paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-06-29 20:28:32 +02:00
Code Issues Packages Projects Releases Wiki Activity
69,141 Commits 100 Branches 76 Tags 295 MiB
c3b2e35a8f
Commit Graph

526 Commits

Author SHA1 Message Date
florian
e0a1bc6675 + optimize (a and b) or (c and not(b)) into c xor ((c xor a) and b) 
+ test

git-svn-id: trunk@48841 -
 2021-02-28 18:20:46 +00:00
florian
272a0e3e26 + apply excess precision also to unary minus nodes 
* made excess precision really working

git-svn-id: trunk@48840 -
 2021-02-28 10:52:11 +00:00
yury
64c586b86d * Removed/ifdefed lots of unused variables. 
git-svn-id: trunk@48384 -
 2021-01-24 12:24:01 +00:00
florian
12adb47d65 * clean up 
git-svn-id: trunk@48011 -
 2021-01-03 15:03:27 +00:00
florian
3109005791 + fold also (string const+(string const+string var)) 
* string tree folding code moved into level 2 opt. block

git-svn-id: trunk@47986 -
 2021-01-02 22:18:38 +00:00
florian
7de94e7a1d * properly fold string+string const+string const., resolves #38267 
git-svn-id: trunk@47977 -
 2021-01-02 17:18:54 +00:00
florian
e6037961e3 * refactored constant folding code 
git-svn-id: trunk@47935 -
 2021-01-01 21:58:12 +00:00
florian
97abf6b495 * first part of fixing #38267: do not bail out early during constant folding 
if the constant is 1 or -1

git-svn-id: trunk@47933 -
 2021-01-01 15:26:13 +00:00
florian
4f3a21eeec * don't do x*x -> optimziation for xtensa as it causes a recursive call in sqr(...) 
git-svn-id: trunk@47688 -
 2020-12-04 22:43:55 +00:00
florian
6380df9b42 * correctly simplify tree transforms of (a+c1+c2) if a is a pointer, resolves #37671 
git-svn-id: trunk@47437 -
 2020-11-16 21:47:19 +00:00
florian
98eab07abd * simplified and fixed code of (v1=const1) and (v2=const2) optimization, resolves #38066 
git-svn-id: trunk@47423 -
 2020-11-15 16:17:54 +00:00
florian
ff31d95a69 * excess precision should not be applied to comp or currency 
git-svn-id: trunk@47246 -
 2020-10-28 21:37:27 +00:00
florian
13f1fb60b4 * directly copy the result of set <op> set into the left side of an assignment 
git-svn-id: trunk@47226 -
 2020-10-27 21:42:51 +00:00
Jonas Maebe
0729340817 * try_make_mul32to64: restore original node order if transformation failed 
o this gets called by taddnode.pass_1, which caches the left/right
   node types/resultdefs

git-svn-id: trunk@47158 -
 2020-10-23 13:34:24 +00:00
Jonas Maebe
099134b545 * fix try_make_mul32to64() if it's called with a constant node on the 
left (mantis #37878)
   o can happen if use_generic_mul32to64 returns false

git-svn-id: trunk@47142 -
 2020-10-19 19:10:16 +00:00
florian
637976e83f * patch by Marģers to unify internal error numbers, resolves #37888 
git-svn-id: trunk@47103 -
 2020-10-13 19:59:01 +00:00
florian
65047cecdb + optimize comparisons of constant pointers 
git-svn-id: trunk@47031 -
 2020-10-01 20:33:03 +00:00
florian
e124b07e86 + evaluate typeinfo(<type1>)=/<>typeinfo(<type2>) at compile time, resolves #30260 
git-svn-id: trunk@47008 -
 2020-09-29 20:14:27 +00:00
Jonas Maebe
aa75d39ab5 * don't perform "(x=y) or (z=u)" -> "(x xor y) or (z xor u)" optimization 
if z or u may raise exceptions (mantis #37780)

