Record thread data when making reentrant transitions.
Calls from managed to native code record the top of the managed stack
for stack crawls. Up calls from native to managed (already in the
runnable managed thread state) need to preserve this data as the managed
code may call back into native code. This change introduces a linked
list of records from native to managed code, that allows the regions of
managed code to have their stack crawled.
2 files changed