Commit Graph

174 Commits

Author SHA1 Message Date
florian
e80eee5da5 * generate better code for length(<dyn. type>)=0, resolves #36060
git-svn-id: trunk@43272 -
2019-10-20 13:03:45 +00:00
Jonas Maebe
14a7429e19 * factored out inlinenode check whether it may have a side effect
o also added register assert as having a side-effect (if asserts are
     enabled)

git-svn-id: trunk@42728 -
2019-08-18 15:55:03 +00:00
Jonas Maebe
1b6425176b * synchronised with trunk till r42049
git-svn-id: branches/debug_eh@42050 -
2019-05-12 18:44:05 +00:00
Jonas Maebe
281b3ad276 * fix case completeness and unreachable code warnings in compiler that would
be introduced by the next commit

git-svn-id: trunk@42046 -
2019-05-12 14:29:03 +00:00
Jonas Maebe
b3a8543a56 * synchronised with trunk till r40942
git-svn-id: branches/debug_eh@40943 -
2019-01-20 17:37:07 +00:00
florian
e24449bcfd * if sub nodes of a commutative node contain conditionally executed nodes, these sub nodes might not be swapped
as this might result in some nodes not being executed, like temp. create nodes with init. code, see e.g. issue #34653, resolves #34653

git-svn-id: trunk@40934 -
2019-01-20 12:55:20 +00:00
Jonas Maebe
9630eb7ce9 * synchronised with trunk till r40466
git-svn-id: branches/debug_eh@40467 -
2018-12-04 19:54:31 +00:00
Jonas Maebe
122d0d36d6 + volatile() expression that marks an expression as volatile
* disable matching volatile references in the assembler optimisers, so they
    can't be removed (more conservative than needed, but better than removing
    too many)
   o the CSE optimiser will ignore them by default, because they're an unknown
     inline node for it
  * also removed no longer used fpc_in_move_x and fpc_in_fillchar_x inline node
    identifiers from rtl/inc/innr.inc, and placed fpc_in_unaligned_x at the
    right place

git-svn-id: trunk@40465 -
2018-12-04 19:53:20 +00:00
Jonas Maebe
03013cf2e1 * make ttryfinallynode inherit from ttertiary node so it has a "third" node
member. This one will be used by the LLVM descendent to keep a copy of
    the finally code, as it must use a separate version for the code that
    gets executed when an exception is caught and when no exception happens

git-svn-id: branches/debug_eh@40416 -
2018-11-29 21:31:34 +00:00
pierre
044fae62ea Fix bug report 34605 and add corresponding test
git-svn-id: trunk@40377 -
2018-11-27 10:19:36 +00:00
Jonas Maebe
ea69a5b372 * finalizetemps nodes should never be considered equal and have side effects
git-svn-id: trunk@40348 -
2018-11-18 11:07:28 +00:00
Jonas Maebe
3b9f5a5e96 * moved finalization of code generator temps to a node, so it can be getcopy'd
(needed for LLVM, where we need two copies of the finally code: one in case
     an exception occurs, and one in case none is raised)
   o also first finalize parameters and only then code generator temps, since
     in theory the former could create more of the latter

git-svn-id: trunk@40345 -
2018-11-17 22:38:36 +00:00
Jonas Maebe
0cd85d73e4 * ttryfinallynode now inherits from tbinarynode instead of tloopnode, as its
t1 field was always set to tnothingnode

git-svn-id: trunk@40142 -
2018-11-01 13:19:33 +00:00
florian
88f83280dc * correctly take care of short boolean evaluation when rearranging nodes during CSE
git-svn-id: branches/laksen/riscv_new@39803 -
2018-09-25 20:31:03 +00:00
Jeppe Johansen
f828d8700c A subscript can also trigger an exception.
git-svn-id: branches/laksen/riscv_new@39718 -
2018-09-09 14:41:37 +00:00
florian
32a58ba7d1 * might_have_sideeffects gets flags: if mhs_exceptions is passed, nodes which might cause an exception are considered as having a side effect
git-svn-id: branches/laksen/riscv_new@39483 -
2018-07-22 13:11:21 +00:00
florian
c59bd8c29a + calculate loop unrolling using node_count_weighted which takes care of nodes generating no code
* optimized unrolling calculation

git-svn-id: trunk@38688 -
2018-04-04 21:39:46 +00:00
florian
3d514856a4 * improve node complexity calculations for type nodes
git-svn-id: trunk@37151 -
2017-09-06 20:34:56 +00:00
nickysn
efc5e339d0 * use an enum instead of integer constants to represent inline numbers
* compinnr.inc include file converted to a unit
* inline number field size stored in ppu increased from byte to longint
* inlines in the parse tree (when written with the -vp option) now printed with
  their enum name, instead of number

