| Age | Commit message (Collapse) | Author |
|---|
|
|
|
By default generated functions are non-static, using the -static flag makes them static which is useful for measuring coverage using generated code. Some utility helper functions will always be static, but the compiled versions of sail functions won't be unless this flag is set.
|
|
Makes sure equality comparisons for variants are compiled correctly. Needed for CHERI and mips, which have structs containing
variants. Also make sure that struct equality works for structs containing other structs.
|
|
Add a test case in test/c/eq_struct.sail. Ensure that the macro EQUAL(type) will always give a valid equality function for any
builtin type in sail.h.
|
|
Non bitvector literals for decreasing vectors were not being reversed
correctly, so the list of capability registers was effectively in
reverse order.
Added a test case to test/c/ based on this aspect of CHERI
|
|
to notice if they are dead or inline them if appropriate, cleaning up coverage reports and potentially improving execution speed.
|
|
We should test before the first iteration in case 'to' starts out as
less than 'from'.
|
|
The code to do this is rather ugly. It would be nice to have a generic
callgraph representation we could just query and not use the rewriter
in a weird way to extract this info.
|
|
|
|
|
|
This is now directly supported from SAIL so we can call the SAIL __SetConfig function
instead.
|
|
Registers can now be marked as configuration registers, for example:
register configuration CFG_RVBAR = 0x1300000
They work like ordinary registers except they can only be set by
functions with the 'configuration' effect and have no effect when
read. They also have an initialiser, like a let-binding. Internally
there is a new reg_dec constructor DEC_config. They are intended to
represent configuration parameters for the model, which can change
between runs, but don't change during execution. Currently they'll
only work when compiled to C. Internally registers can now have custom
effects for reads and writes rather than just rreg and wreg, so the
type signatures of Env.add_register and Env.get_register have changed,
as well as the Register lvar, so in the type checker we now write:
Env.add_register id read_effect write_effect typ
rather than
Env.add_register id typ
For the corresponding change to ASL parser there's a function
is_config in asl_to_sail.ml which controls what becomes a
configuration register for ARM. Some things we have to keep as
let-bindings because Sail can't handle them changing at runtime -
e.g. the length of vectors in other top-level definitions. Luckily
__SetConfig doesn't (yet) try to change those options.
Together these changes allow us to translate the ASL __SetConfig
function, which means we should get command-line option compatibility
with ArchEx for running the ARM conformance tests.
|
|
Also further tweaks to Sail library for C and include sail lib files for tracing
|
|
Refactor the C compilation process by moving out the conversion to
A-normal form into its own file. Also make the A-normal form AST
parameterised by the type of the types annotating it. The idea being we
can have a typ aexp -> ctyp aexp translation, converting to low-level
types at a slightly higher level before mapping into our low-level IR.
This would fix some issues we have where the type of variables change
due to flow typing, because we could map the sail types to low-level
types in the ANF ast where we still have some knowledge about the
structure of the original Sail.
|
|
For example, the MIPS model can boot FreeBSD as
./mips_c --binary=0x100000,/path/to/kernel --image=/path/to/simboot.sailbin
Or with short options as
./mips_c -b 0x100000,/path/to/kernel -i /path/to/simboot.sailbin
The current options are:
-e, --elf, which loads an elf file directly
-n, --entry, which sets the entry point
-i, --image, which loads an image file compiled by "sail -elf" using Linksem
-b, --binary, which loads a plain binary image into memory at a specific address
-l, --cyclelimit, which means the (new) cycle_count() builtin exits the model after a certain number of calls
Also there are the default -? --help and --usage options.
|
|
|
|
This plus changes to bitfield internals is enough to run some MIPS tests at 1Mhz.
|
|
Previously the ANF->IR translation cared too much about how things
were allocated in C, so it had to constantly check whether things
needed to be allocated on the stack or heap, and generate different
cequences of IR instructions depending on either. This change removes
the ialloc IR instruction, and changes iinit and idecl so that the
code generator now generates different C for the same IR instructions
based on the variable types involved.
The next change in this vein would be to merge icopy and iconvert at
the IR level so that conversions between uint64_t and large-bitvectors
are inserted by the code generator. This would be good because it
would make the ANF->IR translation more robust to changes in the types
of variables caused by flow-typing, and optimization passes could
convert large bitvectors to uint64_t as local changes.
|
|
|
|
|
|
Fix a bug involving indentifers on the left hand side of assignment
statements not being shadowed correctly within foreach loops.
Make the different between different types of integer division
explicit in at least the C compilation for now. fdiv_int is division
rounding towards -infinity (floor). while tdiv_int is truncating
towards zero. Same for fmod_int and tmod_int.
|
|
|
|
|
|
|
|
|
|
|
|
This should fix the issue raised in commit 45554f
Adds a test loop_exception that tests throwing exceptions in loops, various looping constructs, and returning values from try/catch blocks. Also modified the test-suite to test C compiled output both with and without optimisations
|
|
|
|
optimizations.
