aboutsummaryrefslogtreecommitdiff
path: root/tccgen.c
Commit message (Collapse)AuthorAgeFilesLines
* avoid "decl after statement" pleasegrischka2013-04-301-5/+6
| | | | for compiling tcc with msc
* Improved variable length array support.James Lyon2013-04-271-12/+66
| | | | | | | | | | | | | | | VLA storage is now freed when it goes out of scope. This makes it possible to use a VLA inside a loop without consuming an unlimited amount of memory. Combining VLAs with alloca() should work as in GCC - when a VLA is freed, memory allocated by alloca() after the VLA was created is also freed. There are some exceptions to this rule when using goto: if a VLA is in scope at the goto, jumping to a label will reset the stack pointer to where it was immediately after the last VLA was created prior to the label, or to what it was before the first VLA was created if the label is outside the scope of any VLA. This means that in some cases combining alloca() and VLAs will free alloca() memory where GCC would not.
* 64-bit tests now pass (well, nearly).James Lyon2013-04-241-2/+26
| | | | | tcctest1-3 fail, but this appears to be due to bugs in GCC rather than TCC (from manual inspection of the output).
* Added CMake build system (to facilitate Win64 builds)James Lyon2013-04-211-1/+1
| | | | | | Win32 build and tests work under CMake, however I haven't added install code yet. Win64 build fails due to chkstk.S failing to assemble.
* Fixed 64-bit integer bug introduced by x86-64 ABI work.James Lyon2013-04-191-6/+5
| | | | Now I need to check that the x86-64 stuff still works.
* Improved x86-64 XMM register argument passing.James Lyon2013-04-191-1/+1
| | | | | | | Also made XMM0-7 available for use as temporary registers, since they are not used by the ABI. I'd like to do the same with RSI and RDI but that's trickier since they can be used by gv() as temporary registers and there isn't a way to disable that.
* Fixed problems with XMM1 use on Linux/x86-64.James Lyon2013-04-191-4/+5
| | | | | | All tests pass. I think I've caught all the cases assuming only XMM0 is used. I expect that Win64 is horribly broken by this point though, because I haven't altered it to cope with XMM1.
* Most x86-64 tests now work; only on error in test1-3.James Lyon2013-04-191-22/+47
| | | | | | I've had to introduce the XMM1 register to get the calling convention to work properly, unfortunately this has broken a fair bit of code which assumes that only XMM0 is used.
* Got test1-3 working on x86-64.James Lyon2013-04-191-13/+5
| | | | | | There are probably still issues on x86-64 I've missed. I've added a few new tests to abitest, which fail (2x long long and 2x double in a struct should be passed in registers).
* x86-64 ABI fixes.James Lyon2013-04-191-3/+11
| | | | | | | | | abitest now passes; however test1-3 fail in init_test. All other tests pass. I need to re-test Win32 and Linux-x86. I've added a dummy implementation of gfunc_sret to c67-gen.c so it should now compile, and I think it should behave as before I created gfunc_sret.
* Tests in abitest.c now work on Win32.James Lyon2013-04-181-39/+56
| | | | | | | I expect that Linux-x86 is probably fine. All other architectures except ARM are definitely broken since I haven't yet implemented gfunc_sret for these, although replicating the current behaviour should be straightforward.
* tcc.h: declare CValue.tab[LDOUBLE_SIZE/4]grischka2013-02-081-11/+7
| | | | | | | | Should fix some warnings wrt. access out of array bounds. tccelf.c: fix "static function unused" warning x86_64-gen.c: fix "ctype.ref uninitialzed" warning and cleanup tcc-win32.txt: remove obsolete limitation notes.
* Revert "Add predictability in CType initialization."Thomas Preud'homme2013-01-311-6/+0
| | | | This reverts commit 93785149ed61b7165510ff30149f297cf0821c88.
* safety: replace occurrences of strcpy by pstrcpygrischka2013-01-311-1/+1
|
* Revert "Check whether structure fields have a type"Thomas Preud'homme2013-01-311-2/+1
| | | | This reverts commit 981eb84d8ab91aafe73d5eeb218ffe31d28b1014.
