occheung
/
artiq
forked from M-Labs/artiq
1
0
Fork 0
Code Pull Requests Activity
8,444 Commits 14 Branches 35 Tags 23 MiB
Commit Graph

794 Commits

Author SHA1 Message Date
David Nadlinger 56a872ccc0 compiler: Insert array binop shape check in caller for location information 2020-08-09 03:54:42 +01:00
David Nadlinger ef260adca8 compiler: Implement matrix multiplication 
LLVM will take care of optimising the loops. This was still
unnecessarily painful; implementing generics and implementing
this in ARTIQ Python looks very attractive right now.
 2020-08-09 03:54:42 +01:00
David Nadlinger 0da4a61d99 compiler: Fix method name typo [nfc] 2020-08-09 03:54:42 +01:00
David Nadlinger 78afa2ea8e compiler: Support MatMult in inferencer 
Still needs actual codegen support.
 2020-08-09 03:54:42 +01:00
David Nadlinger 4d48470320 compiler: Support common numpy.* math functions 
Relies on the runtime to provide the necessary
(libm-compatible) functions.

The test is nifty, but a bit brittle; if this breaks in the
future because of optimizer changes, do not hesitate to convert
this into a more pedestrian test case.
 2020-08-09 03:54:41 +01:00
David Nadlinger d37503f21d compiler: T{C -> External}Function, clarify docs [nfc] 2020-08-09 03:54:41 +01:00
David Nadlinger da255bee1b compiler: Implement element type coercion for arrays 
So far, this is not exposed to the user beyond implicit conversions.

Note that all the implicit conversions, such as triggered by adding
arrays of mismatching types, or dividing integer arrays, are currently
emitted in a maximally inefficient way, where a temporary copy is first
made for the type conversion. The conversions would more sensibly be
implemented during the per-element operations to save on the extra
copies, but the current behaviour fell out of the rest of the IR
generator structure without extra changes.
 2020-08-09 03:54:41 +01:00
David Nadlinger 4426e4144f compiler: Implement unary plus/minus for arrays 
Implementation is needlessly generic to anticipate
coercion/transcendental functions.
 2020-08-09 03:54:41 +01:00
David Nadlinger 4d002c7934 compiler: Explain use of rpc_tag() in array ops, formatting [nfc] 2020-08-09 03:54:41 +01:00
David Nadlinger a7e855b319 compiler.types: Change invalid default value [nfc] 
This wasn't actually ever used, but was a dict instead of a set.
 2020-08-09 03:54:41 +01:00
David Nadlinger 48fb80017f compiler: Implement basic element-wise array operations 2020-08-09 03:54:41 +01:00
David Nadlinger 9af6e5747d compiler: Factor rpc_tag() out of llvm_ir_generator 2020-08-09 03:54:41 +01:00
David Nadlinger e77c7d1c39 compiler: Add inferencer support for array operations 2020-08-09 03:54:41 +01:00
David Nadlinger a9a975e5d4 language: Allow instantating TArray using bare ints 2020-08-09 03:54:41 +01:00
David Nadlinger dea3c0c572 compiler: Don't store redundant ndarray buffer length, match list layout 
This adds `elt` to _TPointer and the ir.Offset IR instruction,
which is like GetElem but without the final load.
 2020-08-09 03:54:41 +01:00
David Nadlinger cb1cadb46a compiler: Fix/test 1D array construction from generic iterables 2020-08-09 03:54:41 +01:00
David Nadlinger 38c17622cc compiler: Axis-wise iteration of ndarrays 
Matches NumPy. Slicing a TList reallocates, this doesn't; offsetting
couldn't be handled in the IR without introducing new semantics
(the Alloc kludge; could/should be made its own IR type).
 2020-08-09 03:54:41 +01:00
David Nadlinger c95a978ab6 compiler: Iteration for 1D ndarrays 2020-08-09 03:54:41 +01:00
David Nadlinger bc17bb4d1a compiler: Parametrize TArray in number of dimensions 2020-08-09 03:54:41 +01:00
David Nadlinger 632c5bc937 compiler: Add ndarray .shape access 2020-08-09 03:54:41 +01:00
David Nadlinger d882f8a3f0 compiler: Implement len() for ndarrays 2020-08-09 03:54:41 +01:00
David Nadlinger 575be2aeca compiler: Basic support for creation of multidimensional arrays 
Breaks all uses of array(), as indexing is not yet implemented.
 2020-08-09 03:54:41 +01:00
David Nadlinger 56010c49fb compiler/inferencer: Detect rectangular array()s 
Still needs support through all the rest of the compiler, and
support for higher-dimensional arrays.