Move the utility functions for graph generation and pretty printing of
intermediate representation instructions into a separate file,
bytecode_util.ml, by analogy with ast_util.ml.
Add an optimization pass that searches for specific patterns of struct
updates and removes uncessary copying of the structs involved. With
this optimisation pass the time taken for u-boot to run approx
57,000,000 instructions goes down from about 11-12 minutes to 8
minutes (about 120,000 IPS).
|
|
Also add an additional -Oz3 flag that uses z3 to optimize some
additional types. This is currently very experimental and doesn't
fully work yet.
|
|
We now store the location where type variables were bound, so we can
use this information when printing error messages.
Factor type errors out into type_error.ml. This means that
Type_check.check is now Type_error.check, as it previously it handled
wrapping the type_errors into reporting_basic
errors. Type_check.check' has therefore been renamed to
Type_check.check.
|
|
- Refactor the flow typing implementation in the type-checker. This
should fix an issue involving riscv_platform. Specifically it should
now work better when an if statement contains multiple conditions
combined with and/or, only some of which imply constraints at the
type level. This change also simplifies the implementation of flow
typing, and removes some obscure features that were hardly used -
specifically, flow typing could modify types, but this was fairly
obscure and doesn't seem to affect any of our specifications. More
testing is needed to ensure that this change hasn't inadvertantly
broken anything, but it does pass all our tests and continue to
typecheck arm, riscv and cheri.
- Also adds a option for generating faster undefined functions for
enum and variant types. Previously I tried to optimise away such
functions in the C backend, because they could be slow and cause
considerable uneccessary allocation, however this was error prone
and it turns out a much simpler solution is to simply make the
functions themselves much faster, at the cost of hard-coding certain
decisions about what undefined means for these types at compile tile
(which is fine for fast emulation). This almost doubles the
performance of the generated C code.
- Add a wrapper for right shift to avoid UB when shifting by 64 or
more places.
|
|
|
|
Also fixes to C backend for compiling MIPS spec to C
- Fix an issue with const correctness in internal_vector_update
functions generated by C backend
- Add builtins for MIPS to sail.h
- Fix an issue where reg_deref didn't work when called on pointers to
large bitvectors, i.e. vectors containing references to large
bitfields as in the MIPS TLB code
- Various bug fixes and changes for running U-boot on ARM model,
including for interpreter and OCaml compilation.
- Fix memory leak issues and incorrect shadowing for foreach loops
- Update C header file. Fixes memory leak in memory read/write
builtins.
- Add aux constructor to ANF representation to hold environment
information.
- Fix undefined behavior caused by optimisation left shifting uint64_t
vectors 64 or more times. Unfortunately there's more issues because
the same happens for X >> 64 right shifts. It would make sense for
this to be zero, because that would guarantee the property that ((X
>> n) >> m) == (X >> (n + m)) but we probably need to do (X >> (n -
1) >> 1) in the optimisation to ensure that we don't cause
UB. Shifting by 63 and then by 1 is well-defined, but shifting by 64
in one go isn't according to the C standard. This issue with
right-shifts only occurs for zero-length vectors, so it's not a huge
deal, but it's still annoying.
- Add versions of print_bits and print_int that print to
stderr. Follows OCaml convention of print/prerr. Should make things
more explicit. Different backends had different ideas about where
print should output to, not every backend needs to have this
(e.g. theorem prover backends don't need to print) but having both
stderr and stdout seperate and clear is useful for executable models
(UART needs to be stdout, debug messages should be stderr).
|
|
|
|
Take into account existential types when determining bounds for the loop
variable
|
|
This commit primarily changes how existential types are bound in
letbindings. Essentially, the constraints on both numeric and
existentially quantified types are lifted into the surrounding type
context automatically, so in
```
val f : nat -> nat
let x = f(3)
```
whereas x would have had type nat by default before, it'll now have
type atom('n) with a constraint that 'n >= 0 (where 'n is some fresh
type variable). This has several advantages: x can be passed to
functions expecting an atom argument, such as a vector indexing
operation without any clunky cast functions - ex_int, ex_nat, and
ex_range are no longer required. The let 'x = something() syntax is
also less needed, and is now only really required when we specifically
want a name to refer to x's type. This changes slightly the nature of
the type pattern syntax---whereas previously it was used to cause an
existential to be destructured, it now just provides names for an
automatically destructured binding. Usually however, this just works
the same.
Also:
- Fixed an issue where the rewrite_split_fun_constr_pats rewriting
pass didn't add type paramemters for newly added type variables in
generated function parameters.
