>>> On 01.02.18 at 16:20, wrote:
> On 31/01/18 11:00, Jan Beulich wrote:
> On 30.01.18 at 20:26, wrote:
>>> On 30/01/18 08:39, Jan Beulich wrote:
>>> On 29.01.18 at 16:38, wrote:
> +/*
> + * We are the CPU performing the patching, and might have ended up
> here by
>>
On 31/01/18 11:00, Jan Beulich wrote:
On 30.01.18 at 20:26, wrote:
>> On 30/01/18 08:39, Jan Beulich wrote:
>> On 29.01.18 at 16:38, wrote:
+/*
+ * We are the CPU performing the patching, and might have ended up
here by
+ * hitting a breakpoint.
+
>>> On 30.01.18 at 20:26, wrote:
> On 30/01/18 08:39, Jan Beulich wrote:
> On 29.01.18 at 16:38, wrote:
>>> +/*
>>> + * We are the CPU performing the patching, and might have ended up
>>> here by
>>> + * hitting a breakpoint.
>>> + *
>>> + * Either way, we need to complet
On 31/01/18 06:07, Juergen Gross wrote:
> On 30/01/18 20:26, Andrew Cooper wrote:
>> However, there is literally nothing we can do to prevent #MC from
>> arriving. We can stop servicing #MC by disabling CR4.MCE, but then the
>> processor will shut down.
> Hmm, there is a way to avoid #MC on other
On 30/01/18 20:26, Andrew Cooper wrote:
> However, there is literally nothing we can do to prevent #MC from
> arriving. We can stop servicing #MC by disabling CR4.MCE, but then the
> processor will shut down.
Hmm, there is a way to avoid #MC on other processors, but this
requires the really big h
On 30/01/18 08:39, Jan Beulich wrote:
On 29.01.18 at 16:38, wrote:
>> +bool init_or_livepatch text_poke_live(const struct cpu_user_regs *regs)
>> +{
>> +struct live_poke_info *i = &live_poke_info;
>> +
>> +if ( unlikely(i->cpu != smp_processor_id()) )
>> +{
>> +ASSERT(regs
>>> On 29.01.18 at 16:38, wrote:
> +bool init_or_livepatch text_poke_live(const struct cpu_user_regs *regs)
> +{
> +struct live_poke_info *i = &live_poke_info;
> +
> +if ( unlikely(i->cpu != smp_processor_id()) )
> +{
> +ASSERT(regs);
> +
> +/*
> + * We hit a br
Patching code which is being executed is problematic, because it impossible to
arrange an atomic update of the instruction stream outside of a few corner
cases. Furthermore, we have no feasible way to prevent execution of the NMI
and #MC exception handlers, but have patch points in them.
Use a br
