M-Labs/artiq
2
0
Fork 2
mirror of https://github.com/m-labs/artiq.git synced 2024-12-23 10:24:01 +08:00
Code Issues Projects Releases Wiki Activity
2,342 Commits 13 Branches 35 Tags 26 MiB
ff0ab736e9
Commit Graph

206 Commits

Author SHA1 Message Date
whitequark
82b470891f transforms.interleaver: handle function calls (as atomic so far). 
This commit solves issue #2 described in 50e7b44; a function call
is now a valid decomposition for a delay instruction, and this
metadata is propagated when the interleaver converts delays.

However, the interleaver does not yet detect that a called function
is compound, i.e. it is not correct.
 2015-11-21 03:34:24 +08:00
whitequark
57dd163d37 transforms.artiq_ir_generator: fix decomposition of explicit delay_mu(). 2015-11-21 03:27:06 +08:00
whitequark
cb3b811fd7 compiler: maintain both the IR and iodelay forms of delay expressions. 
After this commit, the delay instruction (again) does not generate
any LLVM IR: all heavy lifting is relegated to the delay and delay_mu
intrinsics. When the interleave transform needs to adjust the global
timeline, it synthesizes a delay_mu intrinsnic. This way,
the interleave transformation becomes composable, as the input and
the output IR invariants are the same.

Also, code generation is adjusted so that a basic block is split off
not only after a delay call, but also before one; otherwise, e.g.,
code immediately at the beginning of a `with parallel:` branch
would have no choice but to execute after another branch has already
advanced the timeline.

This takes care of issue #1 described in 50e7b44 and is a step
to solving issue #2.
 2015-11-21 03:22:47 +08:00
whitequark
50e7b44d04 compiler: actually implement interleaving correctly (calls are still broken). 
The previous implementation was completely wrong: it always advanced
the global timeline by the same amount as the non-interleaved basic
block did.

The new implementation only advances the global timeline by
the difference between its current time and the virtual time of
the branch, which requires it to adjust the delay instructions.

Previously, the delay expression was present in the IR twice: once
as the iodelay.Expr transformation-visible form, and once as regular
IR instructions, with the latter form being passed to the delay_mu
builtin and advancing the runtime timeline.

As a result of this change, this strategy is no longer valid:
we can meaningfully mutate the iodelay.Expr form but not the IR
instruction form. Thus, IR instructions are no longer generated for
delay expressions, and the LLVM lowering pass now has to lower
the iodelay.Expr objects as well.

This works OK for flat `with parallel:` expressions, but breaks down
outside of `with parallel:` or when calls are present. The reasons
it breaks down are as follows:

  * Outside of `with parallel:`, delay() and delay_mu() must accept
    any expression, but iodelay.Expr's are not nearly expressive
    enough. So, the IR instruction form must actually be kept as well.

