On Mon, Mar 7, 2016 at 10:22 AM, Khalid Aziz [off-list ref] wrote:

On 03/07/2016 11:08 AM, Andy Lutomirski wrote:

quoted

On Mon, Mar 7, 2016 at 10:04 AM, Khalid Aziz [off-list ref]
wrote:

quoted

On 03/07/2016 09:56 AM, David Miller wrote:

quoted

From: Khalid Aziz <redacted>
Date: Mon, 7 Mar 2016 08:07:53 -0700

quoted

PR_GET_SPARC_ADICAPS

Put this into a new ELF auxiliary vector entry via ARCH_DLINFO.

So now all that's left is supposedly the TAG stuff, please explain
that to me so I can direct you to the correct existing interface to
provide that as well.

Really, try to avoid prtctl, it's poorly typed and almost worse than
ioctl().

The two remaining operations I am looking at are:

1. Is PSTATE.mcde bit set for the process? PR_SET_SPARC_ADI provides this
in
its return value in the patch I sent.

2. Is TTE.mcd set for a given virtual address? PR_GET_SPARC_ADI_STATUS
provides this function in the patch I sent.

Setting and clearing version tags can be done entirely from userspace:

         while (addr < end) {
                 asm volatile(
                         "stxa %1, [%0]ASI_MCD_PRIMARY\n\t"
                         :
                         : "r" (addr), "r" (version));
                 addr += adicap.blksz;
         }
so I do not have to add any kernel code for tags.

Is the effect of that to change the tag associated with a page to
which the caller has write access?

No, it changes the tag associated with the virtual address for the caller.
Physical page backing this virtual address is unaffected. Tag checking is
done for virtual addresses. The one restriction where physical address is
relevant is when two processes map the same physical page, they both have to
use the same tag for the virtual addresses that map on to the shared
physical pages.

Slow down, please.  *Why* do the tags for two different VAs that map
to the same PA have to match?  What goes wrong if they don't, and why
is requiring them to be the same a good idea?

quoted

I sense DoS issues in your future.

Are you concerned about DoS even if the tag is associated with virtual
address, not physical address?

Yes, absolutely.

fd = open("/lib/ld.so");
mmap(fd)
stxa to write the tag

*boom*, presumably, because the tags apparently have to match for all mappings.

What data structure or structures changes when this stxa instruction happens?

--Andy