M-Labs
/
artiq
mirror of https://github.com/m-labs/artiq.git
2
0
Fork 1
Code Issues Projects Releases Wiki Activity
4,906 Commits 13 Branches 35 Tags 27 MiB
Commit Graph

608 Commits

Author SHA1 Message Date
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
David Nadlinger bfbdba9205 compiler: Emit all-kernel_invariant objects as LLVM constants 
This enables constant propagation optimisations, as verified by
the included test case. This is only a first stop-gap measure, though;
we should support optimisation based on kernel invariants on a more
fine-grained level.
 2016-11-09 14:50:48 +00:00
whitequark ec8fe6f8bd Revert "Revert "Revert "Revert "Update for LLVM 3.9."""" 
This reverts commit 7b81ed1d18.
 2016-11-08 14:22:47 +00:00
whitequark 7b81ed1d18 Revert "Revert "Revert "Update for LLVM 3.9.""" 
This reverts commit 0d7688017b.
 2016-11-08 12:58:20 +00:00
whitequark 0d7688017b Revert "Revert "Update for LLVM 3.9."" 
This reverts commit 5f5975844a.
 2016-11-08 11:59:16 +00:00
whitequark 798a5f70df Revert "runtime: remove some redundant libm functions copied inline." 
This reverts commit fee75bd50f.

LLVM does not have lround as a libcall and this inhibits LICM.
 2016-11-08 11:54:17 +00:00
David Nadlinger 6e77f65d50 compiler: Clarify recv_rpc value names and documentation [nfc] 
Previously, the phi emitted for the pointer parameter to recv_rpc was –
rather confusingly – called "size", and the pseudo-code in the comment
had bit-rotted.

Signed-off-by: David Nadlinger <code@klickverbot.at>
 2016-11-07 14:38:21 +00:00
David Nadlinger 7dcc987dd7 compiler: Fix break/continue targets in loop else blocks 
Previously, they would still target the respective labels in the
just-exited loop.

Signed-off-by: David Nadlinger <code@klickverbot.at>
 2016-11-05 02:35:01 +00:00
whitequark cd68577dbc compiler: add support for async RPCs. 2016-10-30 00:57:31 +00:00
whitequark 2a9e370840 llvm_ir_generator: make sure RPC allocations are not underaligned. 2016-10-16 16:43:03 +00:00
whitequark 7618907cad runtime: use i64 for watchdog timeout, not i32. 2016-10-16 16:32:43 +00:00
whitequark a8c017bfcc runtime: port ksupport to Rust. 2016-10-16 16:32:43 +00:00
whitequark fee75bd50f runtime: remove some redundant libm functions copied inline. 2016-10-16 16:25:29 +00:00
whitequark 1cbb187136 runtime: eliminate va_list from kernel interface. 2016-09-30 03:07:27 +00:00
whitequark ed60ba8c4c compiler: skip kernel_invariant linting for exception types. 2016-09-14 23:34:57 +00:00
whitequark feeb089505 compiler: warn about unused kernel_invariant entries. 
Fixes #543.
 2016-09-14 23:28:55 +00:00
Sebastien Bourdeauducq c414026b29 import_cache: make sure last line ends with \n as linecache does. Closes #547 2016-09-02 11:01:28 +08:00
whitequark f26f446724 artiq_run: unbreak 2016-08-17 10:20:04 +00:00
whitequark 5f5975844a Revert "Update for LLVM 3.9." 
This reverts commit 3aa7b99b8f.
 2016-08-13 04:43:19 +00:00
whitequark 3aa7b99b8f Update for LLVM 3.9. 2016-08-13 03:28:04 +00:00
whitequark 1a518ea7eb compiler.embedding: implement string concatenation. 
Fixes #526.
 2016-08-08 04:05:52 +00:00
whitequark 5a2306ae5a compiler.embedding: implement type annotations for function arguments. 
Fixes #318.
 2016-08-08 03:28:25 +00:00
Sebastien Bourdeauducq 84f4725015 cache source on import of modules that may contain kernels. Closes #416 2016-08-06 12:01:49 +08:00
whitequark 21bc285604 transforms.llvm_ir_generator: skip RPC values for attribute writeback. 2016-08-03 04:59:14 +00:00
whitequark c50d436f0b ir: `invoke` is a valid `delay` decomposition. 
Fixes #510.
 2016-07-13 08:48:31 +00:00
whitequark c5ba44b8a6 compiler.testbench.perf_embedding: more fine grained reporting. 2016-07-11 17:55:24 +00:00
whitequark c4dc4e7bf5 compiler.testbench.perf_embedding: update for core changes. 2016-07-08 10:48:41 +00:00
whitequark 5a79fcf9ba embedding: reimplement 373578bc properly. 
The core of the problem that 373578bc was attempting to solve is
that diagnostics sometimes should be chained; one way of chaining
is the loc.expanded_from feature, which handles macro-like expansion,
but another is providing context.

