We should either remove them or make them match reality.
|
// Free Build EH Stack Stats: |
|
// -------------------------------- |
|
// currently the maximum stack usage we'll face while handling a SO includes: |
|
// 4.3k for the OS (kernel32!RaiseException, Rtl EH dispatch code, RtlUnwindEx [second pass]) |
|
// 1.2k for the CLR EH setup (NakedThrowHelper*) |
|
// 4.5k for other heavy CLR stack creations (2x CONTEXT, 1x REGDISPLAY) |
|
// ~1.0k for other misc CLR stack allocations |
|
// ----- |
|
// 11.0k --> ~2.75 pages for CLR SO EH dispatch |
|
// |
|
// -plus we might need some more for debugger EH dispatch, Watson, etc... |
|
// -also need to take into account that we can lose up to 1 page of the guard region |
|
// -additionally, we need to provide some region to hosts to allow for lock acquisition in a hosted scenario |
|
// |
We should either remove them or make them match reality.
runtime/src/coreclr/vm/threads.cpp
Lines 6117 to 6130 in 824fdb1