  * A delay instruction is currently inserted after a call to
    a user-defined function; this delay instruction introduces
    a point where basic block reordering is possible as well as
    provides delay information. However, the callee knows nothing
    about the context in which it is called, which means that
    the runtime timeline is advanced twice. So, a new terminator
    instruction must be added that combines the properties of delay
    and call instructions (and another for delay and invoke as well).
 2015-11-21 00:02:47 +08:00
whitequark
73c358a59a Reformat. 2015-11-20 23:33:06 +08:00
whitequark
cc623c13b4 Reformat. 2015-11-20 21:45:34 +08:00
whitequark
d0f86e05d0 transforms.interleaver: add safety check. 2015-11-20 17:27:04 +08:00
whitequark
88b7990714 transforms.iodelay_estimator: fail statements with indeterminate delay inside with parallel. 2015-11-20 17:10:25 +08:00
whitequark
00ec574d73 transforms.interleaver: implement (without inlining). 2015-11-20 00:03:26 +08:00
whitequark
025bfbe746 transforms.llvm_ir_generator: accept delay instructions. 
The delay instruction is just like a branch (discontinuity
in instruction flow), but it also carries metadata: how long
did the execution of its basic block take. This metadata only
matters during inlining and interleaving, so we treat it here
as a mere branch.
 2015-11-19 23:55:52 +08:00
whitequark
9639a831bc transforms.artiq_ir_generator: correctly emit IfExpT with control flow. 
This can happen with nested if expressions, as well as if
the if expression includes delays.
 2015-11-19 23:44:39 +08:00
whitequark
2543daa5cf transforms.artiq_ir_generator: don't emit delay instruction for zero delay. 
Call nodes with iodelay=Const(0) can be generated outside of
`with parallel:`, where Interleaver won't and LLVMIRGenerator can't
lower them.
 2015-11-19 23:24:46 +08:00
whitequark
58db347e01 transforms.iodelay_estimator: fix uninitialized access. 2015-11-19 23:23:39 +08:00
whitequark
48a2bb10d5 transforms.interleaver: add boilerplate. 2015-11-17 05:22:20 +03:00
whitequark
de9d7eb2e4 compiler: add delay IR instruction. 2015-11-17 05:16:43 +03:00
whitequark
e619154c81 transforms.iodelay_estimator: fix handling of with sequential. 2015-11-17 01:22:48 +03:00
whitequark
506725f78a transforms.iodelay_estimator: fix handling of methods. 2015-11-17 01:19:22 +03:00
whitequark
b0c6b70971 transforms.asttyped_rewriter: fix class redefinition diagnostic. 2015-11-17 01:17:25 +03:00
whitequark
9e0a5b9404 transforms.iodelay_estimator: skip statements, not modules on _UnknownDelay. 2015-11-17 01:00:10 +03:00
whitequark
44d0a35409 transforms.iodelay_estimator: actually iterate to fixpoint. 2015-11-17 00:59:40 +03:00
whitequark
10f82ff2c8 transforms.iodelay_estimator: do not unify indeterminate delays. 2015-11-17 00:59:09 +03:00
whitequark
eefa9e2ea6 transforms.inferencer: fix typo. 2015-11-17 00:02:15 +03:00
whitequark
b91ffa1b38 ir: fix default argument fiasco. 2015-10-14 17:02:59 +03:00
whitequark
5d64df829e transforms.artiq_ir_generator: fix devirtualized call annotation. 2015-10-14 16:12:00 +03:00
