refactor: internal low level allocation stuff, removing a branch when…

… allocating (fixes #25)

crates/inspect-asm/out/x86-64/alloc_12_u32/down.asm

@@ -4,12 +4,11 @@ inspect_asm::alloc_12_u32::down:
 	mov rax, qword ptr [rcx]
 	mov rdx, rax
 	sub rdx, qword ptr [rcx + 8]
-	cmp rdx, 48
-	jb .LBB0_1
+	cmp rdx, 47
+	jbe .LBB0_1
 	and rax, -4
 	add rax, -48
 	mov qword ptr [rcx], rax
-	je .LBB0_1
 .LBB0_0:
 	movups xmm0, xmmword ptr [rsi]
 	movups xmm1, xmmword ptr [rsi + 16]

crates/inspect-asm/out/x86-64/alloc_12_u32/down_a.asm

@@ -4,11 +4,10 @@ inspect_asm::alloc_12_u32::down_a:
 	mov rax, qword ptr [rcx]
 	mov rdx, rax
 	sub rdx, qword ptr [rcx + 8]
-	cmp rdx, 48
-	jb .LBB0_1
+	cmp rdx, 47
+	jbe .LBB0_1
 	add rax, -48
 	mov qword ptr [rcx], rax
-	je .LBB0_1
 .LBB0_0:
 	movups xmm0, xmmword ptr [rsi]
 	movups xmm1, xmmword ptr [rsi + 16]

crates/inspect-asm/out/x86-64/alloc_12_u32/up.asm

@@ -6,12 +6,10 @@ inspect_asm::alloc_12_u32::up:
 	add rax, 3
 	and rax, -4
 	sub rdx, rax
-	cmp rdx, 48
-	jb .LBB0_1
+	cmp rdx, 47
+	jbe .LBB0_1
 	lea rdx, [rax + 48]
 	mov qword ptr [rcx], rdx
-	test rax, rax
-	je .LBB0_1
 .LBB0_0:
 	movups xmm0, xmmword ptr [rsi]
 	movups xmm1, xmmword ptr [rsi + 16]

crates/inspect-asm/out/x86-64/alloc_12_u32/up_a.asm

@@ -4,8 +4,8 @@ inspect_asm::alloc_12_u32::up_a:
 	mov rax, qword ptr [rcx]
 	mov rdx, qword ptr [rcx + 8]
 	sub rdx, rax
-	cmp rdx, 48
-	jb .LBB0_1
+	cmp rdx, 47
+	jbe .LBB0_1
 	lea rdx, [rax + 48]
 	mov qword ptr [rcx], rdx
 .LBB0_0:

crates/inspect-asm/out/x86-64/alloc_big/down.asm

@@ -10,8 +10,6 @@ inspect_asm::alloc_big::down:
 	cmp rdi, qword ptr [rcx + 8]
 	jb .LBB0_0
 	mov qword ptr [rcx], rdi
-	test rdi, rdi
-	je .LBB0_0
 	mov edx, 512
 	pop rbx
 	jmp qword ptr [rip + memcpy@GOTPCREL]

crates/inspect-asm/out/x86-64/alloc_big/down_a.asm

@@ -10,8 +10,6 @@ inspect_asm::alloc_big::down_a:
 	cmp rdi, qword ptr [rcx + 8]
 	jb .LBB0_0
 	mov qword ptr [rcx], rdi
-	test rdi, rdi
-	je .LBB0_0
 	mov edx, 512
 	pop rbx
 	jmp qword ptr [rip + memcpy@GOTPCREL]

crates/inspect-asm/out/x86-64/alloc_big/up.asm

@@ -11,9 +11,8 @@ inspect_asm::alloc_big::up:
 	cmovae rdx, r8
 	cmp rdx, qword ptr [rcx + 8]
 	ja .LBB0_0
-	mov qword ptr [rcx], rdx
 	add rdi, 512
-	je .LBB0_0
+	mov qword ptr [rcx], rdx
 	mov edx, 512
 	pop rbx
 	jmp qword ptr [rip + memcpy@GOTPCREL]

crates/inspect-asm/out/x86-64/alloc_big/up_a.asm

@@ -11,9 +11,8 @@ inspect_asm::alloc_big::up_a:
 	cmovae rdx, r8
 	cmp rdx, qword ptr [rcx + 8]
 	ja .LBB0_0
-	mov qword ptr [rcx], rdx
 	add rdi, 512
-	je .LBB0_0
+	mov qword ptr [rcx], rdx
 	mov edx, 512
 	pop rbx
 	jmp qword ptr [rip + memcpy@GOTPCREL]