Alternatively, we could always assume ndarrays of ndarrays
are rectangular (i.e. ban array/list element types), and
detect mismatch at runtime. This might turn out to be
preferrable to be able to construct matrices from rows/columns.

`array()` is disallowed for no particularly good reason but
numpy API compatibility.
 2020-08-09 03:54:41 +01:00
pca006132 3bfd372c20 compiler: linker discard local symbols. 
Fixes exception backtrace problem for ARM.
 2020-08-06 16:07:28 +08:00
David Nadlinger ae999db8f6 compiler: Revert function call lifetime tracking fix 
This reverts commits f8d1506922
and cf19c9512d.

While the commit just fixes a clear typo in the implementation,
it turns out the original algorithm isn't flexible enough to
capture functions that transitively return references to
long-lived data. For instance, while cache_get() is special-cased
in the compiler to be recognised as returning a value of Global()
lifetime, a function just forwarding to it (as seen in the
embedding tests) isn't anymore.

A separate issue is also that this makes implementing functions
that take lists and return references to global data in user code
impossible, which central parts of the Oxford codebase rely on.

Just reverting for now to unblock master; a fix is easily designed,
but needs testing.
 2020-07-30 16:40:39 +01:00
David Nadlinger f8d1506922 compiler: Fix lifetime tracking for function call return values 
GitHub: Fixes #1497.
 2020-07-28 00:33:28 +01:00
David Nadlinger 1c72585c1b compiler: Handle None-returning function calls used as values 
GitHub: Fixes #1493.
 2020-07-25 02:20:53 +01:00
David Nadlinger 57e759a1ed compiler: Consistently use llunit through llvm_ir_generator [nfc] 2020-07-25 02:20:52 +01:00
Sebastien Bourdeauducq 95807234d9 compiler: use binutils for ARM 
This is mostly due to Windoze, where installing anything is a PITA and the LLVM tools won't be available soon.
 2020-06-28 17:33:03 +08:00
Sebastien Bourdeauducq 85e92ae28c compiler: use more LLVM tools on ARM (#733) 2020-04-28 16:21:50 +08:00
Sebastien Bourdeauducq 140a26ad7e compiler: ld -> ld.lld 2020-04-28 16:07:26 +08:00
Sebastien Bourdeauducq 4228e0205c compiler: link with lld on ARM (#733) 2020-04-28 15:00:24 +08:00
Sebastien Bourdeauducq 251a0101a6 compiler: support disabling now-pinning 2020-04-26 12:38:43 +08:00
David Nadlinger d8c81d6d05 compiler: Other types microoptimisations 
Interestingly enough, these actually seem to give a measurable
speedup (if small – about 1% improvement out of 6s whole-program
compile-time in one particular test case).

The previous implementation of is_mono() had also interesting
behaviour if `name` wasn't given; it would test only for the
presence of any keys specified via keyword arguments,
disregarding their values. Looking at uses across the current
ARTIQ codebase, I could neither find a case where this would
have actually been triggered, nor any rationale for it.

