* Improved peephole optimizer to remove these instructions if preceded by flag-setting instruction that operates on same location and followed by conditional jump, and change them back into "test %reg,%reg" otherwise.
git-svn-id: trunk@27617 -
memory models. Enabled by the new directive {$hugecode on}. The directive is
ignored in the near code memory models. When enabled, it forces each procedure
to be in a separate segment and disables mixing near and far procedures (so
'near' and {$F-} are ignored in this mode). Note that {$hugecode on} does not
count as a different memory model, because you can freely link modules (units)
compiled with {$hugecode on} and {$hugecode off}.
git-svn-id: trunk@27615 -
* force the left operand into a register if both the left and
right are constants (can happen because not all set add-nodes
are handled in taddnode.simplify) because the code below
expects at least one register operand. Fixes the compilation of a.o.
taddset2 with -O3 on ppc (constant propagagation -> 2 constant
operands)
git-svn-id: trunk@27607 -
after r27581 in regard to floating point comparisons. The tcmpnan.pp test now
almost works (i.e. fails due to only one error, instead of many).
git-svn-id: trunk@27605 -
otherwise data in initialised packed records gets aligned wrongly. This
should probably also be done for other targets that automatically
align .short/.long/...
git-svn-id: trunk@27600 -
models. The $F directive and the 'near' and 'far' procedure modifiers should
now work as expected in the far code memory models (they are still ignored in
the near code memory models). The compiler defaults to the {$F+} state,
because {$F-} requires adding 'far' to a lot of procedures in the rtl,
packages and tests.
git-svn-id: trunk@27590 -
Currently it is implemented only for x86-CPUs supporting the FMA extension. While it would
be possible to implement it in software or simulate fma(<single>,<single>,<single>) using
double operations, it makes no sense in my opinion to do so.
git-svn-id: trunk@27564 -
+ added helper function to i8086's symcpu is_proc_far that dispatches the call
to the proper is_far method (because we can't make a cpu specific descendant
of tabstractprocdef and add it there)
* all checks for (po_far in procoptions) in the i8086 code generator replaced
with calls to is_proc_far
git-svn-id: trunk@27559 -
6-byte method pointers and other i8086 types that use loc.registerhi
* changed the i8086 specific code in tcgassignmentnode.pass_generate_code to
use hlcg.a_load_loc_ref
git-svn-id: trunk@27555 -
o The checks for types that need special handling moved to separate methods,
because they'll probably be used by the other hlcg.a_load*loc* methods as
well.
o Use voidcodepointertype and voidpointertype to avoid the ifs for checking
the memory model.
git-svn-id: trunk@27554 -
are called, when the 'near' or 'far' directive is parsed in a procedure
definition. This allows overriding and implementing them on architectures,
which support near and far calls.
git-svn-id: trunk@27545 -
following cases:
1) external linking is already enabled
2) target does not support an internal linker
3) the 'omit linking stage' option is used
git-svn-id: trunk@27543 -
nmat.pas:
* correctly disable type checks if one of the arguments for MOD, DIV, SHR, SHL, NOT, -X and +X is a generic type parameter
+ added tests
git-svn-id: trunk@27535 -
symtable.pas, search_objectpascal_helper:
* handle more symbol types than just properties and methods; afterall helpers can have types, static variables and constants as well
+ added test
git-svn-id: trunk@27534 -
ninl.pas, tinlinenode.pass_typecheck:
* don't generate a type error if Length() is used inside generics with a generic argument
+ added test
git-svn-id: trunk@27533 -
ptype.pas, read_named_type.array_dec:
* generate a valid range definition if we are inside a generic and the array bounds are not ordinals (Note: the check for "parse_generic" should be improved to really detect erratic indexes that are not related to generics in any way)
+ added test
git-svn-id: trunk@27532 -
o the maintenance effort will be less - previously we had to test 4
combinations:
tiny program with tiny rtl
tiny program with small rtl
small program with tiny rtl
small program with small rtl
I only tested both tiny and small programs with the small rtl, so I missed
a bug, which caused small programs built with the tiny rtl to always give
a "Nil pointer assignment" error. Now we only need to test two cases:
tiny program with tiny rtl
small program with small rtl
o I'm planning a bug fix for interrupt procedures in the tiny model, which
will make their prolog differ between small and tiny. Currently, they're
not used in the rtl, but that may change. And even if it doesn't,
interrupt procedures may also be used by user units.