git-svn-id: trunk@46905 -
 2020-09-20 12:43:45 +00:00
yury
6878ffbf04 * In the TP mode for 16-bit CPUs uint8+uint8 is extended to sint16+sint16. This is TP compatible. 
git-svn-id: trunk@46468 -
 2020-08-17 09:25:45 +00:00
florian
18b8b9c566 * do not apply range test optimization if the variable is a string being compared against a char constants, resolves #37476 
git-svn-id: trunk@46453 -
 2020-08-15 18:29:26 +00:00
florian
d7ba09f543 * insert type casts when creating new unaryminus nodes during simplification, resolves #37508 
git-svn-id: trunk@46309 -
 2020-08-07 21:43:00 +00:00
florian
e14a3f17e6 * reverted r46220 as it is fixed now properly 
git-svn-id: trunk@46276 -
 2020-08-05 21:15:33 +00:00
yury
719220e8a4 * For 8/16 bit CPUs and TP mode, disable expansion to a larger integer type for subtract operations on unsigned integers. This is TP7 compatible. 
git-svn-id: trunk@46233 -
 2020-08-04 20:45:29 +00:00
svenbarth
c17ad509e2 * fix for Mantis #36775: apply patch by J. Gareth Moreton 
+ added tests

git-svn-id: trunk@46220 -
 2020-08-03 21:53:29 +00:00
Jonas Maebe
346adf7f55 * replace explicit (sometimes incomplete) tests with calls to doshortbooleval() 
* the c-style boolean evaluation of "and" is independent of short-circuit evaluation
    (you always have to consider the complete values in case the first one is <> 0)

git-svn-id: trunk@45990 -
 2020-08-02 14:39:50 +00:00
florian
5f118da600 * optimize a/a and a-a if fast math is on 
git-svn-id: trunk@45630 -
 2020-06-09 20:50:47 +00:00
florian
aafc22bd74 + optimize real operations with 0 and 1 if fast math is turned on 
git-svn-id: trunk@45627 -
 2020-06-08 20:33:27 +00:00
florian
6d3b74f96e * extended r45555 to take care of chars as well 
git-svn-id: trunk@45565 -
 2020-06-02 20:58:44 +00:00
florian
792cf056eb * a nil node is also a constant pointer node 
+ convert (v1=const1) and (v2=const2) into ((v1 xor const1) or (v2 xor const2))=0 if the expressions have no side effects

git-svn-id: trunk@45555 -
 2020-06-01 20:50:38 +00:00
florian
7a60e044e3 * simplify constant expressions after re-ordering, resolves #37062 
git-svn-id: trunk@45360 -
 2020-05-13 20:26:04 +00:00
florian
e5e386f94a * cosmetics 
git-svn-id: trunk@45327 -
 2020-05-10 18:41:10 +00:00
florian
ebcae70d72 * avoid unncessary conversions of constants in 64 bit ints when doing constant folding 
git-svn-id: trunk@45326 -
 2020-05-10 15:25:31 +00:00
florian
bdabf674e1 * propagate forinline 
git-svn-id: trunk@45319 -
 2020-05-09 17:36:36 +00:00
florian
43e6d84f33 * GetCopyAndTypeCheck forces correct type 
* propagate forinline flag through simplify

