Re: [RFC][PATCH 1/2] Add a layer of abstraction to use the memory reserved by device tree for ima buffer pass.

[Prakhar Srivastava]

On 5/12/20 4:09 PM, Rob Herring wrote:
> On Mon, May 04, 2020 at 01:38:28PM -0700, Prakhar Srivastava wrote:
> > Introduce a device tree layer for to read and store ima buffer
> > from the reserved memory section of a device tree.
>
> But why do I need 'a layer of abstraction'? I don't like them.
This is a common path for the all architectures to carry over the
IMA measurement logs. A single layer will avoid any code duplication.

> > Signed-off-by: Prakhar Srivastava <prsriva@xxxxxxxxxxxxxxxxxxx>
> > ---
> >   drivers/of/Kconfig  |   6 ++
> >   drivers/of/Makefile |   1 +
> >   drivers/of/of_ima.c | 165 ++++++++++++++++++++++++++++++++++++++++++++
>
> Who are the users of this code and why does it need to be here? Most
> code for specific bindings are not in drivers/of/ but with the user. It
> doesn't sound like there's more than 1 user.
Currently the path is exercised by arm64 kexec_file_load path. A slight
restructuring is needed on the powerpc side to use the same code path 
and other architectures can follow to add carrying over IMA logs over
kexec with just a few function calls.

I have attempted to bring the code path down to the highest common 
layer, however please do suggest if i can move this some where else.

Thanks,
Prakhar

> >   include/linux/of.h  |  34 +++++++++
> >   4 files changed, 206 insertions(+)
> >   create mode 100644 drivers/of/of_ima.c