[libcxxabi patch] Add Registers_arm class and unw_getcontext implementation

[Nico Weber]

On Wed, Dec 18, 2013 at 9:33 AM, Nick Kledzik <kledzik at apple.com> wrote:

>
> On Dec 17, 2013, at 6:04 PM, Nico Weber wrote:
>
> On Tue, Dec 17, 2013 at 2:12 PM, Nick Kledzik <kledzik at apple.com> wrote:
>
>>
>> On Dec 17, 2013, at 1:40 PM, Nico Weber <thakis at chromium.org> wrote:
>>
>>
>>
>>>
>>> I have not investigated deeply how _Unwind_VRS_Get() and friends work.
>>>  I thought part of the model was that arm could ship one static library
>>> that worked with all processors, and it did so via runtime checks for what
>>> registers the processor supported, and the core unwinder was always
>>> compiled to only use integer registers.  Meaning the float/vector registers
>>> where never saved/restored to the unwind buffer.  That is, calling a
>>> setFloatRegister() would change the real processor register and not
>>> something in the register set struct.
>>>
>>
>> Would that work for unw_set_fpreg / unw_get_fpreg? (I don't know if users
>> call this function directly). Also, 9.3 in
>> http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
>>  (remark d, and instruction 11000111 0000iiii and similar) suggest that
>> the exception bytecode modifies float registers in the virtual register
>> bank, and if those modifications are done in real registers, wouldn't
>> things be in an inconsistent state early during phase 2 unwinding, when the
>> bytecode was processed once for phase 1 already?
>>
>>
>> Yes, it would have to know in phase 1 to not manipulate the real VFP
>> registers.
>>
>> But Section 4.7 in that document sounds like VFP registers are lazily
>> saved.  That initially, only integer registers are saved into the register
>> state set.  Then when unwinding, if it sees a frame that says it saved VFP
>> registers, only at that point are VFP registers saved to the register state
>> set.
>>
>
> I see. How about we leave the float register functions stubbed out for
> now, and we can fill them in once the integer bits of the unwinder work,
> and then we can figure out how to handle VFP registers :-)
>
> Sounds good to me.
>

r197591, thanks!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20131218/b0366bb3/attachment.html>