With the short-circuited implementation from this commit,
is_mono() now checks name/all of params against any specified
conditions.
 2020-01-01 08:49:19 +00:00
David Nadlinger 2c34f0214b compiler: Short-circuit Type.unify() with identical other type 
This considerably improves performance; ~15% in terms of total
artiq_run-to-kernel-compiled duration in one test case.
 2020-01-01 08:49:19 +00:00
David Nadlinger 8f518c6b05 compiler: Allow `None` in type hints 
Similar to how Python itself interprets None as type(None),
make it translate to TNone in ARTIQ compiler type hints.
 2019-12-19 09:36:45 +08:00
David Nadlinger 594ff45750 compiler: Revert support for `None` as `TNone` 
This was mistakenly included in fb2b634c4a, and broke the test
case verifying that using None as an ARTIQ type annotation in fact
generates an error message.
 2019-12-18 13:23:40 +00:00
David Nadlinger fb2b634c4a compiler, language: Implement @kernel_from_string 
With support for polymorphism (or type erasure on pointers to
member functions) being absent in the ARTIQ compiler, code
generation is vital to be able to implement abstractions that
work with user-provided lists/trees of objects with uniform
interfaces (e.g. a common base class, or duck typing), but
different concrete types.

@kernel_from_string has been in production use for exactly
this use case in Oxford for the better part of a year now
(various places in ndscan).

GitHub: Fixes #1089.
 2019-12-18 10:51:04 +08:00
David Nadlinger 611bcc4db4 compiler: Cache expensive target data layout queries 
On one typical experiment, this shaves about 3 seconds (~25%)
off the overall kernel compilation time.

GitHub: Closes #1370.
 2019-10-28 11:09:25 +00:00
David Nadlinger 5d7f22ffa4 compiler: Remove provision for unused four-parameter llptr_to_var() form [nfc] 
`var_type` was presumably intended to convert to a target type,
but wasn't actually acted on in the function body (nor was it
used anywhere in the codebase).
 2019-10-28 11:02:46 +00:00
Sebastien Bourdeauducq c03c35f375 Revert "compiler: armv7-unknown-linux-gnueabihf -> armv7-linux-gnueabihf" 
rustc insists on -unknown.

This reverts commit cf47fa44d8.
 2019-08-26 11:23:00 +08:00
Sebastien Bourdeauducq cf47fa44d8 compiler: armv7-unknown-linux-gnueabihf -> armv7-linux-gnueabihf 2019-08-26 11:12:49 +08:00
Sebastien Bourdeauducq 98cd9a539c compiler: support Cortex A9 target 2019-08-26 10:46:22 +08:00
David Nadlinger cdaf554736 compiler: Implement abs() for scalars 
GitHub: Fixes #1303.
 2019-04-13 13:39:16 +08:00
David Nadlinger 88fd5c8440 compiler: Fix crash in escape analysis for assigning string literals 2019-03-31 17:10:27 +01:00
David Nadlinger 990e0b7dd9 compiler: Fix comparison of tuples of lists 2019-03-31 17:10:27 +01:00
David Nadlinger baf102dbb2 compiler: Fix comparison of nested lists 2019-03-31 17:10:27 +01:00
David Nadlinger f9af058b96 compiler: Quote tuples as TTuple values 
Previously, they would end up being TInstances,
rendering them useless.
 2019-03-31 23:40:21 +08:00
whitequark d6eb2b023a compiler: monomorphize casts first, but more carefully. 
This reverts 425cd7851, which broke the use of casts to define
integer width.

Instead of it, two steps are taken:
  * First, literals are monomorphized, leading to predictable result.
  * Second, casts are monomorphized, in a top-bottom way. I.e.
    consider the expression `int64(round(x))`. If round() was visited
    first, the intermediate precision would be 32-bit, which is
    clearly undesirable. Therefore, contextual rules should take
    priority over non-contextual ones.

Fixes #1252.
 2019-02-07 06:24:32 +00:00
whitequark 425cd7851e compiler: first monomorphize ints, then casts. 
Fixes #1242.
 2019-01-12 13:40:12 +00:00
whitequark 49682d0159 Improve Python 3.7 compatibility. 
async is now a full (non-contextual) keyword.

