Commit Graph

430 Commits

Author SHA1 Message Date
Eric Anholt
43ce26e64b Verify that the instruction encoding matches what the ISA would encode.
Fixes #69
2017-04-23 17:21:32 -07:00
Jakob Stoklund Olesen
c5da572ebb Add a liveness verifier.
The liveness verifier will check that the live ranges are consistent
with the function. It runs as part of the register allocation pipeline
when enable_verifier is set.

The initial implementation checks the live ranges, but not the
ISA-specific constraints and affinities.
2017-04-21 16:01:08 -07:00
Jakob Stoklund Olesen
c4b794f7cf Run the verifier in the Context methods when it is enabled.
The test drivers can stop calling comp_ctx.verify because legalize() and
regalloc() do it themselves now.

This also makes it possible for those two passes to return other
CtonError codes in the future, not just verifier errors.
2017-04-21 12:36:35 -07:00
Jakob Stoklund Olesen
d66a9d196e Implement binary emission of RISC-V return instructions.
The return address is now always supplied in %x1, so the return address
predictor will recognize the jalr as a return and not some indirect
branch.
2017-04-19 16:26:04 -07:00
Jakob Stoklund Olesen
0cb36c9031 Remove the return_reg instruction.
RISC architectures that take a return address in a register can use a
special-purpose `link` return value to do so.
2017-04-19 16:08:16 -07:00
Jakob Stoklund Olesen
e44a5a4391 Append link and sret arguments in legalize_signature.
These special-purpose arguments and return values are only relevant for
the function being compiled, so add a `current` flag to
legalize_signature().

- Add the necessary argument values to the entry block to represent
  the special-purpose arguments.
- Propagate the link and sret arguments to return instructions if the
  legalized signature asks for it.
2017-04-19 15:55:12 -07:00
Jakob Stoklund Olesen
49c1209572 Fix broken test. 2017-04-17 15:45:55 -07:00
Jakob Stoklund Olesen
d424589daa Allow for special purpose function arguments and return values.
Enumerate a set of special purposes for function arguments that general
purpose code needs to know about. Some of these argument purposes will
only appear in the signature of the current function, representing
things the prologue and epilogues need to know about like the link
register and callee-saved registers.

Get rid of the 'inreg' argument flag. Arguments can be pre-assigned to a
specific register instead.
2017-04-17 15:06:30 -07:00
Jakob Stoklund Olesen
3c8bdba15e Don't create value aliases when legalizing ABI boundaries.
When converting from ABI types to original program types, the final
conversion instruction can place its result into the original value, so
it doesn't need to be changed to an alias.
2017-04-13 10:16:58 -07:00
Jakob Stoklund Olesen
d48f79aa72 Avoid creating aliases when expanding legalizer patterns.
Now that we can detach and reuse all values, there is no longer a need
to create a lot of alias values during pattern expansion. Instead, reuse
the values from the source pattern when emitting instructions in the
destination pattern.

If a destination instruction produces the exact same values as a source
instruction, simply leave the values attached and replace the
instruction it. Otherwise, detach the source values, reuse them in the
expansion, and remove the source instruction afterwards.
2017-04-13 09:30:21 -07:00
Jakob Stoklund Olesen
18b567f88e Flatten the Value reference representation.
All values are now references into the value table, so drop the
distinction between direct and table values. Direct values don't exist
any more.

Also remove the parser support for the 'vxNN' syntax. Only 'vNN' values
can be parsed now.
2017-04-12 14:45:22 -07:00
Jakob Stoklund Olesen
28ff7b7925 Move the ctrl_typevar function into dfg.
Soon, InstructionData won't have sufficient information to compute this.

Give TargetIsa::encode() an explicit ctrl_typevar argument. This
function does not require the instruction to be inserted in the DFG
tables.
2017-04-12 14:32:13 -07:00
Jakob Stoklund Olesen
8a2398be08 Stop calling Value::new_direct.
We only ever create table values now.

Simplify legalizer::legalize_inst_results. Instead of calling
detach_secondary_results, just detach all the results and don't treat
the first result specially.
2017-04-12 14:32:13 -07:00
Jakob Stoklund Olesen
94e26a845a Make tests less sensitive to specific value numbers. 2017-04-12 14:32:13 -07:00
Jakob Stoklund Olesen
4ee519e620 Add RISC-V call instruction encodings.
Calls are jal with a fixed %x1 link register.
2017-04-11 10:40:38 -07:00
Jakob Stoklund Olesen
b4ac520332 Extending loads and truncating stores 2017-04-11 10:30:03 -07:00
Jakob Stoklund Olesen
aad6ebebb5 Add load and store instructions.
Define a MemFlags class, currently holding a notrap and aligned flag.
2017-04-11 09:54:55 -07:00
Jakob Stoklund Olesen
0c3771bccb Ensure that the docs examples verify as Cretonne IL.
Any *.cton files in the docs directory are now included when running the
test-all.sh script. This is to ensure that the examples are in fact
correct IL.