* Check whether structure fields have a typeThomas Preud'homme2013-01-311-1/+2
|
* Add predictability in CType initialization.Domingo Alvarez Duarte2013-01-311-0/+6
| | | | | | Initialize the ref field to 0 when manipulating a CType. Signed-off-by: Thomas Preud'homme <robotux@celest.fr>
* Revert "Optimize vswap()"grischka2013-01-141-29/+8
| | | | | | | | | | This reverts commit 63193d1794b037eb4f3b6551596fa8a6d423e0c3. Had some problems (_STATIC_ASSERT) and was too ugly anyway. For retry, I'd suggest to implement a general function static inline void memswap (void *p1, void* p2, size_t n); and then use that. If you do so, please keep the original code as comment.
* Fix C99ism in vswap()Thomas Preud'homme2013-01-131-1/+2
| | | | | Declare vtopl in vswap at the beginning of the function before any assignments. Doing otherwise means C99 is assumed when compiling.
* tccpp: alternative fix for #include_next infinite loop buggrischka2013-01-061-6/+6
| | | | | | | | | | | | | | | This replaces commit 3d409b08893873b917ccb8c34398bc41a4e84d7c - revert old fix in libtcc.c - #include_next: look up the file in the include stack to see if it is already included. Also: - streamline include code - remove 'type' from struct CachedInclude (obsolete because we check full filename anyway) - remove inc_type & inc_filename from struct Bufferedfile (obsolete) - fix bug with TOK_FLAG_ENDIF not being reset - unrelated: get rid of an 'variable potentially uninitialized' warning
* Optimize vswap()Kirill Smelkov2012-12-211-5/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | vswap() is called often enough and shows in profile and it was easy to hand optimize swapping vtop[-1] and vtop[0] - instead of large (28 bytes on i386) tmp variable and two memory to memory copies, let's swap areas by longs through registers with streamlined assembly. For $ ./tcc -B. -bench -DONE_SOURCE -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" -c tcc.c before: # Overhead Command Shared Object Symbol # ........ ........... ................... .............................................. # 15.19% tcc tcc [.] next_nomacro1 5.19% tcc libc-2.13.so [.] _int_malloc 4.57% tcc tcc [.] next 3.36% tcc tcc [.] tok_str_add2 3.03% tcc tcc [.] macro_subst_tok 2.93% tcc tcc [.] macro_subst 2.53% tcc tcc [.] next_nomacro_spc 2.49% tcc tcc [.] vswap 2.36% tcc libc-2.13.so [.] _int_free │ ST_FUNC void vswap(void) │ { 1,96 │ push %edi 2,65 │ push %esi 1,08 │ sub $0x20,%esp │ SValue tmp; │ │ /* cannot let cpu flags if other instruction are generated. Also │ avoid leaving VT_JMP anywhere except on the top of the stack │ because it would complicate the code generator. */ │ if (vtop >= vstack) { 0,98 │ mov 0x8078cac,%eax │ cmp $0x8078d3c,%eax 1,18 │ ┌──jb 24 │ │ int v = vtop->r & VT_VALMASK; 1,08 │ │ mov 0x8(%eax),%edx 0,78 │ │ and $0x3f,%edx │ │ if (v == VT_CMP || (v & ~1) == VT_JMP) 0,78 │ │ cmp $0x33,%edx 0,69 │ │↓ je 54 0,59 │ │ and $0xfffffffe,%edx 0,49 │ │ cmp $0x34,%edx 0,29 │ │↓ je 54 │ │ gv(RC_INT); │ │ } │ │ tmp = vtop[0]; 1,08 │24:└─→lea 0x4(%esp),%edi 0,39 │ mov $0x7,%ecx │ mov %eax,%esi 14,41 │ rep movsl %ds:(%esi),%es:(%edi) │ vtop[0] = vtop[-1]; 9,51 │ lea -0x1c(%eax),%esi 1,96 │ mov $0x7,%cl │ mov %eax,%edi 17,06 │ rep movsl %ds:(%esi),%es:(%edi) │ vtop[-1] = tmp; 10,20 │ mov 0x8078cac,%edi 2,35 │ sub $0x1c,%edi 0,78 │ lea 0x4(%esp),%esi │ mov $0x7,%cl 15,20 │ rep movsl %ds:(%esi),%es:(%edi) │ } 9,90 │ add $0x20,%esp 2,25 │ pop %esi 1,67 │ pop %edi 0,69 │ ret after: # Overhead Command Shared Object Symbol # ........ ........... ................... .............................................. # 15.27% tcc tcc [.] next_nomacro1 5.08% tcc libc-2.13.so [.] _int_malloc 4.57% tcc tcc [.] next 3.17% tcc tcc [.] tok_str_add2 3.12% tcc tcc [.] macro_subst 2.99% tcc tcc [.] macro_subst_tok 2.43% tcc tcc [.] next_nomacro_spc 2.32% tcc libc-2.13.so [.] _int_free . . . 0.71% tcc tcc [.] vswap │ ST_FUNC void vswap(void) │ { 7,22 │ push %eax │ /* cannot let cpu flags if other instruction are generated. Also │ avoid leaving VT_JMP anywhere except on the top of the stack │ because it would complicate the code generator. */ │ if (vtop >= vstack) { 11,34 │ mov 0x8078cac,%eax 2,75 │ cmp $0x8078d3c,%eax 0,34 │ ┌──jb 20 │ │ int v = vtop->r & VT_VALMASK; 0,34 │ │ mov 0x8(%eax),%edx 8,93 │ │ and $0x3f,%edx │ │ if (v == VT_CMP || (v & ~1) == VT_JMP) 2,06 │ │ cmp $0x33,%edx 2,41 │ │↓ je 74 2,41 │ │ and $0xfffffffe,%edx 0,34 │ │ cmp $0x34,%edx 2,41 │ │↓ je 74 │ │ vtopl[-1*VSIZEL + i] = tmpl; \ │ │ } do {} while (0) │ │ │ │ VSWAPL(15); VSWAPL(14); VSWAPL(13); VSWAPL(12); │ │ VSWAPL(11); VSWAPL(10); VSWAPL( 9); VSWAPL( 8); │ │ VSWAPL( 7); VSWAPL( 6); VSWAPL( 5); VSWAPL( 4); 2,06 │20:└─→mov 0x18(%eax),%edx 1,37 │ mov -0x4(%eax),%ecx 2,06 │ mov %ecx,0x18(%eax) 1,37 │ mov %edx,-0x4(%eax) 2,06 │ mov 0x14(%eax),%edx 2,06 │ mov -0x8(%eax),%ecx 2,41 │ mov %ecx,0x14(%eax) 3,09 │ mov %edx,-0x8(%eax) 3,09 │ mov 0x10(%eax),%edx 1,72 │ mov -0xc(%eax),%ecx 2,75 │ mov %ecx,0x10(%eax) 1,72 │ mov %edx,-0xc(%eax) │ VSWAPL( 3); VSWAPL( 2); VSWAPL( 1); VSWAPL( 0); 2,41 │ mov 0xc(%eax),%edx 2,41 │ mov -0x10(%eax),%ecx 2,41 │ mov %ecx,0xc(%eax) 0,69 │ mov %edx,-0x10(%eax) 1,72 │ mov 0x8(%eax),%edx 0,69 │ mov -0x14(%eax),%ecx 1,03 │ mov %ecx,0x8(%eax) 1,37 │ mov %edx,-0x14(%eax) 1,37 │ mov 0x4(%eax),%edx 0,69 │ mov -0x18(%eax),%ecx 3,09 │ mov %ecx,0x4(%eax) 2,06 │ mov %edx,-0x18(%eax) 1,37 │ mov (%eax),%edx 2,41 │ mov -0x1c(%eax),%ecx 1,37 │ mov %ecx,(%eax) 4,12 │ mov %edx,-0x1c(%eax) │ } │ │ # undef VSWAPL │ # undef VSIZEL │ } 1,03 │ pop %eax 3,44 │ ret Overal speedup: # best of 5 runs before: 8268 idents, 47203 lines, 1526763 bytes, 0.148 s, 319217 lines/s, 10.3 MB/s after: 8273 idents, 47231 lines, 1527685 bytes, 0.146 s, 324092 lines/s, 10.5 MB/s Static ASSERT macro taken from CCAN's[1] build_assert[2] which is in public domain. [1] http://ccodearchive.net/ [2] http://git.ozlabs.org/?p=ccan;a=blob;f=ccan/build_assert/build_assert.h;h=24e59c44cd930173178ac9b6e101b0af64a879e9;hb=HEAD
