aboutsummaryrefslogtreecommitdiff
path: root/expr.c
AgeCommit message (Collapse)Author
2022-02-10Remove one last use of _NoreturnMichael Forney
2022-01-22Handle unary minus specially instead of 0 - xMichael Forney
This is necessary to fix unary negation of floating-point 0 (also depends on a pending qbe patch).
2022-01-22Remove most usage of fixed-width integer typesMichael Forney
We only require a type of at least 64 bits, so just use unsigned long long. Only siphash remains as the last user of uint64_t.
2022-01-22Remove unused type propertiesMichael Forney
2022-01-22Port to C99Michael Forney
2022-01-01Remove dependency on signed integer representationMichael Forney
2021-11-25Add __builtin_unreachable stubMichael Forney
2021-10-21expr: Add support for UTF-8 character constantsMichael Forney
2021-10-21expr: Implement binary integer constantsMichael Forney
These are in the latest C23 draft.
2021-10-20expr: Fix octal escapes followed by octal digitsMichael Forney
2021-10-20expr: Add support for wide string literalsMichael Forney
Thanks to Nihal Jere for his initial patches implementing this feature. Fixes #35.
2021-10-18expr: Make sure __builtin_va_end argument is evaluated for side-effectsMichael Forney
2021-10-03expr: Use end pointer to detect string-to-number conversion failuresMichael Forney
We explicitly ignore ERANGE for strtod, and in any other error case, the end pointer is set to the beginning of the string.
2021-10-03qbe: Remove some unused includesMichael Forney
We now hard-code the float precision in the format string instead of using the float.h macros. The other headers were never used (except maybe prior to the first commit).
2021-09-29Rename uint64_t field of constant union to uMichael Forney
This will prepare us for adding a signed int64_t field called i.
2021-09-29Use unsigned long long for sizes and offsetsMichael Forney
We don't need exact-width integer types here.
2021-09-28expr: Skip codegen for unused expression in conditional with constant expressionMichael Forney
This compiles `0 ? e1 : e2` as `e2`, and `1 ? e1 : e2` as `e1` (while still adjusting the type as necessary).
2021-09-28mkexpr: Add a base parameter to mkexprMichael Forney
2021-09-28Skip unnecessary conversion to bool for logical and conditional expressionsMichael Forney
As in ede6a5c9, if an expression is used only to control a jnz, we don't need to convert it to a 0 or 1 value. QBE ignores the upper 32-bits of the argument to jnz, so the conversion is still needed for pointer, long, and floating point types (including float since -0 has non-zero bit representation).
2021-09-13Revert "Add stringconcat function to concatenate adjacent string literals"Michael Forney
This reverts commit c16f07acf655b9f4fb006d8256b4027fb5a13aa8. This incorrectly allows octal escapes to span between adjacent string literals (e.g. "\0" "1" is not the same as "\01").
2021-09-13Make string literal data unsigned charMichael Forney
2021-09-07expr: Fix varargs again and add more testsMichael Forney
2021-09-06Fix type-checking of va_list arguments to varargs built-insMichael Forney
If the argument was a function parameter, its type has already been adjusted. So on x86_64, we can't just ignore the automatic array-to-pointer conversion, since it was never a pointer to begin with. Instead, keep track of the adjusted va_list type, and check that the arguments to varargs built-ins match that type.
2021-09-04Use architecture-specific va_list typeMichael Forney
Previously, cproc effectively used used typedef struct { /* 32 bytes, 8-byte aligned */ } __builtin_va_list[1]; However, this is not quite correct for x86_64 nor aarch64, though it was close enough for both to work in most cases. In actuality, for x86_64 we want typedef struct { /* 24 bytes, 8-byte aligned */ } __builtin_va_list[1]; and for aarch64 we want typedef struct { /* 32 bytes, 8-byte aligned */ } __builtin_va_list; The difference only appears when the size of va_list matters, or when va_list is passed as a parameter. However, the former is not often the case, and the aarch64 ABI replaces aggregate arguments with pointers to caller-allocated memory, which is quite similar to arrays decaying to pointers in C except that the struct is not copied. Additionally, riscv64 simply uses typedef void *__builtin_va_list; which again has a different size and calling convention. To fix this, make the __builtin_va_list type architecture-specific and use architecture-specific tests for varargs-related functionality.
2021-09-04Prepare for supporting architecture-specific va_list typeMichael Forney
2021-06-28Add stringconcat function to concatenate adjacent string literalsMichael Forney
This function also ensures that the string prefixes (if any) are compatible. It should make it easier to implement wide string support.
2021-05-02expr: Include NUL-terminator in string expression dataMichael Forney
This will facilitate the support of wide-string literals. Based on patch from Nihal Jere.
2020-04-05expr: Just ignore decayed operand in unary `&` operatorMichael Forney
Reusing the decayed expression is more complicated, and only saved one malloc.
2020-04-04expr: Add type checking for equality and relational expressionsMichael Forney
2020-04-04expr: Fix type of '&' operator applied to arrayMichael Forney
It should be a pointer to the array, not to the first element (as it would after implicit conversion without the '&' operator).
2020-03-18expr: Slight simplification in function call parsingMichael Forney
2020-01-30expr: String literals have complete typeMichael Forney
2019-12-27expr: Disallow function/incomplete types and bit-fields in sizeof/_AlignofMichael Forney
2019-08-12Revert "expr: Fix cast of same type"Michael Forney
This reverts commit 1a38a5fc4844a0de8729be694a62ba0afce3ff52. This breaks comparisons bitfields in some cases, for instance extern struct {unsigned x:31;} s; int main(void) { return (unsigned)s.x - 1 < 0; } If we discard the cast, then it is a signed comparison because of integer promotion for bit-fields, otherwise it is an unsigned comparison. Additionally, the test case this was meant to fix is not actually ISO C, since casts must be to scalar types or `void`.
2019-07-11expr: Fix cast of same typeMichael Forney
2019-07-03Implement no-op __builtin_expectMichael Forney
2019-06-27Implement prefixed character constantsMichael Forney
2019-05-12eval: Keep track of kind of constant expression we are evaluatingMichael Forney
When we are evaluating an arithmetic constant expression, we don't want to indroduce static data definitions for string or compound literals. Fixes #59.
2019-04-24Use a common member for expression baseMichael Forney
2019-04-24Use a common member for expression opMichael Forney
2019-04-24expr: Free subexpressions in delexprMichael Forney
2019-04-24Implement _Generic selectionMichael Forney
2019-04-23Fix integer promotion on bit-fieldsMichael Forney
Fixes #47.
2019-04-23Improve some error messagesMichael Forney
2019-04-23expr: Add mkincdecexpr for pre/postfix inc/decrement operatorsMichael Forney
2019-04-23expr: Use tokstr in ++/-- operator error messagesMichael Forney
2019-04-23expr: Add some type checking for unary operatorsMichael Forney
2019-04-23expr: Check qualifiers when parsing ++/-- expressionsMichael Forney
2019-04-22expr: Fix check of not enough arguments for function callMichael Forney
2019-04-21Keep track of type properties in typeMichael Forney
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-15 11:44:32 +0000