libc cleanup #1568
libc cleanup #1568
Conversation
| #define M_SQRT2 1.41421356237309504880f | ||
| #define FLT_MAX 340282346638528859811704183484516925440.0f | ||
| #define MAXFLOAT 3.40282347e+38f | ||
| #define SHT_MAX 32767.0f |
There was a problem hiding this comment.
is this a known original name or a standard name for this?
am fine with it if so, but it just goes against our convention for similar things
There was a problem hiding this comment.
This is the standard C name for the largest representable float define in math.h.
FLT_MAX is another standard C name, but is part of float.h instead, which we don't currently have. It might be nice to add additional standard headers in the future, but there's some questions about exactly how much to add so I opted not to add too many new headers for now.
There was a problem hiding this comment.
ISO C has FLT_MAX but not MAXFLOAT
https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3096.pdf
MAXFLOAT seems to be POSIX
https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/math.h.html
There was a problem hiding this comment.
Specifically ISO C has FLT_MAX in float.h, which is not a header we currently have. I think it may be nice to add float.h and limits.h at some point, or I can attempt to add them now if it's much desired?
There was a problem hiding this comment.
I'm fine with the way it is now personally
include/ultra64/libc.h
Outdated
There was a problem hiding this comment.
What is the meaning of this file name?
I see you put bzero/bcmp/bcopy in here which apparently were posix and today are deprecated/removed
There was a problem hiding this comment.
idk, it's a libultra header so I could only speculate
|
|
||
| if (*(u16*)&px->v.ll & 0x8000) { | ||
| if (LDSIGN(px->v.ld)) { |
There was a problem hiding this comment.
why make a macro for this? and is there a reason to not do ld < 0 (besides, for us, matching)
this with the macro isdigit (not IS_DIGIT) makes me wonder if you yoinked this code from somewhere or are looking at a reference?
There was a problem hiding this comment.
I guess they wanted to avoid float operations, possibly for NaN/infinity handling. I tried to sync a bit more with ultralib's xprintf hence the macros
src/libultra/libc/ll.c
Outdated
| return l / r; | ||
| } | ||
|
|
||
| s64 __ll_lshift(s64 l, s64 r) { | ||
| unsigned long long __ll_lshift(unsigned long long l, unsigned long long r) { |
There was a problem hiding this comment.
you changed the types from signed to unsigned here and in other functions below, was that intended?
There was a problem hiding this comment.
I wanted to better match ultralib, but looking at the libultra debug info the current types in ultralib seem a bit wrong
| @@ -46,7 +46,7 @@ s32 Overlay_Load(uintptr_t vromStart, uintptr_t vromEnd, void* vramStart, void* | |||
| // "Clear BSS area (% 08x-% 08x)" | |||
| osSyncPrintf("BSS領域をクリアします(%08x-%08x)\n", end, end + ovlRelocs->bssSize); | |||
| } | |||
| bzero((void*)end, ovlRelocs->bssSize); | |||
| bzero((void*)end, (s32)ovlRelocs->bssSize); | |||
Mostly contains the libc changes from #1468, to try and cut that one down to be more manageable.