* Make tcc work after self-compiling with bounds-check enabledKirill Smelkov2012-12-091-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | For vstack Fabrice used the trick to initialize vtop to &vstack[-1], so that on first push, vtop becomes &vstack[0] and a value is also stored there - everything works. Except that when tcc is compiled with bounds-checking enabled, vstack - 1 returns INVALID_POINTER and oops... Let's workaround it with artificial 1 vstack slot which will not be used, but only serve as an indicator that pointing to &vstack[-1] is ok. Now, tcc, after being self-compiled with -b works: $ ./tcc -B. -o tccb -DONE_SOURCE -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" tcc.c -ldl $ cd tests $ ../tcc -B.. -run tcctest.c >1 $ ../tccb -B.. -run tcctest.c >2 $ diff -u 1 2 and note, tcc's compilation speed is not affected: $ ./tcc -B. -bench -DONE_SOURCE -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" -c tcc.c before: 8270 idents, 47221 lines, 1527730 bytes, 0.152 s, 309800 lines/s, 10.0 MB/s after: 8271 idents, 47221 lines, 1527733 bytes, 0.152 s, 310107 lines/s, 10.0 MB/s But note, that `tcc -b -run tcc` is still broken - for example it crashes on $ cat x.c double get100 () { return 100.0; } $ ./tcc -B. -b -DTCC_TARGET_I386 -DCONFIG_MULTIARCHDIR=\"i386-linux-gnu\" -run \ -DONE_SOURCE ./tcc.c -B. -c x.c Runtime error: dereferencing invalid pointer ./tccpp.c:1953: at 0xa7beebdf parse_number() (included from ./libtcc.c, ./tcc.c) ./tccpp.c:3003: by 0xa7bf0708 next() (included from ./libtcc.c, ./tcc.c) ./tccgen.c:4465: by 0xa7bfe348 block() (included from ./libtcc.c, ./tcc.c) ./tccgen.c:4440: by 0xa7bfe212 block() (included from ./libtcc.c, ./tcc.c) ./tccgen.c:5529: by 0xa7c01929 gen_function() (included from ./libtcc.c, ./tcc.c) ./tccgen.c:5767: by 0xa7c02602 decl0() (included from ./libtcc.c, ./tcc.c) that's because lib/bcheck.c runtime needs more fixes -- see next patches.
* Add support for __builtin_frame_address(level)Kirill Smelkov2012-11-161-6/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Continuing d6072d37 (Add __builtin_frame_address(0)) implement __builtin_frame_address for levels greater than zero, in order for tinycc to be able to compile its own lib/bcheck.c after cffb7af9 (lib/bcheck: Prevent __bound_local_new / __bound_local_delete from being miscompiled). I'm new to the internals, and used the most simple way to do it. Generated code is not very good for levels >= 2, compare gcc tcc level=0 mov %ebp,%eax lea 0x0(%ebp),%eax level=1 mov 0x0(%ebp),%eax mov 0x0(%ebp),%eax level=2 mov 0x0(%ebp),%eax mov 0x0(%ebp),%eax mov (%eax),%eax mov %eax,-0x10(%ebp) mov -0x10(%ebp),%eax mov (%eax),%eax level=3 mov 0x0(%ebp),%eax mov 0x0(%ebp),%eax mov (%eax),%eax mov (%eax),%ecx mov (%eax),%eax mov (%ecx),%eax But this is still an improvement and for bcheck we need level=1 for which the code is good. For the tests I had to force gcc use -O0 to not inline the functions. And -fno-omit-frame-pointer just in case. If someone knows how to improve the generated code - help is appreciated. Thanks, Kirill Cc: Michael Matz <matz@suse.de> Cc: Shinichiro Hamaji <shinichiro.hamaji@gmail.com>
* forbid invalid comparison of structHitoshi Mitake2012-11-051-0/+2
| | | | | Current tcc permits comparison of structs and comparison between struct and other typed values.
