Commit Graph

163 Commits

Author SHA1 Message Date
sergei
fa98e0523d * Safecall handling: forgot to free cgpara before doing the call, fixed.
git-svn-id: trunk@19976 -
2012-01-05 19:12:56 +00:00
sergei
f6761d7939 * Handle safecall exceptions entirely in tcgtryfinallynode pass2. This simplifies things and removes the need in hidden safe_result variable.
git-svn-id: trunk@19950 -
2012-01-02 14:23:31 +00:00
sergei
8c9ad67682 + tcgexitnode,tcgbreaknode,tcgcontinuenode: added possibility to generate specialized code for leaving unwind-protected regions.
git-svn-id: trunk@19636 -
2011-11-14 22:47:37 +00:00
sergei
e2d43fcb30 - tcgonnode: don't store exception object in temp if it is not used.
- tcgtryexceptnode: don't generate code to decrease/compare exception reason if not necessary.

git-svn-id: trunk@19517 -
2011-10-19 20:28:41 +00:00
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.
* Refactored code generation for tcgonnonde and tcgtryexceptnode, simplifying both compiler and generated code. Merged compilerprocs called in sequence into larger ones (fpc_popobjectstack+fpc_destroyexception -> fpc_doneexception, fpc_popsecondobjectstack+fpc_destroyexception+fpc_reraise -> fpc_raise_nested).

git-svn-id: trunk@19506 -
2011-10-18 15:46:15 +00:00
sergei
9ed2fa6f7b - Exception handling: calling FPC_CATCHES(-1) is obsolete, it does not change any state and its return value is not used. Removed.
- Don't generate guard frame around 'except' block if it contains no code (tryexceptnode.t1 is an empty blocknode in this case).

git-svn-id: trunk@19415 -
2011-10-08 12:57:58 +00:00
florian
d9ed5adb21 * set line info markers properly, resolves #17803
git-svn-id: trunk@17476 -
2011-05-16 15:15:35 +00:00
joost
65ab1e820b * Safecall on linux/i386 now behaves like cdecl plus hidden exception support.
* Adapted tests because safecall on linux/i386 now does not include high() 
   support anymore

git-svn-id: trunk@15913 -
2010-08-27 19:07:22 +00:00
florian
ad23569cf1 * renamed pi_has_goto into pi_has_label because the problematic thing is if a procedure has a label as target and not if it has some goto
* moved inclusion of pi_has_label to tlabelnode.pass_1

git-svn-id: trunk@15712 -
2010-08-05 19:22:12 +00:00
Jonas Maebe
283018a3bf * changed tprocdef.funcretloc[] from a tlocation into a tcgpara so it can
represent complex locations (required for full x86-64 ABI support,
    which is not yet implemented) -> lots of special result handling
    code has been removed and replaced by the parameter handling
    routines
  + added support for composite parameters (and hence function
    results) to tcg.a_load_ref_cgpara() (so it can be used for
    handling, e.g., 64 bit parameters on 32 bit platforms)
  * the above fixed writing past the end of allocated memory when
    handling records returned in registers on x86-64 whose size is
    not a multiple of 8 bytes (mantis #16357)
  - removed the x86-64 and PPC specific versions of a_load_ref_cgpara(),
    as they are now handled correctly by the generic version
  * moved the responsibility of allocating tcgpara cpu registers
    (using paramanager.allocparaloc()) from the callers of
    cg.a_load*_cgpara() to the cg.a_load*_cgpara() methods
    themselves (so the register allocation can be done efficiently
    when dealing with function results)
  * for the above, renamed paramanager.alloc/freeparaloc() to
    paramanager.alloc/freecgpara(), and use paramanager.allocparaloc()
    to allocate individual pcgparalocations instead
  * fixed the register size of SSE2 function result registers for
    x86-64 (when used for floating point), which results in removing
    a few superfluous "movs? %xmm0,%xmm0" instructions
  * fixed compilation of paramanagers of avr, m68k and mips after r13695
    and also updated them for these new changes

git-svn-id: trunk@15350 -
2010-05-30 21:12:57 +00:00
Jonas Maebe
9bc15a5f61 * renamed a_param_* to a_load_*_cgpara
git-svn-id: trunk@15305 -
2010-05-22 09:07:21 +00:00
Jonas Maebe
d8a902b4c4 * fixed internalerror in case the from-value of a for-loop has loc_jump
(mantis #15750)

git-svn-id: trunk@15071 -
2010-03-26 22:26:25 +00:00
paul
e2fdfc426f compiler: fix safecall result handling if it is placed to a register
git-svn-id: trunk@14952 -
2010-02-28 10:50:40 +00:00
paul
b884d568e1 compiler: fix 64bit safecall handling
git-svn-id: trunk@14949 -
2010-02-27 16:15:03 +00:00
paul
cf0a1b1af9 compiler: safecall exception handling:
- move safecall exception handling from codegenerator to generate_except_block method (which generates nodes)
  - call SafeCallException for safecall methods which raises an exception