Before this commit, context was provided using an ad-hoc override
of a diagnostic engine, which did not work in cases where diagnostic
engine was not threaded through the call stack. This commit uses
the newly added pythonparser context feature to elegantly handle
the problem.
 2016-07-07 11:49:21 +00:00
whitequark 373578bc02 embedding: fix location for diagnostics on quoted values. 
Fixes #489.
 2016-07-07 08:32:56 +00:00
whitequark d90fd7dc00 compiler: implement numpy.full (#424). 2016-07-07 06:33:38 +00:00
whitequark 7a671fb2fd embedding: treat numpy.{int32,int64,array} specially (#424). 
Also, remove them from prelude to not pollute the namespace.
 2016-07-07 06:33:38 +00:00
whitequark 933ea53c77 compiler: add basic numpy array support (#424). 2016-07-06 09:51:57 +00:00
whitequark 906db876a6 language: replace coredevice int with numpy.{int32,int64}. 
Fixes #453.
 2016-07-06 04:44:21 +00:00
whitequark fa71b40c80 compiler: print the builtin type int(width=...) as np.int... 2016-07-06 04:22:51 +00:00
whitequark ef2af8c331 compiler.embedding: use the builtin print as RPC. 
Fixes #206.
 2016-06-28 04:42:41 +00:00
whitequark 6155f65366 compiler: remove now()/at(). 
Fixes #490.
 2016-06-28 04:39:14 +00:00
whitequark 21574bdfa9 compiler.embedding: rename user-defined types with identical names. 
Fixes #478.
 2016-06-22 01:32:01 +00:00
whitequark 33e8e59cc7 compiler: implement min()/max() as builtins. 
Fixes #239.
 2016-06-22 01:09:41 +00:00
whitequark 77d47c2fdd transforms.artiq_ir_generator: split out finally→reraise control flow. 
This makes it accessible to introspect by local access validator,
making some previously rejected code valid.

Fixes #331.
 2016-06-22 00:57:32 +00:00
whitequark f2ae24da39 compiler: add support for Python modules. 
Fixes #408.
 2016-06-21 23:35:07 +00:00
whitequark c0e42bbfc8 compiler.embedding: always do one final inference pass. 
Fixes #477.
 2016-06-16 14:18:43 +00:00
whitequark 26117e8d93 transforms.inferencer: allow variable as type of `n` in `[]*n`. 
Fixes #473.
 2016-06-16 13:35:00 +00:00
whitequark e47538ca33 analyzer: explicitly delimit messages (with \x1D). 
Fixes #461.
 2016-06-07 11:26:49 +00:00
whitequark 355af3e569 embedding: specialize inherited functions. 
Fixes #414.
 2016-05-16 15:59:26 +00:00
whitequark 640022122b embedding: refactor some more. 2016-05-16 14:38:09 +00:00
whitequark d085d5a372 embedding: refactor. 2016-05-16 14:38:09 +00:00
whitequark c94c411fd5 compiler: demangle symbols. 
In future commits we'll add Itanium C++ ABI style mangling to ARTIQ
emitted function names.
 2016-05-16 14:38:09 +00:00
whitequark 0826ceb8ba Commit missing parts of a5bb4a24. 2016-05-10 02:12:57 +00:00
whitequark a5bb4a24af compiler: support short form of raise. 
Fixes #240.
 2016-05-10 01:41:40 +00:00
whitequark 6d29e768a5 Commit missing parts of 4e5d75295. 2016-05-09 12:51:24 +00:00
whitequark 4e5d752951 compiler: fix quoting of methods (fixes #423). 2016-05-09 12:25:57 +00:00
whitequark f7d83e9bdf compiler: make kernel_invariant an instance, not class, property. 
Fixes #409.
 2016-04-29 13:06:11 +00:00
whitequark aa0882a06f embedding: ignore empty lines, like annotations, before kernel functions. 
Fixes #363.
 2016-04-26 02:25:08 +00:00
whitequark a88425b66b compiler: allow RPCing builtin functions. 
Fixes #366.
 2016-04-26 01:31:17 +00:00
whitequark 1464bae6b7 compiler: don't typecheck RPCs except for return type. 
Fixes #260.
 2016-04-26 01:12:36 +00:00
whitequark 9cc9e8b276 embedding: s/kernel_constant_attributes/kernel_invariants/g 
Requested in #359.
 2016-04-06 22:38:31 +00:00
whitequark 8b1c5e3a26 llvm_ir_generator: add fast-math flags to fcmp. 
This is allowed in 3.8.
 2016-04-05 00:48:41 +00:00
whitequark ad83b092fa llvm_ir_generator: change !{→unconditionally_}dereferenceable. 
Since LLVM 3.8, !dereferenceable is weaker, so we introduce
!unconditionally_dereferenceable (http://reviews.llvm.org/D18738)
to regain its functionality.
 2016-04-05 00:48:41 +00:00
whitequark 72b952e850 llvm_ir_generator: fix DICompileUnit.language. 2016-04-05 00:48:41 +00:00
whitequark a57aabb3ea compiler: purge generated functions from backtraces. 2016-04-02 18:29:36 +00:00
whitequark a1e98a4437 llvm_ir_generator: add TBAA metadata for @now. 2016-04-02 18:19:14 +00:00
whitequark b1f371e578 llvm_ir_generator: update debug info emission for LLVM 3.8. 2016-04-02 18:05:40 +00:00
whitequark df8018f0be compiler: emit verbose assembly via ARTIQ_DUMP_ASM. 2016-04-01 09:34:51 +00:00
whitequark d9dd7df3b2 compiler: add analysis passes from TargetMachine. 
This doesn't have any effect right now, but is the right thing to do.
 2016-04-01 09:34:36 +00:00
whitequark 10108e6d64 compiler: mark loaded pointers as !dereferenceable. 
Also, lower the bound for test_pulse_rate_dds, since we generate
better code for it now.
 2016-03-31 12:04:38 +00:00
whitequark 2c04979727 compiler: update for LLVM 3.7. 2016-03-31 11:23:47 +00:00
whitequark b8bd344e54 compiler: use correct data layout. 2016-03-31 11:22:29 +00:00
whitequark 8a908a7656 llvm_ir_generator: do not use 'coldcc' calling convention. 
First, this calling convention doesn't actually exist in OR1K
and trying to use it in Asserts build causes an UNREACHABLE.

