Re: [RFC PATCH 3/3] fs: detect that the i_rwsem has already been taken exclusively

[Eric W. Biederman]

Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> writes:

> On Thu, Sep 28, 2017 at 6:53 PM, Mimi Zohar <zohar@xxxxxxxxxxxxxxxxxx> wrote:
>>
>> The locking issue isn't with validating the file hash, but with the
>> setxattr, chmod, chown syscalls.  Each of these syscalls takes the
>> i_rwsem exclusively before IMA (or EVM) is called.
>
> Read my email again.
>
>> In setxattr, chmod, chown syscalls, IMA (and EVM) are called after the
>> i_rwsem is already taken.  So the locking would be:
>>
>> lock: i_rwsem
>> lock: iint->mutex
>
> No.
>
> Two locks. One inner, one outer. Only the actual ones that calculates
> the hash would take the outer one. Read my email.

That would require a task_work or another kind of work callback so that
the writes of the xattr are not synchronous with the vfs callback
correct?

Eric