crates/inspect-asm/out/x86-64/alloc_iter_u32/down.asm

@@ -23,7 +23,6 @@ inspect_asm::alloc_iter_u32::down:
 	sub rax, r12
 	and rax, -4
 	mov qword ptr [rcx], rax
-	je .LBB0_10
 .LBB0_0:
 	mov qword ptr [rsp + 8], rax
 	mov qword ptr [rsp + 16], 0

crates/inspect-asm/out/x86-64/alloc_iter_u32/down_a.asm

@@ -6,52 +6,51 @@ inspect_asm::alloc_iter_u32::down_a:
 	push r12
 	push rbx
 	sub rsp, 40
-	mov rbx, rdi
+	mov r14, rdi
 	test rdx, rdx
 	je .LBB0_5
 	mov rax, rdx
 	shr rax, 61
 	jne .LBB0_11
 	mov r15, rsi
 	lea r12, [4*rdx]
-	mov rcx, qword ptr [rbx]
+	mov rcx, qword ptr [r14]
 	mov rax, qword ptr [rcx]
 	mov rsi, rax
 	sub rsi, qword ptr [rcx + 8]
 	cmp r12, rsi
 	ja .LBB0_10
 	sub rax, r12
 	mov qword ptr [rcx], rax
-	je .LBB0_10
 .LBB0_0:
 	mov qword ptr [rsp + 8], rax
 	mov qword ptr [rsp + 16], 0
 	mov qword ptr [rsp + 24], rdx
-	mov qword ptr [rsp + 32], rbx
+	mov qword ptr [rsp + 32], r14
 	xor r13d, r13d
-	lea rbx, [rsp + 8]
-	xor r14d, r14d
+	lea r14, [rsp + 8]
+	xor ebx, ebx
 	jmp .LBB0_2
 .LBB0_1:
-	mov dword ptr [rax + 4*r14], ebp
-	inc r14
-	mov qword ptr [rsp + 16], r14
+	mov dword ptr [rax + 4*rbx], ebp
+	inc rbx
+	mov qword ptr [rsp + 16], rbx
 	add r13, 4
 	cmp r12, r13
 	je .LBB0_3
 .LBB0_2:
 	mov ebp, dword ptr [r15 + r13]
-	cmp qword ptr [rsp + 24], r14
+	cmp qword ptr [rsp + 24], rbx
 	jne .LBB0_1
-	mov rdi, rbx
+	mov rdi, r14
 	call bump_scope::bump_vec::BumpVec<T,A,_,_,_>::generic_grow_cold
 	mov rax, qword ptr [rsp + 8]
-	mov r14, qword ptr [rsp + 16]
+	mov rbx, qword ptr [rsp + 16]
 	jmp .LBB0_1
 .LBB0_3:
 	mov rsi, qword ptr [rsp + 8]
-	mov rbx, qword ptr [rsp + 32]
-	mov rax, qword ptr [rbx]
+	mov r14, qword ptr [rsp + 32]
+	mov rax, qword ptr [r14]
 	cmp rsi, qword ptr [rax]
 	je .LBB0_6
 .LBB0_4:
@@ -60,12 +59,12 @@ inspect_asm::alloc_iter_u32::down_a:
 .LBB0_5:
 	mov qword ptr [rsp + 24], rdx
 	mov esi, 4
-	xor r14d, r14d
-	mov rax, qword ptr [rbx]
+	xor ebx, ebx
+	mov rax, qword ptr [r14]
 	cmp rsi, qword ptr [rax]
 	jne .LBB0_4
 .LBB0_6:
-	lea rdx, [4*r14]
+	lea rdx, [4*rbx]
 	mov rax, qword ptr [rsp + 24]
 	lea rax, [rsi + 4*rax]
 	xor edi, edi
@@ -81,12 +80,12 @@ inspect_asm::alloc_iter_u32::down_a:
 .LBB0_7:
 	call qword ptr [rip + memcpy@GOTPCREL]
 .LBB0_8:
-	mov rcx, qword ptr [rbx]
+	mov rcx, qword ptr [r14]
 	mov rax, r15
 	mov qword ptr [rcx], r15
-	mov qword ptr [rsp + 24], r14
+	mov qword ptr [rsp + 24], rbx
 .LBB0_9:
-	mov rdx, r14
+	mov rdx, rbx
 	add rsp, 40
 	pop rbx
 	pop r12
@@ -96,11 +95,11 @@ inspect_asm::alloc_iter_u32::down_a:
 	pop rbp
 	ret
 .LBB0_10:
-	mov rdi, rbx
+	mov rdi, r14
 	mov rsi, rdx
-	mov r14, rdx
+	mov rbx, rdx
 	call bump_scope::bump_scope::BumpScope<A,_,_,_>::do_alloc_slice_in_another_chunk
-	mov rdx, r14
+	mov rdx, rbx
 	jmp .LBB0_0
 .LBB0_11:
 	call qword ptr [rip + bump_scope::private::capacity_overflow@GOTPCREL]

