ndb - bug#34378
Using more than 16g can cause record-pool ptr.i values to overflow Fix by splitting memory into 2 zones, lo(16g)/hi(rest) When record pools only use zone_lo, and datamemory, buffers etc...can use any storage/ndb/src/kernel/blocks/lgman.cpp: adopt to changed interface for Ndbd_mem_manager storage/ndb/src/kernel/vm/Pool.cpp: Always use ZONE_LO for record pools as they use ptr.i == 19 bit page id + 13 bit page index storage/ndb/src/kernel/vm/ndbd_malloc_impl.cpp: Add zones to Ndbd_mem_manager ZONE_LO = lower 16g ZONE_HI = rest storage/ndb/src/kernel/vm/ndbd_malloc_impl.hpp: Add zones to Ndbd_mem_manager ZONE_LO = lower 16g ZONE_HI = rest
Showing
This diff is collapsed.
Please register or sign in to comment