rseq notes from Cauldron

[Szabolcs Nagy]

On 23/09/2019 16:37, Florian Weimer wrote:
> * Mathieu Desnoyers:
>>> We need a comment somehwere in NPTL about the reliance of the rseq
>>> area on implicit deregistration.  This is currently not a problem
>>> because it lives on the stack (as static TLS), and the stack outlives
>>> the thread.  But if we ever change the TCB allocation, additional
>>> steps may be required.  It is NOT sufficient to block signals because
>>> the kernel may write to the rseq area even without signals.  This
>>> should go into the rseq patch as well.
>>
>> This adds a hard requirement on using the IE TLS model AFAIU. Another
>> action item on my end.
> 
> Yes, but given that glibc already has initial-exec TLS, the rest has
> to be initial-exec, too, I think.  At least in the current
> implementation in the loader.

i think dynamic tls cannot be used for rseq anyway
because rseq wants to be signal safe and lazy alloc
of dynamic tls is not as-safe in glibc.

which means a library that provides the rseq abi
cannot be dynamically loaded.