Second, I tried to introduce it and it does not appear to produce
any measurable benefit: not only OR1K has a ton of CSRs but also
it is quite hard, if not realistically impossible, to produce
the kind of register pressure that would be relieved by sparing
a few more CSRs for our exception raising function calls, since
temporaries don't have to be preserved before a noreturn call
and spilling over ten registers across an exceptional edge
is not something that the code we care about would do.

Third, it produces measurable drawbacks: it inflates code size
of check:* functions by adding spills. Of course, this could be
alleviated by making __artiq_raise coldcc as well, but what's
the point anyway?
 2016-03-29 15:19:46 +00:00
whitequark 586022023b llvm_ir_generator: mark loads as non-null where applicable. 2016-03-29 13:34:37 +00:00
whitequark f6b391bb4e compiler: fix ARTIQ_DUMP_ELF. 2016-03-29 13:03:49 +00:00
whitequark a9489246a4 llvm_ir_generator: don't mark non-constant attribute loads as invariant. 
Oops.
 2016-03-29 13:03:49 +00:00
whitequark 3ed852e077 Commit missing parts of 1d8b0d46. 2016-03-28 21:44:31 +00:00
whitequark 1038f1321f compiler: allow specifying per-function "fast-math" flags. 
Fixes #351.
 2016-03-28 21:44:08 +00:00
whitequark 6f5332f892 compiler: allow flagging syscalls, providing information to optimizer. 
This also fixes a crash in test_cache introduced in 1d8b0d46.
 2016-03-28 19:56:56 +00:00
whitequark 1d8b0d46bc compiler: mark FFI functions as ModRef=Ref using TBAA metadata. 
Fascinatingly, the fact that you can mark call instructions with
!tbaa metadata is completely undocumented. Regardless, it is true:
a !tbaa metadata for an "immutable" type will cause
AliasAnalysis::getModRefBehavior to return OnlyReadsMemory for that
call site.

Don't bother marking loads with TBAA yet since we already place
!load.invariant on them (which is as good as the TBAA "immutable"
flag) and after that we're limited by lack of !nonnull anyway.

Also, add TBAA analysis passes in our pipeline to actually engage it.
 2016-03-28 02:52:15 +00:00
whitequark 418f0a5234 compiler: mark loads of kernel constant attributes as load invariant. 
Also, enable LICM, since it can take advantage of this.
 2016-03-28 00:44:32 +00:00
whitequark ca7463a054 compiler: do not write back kernel constant attributes. 
Fixes #322.
 2016-03-28 00:05:57 +00:00
whitequark f4e6b18690 compiler: implement kernel constant attributes. 
Part of #322.
 2016-03-27 16:10:07 +00:00
whitequark f81930ffd6 compiler: run IPSCCP. 
This doesn't do much, only frees some registers.
 2016-03-27 02:00:05 +00:00
whitequark 7213984330 compiler: raise inliner threshold to the equivalent of -O3. 2016-03-27 01:08:13 +00:00
whitequark e75ad3d1aa compiler: extract runtime checks into separate cold functions. 
This reduces register pressure as well as function size, which
favorably affects the inliner.
 2016-03-27 01:02:15 +00:00
whitequark 20ad7629ac llvm_ir_generator: generate code more amenable to LLVM's GlobalOpt. 
This exposes almost all embedded methods to inlining, with massive
gains.
 2016-03-26 21:35:48 +00:00
whitequark 186a564ba8 compiler: make quoted functions independent of outer environment. 2016-03-26 20:46:45 +00:00
whitequark f5c720c3ee compiler: tune the LLVM optimizer pipeline (fixes #315). 2016-03-26 13:20:23 +00:00
whitequark 3ee9834197 compiler: significantly increase readability of LLVM and ARTIQ IRs. 2016-03-26 12:21:02 +00:00
whitequark 8d0566661a compiler: quote functions directly instead of going through a local. 2016-03-25 22:35:55 +00:00
whitequark 39d23793a4 transforms.llvm_ir_generator: extract class function attributes. 
This should give LLVM more visibility.
 2016-03-25 19:01:39 +00:00