Re: [RFC PATCH v2] iov_iter: Convert iterate*() to inline funcs

[David Howells]

David Laight <David.Laight@xxxxxxxxxx> wrote:

> > That can't be avoided if I convert everything to inline functions and
> > function pointers - but the optimiser can get rid of it where it can
> > inline the step function.
> 
> AFAICT the IOVEC one was only called directly.

I've made some changes that I'll post shortly, and I now get this:

	...
	<+36>:	cmpb   $0x0,0x1(%rdx)	# iter->copy_mc
	...
	<+46>:	je     0xffffffff81779aae <_copy_from_iter+98>
	<+48>:	jmp    0xffffffff81779a87 <_copy_from_iter+59>
	...
	# Handle ->copy_mc == true
	<+59>:	mov    0x38(%rsp),%rax
	<+64>:	sub    %gs:0x28,%rax
	<+73>:	jne    0xffffffff81779db1 <_copy_from_iter+869>
	<+79>:	add    $0x40,%rsp
	<+83>:	pop    %rbx
	<+84>:	pop    %rbp
	<+85>:	pop    %r12
	<+87>:	pop    %r13
	<+89>:	pop    %r14
	<+91>:	pop    %r15
	<+93>:	jmp    0xffffffff81777934 <__copy_from_iter_mc>
	...
	# ITER_UBUF
	<+121>:	mov    (%rdx),%al
	<+123>:	cmp    $0x5,%al
	<+125>:	jne    0xffffffff81779b01 <_copy_from_iter+181>
	...
	<+147>:	call   0xffffffff817777ee <__access_ok>
	<+152>:	test   %al,%al
	<+154>:	je     0xffffffff81779af9 <_copy_from_iter+173>
	<+156>:	nop
	<+157>:	nop
	<+158>:	nop
	<+159>:	mov    %r12,%rdi
	<+162>:	mov    %rdx,%rsi
	<+165>:	rep movsb %ds:(%rsi),%es:(%rdi)
	...
	# ITER_IOVEC
	<+181>:	test   %al,%al
	<+183>:	jne    0xffffffff81779b8d <_copy_from_iter+321>
	...
	<+234>:	call   0xffffffff817777ee <__access_ok>
	<+239>:	test   %al,%al
	<+241>:	je     0xffffffff81779b54 <_copy_from_iter+264>
	<+243>:	nop
	<+244>:	nop
	<+245>:	nop
	<+246>:	lea    (%r12,%r15,1),%rax
	<+250>:	mov    %r8,%rsi
	<+253>:	mov    %rax,%rdi
	<+256>:	rep movsb %ds:(%rsi),%es:(%rdi)
	...
	# ITER_BVEC
	<+321>:	cmp    $0x2,%al
	<+323>:	jne    0xffffffff81779c1f <_copy_from_iter+467>
	...
	<+375>:	call   0xffffffff81777282 <kmap_local_page>
	...
	<+431>:	rep movsb %ds:(%rsi),%es:(%rdi)
	...
	# ITER_KVEC
	<+467>:	cmp    $0x1,%al
	<+469>:	jne    0xffffffff81779c82 <_copy_from_iter+566>
	...
	<+526>:	rep movsb %ds:(%rsi),%es:(%rdi)
	...
	# ITER_XARRAY
	<+566>:	cmp    $0x3,%al
	<+568>:	jne    0xffffffff81779d9d <_copy_from_iter+849>
	...
	<+639>:	call   0xffffffff81126bcf <__rcu_read_lock>
	...
	<+651>:	call   0xffffffff81d5ed97 <xas_find>
	...
	<+764>:	call   0xffffffff817772a7 <kmap_local_folio>
	...
	<+806>:	rep movsb %ds:(%rsi),%es:(%rdi)
	...
	# ITER_DISCARD/default
	<+849>:	sub    %rbx,0x18(%rbp)
	<+853>:	mov    0x38(%rsp),%rax
	<+858>:	sub    %gs:0x28,%rax
	<+867>:	je     0xffffffff81779db6 <_copy_from_iter+874>
	<+869>:	call   0xffffffff81d6578c <__stack_chk_fail>
	<+874>:	add    $0x40,%rsp
	<+878>:	mov    %rbx,%rax
	<+881>:	pop    %rbx
	<+882>:	pop    %rbp
	<+883>:	pop    %r12
	<+885>:	pop    %r13
	<+887>:	pop    %r14
	<+889>:	pop    %r15
	<+891>:	jmp    0xffffffff81d72920 <__x86_return_thunk>

David