This addresses a good comment @pmenon brought up after merging #1423. I misinterpreted the old pointer casting and layout of the old headers, and made the SpinLatch field a pointer to a separately allocated SpinLatch. Instead, it should have been inlined in the header.
This addresses a potential performance issue in the future of having to chase another pointer. Right now we see no difference in TPC-C and YCSB, but we should fix it now.
Coverage increased (+0.04%) to 77.854% when pulling f4fe6795a992db2cf2b7797e2501ee0f0ae59b1f on mbutrovich:tg_header into 29f92eb0f38e8a1cd4baf060d8ebc705541f89e4 on cmu-db:master.
This addresses a good comment @pmenon brought up after merging #1423. I misinterpreted the old pointer casting and layout of the old headers, and made the SpinLatch field a pointer to a separately allocated SpinLatch. Instead, it should have been inlined in the header.
This addresses a potential performance issue in the future of having to chase another pointer. Right now we see no difference in TPC-C and YCSB, but we should fix it now.