There are two more instances:
   - artiq/frontend/artiq_client.py
   - artiq/devices/thorlabs_tcube/driver.py

It is not immediately clear how to fix those, so they are left for
later work.
 2019-01-12 13:17:59 +00:00
Sebastien Bourdeauducq 421834fa3e compiler: document Target.little_endian 2018-12-02 19:07:18 +08:00
Sebastien Bourdeauducq 981a77834a compiler: use default triple to determine data_layout for JIT 2018-12-02 18:52:13 +08:00
Sebastien Bourdeauducq d931967e5c fix previous commits 2018-12-02 18:32:03 +08:00
Sebastien Bourdeauducq dd03fdfd1a typo 2018-12-02 18:26:54 +08:00
Sebastien Bourdeauducq 8940009e1a compiler: pass data_layout string to llvm.create_target_data before determining endianness 2018-12-02 18:26:19 +08:00
Sebastien Bourdeauducq 2e66788c6c compiler: support little endian target when storing now 2018-12-02 17:40:34 +08:00
Sebastien Bourdeauducq 7e14f3ca4e compiler,gateware: atomic now stores 2018-12-02 05:06:46 +08:00
Sebastien Bourdeauducq 8caea0e6d3 gateware,runtime: optimize RTIO kernel interface further 
* now pinning (TODO: atomicity)
* for inputs, merge request and timeout registers
 2018-11-08 18:29:24 +08:00
David Nadlinger 2463e5667d compiler: Fix attribute writeback with skipped fields 
offset wasn't advanced for skipped fields previously,
leading to memory corruption/unaligned accesses at runtime.
 2018-08-14 13:34:32 +01:00
whitequark fab6e5cdff compiler: skip functional values in attribute writeback. 
Fixes #1088.
 2018-08-10 12:02:49 +00:00
whitequark 93af5d2a03 compiler: handle async RPC as last statement in try block. 
Fixes #1107.
 2018-08-07 07:06:53 +00:00
whitequark a74958f01f ksupport: raise RuntimeError on reraise with no inflight exception. 
Fixes #1123.
 2018-08-07 05:53:13 +00:00
David Nadlinger 768b970deb Fixup 4359a437 (tuples of lists), add regression tests 2018-07-10 01:18:51 +01:00
David Nadlinger 4359a43732 compiler: Indirection status of TTuple depends on elements 
For instance, TTuple(TList(TInt32())) has indirections, while
TTuple(TInt32()) does not.

This fixes memory corruption with RPCs that return tuples of lists.

Signed-off-by: David Nadlinger <code@klickverbot.at>
 2018-07-09 18:49:50 +08:00
Sebastien Bourdeauducq 9153c4d8a3 use tokenize.open() to open Python source files 
Fixes encoding issues especially with device databases modified in obscure editors.
 2018-07-07 17:04:56 +08:00
whitequark 9260cdb2e8 compiler: support conversion of list to bytearray and bytes. 
Fixes #1077.
 2018-06-21 00:40:45 +00:00
whitequark 38dac16041 compiler: don't crash when quoting builtin functions. 
Fixes #1051.
 2018-06-05 23:27:23 +00:00
whitequark db4d1878d3 compiler: remove debug print. 2018-05-25 09:37:18 +00:00
whitequark fbf2c9a2fb compiler: hoist loads of kernel invariants to function entry block. 
Addresses intraprocedural cases of #1007.
 2018-05-25 02:18:13 +00:00
whitequark 12d1b9819c compiler: handle direct calls to class methods. 
Fixes #1005.
 2018-05-25 02:02:18 +00:00
whitequark ce3c8fcd3e compiler: add source location to SSA arguments. 
Fixes #1006.
 2018-05-22 18:14:14 +00:00
whitequark 9b4ad8b5af compiler: implement local variable demotion. 2018-05-19 17:05:34 +00:00
whitequark fd110e9848 compiler: sort predecessors in IR dump output. 
Makes diffs more useful.
 2018-05-19 17:02:08 +00:00
whitequark dc1c4ebf1d compiler: use ARTIQ_IR_NO_LOC variable to control IR dump output. 
Useful e.g. in tests or for running diff on irgen output.
 2018-05-19 17:02:08 +00:00
whitequark 21a0f26cd1 compiler: do not emit nonstandard !unconditionally* LLVM metadata. 
This metadata was added by an LLVM patch that was removed because
it was unsound.
 2018-05-19 16:58:03 +00:00
whitequark 3b61b7c30b compiler: fix tests after 4d06c1d8. 2018-05-16 21:53:53 +00:00
whitequark 917a4185eb Unbreak 9dc7efef. 2018-05-05 03:00:30 +00:00
whitequark 9dc7efefe4 compiler: transparently handle Windows newlines in RunTool. 2018-05-05 01:43:00 +00:00
whitequark f7e08ec46b Unbreak f35f1001. 2018-04-28 01:03:08 +00:00
whitequark f35f100110 compiler: don't crash printing locations of specialized functions. 
Fixes #987.
 2018-04-28 00:49:25 +00:00
whitequark 58967f14fd compiler: do not try to re-coerce fully coerced numerics. 2018-04-21 18:25:52 +00:00
whitequark b4e3c30d8c compiler: desugar x != y into not x == y (fixes #974). 2018-04-20 12:20:45 +00:00
whitequark 4b5a78e231 compiler: do not pass files to external tools while they are opened. 
This fixes access violations on Windows that are present both with
input and output files. For some reason, Cygwin-compiled binutils
did not exhibit this problem, but MSYS-compiled binutils do.

