var-parameter, as the size sometimes gets changed in this routine and
high level code generators don't use the location.size but the def
git-svn-id: trunk@40633 -
hlcgllvm in the uses clause of every unit that calls create_hlcodegen
o prevents dependency cycles that can cause llvm codegen units to init
before the cpu variants, which is bad since the llvm versions have to
override the cpu variants in their init code (+ added checks in the
init code that they are in fact initialised later)
git-svn-id: branches/debug_eh@40410 -
(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 -
so that they can still be freed after the reference has been changed
(e.g. in case of array indexing or record field accesses) (mantis #33628)
git-svn-id: trunk@38814 -
code from one unit to another, but in splitting some methods
I made a couple of mistakes :/ (causing global variables to be
initialised three times)
git-svn-id: trunk@35461 -
o separate information for reading and writing, because e.g. in a
try-block, only the writes to local variables and parameters are
volatile (they have to be committed immediately in case the next
instruction causes an exception)
o for now, only references to absolute memory addresses are marked
as volatile
o the volatily information is (should be) properly maintained throughout
all code generators for all archictures with this patch
o no optimizers or other compiler infrastructure uses the volatility
information yet
o this functionality is not (yet) exposed at the language level, it
is only for internal code generator use right now
git-svn-id: trunk@34996 -
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 -
it was AT_NONE, which is invalid and should never be used
* explicitly pass the correct value for all calls to those methods elsewhere
in the compiler
git-svn-id: trunk@34250 -
Use indirect reference for the RTL helpers that require RTTI symbols.
hlcgobj.pas, thlcgobj:
* g_incrrefcount, g_initialize, g_finalize & g_array_rtti_helper: use the indirect symbol of the RTTI if the target supports packages
........
git-svn-id: trunk@34222 -
tasmdata.DefineAsmSymbol() and all routines that call it
o will be used to automatically generate AB_INDIRECT sybols when
necessary
git-svn-id: trunk@34164 -
list, so that the assembler directives generated for Darwin are emitted
correctly for LLVM
o extracted the library init code generation from hlcgobj to pmodules so
we don't add things to another list than what's passed to
gen_proc_symbol_end(), and so it's done where we generate the init
routine for libraries (rather then checking for every routine whether
it's the init routine for a library)
o removed the llvm-specific gen_proc_symbol_end() because the regular one
now does the same
git-svn-id: trunk@34134 -
moved the code to create the procedure start/end etc. that was at the
call sites of g_external_wrapper() into the new g_external_wrapper()
-> got rid of the x86-specific code in expunix, and fixed expunix for
llvm (e.g. tests/test/tlib1{a,b})
git-svn-id: trunk@34131 -
always be different from the old location, since we just allocated
the registers for the new one above. Should the old registers ever
be needed again for tai_varloc: rr.old/rr.oldhi contains it
git-svn-id: trunk@34086 -
it was currently broken as the specialization was declared
with local bind type, which generated no symbol at all
with the internal ogomf writer.
git-svn-id: trunk@34020 -
Switch Mac OS X to indirect entry information.
compiler/systems/t_bsd.pas, TLinkerBSD:
+ implement InitSysInitUnitName
compiler/system.pas:
* systems_internal_sysinit: add Darwin systems
compiler/hlcgobj.pas, thlcgobj:
* gen_proc_symbol_end: for Darwin systems don't directly call PascalMain for libraries, but FPC_LIBMAIN instead
rtl/darwin:
+ add sysinit.pas unit which contains the executable and library entry points for Darwin
rtl/darwin/Makefile.fpc:
* add sysinit unit
rtl/bsd/sysosh.inc:
+ add a Darwin specific TPlatformEntryInformation (could probably be used for all Unix systems...)
rtl/bsd/system.pp:
* define FPC_HAS_INDIRECT_MAIN_INFORMATION for Darwin systems once we're no longer bootstrapping with 2.6.x
+ add EntryInformation variable (this could maybe moved to system.inc...)
+ add new procedure SysEntry that is called from the entrypoint and which sets up necessary information required by other parts of the RTL
* the old FPC_SYSTEMMAIN is still in place for bootstrapping
rtl/inc/systemh.inc, TEntryInformation:
* PascalMain is cdecl on non-Windows systems (ToDo: really?)
rtl/inc/system.inc:
* initialstklen is provided by the indirect entry information, so no external here
........
Fix bootstrapping.
rtl/darwin/sysinit.pas:
* enable the code only when no longer bootstrapping with 2.6.x
........
Fix bootstrapping.
rtl/objpas/objpas.pp:
* define FPC_HAS_INDIRECT_MAIN_INFORMATION for Darwin only while not bootstrapping with 2.6.x
........
rtl/darwin/sysinit.pas:
* reference PASCALMAIN correctly (I wonder how this even worked previously :/ )
git-svn-id: trunk@33949 -
Merged revision(s) 28239-28289 from branches/svenbarth/packages:
Provide the possiblity to switch between the direct and indirect RTTI symbols.
ncgrtti.pas, TRTTIWriter:
+ get_rtti_label, get_rtti_label_ord2str & get_rtti_label_str2ord: add new "indirect" parameter and pass that along to rtti_mangledname
hlcgobj.pas, thlcgobj:
* g_incrrefcount, g_initialize, g_finalize, g_array_rtti_helper: for now pass along False to get_rtti_label()
ncgvmt.pas, TVMTWriter:
* writevmt: for now pass along False to get_rtti_label()
ncgld.pas, tcgrttinode:
* pass_generate_code: for now pass along False to the get_rtti_label*() methods
........