git-svn-id: trunk@14940 -
2010-02-26 10:17:24 +00:00
florian
c1f3d8dcaa * unified names of system_*/systems_* sets
git-svn-id: trunk@14566 -
2010-01-07 18:16:20 +00:00
Jonas Maebe
4838ebe73b * renamed mark_InlineStart/mark_InlineEnd to mark_NoLineinfoStart/
mark_NoLineinfoEnd
  * add "no line info" markers for try/except and try/finally internal cleanup
    code, so the debugger doesn't jump back and forth between the end and start
    of exception blocks when you arrive at the end
  * honour "no line info" markers in dbgdwarf.pas

git-svn-id: trunk@14327 -
2009-12-04 19:37:22 +00:00
Jonas Maebe
d644fc87a1 * handle loc_jump/loc_flags in for-loop upper bound (mantis #15169)
git-svn-id: trunk@14261 -
2009-11-23 21:11:16 +00:00
Jonas Maebe
1f6ec379de * moved field definitions before method/property definitions (see mantis
#13971)

git-svn-id: trunk@13330 -
2009-06-27 11:27:31 +00:00
Jonas Maebe
c1e327c8c4 * insert register allocation/deallocation information for the exception
handling code (g_exception_reason_load() now automatically allocates
    the cpu register that it loads). This fixes the bug reported in
    http://lists.freepascal.org/lists/fpc-devel/2009-May/016773.html
    on i386 with -O2)

git-svn-id: trunk@13105 -
2009-05-06 19:11:54 +00:00
Jonas Maebe
7d459cf12a * the compiler now explicitly keeps track of the minimally guaranteed
alignment for each memory reference (mantis #12137, and
    test/packages/fcl-registry/tregistry1.pp on sparc). This also
    enables better code generation for packed records in many cases.
  o several changes were made to the compiler to minimise the chances
    of accidentally forgetting to set the alignment of memory references
    in the future:
    - reference_reset*() now has an extra alignment parameter
    - location_reset() can now only be used for non LOC_(C)REFERENCE,
      use location_reset_ref() for those (split the tloc enum so the
      compiler can catch errors using range checking)

git-svn-id: trunk@12719 -
2009-02-08 13:00:24 +00:00
Jonas Maebe
718694d1d6 * always specify an explicit alignment for tgobj.gettemp (so e.g.
shortstring temps don't get maximum alignment)
  * changed some gettemptyed() calls into gettemp() calls (gettemptyped
    means that this temp can only be used for temps of that type,
    which is necessary for refcounted types but not for floats)

git-svn-id: trunk@12036 -
2008-11-08 22:20:47 +00:00
Jonas Maebe
a23630260b + "weakexternal" support for imported procedures and variables.
the syntax is exactly the same as for "external", except for
    the keyword. It is currently only active for Darwin targets.
    It should also work at least for Linux targets, but only with
    the GNU assembler (which is why it is not activated there)
  + test for this functionality

git-svn-id: trunk@12009 -
2008-11-01 18:38:32 +00:00
Jonas Maebe
c65f8e7652 * safecall is only supposed to do anything for Windows targets. Removed
partial implementation for other targets and run test only for Windows

git-svn-id: trunk@11593 -
2008-08-16 14:46:20 +00:00
florian
050627ea1d * allow goto inside finally blocks
git-svn-id: trunk@11392 -
2008-07-17 19:49:38 +00:00
Jonas Maebe
2348c7ea00 * allow ssa to happen for outer for-loop control variables
git-svn-id: trunk@11320 -
2008-07-03 19:39:23 +00:00
Jonas Maebe
f36e5411af * split cpu64bit compiler define into
a) cpu64bitaddr, which means that we are generating a compiler which
       will generate code for targets with a 64 bit address space/abi
    b) cpu64bitalu, which means that we are generating a compiler which
       will generate code for a cpu with support for 64 bit integer
       operations (possibly running in a 32 bit address space, depending
       on the cpu64bitaddr define)
   All cpus which had cpu64bit set now have both the above defines set,
   and none of the 32 bit cpus have cpu64bitalu set (and none will
   compile with it currently)
  + pint and puint types, similar to aint/aword (not pword because that
    that conflicts with pword=^word)
  * several changes from aint/aword to pint/pword
  * some changes of tcgsize2size[OS_INT] to sizeof(pint)

git-svn-id: trunk@10320 -
2008-02-13 20:44:00 +00:00
Jonas Maebe
1f1a43d7f0 * fixed crash when compiling inlined procedures containing labels
git-svn-id: trunk@9804 -
2008-01-20 10:29:04 +00:00
daniel
4ca3a590e8 * Labels that are defined or referenced in a different lexical level
than their declaration need to be code generated as global asmlabels.