Always print NaN and Inf floats with a sign. Print the positive ones as
+NaN and +Inf to make them easier to parse.
2017-04-10 15:28:24 -07:00
Jakob Stoklund Olesen
b474485c0d Add heap_load, heap_store, and heap_addr instructions.
These are used when lowering WebAssembly sandbox code.
2017-04-10 15:04:33 -07:00
Jakob Stoklund Olesen
222ae8af22 Define stack_load, stack_store, and stack_addr instructions. 2017-04-10 13:56:57 -07:00
Jakob Stoklund Olesen
eccc2d0dae Add an Offset32 immediate operand kind.
This will be used to represent an immediate 32-bit signed address offset
for load/store instructions.
2017-04-10 11:53:46 -07:00
Jakob Stoklund Olesen
51dea9f784 Syntax tweak: Omit comma after an initial enum immediate.
This affects the comparison instructions which now read "icmp ult a, b".
This mimics LLVM's style and makes it simpler to add instruction flags
in the future, such as "load v1" -> "load aligned v1".

These enumerated operands and flags feel like opcode modifiers rather
than value operands, so displaying them differently makes sense.

Value and numeric operands are still comma separated.
2017-04-10 10:28:37 -07:00
Jakob Stoklund Olesen
896ac935b9 Add jump encodings to RISC-V.
Fix a bug in gen_encoding.py when dealing with non-polymorphic
instructions where the type variable is None in Python, VOID in Rust.
2017-04-06 15:17:57 -07:00
Jakob Stoklund Olesen
fa4f151b9b Add a fallthrough instruction.
Change jumps to fallthroughs in the branch relaxation pass before
computing the EBB offsets.
2017-04-06 14:22:32 -07:00
Jakob Stoklund Olesen
856a67ca3a Use EBB offsets for encoding RISC-V branches.
Stop emitting EBB relocations. Use the offsets computed by
relax_branches() to encode the correct displacements immediately.
2017-04-06 12:55:57 -07:00
Jakob Stoklund Olesen
8a3f538441 Fix a bug in the binemit file test.
Only emit each instruction once, or the offset computations go all
wrong.
2017-04-06 11:17:46 -07:00
Jakob Stoklund Olesen
81251c3005 Add a branch relaxation pass for #72.
Compute exact EBB header offsets and check that branches are in range.

Not implemented yet: Relax branches that are not in range.

Invoke the relax_branches() pass from the 'test binemit' file tests so
they can verify the proper encoding of branch instructions too.
2017-04-05 15:11:39 -07:00
Jakob Stoklund Olesen
72cb6459a9 Add a CodeOffset type and CodeSink::offset() method.
We need to keep track of code offsets in order to compute accurate
branch displacements.
2017-04-05 09:12:12 -07:00
Jakob Stoklund Olesen
1984c96f7c rustfmt 0.8.1 2017-04-05 09:00:11 -07:00
Jakob Stoklund Olesen
e641c97670 Add RISC-V encodings for brz and brnz.
These branches compare a register to zero. RISC-V implements this with
the %x0 hard-coded zero register.
2017-04-03 15:20:57 -07:00
Jakob Stoklund Olesen
479ff156c1 Add conditional branch encodings for RISC-V.
Not all br_icmp opcodes are present in the ISA. The missing ones can be
reached by commuting operands.

Don't attempt to encode EBB offsets yet. For now just emit an EBB
relocation for the branch instruction.
2017-04-03 15:16:25 -07:00
Jakob Stoklund Olesen
1b6a6f4e48 Add the br_icmp instruction.
This instruction behaves like icmp fused with brnz, and it can be used
to represent fused compare+branch instruction on Intel when optimizing
for macro-op fusion.