whitequark
32ce33a1f9 transforms.artiq_ir_generator: emit ir.Parallel for with parallel:. 2015-10-09 03:10:39 +03:00
whitequark
48f1f48f09 transforms.artiq_ir_generator: devirtualize method calls. 2015-10-09 02:27:52 +03:00
whitequark
0bb793199f transforms.artiq_ir_generator: devirtualize closure calls. 2015-10-09 01:32:27 +03:00
whitequark
7043b333a7 transforms.devirtualizer → analyses.devirtualization. 2015-10-09 00:29:31 +03:00
whitequark
962dd9de13 transforms.devirtualizer.MethodResolver: implement. 2015-10-07 22:33:08 +03:00
whitequark
6ac82e1439 transforms.devirtualizer.FunctionResolver: implement. 2015-10-07 22:33:00 +03:00
whitequark
42b0089a4c transforms.asttyped_rewriter: simplify. 2015-10-07 22:03:24 +03:00
whitequark
7a6fc3983c Make delay component of function type unifyable. 2015-09-30 18:41:14 +03:00
whitequark
60c985bf0b Remove debug print. 2015-09-27 17:55:19 +03:00
whitequark
867a0689ad transforms.Inferencer: narrow range() element type. 2015-09-02 17:55:48 -06:00
whitequark
b971cc8cdf compiler.{iodelay,transforms.iodelay_estimator}: implement. 2015-09-02 17:55:48 -06:00
whitequark
156779007a LLVMIRGenerator: implement quoting of lists. 2015-08-31 23:33:33 -06:00
whitequark
956c1985b1 ARTIQIRGenerator: evaluate SubscriptT index in non-assignment context. 2015-08-31 23:33:04 -06:00
whitequark
3ca5967cea LLVMIRGenerator: don't map sret LLVM argument to any ARTIQ arguments. 2015-08-31 14:02:02 -06:00
whitequark
5e0ec3a6ea ARTIQIRGenerator: fix keyword/optional argument codegen in calls. 2015-08-31 13:42:36 -06:00
whitequark
501ba912c2 Implement {delay,now,at}{,_mu} and {mu,seconds}_to_{seconds,mu}. 2015-08-31 09:59:33 -06:00
whitequark
2124ff9e91 Fix tests. 2015-08-28 05:28:58 -05:00
whitequark
37811f690b ARTIQIRGenerator: support comparisons against None. 2015-08-28 05:14:52 -05:00
whitequark
c621b1f275 compiler: handle language.core.int during embedding. 2015-08-28 03:24:15 -05:00
whitequark
83ebb999c8 transforms.Inferencer: improve attribute unification diagnostic. 2015-08-28 03:23:15 -05:00
whitequark
9fd25a1cc4 LLVMIRGenerator: fix syscall emission. 2015-08-28 03:03:12 -05:00
whitequark
9936768603 ARTIQIRGenerator: fix non-nullary method calls. 2015-08-28 02:11:05 -05:00
whitequark
c21387dc09 compiler.embedding: support calling methods marked as @kernel. 2015-08-27 19:46:50 -05:00
whitequark
84e32db622 LLVMIRGenerator: handle self-referential class types. 2015-08-27 15:40:46 -05:00
whitequark
71ebe1778d LLVMIRGenerator: remove debug print. 2015-08-27 15:40:15 -05:00
whitequark
c62b16d5e1 compiler.embedding: support RPC functions as host attribute values. 2015-08-27 05:53:18 -05:00
whitequark
cb225269ff Allow accessing attributes of embedded host objects. 2015-08-27 05:01:04 -05:00
whitequark
9b9fa1ab7c Allow embedding and RPC sending host objects. 2015-08-25 21:56:01 -07:00
whitequark
526d7c4e46 Fix a typo. 2015-08-22 14:01:55 -07:00
whitequark
a557445e05 LocalAccessValidator: assume variables with "$" in name are internal. 
Internal variables are assumed to be scoped correctly
(including not being accessed uninitialized).

