Commit Graph

132 Commits

Author SHA1 Message Date
Damien George
5b65f0c7d3 py: Rename USE_COMPUTED_GOTOS to USE_COMPUTED_GOTO and enable on stmhal.
On stmhal, computed gotos make the binary about 1k bigger, but makes it
run faster, and we have the room, so why not.  All tests pass on
pyboard using computed gotos.
2014-04-17 23:24:13 +01:00
Damien George
ea8d06c39d py: Add MP_OBJ_STOP_ITERATION and make good use of it.
Also make consistent use of MP_OBJ_NOT_SUPPORTED and MP_OBJ_NULL.
This helps a lot in debugging and understanding of function API.
2014-04-17 23:19:36 +01:00
Damien George
729f7b42d6 py: Merge BINARY_OP_SUBSCR and store_subscr (w/ delete) into subscr.
mp_obj_t->subscr now does load/store/delete.
2014-04-17 22:10:53 +01:00
Damien George
66ae8c9f49 py: Tidy up variables in VM, probably fixes subtle bugs.
Things get tricky when using the nlr code to catch exceptions.  Need to
ensure that the variables (stack layout) in the exception handler are
the same as in the bit protected by the exception handler.

Prior to this patch there were a few bugs.  1) The constant
mp_const_MemoryError_obj was being preloaded to a specific location on
the stack at the start of the function.  But this location on the stack
was being overwritten in the opcode loop (since it didn't think that
variable would ever be referenced again), and so when an exception
occurred, the variable holding the address of MemoryError was corrupt.
2) The FOR_ITER opcode detection in the exception handler used sp, which
may or may not contain the right value coming out of the main opcode
loop.

With this patch there is a clear separation of variables used in the
opcode loop and in the exception handler (should fix issue (2) above).
Furthermore, nlr_raise is no longer used in the opcode loop.  Instead,
it jumps directly into the exception handler.  This tells the C compiler
more about the possible code flow, and means that it should have the
same stack layout for the exception handler.  This should fix issue (1)
above.  Indeed, the generated (ARM) assembler has been checked explicitly,
and with 'goto exception_handler', the problem with &MemoryError is
fixed.

This may now fix problems with rge-sm, and probably many other subtle
bugs yet to show themselves.  Incidentally, rge-sm now passes on
pyboard (with a reduced range of integration)!

Main lesson: nlr is tricky.  Don't use nlr_push unless you know what you
are doing!  Luckily, it's not used in many places.  Using nlr_raise/jump
is fine.
2014-04-17 16:50:23 +01:00
AZ Huang
9413ca02fb Rename header file. 2014-04-15 21:29:08 +08:00
AZ Huang
9309d9982f Move entry_table to separated header file. 2014-04-15 17:08:42 +08:00
Damien George
3bb8bd899b Make USE_COMPUTED_GOTO a config option in mpconfig.h.
Disabled by default.  Enabled in unix port.
2014-04-14 21:20:30 +01:00
AZ Huang
b1f692e82e Use computed goto instead of switching op-codes. 2014-04-15 02:40:28 +08:00
Damien George
df8127a17e py: Remove unique_codes from emitglue.c. Replace with pointers.
Attempt to address issue #386.  unique_code_id's have been removed and
replaced with a pointer to the "raw code" information.  This pointer is
stored in the actual byte code (aligned, so the GC can trace it), so
that raw code (ie byte code, native code and inline assembler) is kept
only for as long as it is needed.  In memory it's now like a tree: the
outer module's byte code points directly to its children's raw code.  So
when the outer code gets freed, if there are no remaining functions that
need the raw code, then the children's code gets freed as well.

This is pretty much like CPython does it, except that CPython stores
indexes in the byte code rather than machine pointers.  These indices
index the per-function constant table in order to find the relevant
code.
2014-04-13 11:04:33 +01:00
Damien George
6ce4277551 py: Make all LOAD_FAST ops check for unbound local.
This is necessary to catch all cases where locals are referenced before
assignment.  We still keep the _0, _1, _2 versions of LOAD_FAST to help
reduced the byte code size in RAM.