git-svn-id: trunk@45213 -
 2020-05-01 20:50:18 +00:00
nickysn
55254ff64b + set expectloc in the first pass of the addnode for small CPUs for procvardef comparisons 
git-svn-id: branches/z80@45044 -
 2020-04-23 16:05:23 +00:00
nickysn
96a714b20e * set expectloc correctly for procvardef comparisons with nil for small CPUs 
git-svn-id: branches/z80@45043 -
 2020-04-23 15:52:22 +00:00
nickysn
073afe3735 + set expectloc correctly for small CPUs in the first pass of the add node for classrefdefs 
git-svn-id: branches/z80@45042 -
 2020-04-23 15:48:59 +00:00
nickysn
2ee28809d8 * set expectloc correctly for small CPUs for implicit pointer object types in the first pass of the add nodes 
git-svn-id: branches/z80@45041 -
 2020-04-23 15:46:27 +00:00
nickysn
4ae5b81bff * set expectloc correctly for enumdef comparisons on small CPUs 
git-svn-id: branches/z80@45040 -
 2020-04-23 15:31:50 +00:00
nickysn
c0ea4d826c + also support the 8-bit mul helpers in the generic code of nadd 
git-svn-id: branches/z80@44990 -
 2020-04-22 02:45:49 +00:00
florian
af8202be3d * compilation for non-Xtensa targets fixed 
git-svn-id: trunk@44543 -
 2020-04-03 20:15:26 +00:00
florian
66cbee5e31 * factor out first_addfloat_soft 
git-svn-id: trunk@44540 -
 2020-04-03 20:15:24 +00:00
florian
f8339a43ea * do not do full boolean evaluation on LOC_FLAGS 
git-svn-id: trunk@44270 -
 2020-03-06 16:16:37 +00:00
florian
93521ae8e8 * fold also pointer additions/subtractions 
git-svn-id: trunk@44199 -
 2020-02-16 20:14:28 +00:00
florian
27e89e07a3 * in case a 32x32 -> 64 multiplication is converted into shl, the type casts need to be restored 
git-svn-id: trunk@44197 -
 2020-02-16 20:14:27 +00:00
florian
251b18b966 * do not fold constants if the resulttypes do not match, improved fix for #tw26587 
git-svn-id: trunk@44165 -
 2020-02-13 20:34:27 +00:00
yury
39c48f0d01 * Fixed r43577 for cases when the size of the new result of a simplified node after typecheckpass is smaller than the size of the original node's result. This caused internal errors or invalid assembly in some cases. Issue #36587. 
git-svn-id: trunk@44151 -
 2020-02-11 12:49:12 +00:00
florian
77b54f501f * when optimizing boolean expressions, might_have_sideeffects is only relevant if the expression would be optimized away 
git-svn-id: trunk@44084 -
 2020-02-01 18:02:17 +00:00
florian
051b9840a5 * write a proper error message if <int>/<int> is encountered while no 
floating point support is available, resolves #32006

git-svn-id: trunk@43794 -
 2019-12-25 18:09:30 +00:00
florian
0cb85eef9b * more currency fixes, should resolve #36176 
git-svn-id: trunk@43635 -
 2019-12-03 19:31:50 +00:00
pierre
c87facbae1 Use PInt64(@value_currency)^ construct to avoid internal error when starting from 3.0.4 ppcarm compiler 
git-svn-id: trunk@43634 -
 2019-12-03 16:05:30 +00:00
pierre
91e550edb6 Remove assigned but unused local variable i64 in taddnode.pass_typecheck_internal as suggested by Ondrej Pokorny 
This also solves the "Invalid TypeCast Error" in rtl-objpas when compiled with -CriotR option.

git-svn-id: trunk@43631 -
 2019-12-02 23:11:21 +00:00
florian
86d8e79e86 * fix some currency issues, resolves #33963 and #36179 
git-svn-id: trunk@43620 -
 2019-12-01 20:24:00 +00:00
florian
acdff47554 + optimize integer expressions like 2*x*4; which were not found by the node simplifier yet 
git-svn-id: trunk@43577 -
 2019-11-24 20:21:33 +00:00
Jonas Maebe
d23c7980aa * properly perform (or not) range/overflow checking when performing constant 
propagation into inlined routines, regardless of the range/overflow
    checking state at the caller site