Fixes #961.
 2018-03-15 22:21:29 +00:00
whitequark 86ceee570f compiler: reject calls with unexpected keyword arguments. 
Fixes #924.
 2018-02-21 11:37:12 +00:00
whitequark 34a5445802 compiler: don't die if addr2line cannot extract line from backtrace. 
Fixes #885.
 2018-01-08 21:53:04 +00:00
whitequark 4d915ad15b compiler: do not permit str(...). (#878) 2017-12-27 03:10:50 +00:00
whitequark 355acb5e13 compiler: fix typo in a0a2650f. 2017-12-26 21:55:22 +00:00
whitequark 9aec64093b compiler: do not ever emit !tbaa on invoke instructions. 2017-12-26 21:35:08 +00:00
whitequark b31b59f7ca compiler: update for llvmlite 0.20.0. 2017-12-26 16:25:19 +00:00
whitequark a0a2650fca compiler: do not use invoke for calls to nounwind ffi functions. 
Otherwise, declarations such as:

  @syscall(flags={"nounwind", "nowrite"})
  def foo(...):

trip an LLVM assert because the invoke instruction and the !tbaa
metadata are no longer compatible since LLVM 4.0.
 2017-12-26 16:24:57 +00:00
whitequark c939c6183e Update to LLVM 4.0. 2017-12-26 14:09:17 +00:00
Sebastien Bourdeauducq 4387b0be1e clean up rtio_log 2017-11-03 00:52:53 +08:00
whitequark ca254ec55e compiler: disallow op= on mutable lists only (fix #835). 
This only really applies to lists since those use fat pointers.
`x.y += z` is fine.
 2017-10-02 07:21:14 +00:00
whitequark e59b301a3b compiler: disallow op= on mutable values (fix #835). 
In general, we can't reallocate a value in earlier stack frames,
or fixup references to it. This mainly impacts lists.
 2017-10-02 01:55:26 +00:00
whitequark f86744c65c compiler: implement ~ operator (fix #836). 2017-10-01 19:33:18 +00:00
whitequark 843786a091 compiler: minor intrinsic refactoring. 2017-10-01 19:00:10 +00:00
whitequark 491c7ef898 compiler: correct semantics of floating point % operator (fix #830). 2017-10-01 18:57:45 +00:00
whitequark 62f2693e36 compiler: correct semantics of integer % operator (#830). 2017-10-01 18:32:57 +00:00
whitequark ea7549cfa4 compiler: coerce `while` condition to bool. 
Fixes #768.
 2017-07-01 18:59:07 +00:00
whitequark f7254dd3ce compiler.validators.constness: take AugAssign into account. 2017-06-09 07:31:08 +00:00
whitequark ad2ee714c2 compiler: do not permit iterating str values. 
This currently breaks badly on UTF-8, and doesn't even return
a value of a correct type.
 2017-06-09 07:29:31 +00:00
whitequark 4e7493843a compiler: Constness is a validator, not analysis. 2017-06-09 07:29:31 +00:00
whitequark 5d841d08e9 compiler: do not permit mutation of bytes values (#714). 2017-06-09 07:29:28 +00:00
whitequark 284382b1f5 compiler: add support for bytearray values in RPC (#714). 2017-06-09 07:15:25 +00:00
whitequark 9ed4e9c1cd compiler: add support for printing of bytearray values (#714). 2017-06-09 07:15:25 +00:00
whitequark e9564b15c8 compiler: add support for bytearray type (#714). 2017-06-09 07:15:24 +00:00
whitequark 5b4fde30a8 compiler: unbreak subscripts for bytes values (#714). 2017-06-09 07:10:48 +00:00
whitequark 66a683f583 compiler: add support for bytes values in RPC (#714). 2017-06-09 07:10:48 +00:00
whitequark 778e7dc2ab compiler: add support for concatenating bytes values (#714). 2017-06-09 07:10:48 +00:00
whitequark 7b2da5294f compiler: add support for printing of bytes values (#714). 2017-06-09 07:10:48 +00:00
whitequark dba4e1a28b compiler: add support for bytes type and b"x" literals (#714). 2017-06-09 07:10:48 +00:00
Sebastien Bourdeauducq cd757c0f16 generate device database from executable python file 2017-05-18 23:14:55 +08:00
whitequark bb64992395 compiler: remove dead code. 2017-04-21 18:38:43 +00:00
whitequark c5d7445973 compiler: reject reachable implicit return if not returning TNone. 
Fixes #718.
 2017-04-21 18:11:14 +00:00
whitequark ed2b10c5aa compiler: in codegen for delay(), round fp instead of truncating. 
Consider delay(8*us). It results in the following computation...
  >>> 8*1e-06/1e-09
  7999.999999999999
with the result promptly getting truncated to 7999.

Fixes #706.
 2017-04-21 17:36:44 +00:00
whitequark 3a1fc729cf compiler: refactor type annotations recognizing in kernels. 
The new implementation is much more generic, more robust,
and shares code with the same for syscalls as well as RPCs.