Addresses issue #457.
2014-04-12 18:20:40 +01:00
Damien George
69b89d21b2 py: Change compile order for default positional and keyword args.
This simplifies the compiler a little, since now it can do 1 pass over
a function declaration, to determine default arguments.  I would have
done this originally, but CPython 3.3 somehow had the default keyword
args compiled before the default position args (even though they appear
in the other order in the text of the script), and I thought it was
important to have the same order of execution when evaluating default
arguments.  CPython 3.4 has changed the order to the more obvious one,
so we can also change.
2014-04-11 13:38:30 +00:00
Damien George
a1ef441d18 py: Fix VM stack overflow detection. 2014-04-10 16:59:44 +00:00
Damien George
e90be0ddf5 py: Add option to VM to detect stack overflow. 2014-04-10 16:21:34 +00:00
Damien George
d99944acdd py: Clear state to MP_OBJ_NULL before executing byte code. 2014-04-09 19:53:31 +01:00
Damien George
2bf7c09222 py: Properly implement deletion of locals and derefs, and detect errors.
Needed to reinstate 2 delete opcodes, to specifically check that a local
is not deleted twice.
2014-04-09 15:26:46 +01:00
Damien George
f4c9b33abf py: Remove DELETE_SUBSCR opcode, combine with STORE_SUBSCR.
This makes the runtime and object APIs more consistent.  mp_store_subscr
functionality now moved into objects (ie list and dict store_item).
2014-04-08 21:32:29 +01:00
Damien George
1d24ea5207 py: Finish implementation of all del opcodes.
At this point, all opcodes are now implemented!

Some del opcodes have been combined with store opcodes, with the value
to store being MP_OBJ_NULL.
2014-04-08 21:11:49 +01:00
Damien George
495d781a36 py: implement UNPACK_EX byte code (for: a, *b, c = d) 2014-04-08 17:51:47 +01:00
Damien George
e753d916c0 py: Raise exception for unimplemented byte codes. 2014-04-08 16:49:28 +01:00
Damien George
cdd96dff2c py: Implement more features in native emitter.
On x64, native emitter now passes 70 of the tests.
2014-04-06 12:58:40 +01:00
Damien George
ea13f407a3 py: Change nlr_jump to nlr_raise, to aid in debugging.
This does not affect code size or performance when debugging turned off.

To address issue #420.
2014-04-05 18:32:08 +01:00
Damien George
09a4d8305d py: Fix bug in DELETE_SUBSCR bytecode, decreasing sp too much. 2014-04-05 13:47:41 +01:00
Damien George
66edc5d899 py: Implement DELETE_SUBSCR bytecode; implement mp_obj_dict_delete. 2014-04-05 13:25:13 +01:00
Damien George
6902eeda25 py: Add m_malloc_fail function to handle memory allocation error.
A malloc/realloc fail now throws MemoryError.
2014-04-04 10:52:59 +00:00
Damien George
43e92cfb52 Merge branch 'master' of github.com:micropython/micropython 2014-03-31 16:28:58 +01:00
Damien George
15d18069c5 py: Remove old "run time" functions that were 1 liners. 2014-03-31 16:28:13 +01:00
Paul Sokolovsky
7da0660516 mp_resume: Dare to pass send_value of NULL.
There was thinkos that either send_value or throw_value is specified, but
there were cases with both. Note that send_value is pushed onto generator's
stack - but that's probably only good, because if we throw exception into
gen, it should not ever use send_value, and that will be just extra "assert".
2014-03-31 17:22:37 +03:00
Damien George
e337f1ef5e py: Towards default keyword arguments.
These are default arguments after a bare *.
2014-03-31 15:18:37 +01:00
Damien George
523b575039 py: Add LOAD_NULL bytecode and use it to simplify function calls.
Adding this bytecode allows to remove 4 others related to
function/method calls with * and ** support.  Will also help with
bytecodes that make functions/closures with default positional and
keyword args.
2014-03-31 11:59:23 +01:00
Paul Sokolovsky
f39d3b93da py: Implement support for generalized generator protocol.
Iterators and ducktype objects can now be arguments of yield from.
2014-03-30 23:30:16 +03:00
Damien George
230fec77d7 py: Implement positional and keyword args via * and **.
Extends previous implementation with * for function calls to * and **
for both function and method calls.
2014-03-30 21:21:24 +01:00
Damien George
f6a820903a Merge pull request #396 from pfalcon/call-star
vm: Implement CALL_FUNCTION_VAR opcode (foo(*(1, 2, 3))).
2014-03-30 19:09:16 +01:00
Paul Sokolovsky
14b8203a99 vm: Implement DELETE_FAST_N bytecode. 2014-03-30 17:49:56 +03:00
Paul Sokolovsky
55ca075cab vm: Implement CALL_FUNCTION_VAR opcode (foo(*(1, 2, 3))). 2014-03-30 17:47:16 +03:00
Damien George
d17926db71 Rename rt_* to mp_*.
Mostly just a global search and replace.  Except rt_is_true which
becomes mp_obj_is_true.