git-svn-id: trunk@36174 -
2017-05-10 14:41:43 +00:00
florian
b1dff29cbf * removed unused units
git-svn-id: trunk@36165 -
2017-05-09 19:53:14 +00:00
nickysn
b6d05dba9e * mark the sar/shl/shr/rol/ror_assign nodes as having side effects
git-svn-id: trunk@35781 -
2017-04-12 12:17:08 +00:00
nickysn
7b40bdaa6e + treat the new modify-in-place inline nodes as having side effects in
check_for_sideeffect()

git-svn-id: trunk@35774 -
2017-04-11 14:59:38 +00:00
Károly Balogh
5426c11611 marked a few uses of TConstExprInt as const during parameter passing, to avoid a few unnecessary record copying
git-svn-id: trunk@35509 -
2017-03-03 17:29:59 +00:00
florian
9c81e90e08 + extension to iso mode: reset/rewrite can take a file name as a second parameter
git-svn-id: trunk@34726 -
2016-10-16 07:55:08 +00:00
pierre
1e9491550d nutils.pas: Add new procedure:
procedure node_change_local_switch(p : tnode;cs : tlocalswitch;enable : boolean);
    to add/remove cs local switch from all p tnode tree localsettings.

    modify second_prefetch implementation in arm, ppcgen and x86 subdirectories
    to use node_change_local_settings to temporarily disabled
    checkpointer local switch, as prefetch is allowed even on unaccessible addresses.
    (previous code was buggy because I frogot that each node has it own localswitches).

git-svn-id: trunk@34576 -
2016-09-29 21:34:34 +00:00
Jonas Maebe
179c1ab328 * changed ttemp*node.tempinfo^.flags to a private field, and added setters
and getters for the ttemp*node classes instead
   o this will allow descendants to prevent certain flags from being added
     or removed. E.g. for LLVM, certain temps must never be put in registers
     because it cannot typecast a value in a register from a non-record/array
     type to an array type without forcing it to memory (so if that is done
     on an lvalue, the result will be written to the memory temp instead of
     to the register)

git-svn-id: trunk@34358 -
2016-08-21 14:23:09 +00:00
Jonas Maebe
8df1d1f9b8 * fixed compilation of tw15391 with range checking enabled after r34034:
support maybe_call_procvar() on internal block nodes, which return
    their result via the last statement

git-svn-id: trunk@34051 -
2016-07-02 21:09:59 +00:00
Károly Balogh
e0c21b6f8b ARM: avoid some range check errors, while running a compiler compiled with -CR
git-svn-id: trunk@33257 -
2016-03-16 10:08:44 +00:00
nickysn
96f331038d * fixed crash in node_complexity() on the callparan nodes of SetLength() (and
perhaps calaparans of other inline nodes as well?)

git-svn-id: trunk@33256 -
2016-03-15 23:46:46 +00:00
Jonas Maebe
9ea38f4577 * support static fields in nested types in records, by always including
support for nested types when generating an access to a static fied;
    we also always do that when generating the mangled name of a static
    field declaration in symcreat.make_field_static() (mantis #29030)

git-svn-id: trunk@32517 -
2015-11-24 16:04:19 +00:00
florian
27828d41e7 * extended nodes which return true from is_const
git-svn-id: trunk@31472 -
2015-08-31 20:03:52 +00:00
Jonas Maebe
f402b0d7df * changed getpointerdef() into a tpointerdef.getreusable() class method
o allows removing the ugly x86 hacks

git-svn-id: trunk@31144 -
2015-06-22 08:17:49 +00:00
florian
919c539ff5 * better estimation of callparanode complexity
git-svn-id: trunk@31038 -
2015-06-13 16:50:55 +00:00
Jonas Maebe
3f736f6114 * handle the loading of VMT entries at the node level, so it's done in a
type-safe way (for LLVM, and also internal consistency checking between
    the VMT as generated in nobj.pas and ncgvmt.pas)
   o also converted the VMT validity checking to the node level

git-svn-id: trunk@30950 -
2015-05-31 16:50:47 +00:00
Jonas Maebe
caea5ac8be * record a load node for the self/vmt tree of the current routine in the
tcallnode constructor, so that when it's needed later during pass 1,
    its value doesn't depend on the context in which pass 1 is executed
    (e.g. when inlining) (mantis #18121)

git-svn-id: trunk@30908 -
2015-05-25 12:55:40 +00:00
florian
0dfbba86f3 * properly firstpass nodes when inc(...) is converted into ...:=...+x, resolves #27471
git-svn-id: trunk@29742 -
2015-02-17 21:14:45 +00:00
svenbarth
5115c3e680 Fix for Mantis #26481. This is a regression.
nutils.pas, handle_staticfield_access:
  * generics don't have staticvarsyms for their static fieldvarsyms so we need to simulate a non-static access to avoid 1) an exception and 2) incorrect errors that instance methods can't be accessed

