[Lldb-commits] [PATCH]: Fix inline stepping test case (and possibly improve remote step performance)

[Malea, Daniel]

Hi all,

Please review the attached patch that fixes the TestInlineStep case on Linux which was broken after r181501. The fix involves tweaking Thread::GetPrivateStopReason() to not reset the m_stop_info_sp in the case of a "virtual" inline step-in (that is, a step that does not cause the inferior process to resume, but rather just pretends to step into an inlined function call).

I also removed a double-initialization and check for a Process at the start of the function. In addition, with my change, I expect that when remotely stepping into an inlined function call, LLDB will avoid the needless call into ThreadGDBRemote::CalculateStopReason, which if I read it correctly, causes some packets to be sent over the wire.


An alternative way to implement what I'm doing here would be to introduce a new stop reason as per the comment in ThreadPlanStepInRange.cpp:458 but I expect that will require a lot more changes to accomplish essentially the same thing.

Reviews are welcome,
Dan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: handle_virtual_inline_step.patch
Type: application/octet-stream
Size: 2079 bytes
Desc: handle_virtual_inline_step.patch
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20130513/bfa13444/attachment.obj>