* Fix commit 85f6fad3a6acbfa07a3dc45b673965fc26890d8eThomas Preud'homme2012-10-251-1/+2
| | | | | Don't reset nocode_wanted with saved_nocode_wanted if it hasn't been modified (and hence saved_nocode_wanted is uninitialized).
* Error out in case of variable name clashThomas Preud'homme2012-10-251-3/+21
| | | | | Error out when two local variable with same name are defined in the same scope. This fixes bug #15597 in savannah's BTS.
* Forbid VLA as static variablesThomas Preud'homme2012-10-251-1/+6
| | | | | | | | | | | Currently, VLA are not forbidden for static variable. This leads to problems even if for fixed-size array when the size expression uses the ternary operator (cond ? then-value : else-value) because it is parsed as a general expression which leads to code generated in this case. This commit solve the problem by forbidding VLA for static variables. Although not required for the fix, avoiding code generation when the expression is constant would be a nice addition though.
* get_reg(): try to free r2 for an SValue firstThomas Preud'homme2012-07-111-4/+5
| | | | | | | | | | To be able to load a long long value correctly on i386, gv() rely on the fact that when get_reg() look at an SValue it tries first to free the register in r2 and then r. More information about the context can be found at http://lists.nongnu.org/archive/html/tinycc-devel/2012-06/msg00017.html and later at http://lists.nongnu.org/archive/html/tinycc-devel/2012-07/msg00021.html
* Incorrect shift result type on unsigned short first argument.Vincent Lefevre2012-07-061-1/+3
| | | | | | The code for shifts is now similar to code for binary arithmetic operations, except that only the first argument is considered, as required by the ISO C standard.
* Incorrect shift result type with 64-bit ABIVincent Lefevre2012-06-271-0/+3
| | | | | | | | | | | | | | | On 2012-06-26 15:07:57 +0200, Vincent Lefevre wrote: > ISO C99 TC3 says: [6.5.7#3] "The integer promotions are performed on > each of the operands. The type of the result is that of the promoted > left operand." I've written a patch (attached). Now the shift problems no longer occur with the testcase and with GNU MPFR's "make check". -- Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/> Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)
* Add support for arm hardfloat calling conventionThomas Preud'homme2012-06-051-6/+14
| | | | | See Procedure Call Standard for the ARM Architecture (AAPCS) for more details.
* Fix removal of vnrottThomas Preud'homme2012-06-051-1/+1
| | | | Make vrotb ST_FUNC so that arm-gen.c can use vrotb.
* Fix comparing comparisonsMichael Matz2012-04-181-0/+8
| | | | | | | | | | | Sometimes the result of a comparison is not directly used in a jump, but in arithmetic or further comparisons. If those further things do a vswap() with the VT_CMP as current top, and then generate instructions for the new top, this most probably destroys the flags (e.g. if it's a bitfield load like in the example). vswap() must do the same like vsetc() and not allow VT_CMP vtops to be moved down.
* Make sizeof() be of type size_tMichael Matz2012-04-181-3/+14
| | | | | | | | This matters when sizeof is directly used in arithmetic, ala "uintptr_t t; t &= -sizeof(long)" (for alignment). When sizeof isn't size_t (as it's specified to be) this masking will truncate the high bits of the uintptr_t object (if uintptr_t is larger than uint).
