Re: IMA + O_DIRECT (Re: [PATCH 0/1] fix IMA + Apparmor kernel panic)

[Dmitry Kasatkin <d.kasatkin@samsung.com>]

On 09/05/14 06:10, J. R. Okajima wrote:
> Dmitry Kasatkin:
>> Following patch replaces IMA usage of kernel_read() with special
>> version which skips security check that triggers kernel panic
>> when Apparmor and IMA appraisal are enabled together.
> I know this is related to exit(2), but this behaviour of IMA is related
> to open(2) too.
> When O_DIRECT is specified, some filesystems (for example, ext2) call
> do_blockdev_direct_IO() which acquires i_mutex. But
> IMA:process_measurement() already acquires i_mutex before kernel_read().
> It causes a deadlock even if you replace kernel_read() by a simpler one.

Hi,

It is a different issue.

I made patch more than a year ago which fix the problem

https://lkml.org/lkml/2013/2/20/601

I think we had to declare the purpose of the patch in a bit different way.
IMA really does not need direct-io, and can temporarily drop the flag.
As side affect, it would fix the deadlock problem

But I have a different patch now. I will post it today.


> How can we stop reading the file from IMA?

It is actually very interesting question...

1) if you would like to use IMA without it reading a file, then I think
I must disappoint you.
It is not possible.. IMA needs reading a file.

2) if you do not use IMA, then there is no problem for you, because IMA
will not read file if it is not used...

Have a nice day.

- Dmitry

>
> J. R. Okajima
>