| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Now compiles to C and builds a working executable. Just need to
correctly implement all the library builtins (some are still stubs),
and it should work.
|
|
Isabelle does not like nested annotations like "((exp :: typ) :: typ)".
|
|
Solves a problem where generated kids crept into type annotations during
rewriting and caused later typechecking passes to fail.
|
|
|
|
|
|
|
|
This was technically allowed previously but the rules for type
variable names in function types were too strict so it didn't work.
Also fixed a bug where Nexp_app constructors were never considered identical
and fixed a bug where top-level let bindings got annotated with the
wrong environment
|
|
|
|
Look deep into sub-patterns for identifiers and literals instead of relying on
assumptions about possible nestings
|
|
Make rewriter look into P_typ patterns instead of throwing them away.
|
|
The internal function _reg_deref is declared as pure, so that bitfield setters
can be implemented as read-modify-write, while only having a wreg effect.
However, for the Lem shallow embedding, the read step of those setters needs to
be embedded into the monad. This could be special-cased in the Lem pretty
printer, but then the pretty printer would have to replicate some logic of the
letbind_effects rewriting step. It seems simplest to add the effect annotation
early in the Lem rewriting pipeline, in the fix_val_specs step. This means
that this rewriting step can only be used for other backends if these
additional effects are acceptable.
|
|
|
|
|
|
|
|
Only add bindings for sub-patterns if they are actually used in the pattern
guard or expression, respectively.
|
|
Seems to increase compilation speed significantly for OCaml 4.05.
|
|
Use consistent nesting of tuples when adding updated local mutable variables to
expressions. Add test case.
|
|
|
|
Uses the typechecker tests for now. Also fix two minor bugs in pretty-printer
and rewriter uncovered by the tests.
|
|
|
|
- Remove vector start indices
- Library refactoring: Definitions in sail_operators.lem now use Bitvector
type class and work for both bit list and machine word representations
- Add Lem bindings to AArch64 and RISC-V preludes
TODO: Merge specialised machine word operations from sail_operators_mwords into
sail_operators.
|
|
Changed -mono-split to -mono_split to be consistent with other options
that use underscores, -mono-split still works but gives a warning
message, just so nothing breaks immediately because of this.
Removed this sil commands since they really don't do anything right
now.
|
|
|
|
This change allows the AST to be type-checked after sizeof
re-writing. It modifies the unification algorithm to better support
checking multiplication in constraints, by using division and modulus
SMT operators if they are defined.
Also made sure the typechecker doesn't re-introduce E_constraint
nodes, otherwise re-checking after sizeof-rewriting will re-introduce
constraint nodes.
|
|
Also drop redundant unit expressions when concatenating with an empty block.
|
|
In particular, there is an ASL pattern with single-branch if-expressions
containing an early return (and an empty else-branch), e.g.
{
...
if (error) then return(Fault) else ();
...
return(Success);
}
The rewriting pass now tries to fold the rest of the block into the
else-branch, since it is unreachable after the then-branch, e.g.
{
...
if (error) then return(Fault) else {
...
return(Success);
}
}
In combination with other rewriting rules, this allows the rewriting pass to
pull out the early returns if *all* branches end in a return statement. If it
is possible to pull out the return all the way, i.e., so that the function body
is a single return statement, then the return can be removed. If that is not
possible, then the function body is left as it was originally.
|
|
Monomorphisation sometimes produces pattern guard with let-bindings, e.g.
| ... if (let regsize = size_itself(regsize) in eq(regsize, 32)) -> ...
Previously, the rewriting pass for let-bindings (and pattern guards) pulled
these out of the guard condition and into the same scope as the case
expression, which potentially clashed with let-bindings for the same variables
in that case expression. The rewriter now leaves let-bindings in place within
pure if-conditions, solving this problem.
|
|
Keep track of which type variables have been bound in the function declaration,
and allow those to be pretty-printed
|
|
Previously we only did top-level literal pattern to guard conversion, this
does it throughout any pattern
|
|
|
|
New testcase for bitfield syntax
Updated to work with latest lem and linksem
|
|
For example:
bitfield cr : vector(8, dec, bit) = {
CR0 : 7 .. 4,
LT : 7,
CR1 : 3 .. 2,
CR2 : 1,
CR3 : 0,
}
The difference this creates a newtype wrapper around the vector type,
then generates getters and setters for all the fields once, rather
than having to handle this construct separately in every backend.
|
|
|
|
* Changed comment syntax to C-style /* */ and //
* References to registers and mutable variables are never created
implicitly - a reference to a register or variable R is now created
via the expression "ref R". References are assigned like "(*Y) = X",
with "(*ref R) = X" being equivalent to "R = X". Everything is always
explicit now, which simplifies the logic in the typechecker. There's
also now an invariant that every id directly in a LEXP is mutable,
which is actually required for our rewriter steps to be sound.