Fixes #713.
 2017-04-13 08:26:10 +00:00
whitequark 31048f4b6a compiler: fix monomorphization of coerced integer literals. 
Fixes #703.
 2017-04-12 04:11:47 +00:00
whitequark 1bd4d13391 artiq_compile: make print() write to core log, not an invalid op. 
Fixes #710.
 2017-04-11 03:16:34 +00:00
whitequark 86dea869b3 compiler.types: add missing TTuple.__hash__ implementation. 2017-03-03 03:27:49 +00:00
whitequark 7e886ddc89 transforms.inferencer: do not unnecessarily mutate typedtree. 
This caused the hash to change after every iteration of inference,
if the result of any coercion was an integer of indefinite width.
 2017-03-02 16:15:37 +00:00
whitequark 8c9f157563 compiler.embedding: fix an overly lax hasher. 2017-03-02 15:28:24 +00:00
whitequark a79c3c2cff compiler.transforms: implement a typedtree printer. 2017-03-02 15:28:04 +00:00
whitequark 4107938fd8 compiler.embedding: use attribute count in fixpoint condition. 
Before this commit, we had a hack where inferencing would run
one more time, in case we can still infer more attributes even
after all the type information is there. This doesn't work
if *that* round of inferencing brings even more code, e.g.
this is easy to trigger with context managers.
 2017-02-28 21:28:40 +00:00
whitequark 252cda6e7f compiler.llvm_ir_generator: fix string quoting. 2017-02-28 21:28:38 +00:00
whitequark de015b994d compiler: allow dumping the object file with ARTIQ_DUMP_OBJ. 2017-02-26 17:09:21 +00:00
whitequark 3a1f14c16c compiler: fix overly strict constness analysis. 
Before this commit, the following code would fail to compile...
  obj.foo.bar = True
... if foo is marked kernel_invariant in obj, even if bar is not
marked as such in obj.foo.
 2017-02-26 01:58:21 +00:00
whitequark d04e611232 firmware, compiler: rename rpc functions to be more consistent. 2017-02-25 14:12:58 +00:00
whitequark c39394b4d5 firmware: port the exception handling routines to Rust. 2017-02-04 08:21:07 +00:00
whitequark fd8b11532f compiler, firmware: use Pascal strings everywhere. 
This removes a large amount of very ugly code, and also simplifies
the compiler and runtime.
 2017-02-03 11:53:27 +00:00
whitequark ee1b2fa3ea compiler: swap the order of ptr/len fields in lists. 
This is to make them match the Rust slices.
Once the cslice crate becomes #![no_std] we'll also use its (tidier)
conversion functions.
 2017-01-31 22:11:13 +00:00
whitequark ab71c9a0ba compiler: the len builtin is not polymorphic, coerce the result. 