Provide possibility to select between using a direct and an indirect RTTI reference. This way the references can be changed for selected cases.
ncgrtti, TRTTIWriter:
* ref_rtti: new "indirect" parameter that's relayed to rtti_mangledname()
* write_rtti_reference: new "indirect" paramater that's relayed to ref_rtti()
........
Switch properties to use the indirect type information without breaking backwards compatiblity. :)
compiler/ncgrtti.pas, TRTTIWriter:
* published_properties_write_rtti_data: use the indirect reference, not the direct one
rtl/objpas/typinfo.pp:
* to allow compilation with both 2.6.x and 2.7.1 and too avoid too many ifdefs at least in the declarations we define a macro TypeInfoPtr which is either PTypeInfo (2.6.x) or PPTypeInfo (2.7.1 and newer)
* TPropInfo: rename PropType to PropTypeRef and change type to TypeInfoPtr
+ TPropInfo: add a new property PropType which returns a PTypeInfo out of the PropTypeRef depending on the compiler version
........
Switch further simple types (sets, enums, class references, pointers) to indirect type information (again without breaking backwards compatibility).
compiler/ncgrtti.pas, TRTTIWriter.write_rtti_data:
* enumdef_rtti, setdef_rtti, classrefdef_rtti & pointerdef_rtti: write an indirect RTTI reference
rtl/objpas/typinfo.pp, TTypeData:
* rename BaseType to BaseTypeRef, CompType to CompTypeRef, InstanceType to InstanceTypeRef and RefType to RefTypeRef and change their type to TypeInfoPtr
+ add properties BaseType, CompType, InstanceType & RefType which return a PTypeInfo out of the corresponding *Ref field depending on the compiler version
........
Switch class and interface parent as well as extended type to indirect type information
compiler/ncgrtti.pas, TRTTIWriter:
* write_rtti_data.objectdef_rtti.objectdef_rtti_class_full: use indirect reference for class parent and extended type
* write_rtti_data.objectdef_rtti.objectdef_rtti_interface_full: use indirect reference for interface parent
rtl/objpas/typinfo.pp, TTypeData:
+ add new method DerefTypeInfoPtr which returns Nil if the indirect reference is Nil and otherwise returns the dereferences indirect reference (for 2.6.x the direct reference is returned as is)
* rename ParentInfo to ParentInfoRef, HelperParent to HelperParentRef, ExtendedInfo to ExtendedInfoRef, IntfParent ot IntfParentRef and RawIntfParent to RawIntfParentRef and change their type to TypeInfoPtr
+ introduce ParentInfo, HelperParent, ExtendedInfo, IntfParent and RawIntfParent properties that return a PTypeInfo and use the new DerefTypeInfoPtr to return the correct type info value
* change the other newly introduced properties of TTypeData to use DerefTypeInfoPtr as well to be on the safe side
........
Switch record/object fields to indirect type information references.
compiler/ncgrtti.pas, TRTTIWriter.fields_write_rtti_data:
* use the indirect reference for the object parent type
* use the indirect reference for the field type
rtl/inc/rtti.inc:
* TRecordElement: change TypeInfo to PPointer for 2.7.1 and newer
* RecordRTTI: correctly dereference the element type for 2.7.1 and newer
* fpc_copy: correctly reference the element type for 2.7.1 and newer
........
Switch static arrays to indirect RTTI reference.
compiler/ncgrtti.pas, TRTTIWriter.write_rtti_data:
* arraydef_rtti: write the dimension types and the final field type as indirect references
rtl/inc/rtti.inc:
* TArrayInfo: switch ElInfo to PPointer for 2.7.1+
* ArrayRTTI & fpc_copy: correctly dereference ElInfo for 2.7.1+
rtl/objpas/typinfo.pp, TArrayTypeData:
* switch ElType and Dims to PPTypeInfo; no backwards compatibility needed here as TArrayTypeData was added in 2.7.1 only
tests/test/trtti8.pp: fix test
........
Switch dynamic arrays to indirect RTTI references.
compiler/ncgrtti.pas, TRTTIWriter.write_rtti_data.arraydef_rtti:
* write indirect references for the two element entries
rtl/inc/dynarr.inc:
* tdynarraytypedata: change elType2 to PPointer for 2.7.1+
* fpc_dynarray_clear, fpc_dynarray_setlength & fpc_dynarray_copy: correctly dereference elType2 for 2.7.1+
rtl/objpas/typinfo.pp, TTypeData:
* rename elType and elType2 to elTypeRef and elType2Ref respectively and change type to TypeInfoPtr
* add properties elType and elType2 which return PTypeInfo by dereferencing elTypeRef and elType2Ref respecively correctly
* remove a few stray "inline" directives in the implementation
........