* More flexible syntax for L-expressions to better support wierd
power-idioms, some syntax sugar means that:
X.GET(a, b, c) ==> _mod_GET(X, a, b, c)
X->GET(a, b, c) ==> _mod_GET(ref X, a, b, c)
for setters, this can be combined with the (still somewhat poorly
named) LEXP_memory construct, such that:
X->SET(a, b, c) = Y ==> _mod_SET(ref X, a, b, c, Y)
Currently I use the _mod_ prefix for these 'modifier' functions, but
we could omit that a la rust.
* The register bits typedef construct no longer exists in the
typechecker. This construct never worked consistently between backends
and inc/dec vectors, and it can be easily replaced by structs with
fancy setters/getters if need be. One can also use custom type operators to mimic the syntax, i.e.
type operator ... ('n : Int) ('m : Int) = slice('n, 'm)
struct cr = {
CR0 : 32 ... 35,
/* 32 : LT; 33 : GT; 34 : EQ; 35 : SO; */
CR1 : 36 ... 39,
/* 36 : FX; 37 : FEX; 38 : VX; 39 : OX; */
CR2 : 40 ... 43,
CR3 : 44 ... 47,
CR4 : 48 ... 51,
CR5 : 52 ... 55,
CR6 : 56 ... 59,
CR7 : 60 ... 63,
}
This greatly simplifies a lot of the logic in the typechecker, as it
means that E_field is no longer ambiguously overloaded between records
and register bit typedefs. This also makes writing semantics for these
constructs much simpler.
|
|
Experimenting with porting riscv model to new typechecker
|
|
|
|
The type-checker already supports a user-defined "exception" type that can be
used in throw and try-catch expressions. This patch adds support for that to
the Lem shallow embedding by adapting the existing exception mechanisms of the
state and prompt monads. User-defined exceptions are distinguished from
builtin exception cases. For example, the state monad uses
type ex 'e =
| Exit
| Assert of string
| Throw of 'e
to distinguish between calls to "exit", failed assertions, and user-defined
exceptions, respectively. Early return is also handled using the exception
mechanism, by lifting to a monad with "either 'r exception" as the exception
type, where 'r is the expected return type and "exception" is the user-defined
exception type.
|
|
Pattern matching against literal numbers is not well supported by Lem or
Isabelle. This rewriting step turns them into guarded patterns (which can be
picked up by the guarded pattern rewriting to if-expressions).
|
|
Works with the vector branch of asl_parser
|
|
|
|
|
|
Improved handling of try/catch
Better handling of unprovable constraints when the environment contains
false
|
|
Recent patches have made the rewriter more strict about performing
type correct rewrites. This is mostly a good thing but did cause some
problems with the ocaml backend.
Currently the sizeof rewriter doesn't seem to preserve type
correctness - I suspect this is because when it resolves the sizeofs,
it generates constraints that are true, but not in a form where the
typechecker can see that they are true. I disabled the re-check after
the sizeof rewriting pass to fix this. Maybe we don't want to do this
anyway because it's slow.
Changes to function clauses with guards + monomorphisation changed how
the typechecker handles literal patterns. I added a rewriting pass to
rewrite literals to guarded equality checks, which is run before
generating ocaml.
The rewriter currently uses Env.empty in a view places. This can cause
bugs because Env.empty is a totally unitialised environment that
doesn't satisfy invariants we expect of an environment. This should be
changed to initial_env and it shouldn't be exported, I fixed a few
cases where this caused things to go wrong, but it should probably not
be exported from Type_check.ml.
|
|
|
|
Also fix bug in mono analysis with generated variables
Breaks lots of typechecking tests because it generates unnecessary
equality tests on units (and the tests don't have generic equality),
which I'll fix next.
|
|
|
|
|
|
|
|
steps
Parser now has syntax for mutual recusion blocks
mutual {
... fundefs ...
}
which is used for parsing and pretty printing
DEF_internal_mutrec. It's stripped away by the initial_check, so the
typechecker never sees DEF_internal_mutrec. Maybe this could change,
as forcing mutual recursion to be explicit would probably be a good
thing.
Added record syntax to the new parser
New option -dmagic_hash is similar to GHC's -XMagicHash in that it
allows for identifiers to contain the special hash character, which is
used to introduce new autogenerated variables in a way that doesn't
clash with existing names.
Option -sil compiles sail down to the intermediate language defined in
sil.ott (not complete yet).
|
