| Commit message (Collapse) | Author | Age | Files | Lines | ||
|---|---|---|---|---|---|---|
| ... | ||||||
| * | lexer: more small optimizations | 2025-12-21 | 1 | -14/+33 | ||
| | | ||||||
| * | rega: fix 3ff0bfcb | 2025-12-21 | 1 | -4/+1 | ||
| | | ||||||
| * | lexer: note source of #include when error occurs in #included file | 2025-12-20 | 2 | -1/+8 | ||
| | | ||||||
| * | isel: fix returning fixarg()'d temp | 2025-12-20 | 1 | -4/+4 | ||
| | | ||||||
| * | c: small typechecking bugfixes | 2025-12-20 | 1 | -3/+3 | ||
| | | ||||||
| * | driver: -fsyntax-only | 2025-12-20 | 2 | -8/+12 | ||
| | | ||||||
| * | rega: fix infinite loop when compiling infinite loop | 2025-12-20 | 2 | -1/+9 | ||
| | | ||||||
| * | some static assertions for packed type sizes | 2025-12-20 | 3 | -6/+11 | ||
| | | ||||||
| * | backend: unify pass memory allocation strategies | 2025-12-20 | 10 | -39/+24 | ||
| | | | | | | | It was all over the place for temporary data structures used by individual passes. Now there is an arena specifically for that, which is nicer. | |||||
| * | bootstrap.sh options | 2025-12-20 | 1 | -7/+8 | ||
| | | ||||||
| * | backend: general simplification pass skeleton | 2025-12-20 | 4 | -2/+155 | ||
| | | ||||||
| * | c: support CRLF newlines | 2025-12-20 | 1 | -3/+11 | ||
| | | ||||||
| * | c: fix a silly C edge case with function redeclaration storage class | 2025-12-20 | 1 | -14/+21 | ||
| | | | | | | | C (or at least clang and gcc) allows declaring `static int f(...);` and defining it later as `int f(...) {...}`, omitting the 'static' in the latter definition. | |||||
| * | copyopt: optimize same-arg phis with multiple preds | 2025-12-20 | 1 | -2/+6 | ||
| | | ||||||
| * | lexer: fix remnant use of TKEOF for character | 2025-12-20 | 1 | -9/+7 | ||
| | | ||||||
| * | quit after emitting too many errors | 2025-12-20 | 1 | -0/+5 | ||
| | | ||||||
| * | optimize lexer a bit more | 2025-12-20 | 4 | -68/+100 | ||
| | | ||||||
| * | lexer: convert pp-idents to keywords only after preprocessing | 2025-12-20 | 2 | -71/+75 | ||
| | | ||||||
| * | ir/regalloc: struct alloc -> union alloc | 2025-12-20 | 1 | -16/+15 | ||
| | | ||||||
| * | c: factor out to tldecl(), improve error recovery, | 2025-12-20 | 1 | -73/+97 | ||
| | | ||||||
| * | emit: fix errata in flagslivep() | 2025-12-20 | 1 | -1/+1 | ||
| | | | | | This little typo inhibited a small optimization. Oops | |||||
| * | test/run.sh: shellcheck | 2025-12-20 | 1 | -11/+10 | ||
| | | ||||||
| * | x86-64: peephole optimization for 3-address shift -> lea | 2025-12-20 | 2 | -1/+15 | ||
| | | ||||||
| * | c.h init zero bitset portability | 2025-12-19 | 1 | -1/+1 | ||
| | | ||||||
| * | driver: -w, update help | 2025-12-19 | 3 | -2/+9 | ||
| | | ||||||
| * | c: support 0-length arrays as syntax for flexible array members | 2025-12-19 | 1 | -19/+29 | ||
| | | | | | | Keeping the expr around in the decllist paves the way for VLA support later on | |||||
| * | ir: move some filluses() to ir.c, rename optmem.c -> mem2reg.c | 2025-12-19 | 5 | -31/+30 | ||
| | | ||||||
| * | c: fix incr/decr codegen for floats | 2025-12-19 | 1 | -2/+2 | ||
| | | ||||||
| * | isel: fix wrong instr cls being used in some cases | 2025-12-19 | 1 | -2/+2 | ||
| | | | | | Specifically comparisons where operation cls != result cls | |||||
| * | c: hack to support __FUNCTION__ GNU extension | 2025-12-19 | 1 | -4/+10 | ||
| | | ||||||
| * | cpp: GNU extension for named variadic arguments | 2025-12-18 | 2 | -4/+11 | ||
| | | ||||||
| * | cpp: better error recovery for #define syntax error | 2025-12-18 | 1 | -6/+7 | ||
| | | ||||||
| * | x86-64 abi: fix wrong class being used in float vararg | 2025-12-18 | 1 | -1/+1 | ||
| | | ||||||
| * | ir: move cls2load to interface | 2025-12-18 | 3 | -7/+7 | ||
| | | | | | | There's plenty of code duplication like this around I'm looking to reduce. | |||||
| * | x86-64/emit: remove some superfluous rex.W prefixes | 2025-12-18 | 1 | -2/+2 | ||
| | | ||||||
