// atomic functions (types [Int64] and [Uint64] are automatically aligned).
// The first word in an allocated struct, array, or slice; in a global
// variable; or in a local variable (because the subject of all atomic operations
// will escape to the heap) can be relied upon to be 64-bit aligned.

// SwapInt32 atomically stores new into *addr and returns the previous *addr value.

//   ((unknown): f32[1,2], (unknown): s64[3,4], (unknown): f32[1], (unknown): f32[1], (unknown): s32[5]) -> (u32[5,6], f32[1], s32[5])
//
// Memory stats:
//   arg bytes total:    392
//   arg bytes aligned:  576
//   temp bytes total:   171
//   temp bytes aligned: 512
class MyClass final : public tensorflow::XlaCompiledCpuFunction {
 public:
  // Number of input arguments for the compiled computation.
  static constexpr size_t kNumArgs = 5;

}

// Signal handler for SIGSEGV on a C thread.
static void segvHandler(int signo, siginfo_t* info, void* ctxt) {
	sigset_t mask;
	int i;

	// Call an arbitrary function that requires the stack to be properly aligned.
	callWithVarargs("dummy arg", 3.1415);

	if (sigemptyset(&mask) < 0) {
		die("sigemptyset");
	}
	if (sigaddset(&mask, SIGSEGV) < 0) {
		die("sigaddset");
	}
	i = sigprocmask(SIG_UNBLOCK, &mask, NULL);

	// Typically a non-fixed seed should be used, such as Uint64(), Uint64().
	// Using a fixed seed will produce the same output on every run.
	r := rand.New(rand.NewPCG(1, 2))

	// The tabwriter here helps us generate aligned output.
	w := tabwriter.NewWriter(os.Stdout, 1, 1, 1, ' ', 0)
	defer w.Flush()
	show := func(name string, v1, v2, v3 any) {
		fmt.Fprintf(w, "%s\t%v\t%v\t%v\n", name, v1, v2, v3)
	}

	// On some platforms we need to provide physical page aligned stack
	// allocations. Where the page size is less than the physical page
	// size, we already manage to do this by default.
	needPhysPageAlign := physPageAlignedStacks && typ == spanAllocStack && pageSize < physPageSize

	// If the allocation is small enough, try the page cache!
	// The page cache does not support aligned allocations, so we cannot use

	pgms := []struct {
		text   []byte
		align  string
		hasNop bool
	}{
		{[]byte(x0pgm), "align=0x0", false},     // No alignment or nop adjustments needed
		{[]byte(x16pgm), "align=0x20", false},   // Increased alignment needed
		{[]byte(x32pgm), "align=0x40", false},   // Worst case alignment needed
		{[]byte(x64pgm), "align=0x0", true},     // 0 aligned is default (16B) alignment

				}
			} else {
				// non-empty cell
				useTabs = false
				if b.flags&AlignRight == 0 { // align left
					b.write0(b.buf[pos : pos+c.size])
					pos += c.size
					if j < len(b.widths) {
						b.writePadding(c.width, b.widths[j], false)
					}
				} else { // align right
					if j < len(b.widths) {
						b.writePadding(c.width, b.widths[j], false)
					}

		maxPageCache = (8 << 10) * 64
	)

	// How big the allocations are needs to depend on the page size.
	// If the page size is too big and the allocations are too small,
	// they might not be aligned to the physical page size, so the scavenger
	// will gloss over them.
	pageSize := os.Getpagesize()
	var allocChunk int
	if pageSize <= 8<<10 {
		allocChunk = 64 << 10
	} else {
		allocChunk = 512 << 10
	}

// Arguments are passed in DI, SI, DX, CX, R8, R9, the rest is on stack.
// Callee-saved registers are: BX, BP, R12-R15.
// SP must be 16-byte aligned.
// On Windows:
// Arguments are passed in CX, DX, R8, R9, the rest is on stack.
// Callee-saved registers are: BX, BP, DI, SI, R12-R15.
// SP must be 16-byte aligned. Windows also requires "stack-backing" for the 4 register arguments:
// https://learn.microsoft.com/en-us/cpp/build/x64-calling-convention

		// we don't want to do it every time. But if _stext happens to be
		// page-aligned but isn't the same as Vaddr, we would symbolize
		// wrong. So if the name the addresses aren't page aligned, or if
		// the name is "vmlinux" we read _stext. We can be wrong if: (1)
		// someone passes a kernel path that doesn't contain "vmlinux" AND
		// (2) _stext is page-aligned AND (3) _stext is not at Vaddr
		symbols, err := ef.Symbols()