* Fix detection of labels with a typedef nameMichael Matz2012-04-181-1/+20
| | | | | | | | | | | | | | This needs to be accepted: typedef int foo; void f (void) { foo: return; } namespaces for labels and types are different. The problem is that the block parser always tries to find a decl first and that routine doesn't peek enough to detect this case. Needs some adjustments to unget_tok() so that we can call it even when we already called it once, but next() didn't come around restoring the buffer yet. (It lazily does so not when the buffer becomes empty, but rather when the next call detects that the buffer is empty, i.e. it requires two next() calls until the unget buffer gets switched back).
* Fix bitfield loads into char/short.Michael Matz2012-04-181-2/+3
| | | | | | Removes a premature optimization of char/short loads rewriting the source type. It did so also for bitfield loads, thereby removing all the shifts/maskings.
* Fix conversion in a?0:ptr.Michael Matz2012-04-181-5/+14
| | | | | | (cond ? 0 : ptr)->member wasn't handled correctly. If one arm is a null pointer constant (which also can be a pointer) the result type is that of the other arm.
* Remove vnrott (duplicate vrotb)Thomas Preud'homme2012-03-141-16/+0
|
* Error out when assigning void value.Thomas Preud'homme2012-01-221-0/+2
| | | | | | | | tcc should now error out when compiling code like: VOID ExitProcess(UINT uExitCode); (…) retCode = ExitProcess(pi.dwProcessId);
* rename error/warning -> tcc_(error/warning)grischka2011-08-111-75/+75
|
* x86-64: fix flags and zero-pad long doublesgrischka2011-08-061-1/+5
| | | | | | | | | | This fixes a bug introduced in commit 8d107d9ffd8126d82b1c56be47431a6bcef39f08 that produced wrong code because of interference between 0x10 bits VT_CONST and x86_64-gen.c:TREG_MEM Also fully zero-pad long doubles on x86-64 to avoid random bytes in output files which disturb file comparison.
* Accept colon separated paths with -L and -Igrischka2011-08-011-0/+9
| | | | | | | | | | | | | | | | | | This allows passing colon separated paths to tcc_add_library_path tcc_add_sysinclude_path tcc_add_include_path Also there are new configure variables CONFIG_TCC_LIBPATH CONFIG_TCC_SYSINCLUDE_PATHS which define the lib/sysinclude paths all in one and can be overridden from configure/make For TCC_TARGET_PE semicolons (;) are used as separators Also, \b in the path string is replaced by s->tcc_lib_path (CONFIG_TCCDIR rsp. -B option)
* Revert "Force const. expr. in expr_cond outside function"Thomas Preud'homme2011-07-311-1/+1
| | | | | This reverts commit b2f5ee9b2de40e26ba3d92cf3950be6da7766b7d as it's useless on mob.
* Force const. expr. in expr_cond outside functionThomas Preud'homme2011-07-311-1/+1
| | | | | | Since no code should be generated outside a function, force expr_cond to only consider constant expression when outside a function since the generic code can generate some code.
* re-added negative-array-size testcase and fixed fix for itJoe Soroka2011-07-221-3/+3
|
* Revert "better constant handling for expr_cond"grischka2011-07-161-4/+1
| | | | | It produced wrong code with one of my test projects. This reverts commit cd3d1a45f34be3c9a358f1743bcaf8aac897992e.
* win64: va_arg with structuresgrischka2011-07-141-0/+4
|
* tccgen: reset aligned attribute for next typegrischka2011-07-141-0/+1
| | | | | | | | | | | | Basically, with: typedef __attribute__((aligned(16))) struct _xyz { ... } xyz, *pxyz; we want the struct aligned but not the pointer. FIXME: This patch is a hack, waiting for someone in the knowledge of correct __attribute__ semantics.
* handle arrays with a flexible member but no initializerJoe Soroka2011-07-111-1/+1
|
* better constant handling for expr_condJoe Soroka2011-07-111-1/+4
|