Switch procedure parameters and result type to indirect RTTI references.
compiler/ncgrtti.pas, TRTTIWriter.write_rtti_data.procvardef_rtti:
* write_procedure_param: use indirect RTTI reference
* write result info for both methods and procvars as indirect RTTI reference
* write parameters for procvars as indirect RTTI reference
rtl/objpas/typinfo.pp:
* TProcedureParam: change type of ParamType to PPTypeInfo (no backwards compatiblity needed here; was added for 2.7.1)
* TProcedureSignature: change type of RseultType to PPTypeInfo (no backwards compatibility needed here; was added for 2.7.1)
* TTypeData: remark in the comments of tkMethod that ResultTypeRef and ParamTypeRefs are of type PPTypeInfo and not PTypeInfo
tests/test/trtti9.pp:
* fix test
........
Remove no longer needed "indirect" parameter for TRTTIWriter methods.
ncgrtti.pas, TRTTIWriter:
- ref_rtti & write_rtti_reference: remove "indirect" parameter
* ref_rtti: call rtti_mangledname with "indirect" always set to "true"
- remove "true" parameter on callsites of write_rtti_reference & ref_rtti
........
git-svn-id: trunk@33944 -
* g_call_system_proc: if we directly call a symbol of the system unit from another unit then it must be considered as imported as well
git-svn-id: trunk@33886 -
fpc_addref_array/fpc_finalize_array (those helpers expect a pointer
to the elements of the array, not a pointer to the array)
git-svn-id: trunk@32779 -
a bitcast that reads from and writes to the same register doesn't
work well
* changed pointerdef parameters of g_ptrtypecast_reg()/ref() to regular tdef
so that these routines can also be used with procvardefs (which are pointers,
but not pointerdefs), except for the fromdef of g_ptrtypecast_ref() as this
one has to call a_loadaddr_ref_reg() (which needs the pointeddef, and getting
that is not possible for non-pointerdefs)
git-svn-id: trunk@32417 -
thlcgobj.g_set_addr_nonbitpacked_field_ref() and adjusted it so that
it also works for objects/classes (if the incoming reference points
to the start of the class contents)
o factored out the related code from tllvmsubscriptnode
git-svn-id: trunk@32411 -
labels of LOC_JUMP in the node's location. This generates some extra jumps
for short circuit boolean and/or-expressions if optimizations are off, but
with optimisations enabled the generated code is the same (except for JVM
because the jump threading optimisation isn't enabled there yet).
git-svn-id: trunk@31431 -
gen_load_para_value(), as this may be different depending on
the calling conventions
* adapt llvm's gen_load_cgpara_loc() to use this passed def rather
than hardcoding the para's def (although these will normally match)
git-svn-id: trunk@31050 -
platforms that normally must copy value parameters on the caller side,
because
o if we copy them on the caller side, then the behaviour will be different
compared to other platforms if the called routine is implemented in C
(since they are passed by reference, changes to the array contents will
be visible on other platforms, while they wouldn't be visible if the
copy is made on the caller side)
o if we don't copy them on the caller nor on the callee side, then behaviour
will be different compared to other platforms if the called routine is
implemented in Pascal and modifies the array (since then changes won't be
visible on the caller side, except on platforms that normally perform
the copy on the caller side)
git-svn-id: trunk@29873 -
be copied on the caller instead of callee side
o mark Darwin/Aarch64 as such a target (any AArch64 target will be like
that normally, as its ABI specifies this behaviour)
o don't mark by-reference value parameters on such targets as
vo_has_local_copy, since a) they don't have one (the copy is on the
caller side), and b) this ensures that all code handling such
parameters automatically knows that they are still by reference
after the init code has run
o when making the copies on the caller side, don't increase the
reference count for managed types except for variants, just like
is done when making the copies on the callee side. This is because
the reference count increasing code on the callee side only runs
for non-assembler functions, and we cannot know 100% certain on the
caller side whether the called function is assembler or not (e.g. in
case of externally declared functions)
o maybe over time we can reuse the Pascal code in
tcallparanode.copy_value_by_ref_para to replace the equivalent code
in hlcgobj and ncgutil also on the caller side for other targets
git-svn-id: trunk@29870 -
hardcoded registers for the most part)
+ added extra g_exception_discard_reason() that can be called when we only
want to get rid of the currently pushed exception reason, and don't have
to load it (so it can do nothing on platforms that don't use push/pop)
git-svn-id: branches/hlcgllvm@28481 -
thlcgobj.paravarsym_set_initialloc_to_paraloc(): it has to be the type of
pointer in the base register, not the type of the variable (and since this
is a parameter on the stack, use voidstackpointertype rather than
getpointerdef() for i8086 compatibility)
git-svn-id: branches/hlcgllvm@28351 -
so that platform-specific versions of g_call_system_proc_intern() don't
have to take care of this (since these methods call random routines from
the system unit, this info may not yet be available)
git-svn-id: branches/hlcgllvm@28203 -