Fixes #659.
 2017-01-31 21:28:20 +00:00
whitequark 6acdcbb82a Fix 2f37b1d. 2017-01-30 10:25:11 +00:00
whitequark 2f37b1d1c0 compiler: support methods defined on singleton instances. 
Fixes #638.
 2017-01-30 10:10:21 +00:00
whitequark 57f54dbd12 llvm_ir_generator: recognize inline and forceinline flags. 2017-01-25 10:25:00 +00:00
whitequark f68e4ae519 compiler: rein in overzealous cast monomorphization. 2016-12-05 05:08:57 +00:00
whitequark 218720cfa7 Revert "compiler: rein in overzealous cast monomorphization." 
This reverts commit 4305903dde.

This broke the monomorphizer/round.py test.
 2016-12-05 05:04:26 +00:00
whitequark 4305903dde compiler: rein in overzealous cast monomorphization. 
This caused failures on e.g. "int32(var64a >> var64b)", where
the type of the argument is already fully known, but was unified
with the result of the cast anyway.
 2016-12-05 04:38:25 +00:00
whitequark b5a684830d compiler: fix parsing of TList annotations (fixes #632). 2016-12-05 03:18:56 +00:00
whitequark 68de724554 compiler: monomorphize int64(round(x)) to not lose precision. 
This applies to any expression with an indeterminate integer type
cast to int64(), not just round().
 2016-12-02 15:02:44 +00:00
whitequark ac997daf95 compiler: disable remarks. 2016-11-21 19:12:11 +00:00
whitequark 55ea68da7f compiler: unbreak casts to int32/int64. 2016-11-21 14:20:28 +00:00
whitequark 53b7d59b6a analyses.constness: fix false positive on x[...]. 2016-11-21 14:20:26 +00:00
whitequark 35f4449ef2 inferencer: significantly improve the op-assignment diagnostic. 
Before this commit, it displayed incorrect output if an error
appeared on 2nd run and beyond, and displayed messages for trying
to do "num32 -= num64" that made very little sense.
 2016-11-21 14:20:24 +00:00
whitequark 009d396740 Move mu_to_seconds, seconds_to_mu to Core. 2016-11-21 05:37:30 +00:00
whitequark 7af41bd29c llvm_ir_generator: handle no-op coercions. 2016-11-21 02:25:34 +00:00
whitequark cdb29f9caa Revert accidentally committed code. 2016-11-20 14:32:59 +00:00
whitequark f5cca6b09e analyses.invariant_detection: implement (#622). 2016-11-20 12:48:26 +00:00
whitequark d3ee858d16 llvm_ir_generator: use !{→unconditionally.}invariant.load metadata. 
This helps LICM, among other things.
 2016-11-12 04:08:58 +00:00
whitequark 0e76cbc414 artiq_compile: actually disable attribute writeback. 
I wrote both halves of this condition but forgot to hook
them together.

Fixes #586.
 2016-11-10 01:04:36 +00:00