This unit was empty unless OLDREGVARS macro was set,
but this does not compile and no change has been made since 2011.
* Remove regvars from all _USES clauses.
git-svn-id: trunk@34808 -
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 -
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 -
initialloc was already set afterwards, now it's also done in the final
case (and allocating the initial location should be different from
setting the current location)
git-svn-id: trunk@34085 -
Merged revision(s) 28793, 28903, 31951 from branches/svenbarth/packages:
psub.pas, read_proc:
* don't generate assembler symbols for compiler internal code
........
Ensure that unneeded initialization/finalization procedures don't leave assembler symbols.
ncgutil.pas:
+ new function release_proc_symbol which removes all assembler symbols of a procdef
pmodules.pas, proc_unit:
* if the initialization or finalization procedure is empty also remove their assembler symbols so they are exported
........
Ensure that readonly sections that need relocations are indeed declared as such.
Thanks to Jonas for this.
ncgvmt.pas, TVMTWriter:
* writeinterfaceids: use sec_rodata instead of sec_rodata_norel
........
git-svn-id: trunk@32982 -
The bug has been triggered only with the following conditions:
- tf_safecall_exceptions, tf_safecall_clearstack in target flags;
- fixed stack is used.
Since only the i386-android target meets these conditions atm, the bug has been unnoticed.
git-svn-id: trunk@32005 -
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 -
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 -
the stack pointer during the prolog. This was done previously in
tcg.g_proc_entry(), but that routine is called after register allocation
and hence has no influence. Also cleaned up the deallocation of that
register by moving the previously ifdef'd code to thlcgppcgen
(mantis #27634)
git-svn-id: trunk@30164 -
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 -
single register, by first storing everything consecutively to memory and
then loading it. Required for "homogeneous float aggregates" on AArch64
git-svn-id: trunk@29815 -
the code written in hlcg.gen_proc_symbol to avoid change from AB_LOCAL
to AB_GLOBAL, which generates erroneous code (at least for targets using GOT)
git-svn-id: trunk@29450 -
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 -
be included at the start of the implementation of the system unit (before
the rest of except.inc)
* catch declarations in/loading from the system unit of the TExceptAddr type
* use this type instead of hardcoded size constants in the compiler
* in generic code that is active for all targets, puts its use in a virtual
method since it's only valid for targets using setjmp/longjmp-style
exception handling (and the record is not defined at all in the JVM RTL)
git-svn-id: branches/hlcgllvm@28376 -