git-svn-id: trunk@27516 -
preserves the reference segment (because reference_reset_base, which is
sometimes used by the base method, kills the segment)
git-svn-id: trunk@27480 -
+ Test.
+ Check duplicate labels in assembler blocks. This was impossible due to above bug and duplicate labels were detected only at assembling stage where no location information could be provided.
git-svn-id: trunk@27472 -
symsym.pas, tfieldvarsym:
+ add new field fieldvarsym which holds a reference to a tfieldvarsym if the static sym was created based on such a symbol
+ add necessary methods and code to correctly load from and store to PPU
+ add new constructor create_from_fieldvar
symcreat.pas, make_field_static:
* use new create_from_fieldvar constructor instead of the default one
hlcgobj.pas, finalize_static_data:
* check whether the static var is based on a generic's class var
ppu.pas:
* increase PPU version
+ added test
git-svn-id: trunk@27466 -
* ptype.pas, id_type:
generate the "identifier not found" message only if the identifier was really not found and not when the identifier is a non-type
* pexpr.pas, statement_syssym:
don't generate the "type id expected" error message for Default() anymore; this is already handled by single_type
+ added test (though this again will not help much as output is not parsed...)
git-svn-id: trunk@27464 -
* pdecsub.pas, parse_proc_direc:
when a hint directive is parsed we need to check whether the new token can still be a directive
+ added test (though that won't help much as we don't check the compiler output yet... :/ )
git-svn-id: trunk@27463 -
cpu-specific descendants (unfortunately causes some duplication, but the
code is trivial and there is no easy way to avoid it)
* also moved the use of the field in ncal to cpu-specific files (with same
caveat)
git-svn-id: trunk@27438 -
Background: these operations are defined as flipping or clearing the upper bit of number, respectively, and never result in precision loss or raise floating-point exceptions.
git-svn-id: trunk@27411 -
specific descendent classes and moved the code that deals with this in the
code generator also to target-specific classes -> only ifdefs left in
pdecvar
git-svn-id: trunk@27379 -
override them safely in subclasses to add more data (they finalise
writing their ibentry)
* removed "virtual" from all ppuload constructors for similar reason
as above
+ added virtual (empty) ppuwrite_plaform() method that is called from
tstoreddef/tstoredsym.ppuwrite(), which can be safely overridden in
child classes to add extra data (called centrally from a new
tstoreddef/sym.writeentry() method)
+ added virtual (empty) ppuload_platform() method that is called from
tstoreddef/tstoredsym.ppuload(), which can be safely overridden in
child classes to read the extra data (unfortunately cannot be called
from one single location, and these calls had to be added for each
generic subclass separately)
git-svn-id: trunk@27378 -
o made all (non-abstract) tdef and tsym constructors virtual
o added c*def/c*sym classref types for every (non-abstract) t*def/t*sym
class
o added cpusym unit for every architecture that derives a tcpu*def/tcpu*sym
class from the base classes, and initialises the c*def/c*sym classes with
them. This is done so that the llvm target will be able to derive from
the tcpu*def/sym classes without umpteen ifdefs, and it also means that
the WPO can devirtualise everything because the c* variables are only
initialised with one class type
o replaced all t*def/t*sym constructor calls with c*def/c*sym constructor
calls
git-svn-id: trunk@27361 -
to use the hlcg in order to support the i8086 far data memory models. Only the
instructions that handle the case when there's no thread manager installed
(FPC_THREADVAR_RELOCATE=nil) are converted, since that's what's used on the
msdos target. The rest still uses the low level code generator.
git-svn-id: trunk@27336 -