This was changed from "." because artiq.compiler.embedding uses
"." in module prefix of function names.
 2015-08-22 13:56:17 -07:00
whitequark
0e26cfb66e LocalAccessValidator: relax restrictions to accept def f(); def g(). 2015-08-22 13:31:09 -07:00
whitequark
b39e76ae28 Remove debug print. 2015-08-22 12:22:26 -07:00
whitequark
27a697920a LLVMIRGenerator: use sret when returning large structures. 2015-08-19 15:06:03 -07:00
whitequark
afc3f36104 ARTIQIRGenerator: fix polymorphic print on closures. 2015-08-19 12:37:22 -07:00
whitequark
53b4d87647 LLVMIRGenerator: attach debug metadata to all emitted LLVM instructions. 2015-08-19 12:18:20 -07:00
whitequark
51c591f01a Unbreak tests. 2015-08-18 22:44:09 -07:00
whitequark
6c8de9b6d4 Implement methods. 2015-08-18 22:39:22 -07:00
whitequark
94a2d5f5fa Implement class attribute access through instances. 2015-08-15 11:07:54 -04:00
whitequark
00efc8c636 Implement class definitions and class attribute access. 2015-08-15 09:45:16 -04:00
whitequark
786fde827a Unbreak tests. 2015-08-11 00:41:31 +03:00
whitequark
c72267ecf5 Implement syscalls for the new compiler. 2015-08-10 20:26:06 +03:00
whitequark
b28a874274 Inferencer: range() does not accept a float argument. 2015-08-10 17:06:55 +03:00
whitequark
f53a5ff202 Remove syscall builtin. 2015-08-10 16:44:29 +03:00
whitequark
c63ec70c53 LLVMIRGenerator: emit debug information. 2015-08-10 15:11:52 +03:00
whitequark
8f510a4407 compiler.ir.Function: add loc field. 2015-08-10 13:14:52 +03:00
whitequark
22570afbda LLVMIRGenerator: allocate less. 2015-08-10 09:12:34 +03:00
whitequark
dfc91a35f2 ARTIQIRGenerator.polymorphic_print: str([x]) uses repr(x), not str(x). 2015-08-09 20:27:04 +03:00
whitequark
f7b64db8f4 LLVMIRGenerator: fixup phis on expansion of ARTIQ instructions. 2015-08-09 20:24:16 +03:00
whitequark
d4270cf66e Implement receiving data from RPCs. 2015-08-09 20:17:00 +03:00
whitequark
02b1543c63 Implement receiving exceptions from RPCs. 2015-08-09 16:16:41 +03:00
whitequark
9c5ca2ae29 LLVMIRGenerator: add target data layout to LLVM modules. 2015-08-09 14:39:21 +03:00
whitequark
153592f1cc Naming. 2015-08-09 02:25:58 +03:00
whitequark
b26af5df60 Implement sending RPCs. 2015-08-09 02:17:19 +03:00
whitequark
22457bc19c Ensure uwtable is added to all generated functions. 2015-08-08 21:48:21 +03:00
whitequark
ee3f35c608 Improve error message on passing an argument twice. 2015-08-08 21:06:13 +03:00
whitequark
bdcf7f100b ARTIQIRGenerator: add semantic locs to all other implicitly raised exceptions. 2015-08-08 16:26:53 +03:00
whitequark
acd8d6355f transforms.ARTIQIRGenerator: IndexError loc should point to "[". 2015-08-08 16:18:57 +03:00
whitequark
27d2390fed Add zero-cost exception support to runtime and host. 2015-08-08 16:01:31 +03:00
whitequark
a7633f75c7 Show origin of expansion in diagnostics for synthesized code. 2015-08-07 13:56:18 +03:00
whitequark
50448ef554 Add support for referring to host values in embedded functions. 2015-08-07 13:24:01 +03:00
whitequark
353f454a29 Add basic support for embedded functions with new compiler. 2015-08-07 11:44:49 +03:00
whitequark
1a969aa9e4 compiler.transforms.inferencer: accept and ignore @kernel decorator. 2015-08-07 07:54:35 +03:00
whitequark
7562d39750 compiler.module: split off inferencing from Module.__init__. 2015-08-06 08:25:53 +03:00
whitequark
ca52b2fdd0 compiler.transforms.ARTIQIRGenerator: fix typo. 2015-08-06 08:25:53 +03:00
whitequark
e8c107925c Implement shared object linking. 2015-07-29 20:35:16 +03:00
whitequark
3378dd57b8 Fold llvmlite patches into m-labs/llvmlite repository. 2015-07-29 13:54:00 +03:00
whitequark
2939d4f0f3 Add tests for finally clause and reraising. 2015-07-27 12:36:21 +03:00
whitequark
a83e7e2248 Add tests for exceptional control flow. 2015-07-27 10:22:28 +03:00
whitequark
90be44c596 Add tests for non-exceptional control flow across finally. 2015-07-27 10:13:22 +03:00
whitequark
7c77dd317a Implement __artiq_personality. 2015-07-27 09:10:20 +03:00
whitequark
47f13bf921 Always load the personality library in JIT testbench, if available. 2015-07-27 04:44:40 +03:00
whitequark
692791f0bd Make sure a landing pad returns {i8*} to soothe LLVM codegen. 2015-07-25 07:01:25 +03:00
whitequark
ece52062f2 Implement code generation for exception handling. 2015-07-25 05:37:37 +03:00
whitequark
65121b437f Rework internal logic of slices. 2015-07-23 07:57:49 +03:00
whitequark
9db199cad9 Handle closure effects appropriately in LocalAccessValidator. 2015-07-23 03:15:36 +03:00
whitequark
acb8810e62 Add tests for lambdas and functions. 2015-07-23 03:07:30 +03:00