RFR: 8311542: Consolidate the native stack printing code

[David Holmes]

On Fri, 6 Dec 2024 02:39:34 GMT, Kim Barrett <kbarrett at openjdk.org> wrote:

>> We now print native stacks in a number of contexts, not just VMError reporting, and we have to try `os::platform_print_native_stack` else fall back to `VMError::print_native stack`. 
>> 
>> The refactoring adds a new `NativeStackPrinter` (NSP) helper class which can be constructed with some of the context information for the printing that will follow. This avoids the need for the print functions to have a large number of parameters. We have to expose both the top-level printing functionality and the "lower-level" frame-based stack walk as the error reporter needs access to that directly. To maintain the exact same format of output the NSP has to be aware of some error reporter usage requirements.
>> 
>> I also had to expose a direct `frame` taking print function for the Debug.cpp `pns` case.
>> 
>> Testing:
>>  - tiers 1 - 4
>> 
>> Some frame management code was also moved from `VMError` to the `frame` class.
>
> src/hotspot/share/runtime/frame.cpp line 1566:
> 
>> 1564:  * @returns an invalid frame (i.e. fr.pc() === 0) if the caller cannot be obtained
>> 1565:  */
>> 1566: frame frame::next_frame(frame fr, Thread* t) {
> 
> `frame` is a non-trivial type to copy, so it seems like it would be better to pass it by const-ref rather
> than by value.  That might trip over lots of const-correctness issues when accessing it though.  If
> pulling that thread looks like an excessive tangle then clean this up as a followup.

It seems we often pass frames around rather than pointers or references to them, but I don't know why. So I would not want to be changing any existing signatures in case there is some nuance I am not aware of.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/22472#discussion_r1872548335