+ added tests

git-svn-id: trunk@29484 -
2015-01-16 16:05:53 +00:00
florian
6dd2edb10d * better complexity estimation for some inline nodes
git-svn-id: trunk@27674 -
2014-04-27 14:28:01 +00:00
florian
701cdf8ea2 * fix for #25956: left is not necessarily the array load node when accessing an open array/array of const element
git-svn-id: trunk@27487 -
2014-04-06 15:15:16 +00:00
svenbarth
9d48bc0baf Implement cross unit type overloading of generics. This fixes the regression introduced with revision 25498.
symtable.pas:
  + add new tsymbol_search_flag type which can be passed to various searchsym* routines
  + add support to not call "addsymref"
  + add new searchsym_with_flags function that calls searchsym_maybe_with_symoption
  * adjust searchsym_maybe_with_symoption, searchsym_in_class & searchsym_in_helper to use new flag type instead of Boolean arguments
  * adjust searchsym & searchsym_with_symoption which call the modified functions
nutils.pas, handle_staticfield_access:
  * adjust searchsym_in_class call
pexpr.pas, handle_factor_typenode, postfixoperators, factor:
  * adjust searchsym_in_helper and searchsym_in_class calls
pinline.pas, new_function:
  * adjust searchsym_in_class call
scanner.pas, try_consume_nestedsym:
  * adjust searchsym_in_class call
fmodule.pas, tmodule:
  + add genericdummysyms field which is a TFPHashObjectList that contains TFPObjectList instances per generic dummy that in turn contains tgenericdummysyms instances
pgenutil.pas:
  + add function split_generic_name to split a generic name into non-generic name and count value of type parameters
  + add function resolve_generic_dummysym which tries to use the new genericdummysyms field to find the real symbol of a dummy sym
  * generate_specialization: adjust searchsym_in_class call
  * specialization_init/specialization_done: save/restore genericdummysyms of module
symdef.pas, tdefawaresymtablestack:
  + add new intermediate method pushcommon which is used by both push and pushafter
  + add new intermediate method remove_helpers_and_generics (which calls remove_generics and remove_helpers if necessary)
  * rename removehelpers to remove_helpers
  * rename addhelpers to add_helpers_and_generics and extend it to correctly fill current_module.genericdummysyms
  * call remove_helpers_and_generics from pop instead of remove_helpers
ptype.pas, single_type, read_named_type.expr_type, read_named_type:
  * try to resolve symbols with sp_generic_dummy with resolve_generic_dummysym

+ added test

git-svn-id: trunk@25519 -
2013-09-18 14:28:46 +00:00
florian
5d4a094799 * calculate complexity of exit nodes correctly
git-svn-id: trunk@25245 -
2013-08-11 21:13:20 +00:00
florian
be2ab84474 * make actualtargetnode a normal procedure using pointers to node so it can be used also when replacing nodes
+ replacenode to replace nodes inline

git-svn-id: trunk@25013 -
2013-06-29 22:20:30 +00:00
florian
541d67771b * keep managed types in registers if possible. Under certain circumstances (if they don't require init/final code,
e.g. being a const parameter or immutable temp. values), managed types like dyn. arrays, new string types and interfaces can be kept in registers.

git-svn-id: trunk@24953 -
2013-06-23 15:16:30 +00:00
florian
18aaeff1d7 + calculate node complexity for addrnodes and labelnodes
git-svn-id: trunk@24358 -
2013-04-28 16:36:10 +00:00
Jonas Maebe
3cde2b2d84 * fixed check for bitpacked accesses (mantis #24007)
git-svn-id: trunk@23705 -
2013-03-07 15:44:41 +00:00
florian
b43b4bb455 + pseudo procedure aligned: tells the compiler to assume that the given parameter is naturally aligned, counterpart of unaligned
git-svn-id: trunk@23310 -
2013-01-04 19:13:24 +00:00
Jonas Maebe
5929ec5592 * turned getpropaccesslist, add_parameters and add_index_parameter into
methods of tpropertysym (refactoring)
  * turned code to copy the contents of a property into another one
    into a method of tpropertysym (refactoring)

git-svn-id: trunk@22955 -
2012-11-08 20:17:48 +00:00
Jonas Maebe
33f5de990d * fixed indentation
git-svn-id: trunk@22255 -
2012-08-26 21:04:51 +00:00
florian
8c39b1eba8 * document tforeachprocmethod
git-svn-id: trunk@22134 -
2012-08-19 21:18:28 +00:00
florian
3987c82bab * assign a complexity of 2 to float nodes because they usually involve a loading from memory
git-svn-id: trunk@22097 -
2012-08-15 18:28:47 +00:00