RISC-V provides compare-and-branch instructions directly, and it is
needed there too.
2017-04-03 15:04:42 -07:00
Jakob Stoklund Olesen
0530e822d4 Add RISC-V encodings for lui.
This instruction can materialize constants with the low 12 bits clear.
2017-04-03 12:27:22 -07:00
Jakob Stoklund Olesen
4e398164d2 Add icmp_imm encodings for RISC-V.
The ISA has icmp_imm slt/ult with 12-bit signed immediate operands.
2017-04-03 10:59:28 -07:00
Jakob Stoklund Olesen
3176ccd131 Use the right operand when documenting type variable inference.
The meaning of format.typevar_operand changes recently to be relative to
value operands only instead of all operands. The Sphinx cton domain
wasn't updated.
2017-04-03 09:56:47 -07:00
Jakob Stoklund Olesen
e23d12bbc7 Add an icmp_imm instruction.
Compare a scalar integer to an immediate constant. Both Intel and RISC-V
ISAs have this operation.

This requires the addition of a new IntCompareImm instruction format.
2017-04-03 09:49:44 -07:00
Jakob Stoklund Olesen
f70ae03b47 Skip the Python checks if the Python files haven't changed. 2017-03-31 15:45:21 -07:00
Jakob Stoklund Olesen
9ddbc5d861 Emit Rshamt-type instructions for RISC-V.
These are the shift-by-immediate instructions.
2017-03-31 15:33:21 -07:00
Jakob Stoklund Olesen
c9a14448b8 Emit I-type instructions for RISC-V.
These are the BinaryImm formats.
2017-03-31 15:15:20 -07:00
Jakob Stoklund Olesen
e9907fbdd6 Add RISC-V encodings for supported icmp variants.
Only slt and ult variants are in the instruction set. Other condition
codes must be synthesized.
2017-03-31 13:47:07 -07:00
Jakob Stoklund Olesen
cb33f93fcd Use pretty_verifier_error in runone too. 2017-03-29 15:30:44 -07:00
Jakob Stoklund Olesen
4c3e590bb9 Run verifier after legalizer and regalloc file tests.
Run the verify_contexti() function after invoking the legalize() and
regalloc() context functions. This will help catch bad code produced by
these passes.
2017-03-29 15:19:03 -07:00
Jakob Stoklund Olesen
b5d4046591 Fix a type error in the legalizer patterns.
The carry and borrow values are boolean, so we have to convert them to
an integer type with bint(c) before we can add them to the result.

Also tweak the default legalizer action for unsupported types: Only
attempt a narrowing pattern for lane types > 32 bits.

This was found by @angusholder's new type checks in the verifier.
2017-03-29 15:00:27 -07:00
Jakob Stoklund Olesen
4613cbcf18 Test binary encodings of some RV32I instructions.
These are the R-format instructions in RV32I.
2017-03-29 13:16:12 -07:00
Jakob Stoklund Olesen
6ad34f90aa Allow for unencoded instructions in the binemit tests.
If an instruction doesn't have an associated encoding, use the standard
TargetIsa hook to encode it.

The test still fails if an instruction can't be encoded. There is no
legalization step.
2017-03-29 09:55:49 -07:00
Jakob Stoklund Olesen
f540cb0664 Add a binemit test command.
This makes it possible to write file tests that verify the binary
encoding of machine code.
2017-03-28 15:56:30 -07:00
Jakob Stoklund Olesen
d9f325f6cd Simplify branch arguments after splitting EBB arguments.
The EBB argument splitting may generate concat-split dependencies when
it repairs branch arguments in EBBs that have not yet been fully
legalized. Add a branch argument simplification step that can resolve
these dependency chains.

This means that all split and concatenation instructions will be dead
after legalization for types that have no legal instructions using them.
2017-03-22 15:29:06 -07:00
Keith Yeung
c7e25e6a35 Limit type inference for controlling type variables in write.rs 2017-03-22 14:45:38 -07:00
Jakob Stoklund Olesen
272df6489c Iteratively split EBB arguments.
When the legalizer splits a value into halves, it would previously stop
if the value was an EBB argument. With this change, we also split EBB
arguments and iteratively split arguments on branches to the EBB.

The iterative splitting stops when we hit the entry block arguments or
an instruction that isn't one of the concatenation instructions.
2017-03-22 13:12:19 -07:00
Jakob Stoklund Olesen
ca6e402b90 Add a ControlFlowGraph argument to legalize_function.
Legalizing some instructions may require modifications to the control
flow graph, and some operations need to use the CFG analysis.

The CFG reference is threaded through all the legalization functions to
reach the generated expansion functions as well as the legalizer::split
module where it will be used first.
2017-03-21 16:00:28 -07:00