 sergei
|
31fddaafe8
+ New subtype of tai_regalloc, "ra_markused". It marks a physical register as used in procedure, triggering saving/restoring that register if it is non-volatile.
|
10 years ago |
|
sergei
|
9b9ea15d68
+ m68k: initial implementation of do_spill_replace (supports only MOVE instructions for now). Tested with qemu for Coldfire target.
|
11 years ago |
 Jonas Maebe
|
246caf3526
* never alias/coalesce cpu registers that are not usable by the register
|
11 years ago |
 florian
|
23c724f885
* prevent a temp. register allocated during spilling being immediately spilled again, resolves #25164
|
11 years ago |
|
florian
|
829764e96b
+ replace registers to spill in mov instructions and convert them into ldr/str
|
11 years ago |
 Károly Balogh
|
587acf6452
fix infinite spilling on m68k, patch by Florian, additional IE workaround by me, based on a similar hack in the ARM cg...
|
11 years ago |
|
Károly Balogh
|
70f9b098e0
no longer define a segment field for m68k
|
11 years ago |
|
sergei
|
d2a9308181
+ SPARC: implemented register spill replacement.
|
11 years ago |
 svenbarth
|
32a18199c5
Print aX or aregX for address registers if DEBUG_REGISTERLIFE is defined.
|
11 years ago |
|
florian
|
dc567acc9a
* accumulate register weight instead of taking only the maximum
|
12 years ago |
|
florian
|
c2f7ccb253
* coalesce more aggressive
|
12 years ago |
|
sergei
|
8b8553991a
+ MIPS: prevent coalescing written-to registers with $sp,$fp,$zero and $at.
|
12 years ago |
|
florian
|
23c84fd27c
* more fixes for segment register usage
|
12 years ago |
|
florian
|
b85addb451
+ color segment register
|
12 years ago |
|
svenbarth
|
d15304c25e
Also check for the type of the register when replacing it. In certain circumstances this can still lead to access to invalid memory resulting either in an access violation or invalid registers.
|
12 years ago |
 pierre
|
322b793506
Try to do something for m68k integer/address registers; not working yet :!(
|
12 years ago |
|
svenbarth
|
2ada9a528b
Fix a critical bug in the register allocator (at least for CPUs with seperate address
|
12 years ago |
|
florian
|
95830d5a4a
+ DEBUG_REGISTERLIFE define
|
13 years ago |
|
florian
|
f75e39e660
+ aliasing of registers, allows to split live ranges
|
13 years ago |
|
florian
|
f9dd5bf5e6
+ live_start and live_end_properties
|
13 years ago |
|
florian
|
c2d067940a
* get rid of hash chars, modern tools and IDEs can use normal comments as documentation if needed
|
13 years ago |
|
florian
|
b78f9826fb
* call AllocMem instead of GetMem/FillChar
|
13 years ago |
|
florian
|
ecdec2e431
* disable reg. allocator optimization introduced in r21812 for now because it caused some trouble, see #22405
|
13 years ago |
 masta
|
f9cdf3d4ca
Be more specific in spilling debug messages
|
13 years ago |
|
masta
|
aef81cad68
Add debug information about register spilling
|
13 years ago |
|
florian
|
9833f642fe
* coalesce if either u or v are precolored, tested on arm and i386, it gives for both a small advantage
|
13 years ago |
|
Jonas Maebe
|
0fc422f244
* moved definition of maxcpuregister and tcpuregisterset from cgbase to
|
13 years ago |
|
Jonas Maebe
|
6f336a5d13
* the register allocator always knows the exact number of the first imaginary
|
13 years ago |
|
Jonas Maebe
|
aee5380ae0
* merged trunk up to r20882
|
13 years ago |
|
florian
|
6211a83865
* write variable locations of 64 bit integer values on 32 bit targets correctly
|
14 years ago |