git-svn-id: trunk@43436 -
 2019-11-10 12:00:05 +00:00
florian
cb67577002 * fix length(<dyn. type>)=0 optimization, resolves #36215 
git-svn-id: trunk@43343 -
 2019-10-31 21:29:41 +00:00
florian
ee3258c0da * optimizing length(<dyn. type>)=0 is not possible for jvm 
git-svn-id: trunk@43280 -
 2019-10-20 21:10:56 +00:00
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
pierre
f82cc96cb9 Check that inserttypeconv really change node type to ttypeconvnode before setting convtype field (should fix taddbool.pp bytebool failure on powerpc) 
git-svn-id: trunk@42692 -
 2019-08-14 23:00:35 +00:00
Jonas Maebe
40082100e1 * fixed range test optimization for signed types and enabled it for them 
git-svn-id: trunk@42332 -
 2019-07-06 09:57:32 +00:00
Jonas Maebe
faf75095cd * synchronised with trunk till r42189 
git-svn-id: branches/debug_eh@42190 -
 2019-06-07 18:24:38 +00:00
Jonas Maebe
eab079c7bd * fixed cbool "and" with full boolean evaluation, and cbool "xor" 
(mantis #35272)

git-svn-id: trunk@42167 -
 2019-06-02 20:04:06 +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
florian
38be96e586 * allow pointer+ord. const always if the node is created internally 
git-svn-id: branches/debug_eh@41381 -
 2019-02-19 22:16:05 +00:00
Jonas Maebe
7af95cbf06 * LLVM does not support 128 bits arithmetic on 32 bit platforms, which we 
need for overflow checking -> use RTL built-ins

git-svn-id: trunk@41205 -
 2019-02-03 21:01:02 +00:00
Jonas Maebe
04d295f186 * first step towards supporting 32 bit targets with the LLVM code generator: 
use the generic code in more cases when cpuhighleveltarget is defined

git-svn-id: trunk@41133 -
 2019-01-29 21:39:09 +00:00
Jonas Maebe
d1361ca6ed * fixed adjustments of lower/upper bounds in range test optimization in case 
of strictly smaller/greater comparisons (mantis #34385)

git-svn-id: trunk@40344 -
 2018-11-17 22:38:31 +00:00
yury
4357caaad8 * Removed unused local vars. 
git-svn-id: trunk@40183 -
 2018-11-02 18:44:29 +00:00
florian
116f36fe56 * avoid range check errors 
git-svn-id: trunk@39990 -
 2018-10-20 10:17:34 +00:00
Jonas Maebe
0b246f3dbd * converted Boolean8 to an internal type, and mapped Boolean to the 
new internal pasbool1(type) (part of mantis #34411)
   o apply the _Bool x86-64 parameter passing rules only to pasbool1

git-svn-id: trunk@39949 -
 2018-10-16 21:14:18 +00:00
pierre
293973b306 Add missing setting of expectloc for multiplications 
git-svn-id: trunk@39935 -
 2018-10-14 22:18:50 +00:00
florian
352fb866ef * is_range_test returns false if "lower" bound is greater than "upper" bound 
git-svn-id: trunk@39837 -
 2018-09-27 20:24:52 +00:00
florian
a34d4e715c Merged riscv_new branch 
git-svn-id: trunk@39813 -
 2018-09-26 19:49:08 +00:00
florian
a21397e356 + based on a patch by Laksen, optimize (v>=x) and (v<=y) into (v-x)<(y-x) 
git-svn-id: trunk@39759 -
 2018-09-16 13:09:36 +00:00
svenbarth
32c307e9ce * fix for Mantis #34021: if one of the two operators is an array constructor try to use an operator overload for that first before converting it to a set 
+ added tests

git-svn-id: trunk@39554 -
 2018-08-03 15:24:59 +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
52846df241 * fixed condition in getbestreal 
git-svn-id: trunk@39446 -
 2018-07-13 19:53:47 +00:00
florian
0c6cf12fbf + support for the directive $EXCESSPRECISION 
git-svn-id: trunk@39443 -
 2018-07-12 21:39:50 +00:00
svenbarth
fa78bb0e89 * warn if the compiler hides an eventual "+" operator overload if modeswitch ArrayOperators is active 
git-svn-id: trunk@39261 -
 2018-06-20 19:53:56 +00:00
svenbarth
97acf24290 + add a new modeswitch ArrayOperators which is set by default in Delphi modes that enables the internal array operators (currently only "+") 
git-svn-id: trunk@39259 -
 2018-06-20 19:53:43 +00:00
svenbarth
9b45f58c0b + add flags that allow checking for overloads with isbinaryoverloaded() and isunaryoverloaded() without modifying the passed in node or even checking for normally non-overloadable operators 
git-svn-id: trunk@39258 -
 2018-06-20 19:53:34 +00:00
svenbarth
199b5809a3 * optimize "dynarr := dynarr + [elem]" to "Insert(elem, dynarr, High(SizeInt))" and "dynarr := [elem] + dynarr" to "Insert(elem, dynarr, 0)" (we need to do this in the typecheck of taddnode as otherwise the array constructor is already converted) 
+ added test
* adjusted test for Mantis #30463 as p1 triggers the optimization as well

