[PATCH v4 5/8] ima: based on policy require signed firmware (sysfs fallback)

Mimi Zohar zohar at linux.vnet.ibm.com
Fri Jun 1 16:04:45 PDT 2018


On Sat, 2018-06-02 at 00:46 +0200, Luis R. Rodriguez wrote:
> On Fri, Jun 01, 2018 at 06:39:55PM -0400, Mimi Zohar wrote:
> > On Fri, 2018-06-01 at 20:21 +0200, Luis R. Rodriguez wrote:
> > > On Tue, May 29, 2018 at 02:01:57PM -0400, Mimi Zohar wrote:
> > > > Luis, is the security_kernel_post_read_file LSM hook in
> > > > firmware_loading_store() still needed after this patch?  Should it be
> > > > calling security_kernel_load_data() instead?
> > > 
> > > That's up to Kees to decide as he added that hook, and knows
> > > what LSMs may be doing with it. From my perspective it is confusing
> > > to have that hook there so I think it could be removed now.
> > > 
> > > Kees?
> > 
> > Commit 6593d92 ("firmware_class: perform new LSM checks") references
> > two methods of loading firmware -  filesystem-found firmware and
> > demand-loaded blobs.  I assume this call in firmware_loading_store()
> > is the demand-loaded blobs.  Does that method still exist?  Is it
> > still being used?
> 
> Yeah its the stupid sysfs interface. So likely loadpin needs porting
> as you IMA as you did.

In this case, it doesn't look like the call to
security_kernel_post_read_file() should be changed, which means that
all the LSMs and IMA still need to support !file.
 
Mimi




More information about the kexec mailing list