crates/inspect-asm/out/x86-64/alloc_iter_u32/exact_down.asm

@@ -1,55 +1,53 @@
 inspect_asm::alloc_iter_u32::exact_down:
+	push r15
 	push r14
 	push rbx
-	push rax
 	mov rax, rdx
 	shr rax, 61
 	jne .LBB0_7
-	lea r8, [4*rdx]
+	lea rbx, [4*rdx]
 	mov rcx, qword ptr [rdi]
 	mov rax, qword ptr [rcx]
-	mov r9, rax
-	sub r9, qword ptr [rcx + 8]
-	cmp r8, r9
+	mov r8, rax
+	sub r8, qword ptr [rcx + 8]
+	cmp rbx, r8
 	ja .LBB0_5
-	sub rax, r8
+	sub rax, rbx
 	and rax, -4
 	mov qword ptr [rcx], rax
-	je .LBB0_5
 	test rdx, rdx
 	je .LBB0_4
 .LBB0_0:
-	lea rcx, [rdx - 1]
-	movabs r8, 4611686018427387903
-	and r8, rcx
-	cmp rdx, r8
-	cmovb r8, rdx
+	add rbx, -4
+	shr rbx, 2
+	cmp rdx, rbx
+	cmovb rbx, rdx
 	mov rcx, rsi
 	mov rdi, rax
-	cmp r8, 8
+	cmp rbx, 8
 	jb .LBB0_2
-	mov r9, rax
-	sub r9, rsi
+	mov r8, rax
+	sub r8, rsi
 	mov rcx, rsi
 	mov rdi, rax
-	cmp r9, 31
+	cmp r8, 31
 	jbe .LBB0_2
-	inc r8
-	mov ecx, r8d
+	inc rbx
+	mov ecx, ebx
 	and ecx, 7
 	mov edi, 8
 	cmovne rdi, rcx
-	sub r8, rdi
-	lea rcx, [rsi + 4*r8]
-	lea rdi, [rax + 4*r8]
-	xor r9d, r9d
+	sub rbx, rdi
+	lea rcx, [rsi + 4*rbx]
+	lea rdi, [rax + 4*rbx]
+	xor r8d, r8d
 .LBB0_1:
-	movups xmm0, xmmword ptr [rsi + 4*r9]
-	movups xmm1, xmmword ptr [rsi + 4*r9 + 16]
-	movups xmmword ptr [rax + 4*r9], xmm0
-	movups xmmword ptr [rax + 4*r9 + 16], xmm1
-	add r9, 8
-	cmp r8, r9
+	movups xmm0, xmmword ptr [rsi + 4*r8]
+	movups xmm1, xmmword ptr [rsi + 4*r8 + 16]
+	movups xmmword ptr [rax + 4*r8], xmm0
+	movups xmmword ptr [rax + 4*r8 + 16], xmm1
+	add r8, 8
+	cmp rbx, r8
 	jne .LBB0_1
 .LBB0_2:
 	lea rsi, [rsi + 4*rdx]
@@ -64,17 +62,17 @@ inspect_asm::alloc_iter_u32::exact_down:
 	cmp rdi, r8
 	jne .LBB0_3
 .LBB0_4:
-	add rsp, 8
 	pop rbx
 	pop r14
+	pop r15
 	ret
 .LBB0_5:
-	mov rbx, rsi
+	mov r14, rsi
 	mov rsi, rdx
-	mov r14, rdx
+	mov r15, rdx
 	call bump_scope::bump_scope::BumpScope<A,_,_,_>::do_alloc_slice_in_another_chunk
-	mov rsi, rbx
-	mov rdx, r14
+	mov rsi, r14
+	mov rdx, r15
 	test rdx, rdx
 	jne .LBB0_0
 	jmp .LBB0_4