 Jonas Maebe
|
834026bfb5
* synchronised with trunk up to r21067
|
před 13 roky |
 sergei
|
5f93d9992a
* Ignore FPU/MM registers in tcg.(de)allocallcpuregisters if they are not used in current procedure. This is consistent with logic of tcgcallnode and removes a lot of useless xmm regallocs on win64.
|
před 13 roky |
|
Jonas Maebe
|
b512744bfb
* don't overwrite the reg to load in a parameter in tcg.a_load_reg_cgpara()
|
před 13 roky |
|
Jonas Maebe
|
aee5380ae0
* merged trunk up to r20882
|
před 13 roky |
|
Jonas Maebe
|
dc70db9402
* parameter passing support for AIX:
|
před 13 roky |
|
Jonas Maebe
|
260958eb45
* changed last parameter of g_indirect_sym_load() to a set and also
|
před 13 roky |
|
Jonas Maebe
|
dabcfc2806
+ LOC_(C)LOC_SUBSETREF/G support for tcg.a_loadmm_loc_reg()
|
před 13 roky |
|
sergei
|
06192a8137
Values of managed variables are never ever used after decrementing reference on the variable, so there is no point of having a 'decrement reference' as a separate operation. We can always do 'finalize', i.e. clear the contents after decref.
|
před 13 roky |
|
sergei
|
8c9ad67682
+ tcgexitnode,tcgbreaknode,tcgcontinuenode: added possibility to generate specialized code for leaving unwind-protected regions.
|
před 14 roky |
|
sergei
|
a32fdc3ae6
+ Added a simple helper tcg.g_call, intended to replace allocallcpuregisters+a_call_name+deallocallcpuregisters sequence which is used all over the place.
|
před 14 roky |
|
sergei
|
6e23565edf
+ Initialize/finalize variants using dedicated helpers, produces both smaller (no RTTI loading necessary) and faster (direct path taken) code.
|
před 14 roky |
 florian
|
4868b09844
* patch by alexvin to replace $ifopt by $push/$pop, resolves #20332
|
před 14 roky |
|
florian
|
6d996bf832
* made a_call_ref of PowerPC generic, we need to allocate a register after all on risc platforms
|
před 14 roky |
|
Jonas Maebe
|
7f22a2f223
+ support for range checking calculations with hlcgobj
|
před 14 roky |
|
florian
|
77f2d6cc0d
* introduce usage of TCGInt in the code generator units
|
před 14 roky |
|
florian
|
fb84cb865b
* rebase to trunk@17295
|
před 14 roky |
|
florian
|
8cfbb70326
+ generic a_cmp_const_reg_label
|
před 14 roky |
|
sergei
|
7f62d5cfb0
* tcg.g_array_rtti_helper: if high-value is a constant, handle it without using a register.
|
před 14 roky |
|
sergei
|
37d5a6c354
* tcg.g_array_rtti_helper(): When converting 'high' to 'length', use an intermediate register to do the increment, so that the original value of 'high' parameter is preserved regardless of platform and optimization settings. Should fix regressions introduced by r17068.
|
před 14 roky |
|
sergei
|
54bc8efa0b
* Fixed reference counting of open array parameters passed by value. RTTI of the array itself is useless in this case, as it does not provide correct element count. Now using dedicated helpers which take RTTI of array element and the element count. Resolves #18859.
|
před 14 roky |
|
florian
|
726b24c591
* (d)allocallcpuregisters must not handle fpu registers on avr
|
před 14 roky |
|
florian
|
26fbfaf5a7
+ introduce the usage of asizeint/asizeuint for cpus with sizeof("alu")<>sizeof(pointer)
|
před 14 roky |
 paul
|
dad8313512
compiler: handle dispinterfaces the same way as com interfaces because they are IDispatch descendants: increment/decrement they references in assignments and parameters passing by _AddRef, _Release
|
před 14 roky |
|
florian
|
5dae691c96
* implementation of bit scan intrinsics by Richard Vida, resolves #17592
|
před 15 roky |
 mazen
|
b127fc154a
* Fixed spell error revealed by lintian.
|
před 15 roky |
|
Jonas Maebe
|
8b0301409a
+ i386/iphonesim target for the new iPhoneSimulator in Xcode 3.2.4 and
|
před 15 roky |
|
Jonas Maebe
|
f13f6627c4
* moved use_fixed_stack from cgutils to a method in paramgr so it can
|
před 15 roky |
|
florian
|
66571ee4c8
* use extended as default real type in calculations only if fputype=x87
|
před 15 roky |
|
Jonas Maebe
|
99aece6230
* cleaned up handling of Darwin non-lazy pointer sections: create sections
|
před 15 roky |
|
Jonas Maebe
|
ea4bb9d752
* don't treat variant records with one element as "records with one element"
|
před 15 roky |