- Updated string_of_ functions in ast_util to reflect syntax changes
- Fixed a C compilation issue where elements of union type
constructors were not being coerced between big integers and 64-bit
integers where appropriate
- Type annotations in patterns now generalise, rather than restrict
the type of the pattern. This should be safer and easier to handle
in the various backends. I don't think any code we had was relying
on this behaviour anyway.
- Add inequality operator to lib/flow.sail
- Fix an issue whereby top-level let bindings with annotations were
checked incorrectly
|
|
First, the specialisation of option types has been fixed by allowing
the specialisation of constructor return types - this essentially
means that a constructor, such as Some : 'a -> option('a) can get
specialised to int -> option(int), rather than int -> option('a). This
means that these constructors are treated like GADTs internally. Since
this only happens just before the C translation, I haven't put much
effort into making this very robust so far.
Second, there was a bug in C compilation for the typing of return
expressions in non-unit contexts, which has been fixed.
Finally support for vector literals that are non-bitvectors has been
added.
|
|
Can now compile RISCV. Requires some library tweaks before it'll pass any tests,
Also adds hyperlinks to wip latex output
|
|
Comment out partially working optimisation passes for now
|
|
Fixed an issue whereby an option constructor that was never
constructed, but only matched on, would cause compilation to
fail. Temporarily fixed an issue where union types that can be
entirely stack-allocated were not being treated as such, by simply
heap-allocating all unions. Need to adapt the code generator to handle
this case properly. Fixed a further small issue whereby multiple union
types would confuse the type specialisation pass.
Added a test case for compiling option types.
RISCV now generates C code, but there are still some bugs that need to
be squashed before it compile and work.
|
|
Add a flag to Sail that allows for an image of an elf file to be
dumped in a simple format using linksem, used as
sail -elf test.elf -o test.bin
This image file can then be used by a compiled C version of a sail
spec as with ocaml simply by
./a.out test.bin
|
|
Also work on making C backend compile RISC-V
|
|
Previously union types could have no-argument constructors, for
example the option type was previously:
union option ('a : Type) = {
Some : 'a,
None
}
Now every union constructor must have a type, so option becomes:
union option ('a : Type) = {
Some : 'a,
None : unit
}
The reason for this is because previously these two different types of
constructors where very different in the AST, constructors with
arguments were used the E_app AST node, and no-argument constructors
used the E_id node. This was particularly awkward, because it meant
that E_id nodes could have polymorphic types, i.e. every E_id node
that was also a union constructor had to be annotated with a type
quantifier, in constrast with all other identifiers that have
unquantified types. This became an issue when monomorphising types,
because the machinery for figuring out function instantiations can't
be applied to identifier nodes. The same story occurs in patterns,
where previously unions were split across P_id and P_app nodes - now
the P_app node alone is used solely for unions.
This is a breaking change because it changes the syntax for union
constructors - where as previously option was matched as:
function is_none opt = match opt {
Some(_) => false,
None => true
}
it is now matched as
function is_none opt = match opt {
Some(_) => false,
None() => true
}
note that constructor() is syntactic sugar for constructor(()), i.e. a
one argument constructor with unit as it's value. This is exactly the
same as for functions where a unit-function can be called as f() and
not as f(()). (This commit also makes exit() work consistently in the
same way) An attempt to pattern match a variable with the same name as
a union-constructor now gives an error as a way to guard against
mistakes made because of this change.
There is probably an argument for supporting the old syntax via some
syntactic sugar, as it is slightly prettier that way, but for now I
have chosen to keep the implementation as simple as possible.
The RISCV spec, ARM spec, and tests have been updated to account for
this change. Furthermore the option type can now be included from
$SAIL_DIR/lib/ using
$include <option.sail>
|
|
Rather than just using strings to represent literals, now use value
types from sail_lib.lem to represent them. This allows for expressions
to be evaluated at compile time, which will be useful for future
optimisations involving constant folding and propagation, and allows
the intermediate bytecode to be interpreted using the same lem
builtins that the shallow embedding uses.
To get this to work I had to tweak the build process slightly to allow
ml files to import lem files from gen_lib/. Hopefully this doesn't
break anything!
|
|
|
|
Fix some issues where some early returns in functions would cause
memory leaks, and optimize struct updates so the struct is not copied
uneccesarily.
Also make C print_bits match ocaml version output, and update tests.
|
|
With these optimisations on, now get about 10x performance over OCaml.
|
|
Fixed an issue with pattern matching on enums
Fixed an issue whereby fix_early_returns would cause memory leaks
Added optimizations for some of the builtins used in the decode
function. Optimizations are turned on with the -O flag.
|
|
Add support for short-ciruiting and/or. I forgot about this in the
original ANF specification and not having it causes problems for the
ARM spec.
|