git-svn-id: trunk@9791 -
2008-01-19 13:50:58 +00:00
peter
6058b2c247 * renamed t_times to executionweight and moved it to
code generation pass
  * include executionweight in the spilling decision

git-svn-id: trunk@9330 -
2007-11-25 16:38:05 +00:00
peter
1dd62b8686 * write also asmblocklabel for a connected labelsym
git-svn-id: trunk@8159 -
2007-07-23 21:01:32 +00:00
yury
9844174e5e * It is needed to destroy exception object in safecall procedure as well.
git-svn-id: trunk@7722 -
2007-06-18 17:26:15 +00:00
yury
78fe68859b * Pop the last exception object in safecall procedure.
git-svn-id: trunk@7721 -
2007-06-18 16:12:49 +00:00
yury
a7d1508959 * Fixed safecall procedures by generating implicit try/finally and setting correct return value if exception was occurred or not. Now safecall is fully Delphi compatible.
git-svn-id: trunk@7720 -
2007-06-18 14:26:08 +00:00
daniel
9adb202a92 * Rework the constexprint to allow operations from low(int64) to high(qword).
+ Some initial work on a formaldef which also carries the typinfo of a parameter.

git-svn-id: trunk@7639 -
2007-06-13 07:41:18 +00:00
Jonas Maebe
fdc813db9a * no need to synchronise the upper bound of a for-loop after the loop
anymore if it is a regvar, because it cannot be a regvar anymore
    after the fix for tw8883

git-svn-id: trunk@7358 -
2007-05-16 12:13:28 +00:00
Jonas Maebe
2c5a284060 * upper bound of for-loop cannot be a regvar, because the regvar may
be modified inside the loop (mantis #8883)

git-svn-id: trunk@7352 -
2007-05-15 20:25:30 +00:00
Jonas Maebe
21f87ef1de * fixed instances of passing properties as var parameter
git-svn-id: trunk@7249 -
2007-05-03 11:02:38 +00:00
peter
658c46b903 * remove tdictionary and tindexarray
* symtables based on TFPHashObjectList and TFPObjectList
  * rename torddef.typ to torddef.ordtype
  * rename tfloatdef.typ to tfloatdef.floattype
  * rename tdef.deftype to tdef.typ
  * remove obsolete browser code, browcol is kept so the ide
    can still be compiled

git-svn-id: trunk@5192 -
2006-11-03 00:30:30 +00:00
Jonas Maebe
97cd8c4f62 * no SSU in try/except/finally (could be enabled in finally in theory,
but requires some changes)

git-svn-id: trunk@5182 -
2006-11-02 17:02:03 +00:00
florian
ef7bd58c54 * germanism removed (aktfilepos -> current_filepos)
git-svn-id: trunk@5099 -
2006-10-30 22:37:31 +00:00
florian
85d63d9fa9 * settings refactored
git-svn-id: trunk@5094 -
2006-10-30 18:02:58 +00:00
peter
3078a1927f * remove ttype
* rename old ttype variables *type to *def
  * rename resulttypepass to pass_typecheck
  * rename pass_2 to pass_generate_code

git-svn-id: trunk@5077 -
2006-10-29 22:19:39 +00:00
Jonas Maebe
c25aba7592 * cleaned up subsetreg support (put everything in a record)
* prepared support for elements with arbitrary bit length (as opposed
    to a multiple of 8)

git-svn-id: trunk@4324 -
2006-08-01 20:39:53 +00:00
Jonas Maebe
e531c4f125 + very limited SSA support (only outside control flow
constructs for now)

git-svn-id: trunk@3671 -
2006-05-25 11:45:08 +00:00
Jonas Maebe
4c41e2c677 * fixed IE when using a field of a record regvar as loop control variable
(the generated code is far from optimal, need to find a way to make such
     records non regable)

git-svn-id: trunk@3639 -
2006-05-23 08:55:45 +00:00
peter
0ec2921bbe * split newasmsymbol to refasmsymbol and defineasmsymbol
git-svn-id: trunk@3057 -
2006-03-27 11:45:18 +00:00
peter
b7fe6797bf Merged revisions 2921-2922,2925 via svnmerge from
http://svn.freepascal.org/svn/fpc/branches/linker/compiler

........
r2921 | peter | 2006-03-15 08:35:00 +0100 (Wed, 15 Mar 2006) | 2 lines

  * pass ObjectWriter to ObjectOuput

........
r2922 | peter | 2006-03-15 12:40:30 +0100 (Wed, 15 Mar 2006) | 2 lines

  * refactor asmdata

........
r2925 | peter | 2006-03-15 16:09:39 +0100 (Wed, 15 Mar 2006) | 3 lines

  * add cfi to asmdata
  * move asmlist, asmcfi, asmdata to own unit

........

git-svn-id: trunk@2932 -
2006-03-16 08:52:22 +00:00
peter
b6e35a200e * rewrite of optimizer options
git-svn-id: trunk@2901 -
2006-03-13 09:05:50 +00:00
Jonas Maebe
ac24965242 * procedures which use goto/label revert to the older and less
efficient regvar allocation scheme
  * also allocate regvars before the loops, not just afterwards

git-svn-id: trunk@2192 -
2006-01-05 23:45:00 +00:00