Still would like to tidy up some of the names, but this will do for now.
2014-03-30 13:35:08 +01:00
Damien George
89f94b51ab py: Rename mp_exc_stack to mp_exc_stack_t. 2014-03-30 00:57:09 +00:00
Damien George
d7592a1c3f py: Fix reraise logic. 2014-03-30 00:54:48 +00:00
Paul Sokolovsky
0c904df8e6 vm: Save current active exception on opening new try block.
Required to reraise correct exceptions in except block, regardless if more
try blocks with active exceptions happen in the same except block.

P.S. This "automagic reraise" appears to be quite wasteful feature of Python
- we need to save pending exception just in case it *might* be reraised.
Instead, programmer could explcitly capture exception to a variable using
"except ... as var", and reraise that. So, consider disabling argless raise
support as an optimization.
2014-03-30 01:01:35 +02:00
Paul Sokolovsky
69975df3ff vm: WITH_CLEANUP: use POP_EXC_BLOCK(). 2014-03-30 01:00:51 +02:00
Paul Sokolovsky
a0ad77ba08 vm: Establish macros PUSH_EXC_BLOCK & POP_EXC_BLOCK to deal with exc stack.
E.g. to handle currently_in_except_block restoring properly.
2014-03-29 23:18:59 +02:00
Paul Sokolovsky
d109676ec0 py: Reraising exception possible only in except block. 2014-03-29 23:18:59 +02:00
Paul Sokolovsky
40d6d29af6 vm: Elaborate comments for WITH_CLEANUP, other cosmetic fixes. 2014-03-29 18:46:04 +02:00
Damien George
c689c19471 py: Make MP_BC_SETUP_WITH use the bytecode stack for load_method.
The compiler allocates 7 entries on the stack for a with statement
(following CPython, but probably can be reduced).  This is enough for
the method load and call in SETUP_WITH.
2014-03-29 14:06:14 +00:00
Damien George
21a07dc50f Merge pull request #389 from pfalcon/with-statement
With statement implementation
2014-03-29 14:00:03 +00:00
Damien George
b04be056fe py: Fix regress with GeneratorExit object becoming truly const. 2014-03-29 13:52:51 +00:00
Damien George
07ddab529c py: Change mp_const_* objects to macros.
Addresses issue #388.
2014-03-29 13:15:08 +00:00
Damien George
da51a399cf Merge pull request #383 from pfalcon/yield-from
Implement "yield from"
2014-03-29 12:18:14 +00:00
Damien George
d1e443d0bc py: Free unique_code slot for outer module.
Partly (very partly!) addresses issue #386.  Most importantly, at the
REPL command line, each invocation does not now lead to increased memory
usage (unless you define a function/lambda).
2014-03-29 11:39:36 +00:00
Paul Sokolovsky
44307d5ef8 vm: Implement "with" statement (SETUP_WITH and WITH_CLEANUP bytecodes). 2014-03-29 04:39:24 +02:00
Paul Sokolovsky
682f9e639d vm: Make sure that exception triple is <type, instance, traceback>.
This reduntant triple is one of the ugliest parts of Python, which they
chickened out to fix in Python3. We really should consider passing just
as single exception instance (without breaking Python-level APIs of course),
but until we do, let's follow CPython layout.
2014-03-29 04:35:36 +02:00