git-svn-id: trunk@39119 -
 2018-05-25 16:03:56 +00:00
florian
71e71ad267 * fix currency division on non x86 32 bit targets 
* disable fix for #33439 during bootstrapping with 3.0.x, as 3.0.x cannot compile the currency division without the fix above

git-svn-id: trunk@38558 -
 2018-03-17 22:44:44 +00:00
florian
b2825f2467 * scale constants if possible before currency multiplications to avoid overflows, resolves #33439 
git-svn-id: trunk@38555 -
 2018-03-17 21:33:07 +00:00
florian
c230f81719 + support for <dyn. array>+<dyn. array>, resolves #30463 
git-svn-id: trunk@38406 -
 2018-03-04 16:12:43 +00:00
Jonas Maebe
1934285c20 * don't convert "int64(sint32) * cardinal_const" into a 32x32->64 
multiplication (mantis #32046)

git-svn-id: trunk@37921 -
 2018-01-06 13:44:03 +00:00
pierre
59edd81a26 Disable some code using bestrealrec if macro FPC_SOFT_FPUX80 is defined 
git-svn-id: trunk@37296 -
 2017-09-21 14:39:14 +00:00
Károly Balogh
b755524c38 made it possible to override the decision from the cgs, if a certain muln needs a helper 
git-svn-id: trunk@36347 -
 2017-05-26 18:42:21 +00:00
nickysn
34113b930b * removed the checkoverflow: boolean parameter of the software mul helpers; 
replaced it with a separate "_checkoverflow" version of each helper

git-svn-id: trunk@36344 -
 2017-05-26 15:34:26 +00:00
Károly Balogh
3e8ee48458 m68k: define cpurox and enabled the rotate-related optimizations on CPUs which support rotate instructionsoptions.pas 
git-svn-id: trunk@36302 -
 2017-05-23 00:16:29 +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
546e993c25 * use resultdef for the type of the resulting zero const from the '0*x' and 
'0 and x' optimization

git-svn-id: trunk@36040 -
 2017-05-01 18:11:12 +00:00
nickysn
f26a844b19 * only do the '0*x' and '0 and x' optimization if x has no side effects (except 
at -O4, where it is done regardless of whether there are side effects or not)

git-svn-id: trunk@36039 -
 2017-05-01 17:34:19 +00:00
nickysn
f44f2f9194 * do the optimizations in r36027 at level -O3 or higher 
git-svn-id: trunk@36034 -
 2017-05-01 13:14:32 +00:00
nickysn
6d589f415b * do the left.equals(right) check before might_have_sideeffects, because 
left.equals(right) is usually faster

git-svn-id: trunk@36033 -
 2017-05-01 12:57:15 +00:00
nickysn
217d39abf1 * fixed bug in r36027 (I had it fixed already, but accidentally committed the 
wrong file in r36027)

git-svn-id: trunk@36030 -
 2017-05-01 11:43:16 +00:00
nickysn
5aeb73b940 + optimizations (at -O2 level) for (where 'a' is an integer expression, without 
side effects):
  * a - a   -> 0
  * a xor a -> 0
  * a and a -> a
  * a or a  -> a
  * a <> a  -> false
  * a < a   -> false
  * a > a   -> false
  * a = a   -> true
  * a <= a  -> true
  * a >= a  -> true

git-svn-id: trunk@36027 -
 2017-05-01 10:17:50 +00:00