[GPU] Remove register reinterpret_casts + WAIT_REG_MEM volatility

Hopefully prevents some potential #1971-like situations.

WAIT_REG_MEM's implementation also allowed the compiler to load the value
only once, which caused an infinite loop with the other changes in the
commit (even in debug builds), so it's now accessed as volatile. Possibly
it would be even better to replace it with some (acquire/release?) atomic
load/store some day at least for the registers actually seen as
participating in those waits.

Also fixes the endianness being handled only on the first wait iteration in
WAIT_REG_MEM.
This commit is contained in:
Triang3l
2024-05-12 17:17:30 +03:00
parent f0ad4f4587
commit 376bad5056
19 changed files with 336 additions and 318 deletions

View File

@@ -17,6 +17,7 @@
#include "xenia/base/assert.h"
#include "xenia/base/math.h"
#include "xenia/base/memory.h"
namespace xe {
namespace gpu {
@@ -1102,10 +1103,10 @@ struct Src : OperandAddress {
}
static Src LI(int32_t x) { return LI(x, x, x, x); }
static Src LF(float x, float y, float z, float w) {
return LU(*reinterpret_cast<const uint32_t*>(&x),
*reinterpret_cast<const uint32_t*>(&y),
*reinterpret_cast<const uint32_t*>(&z),
*reinterpret_cast<const uint32_t*>(&w));
return LU(xe::memory::Reinterpret<uint32_t>(x),
xe::memory::Reinterpret<uint32_t>(y),
xe::memory::Reinterpret<uint32_t>(z),
xe::memory::Reinterpret<uint32_t>(w));
}
static Src LF(float x) { return LF(x, x, x, x); }
static Src LP(const uint32_t* xyzw) {
@@ -1222,12 +1223,10 @@ struct Src : OperandAddress {
bool negate) {
if (is_integer) {
if (absolute) {
*reinterpret_cast<int32_t*>(&value) =
std::abs(*reinterpret_cast<const int32_t*>(&value));
value = uint32_t(std::abs(int32_t(value)));
}
if (negate) {
*reinterpret_cast<int32_t*>(&value) =
-*reinterpret_cast<const int32_t*>(&value);
value = uint32_t(-int32_t(value));
}
} else {
if (absolute) {