Re: [PATCH] block: Annotate a racy read in blk_do_io_stat()

[Marco Elver]

On Wed, 15 May 2024 at 23:51, Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
[...]
> > And I just checked the current access-marking.txt to see where we
> > might add more, and found the section "ACCESS-DOCUMENTATION OPTIONS"
> > already exists. I think that section is perfectly reasonable as is,
> > and it does explicitly talk about ASSERT_EXCLUSIVE* macros.
> >
> > Did you want to add it more prominently at the top? If so, maybe a
> > brief forward-reference to that section might be helpful.
>
> How about like this?
>
> ------------------------------------------------------------------------
>
> The Linux kernel provides the following access-marking options:
>
> 1.      Plain C-language accesses (unmarked), for example, "a = b;"
>
> 2.      Data-race marking, for example, "data_race(a = b);"
>
> 3.      READ_ONCE(), for example, "a = READ_ONCE(b);"
>         The various forms of atomic_read() also fit in here.
>
> 4.      WRITE_ONCE(), for example, "WRITE_ONCE(a, b);"
>         The various forms of atomic_set() also fit in here.
>
> 5.      __data_racy, for example "int __data_racy a;"
>
> 6.      KCSAN's negative-marking assertions, ASSERT_EXCLUSIVE_ACCESS()
>         and ASSERT_EXCLUSIVE_WRITER(), are desccribed in the
>         "ACCESS-DOCUMENTATION OPTIONS" section below.

s/desccribed/described/

> ------------------------------------------------------------------------
>
> Would that work?

It works for me, if we agree that "negative marking" makes sense: if
the other markings indicate the access is happening concurrently with
others, a negative marking does the opposite.

Thanks,
-- Marco