llama.cpp

server : handle context overflow during decode

#17267

Merged