| * | regalloc+emit: get rid of xsave/xrestore hack | 2025-12-18 | 4 | -58/+64 | ||
| | | | | | | | | Was used for situation where we needed to spill more than 1 temporary and have to use a register that is already used. Instead of push/pop, we can just allocate and set aside specific stack slots for this purpose. Also, reworked linearscan() interval sets to separate FPR/GPR intervals. | |||||
| * | rega: implement stack<->stack swap for parallel moves | 2025-12-18 | 1 | -29/+34 | ||
| | | ||||||
| * | x86_64: for vararg calls, write to EAX in isel | 2025-12-18 | 3 | -22/+30 | ||
| | | | | | Also, in regalloc ensure fixed intervals are sorted | |||||
| * | c: fix uninitiliazed field in decltypes | 2025-12-17 | 1 | -0/+1 | ||
| | | ||||||
| * | nicer defaults and facilities for cross-compilation | 2025-12-17 | 7 | -32/+195 | ||
| | | ||||||
| * | emit: don't put the exit block always at the end of the function | 2025-12-17 | 1 | -1/+6 | ||
| | | ||||||
| * | x86-64/emit: implement single-exit-point ret with jump threading | 2025-12-16 | 3 | -26/+40 | ||
| | | ||||||
| * | bitset: better implementation of bsiter() and stuff | 2025-12-16 | 3 | -11/+17 | ||
| | | | | | Also changed the type to size_t for portability | |||||
| * | mem2reg: fix obvious inefficiency | 2025-12-16 | 1 | -16/+10 | ||
| | | | | | | | | deltrivialphis() was iterating over every variably instead of just looking at the variable being examined. And I'd been wondering why mem2reg was such a bottleneck for a testcase like sqlite3 amalgamation.. it's easy to miss the forest for the trees. | |||||
| * | c: improve some type error diagnostics | 2025-12-15 | 2 | -14/+24 | ||
| | | ||||||
| * | create distinct interned string type | 2025-12-15 | 16 | -129/+131 | ||
| | | | | | | | | | | | | | | | Interned strings are used pervasively, so it's a good idea to add a layer of type safety to differentiate them from general cstrs and avoid potential bugs from comparing non-interned and interned strings. Not that that's happened so far that I can remember, but it could. I'm 90% sure it's legal to alias `struct {char c;}` pointers with `char` pointers. This specific typedef gives type safety but with a simple one-way `internstr -> const char *` typecast (with `&istr->c`). Converting the other way around is more intentional: a straight up cast `(internstr)cstr` which sticks out as unchecked and probably wrong, or calling the intern(cstr) function, which is the right way. | |||||
| * | move intern() to mem.c | 2025-12-15 | 6 | -49/+52 | ||
| | | | | | | Being in lex.c was vestigial, since it was being used all over the frontend and backend. | |||||
| * | regalloc: fix lifetime construction for nested loops | 2025-12-15 | 1 | -17/+34 | ||
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously, given something like ``` 1 a = ... 2 loop { // outer 3 b = do something with a 4 loop { // inner 5 ... 6 if (b < 0) 7 break 'inner; 8 if (b == 0) 9 return; 10 ... 11 } 12 } ``` Regalloc thought outer goes from 2..6, because 6 is the last place where flow jumps directly back to 2. So `a` would have the lifetime [1,7). However if neither the break nor return are taken, the inner loop repeats and then control could flow back to 7 -> 3. But now the physical location for `a` might have been clobbered between 8..10, which is wrong. This fixes that by making sure the outer loop is considered to span 2..10. The way I went about it might not be the best way of doing it. I'm not 100% certain that it's fully correct and will always find the correct loopend, either. It's surprising it took this long to hit this edge case. | |||||
| * | lex: loop with goto instead of tail recursion | 2025-12-15 | 1 | -7/+8 | ||
| | | | | | | This uncovered a edgecase with loops and lifetime construction in regalloc! | |||||
| * | lexer: use a hashmap to lookup keywords | 2025-12-15 | 3 | -28/+29 | ||
| | | ||||||
 |
index : antcc | |
| A little C compiler |
| aboutsummaryrefslogtreecommitdiffhomepage |