Re: Oops caused by race between livepatch and ftrace

[Joe Lawrence <joe.lawrence@redhat.com>]

On 5/20/19 5:19 PM, Joe Lawrence wrote:
> On 5/20/19 5:09 PM, Johannes Erdfelt wrote:
>> On Mon, May 20, 2019, Joe Lawrence <joe.lawrence@redhat.com> wrote:
>>> These two testing scenarios might be interesting to add to our selftests
>>> suite.  Can you post or add the source(s) to livepatch-test<n>.ko to the
>>> tarball?
>>
>> I made the livepatches using kpatch-build and this simple patch:
>>
>> diff --git a/fs/proc/version.c b/fs/proc/version.c
>> index 94901e8e700d..6b8a3449f455 100644
>> --- a/fs/proc/version.c
>> +++ b/fs/proc/version.c
>> @@ -12,6 +12,7 @@ static int version_proc_show(struct seq_file *m, void *v)
>>    		utsname()->sysname,
>>    		utsname()->release,
>>    		utsname()->version);
>> +	seq_printf(m, "example livepatch\n");
>>    	return 0;
>>    }
>>
>> I just created enough livepatches with the same source patch so that I
>> could reproduce the issue somewhat reliably.
>>
>> I'll see if I can make something that uses klp directly.
> 
> Ah ok great, I was hoping it was a relatively simply livepatch.  We
> could probably reuse lib/livepatch/test_klp_livepatch.c to do this
> (patching cmdline_proc_show instead).
> 
>> The rest of the userspace in the initramfs is really straight forward
>> with the only interesting parts being a couple of shell scripts.
> 
> Yup.  I'll be on PTO later this week, but I'll see about extracting the
> scripts and building a pile of livepatch .ko's to see how easily it
> reproduces without qemu.
> 

D'oh -- I just remembered that klp doesn't create those klp (arch) 
relocation sections just yet!  Without those, the window for module RO 
-> RW -> RO in klp_init_object_loaded is going to be really small... at 
least I can't reproduce it yet without those special sections.  So maybe 
such selftests need to wait post klp-convert.


BTW, livepatching folks -- speaking of this window, does it make sense 
for klp_init_object_loaded() to unconditionally frob the module section 
permissions?  Should it only bother iff it's going to apply 
relocations/alternatives/paravirt?

-- Joe