forked from OSchip/llvm-project
When supplying memory to expedite the unwinds in the T packet,
include two stack frames worth of unwind information instead of just one -- the unwinder is trying to fetch two stack frames in more instances now and we're sending extra memory reads resulting in a performance degredation while stepping. llvm-svn: 255417
This commit is contained in:
parent
dd4c71ca6e
commit
0071be6590
|
@ -2812,7 +2812,7 @@ RNBRemote::SendStopReplyPacketForThread (nub_thread_t tid)
|
|||
// Add expedited stack memory so stack backtracing doesn't need to read anything from the
|
||||
// frame pointer chain.
|
||||
StackMemoryMap stack_mmap;
|
||||
ReadStackMemory (pid, tid, stack_mmap, 1);
|
||||
ReadStackMemory (pid, tid, stack_mmap, 2);
|
||||
if (!stack_mmap.empty())
|
||||
{
|
||||
for (const auto &stack_memory : stack_mmap)
|
||||
|
|
Loading…
Reference in New Issue