e3f [qfcc] Encode the new vector types
Thanks to the size of the type encoding being explicit in the encoding,
anything that tries to read the encodings without expecting the width
will simply skip over the width, as it is placed after the ev type in
the encoding.

Any code that needs to read both the old encodings and the new can check
the size of the basic encodings to see if the width field is present.
2022-01-30 16:00:49 +09:00