Re: [PATCH v2 6/9] s390/module: Use s390_kernel_write() for late relocations

[Gerald Schaefer]

On Fri, 17 Apr 2020 09:04:31 -0500
Josh Poimboeuf <jpoimboe@xxxxxxxxxx> wrote:

> From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> 
> Because of late module patching, a livepatch module needs to be able to
> apply some of its relocations well after it has been loaded.  Instead of
> playing games with module_{dis,en}able_ro(), use existing text poking
> mechanisms to apply relocations after module loading.
> 
> So far only x86, s390 and Power have HAVE_LIVEPATCH but only the first
> two also have STRICT_MODULE_RWX.
> 
> This will allow removal of the last module_disable_ro() usage in
> livepatch.  The ultimate goal is to completely disallow making
> executable mappings writable.
> 
> Also, for the late patching case, use text_mutex, which is supposed to
> be held for all runtime text patching operations.
> 
> [ jpoimboe: Split up patches.  Use mod state to determine whether
> 	    memcpy() can be used.  Add text_mutex.  Make it build. ]
> 
> Cc: linux-s390@xxxxxxxxxxxxxxx
> Cc: heiko.carstens@xxxxxxxxxx
> Suggested-by: Josh Poimboeuf <jpoimboe@xxxxxxxxxx>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> Signed-off-by: Josh Poimboeuf <jpoimboe@xxxxxxxxxx>
> ---
>  arch/s390/kernel/module.c | 125 ++++++++++++++++++++++++--------------
>  1 file changed, 79 insertions(+), 46 deletions(-)

Sorry, just noticed this. Heiko will return next month, and I'm not
really familiar with s390 livepatching. Adding Vasily, he might
have some more insight.

So, I might be completely wrong here, but using s390_kernel_write()
for writing to anything other than 1:1 mapped kernel, should go
horribly wrong, as that runs w/o DAT. It would allow to bypass
DAT write protection, which I assume is why you want to use it,
but it should not work on module text section, as that would be
in vmalloc space and not 1:1 mapped kernel memory.

Not quite sure how to test / trigger this, did this really work for
you on s390?

Regards,
Gerald