On Mon, Jun 6, 2016 at 11:44 AM, Sergey Fedorov <serge.f...@gmail.com> wrote: > On 03/06/16 21:27, Pranith Kumar wrote: >> On Thu, Jun 2, 2016 at 5:18 PM, Richard Henderson <r...@twiddle.net> wrote: >>> >>> What if we have tcg_canonicalize_memop (or some such) split off the barriers >>> into separate opcodes. E.g. >>> >>> MO_BAR_LD_B = 32 // prevent earlier loads from crossing current op >>> MO_BAR_ST_B = 64 // prevent earlier stores from crossing current op >>> MO_BAR_LD_A = 128 // prevent later loads from crossing current op >>> MO_BAR_ST_A = 256 // prevent later stores from crossing current op >>> MO_BAR_LDST_B = MO_BAR_LD_B | MO_BAR_ST_B >>> MO_BAR_LDST_A = MO_BAR_LD_A | MO_BAR_ST_A >>> MO_BAR_MASK = MO_BAR_LDST_B | MO_BAR_LDST_A >>> >>> // Match Sparc MEMBAR as the most flexible host. >>> TCG_BAR_LD_LD = 1 // #LoadLoad barrier >>> TCG_BAR_ST_LD = 2 // #StoreLoad barrier >>> TCG_BAR_LD_ST = 4 // #LoadStore barrier >>> TCG_BAR_ST_ST = 8 // #StoreStore barrier >>> TCG_BAR_SYNC = 64 // SEQ_CST barrier >> I really like this format. I would also like to add to the frontend: >> > > Actually, the acquire barrier is a combined load-load + load-store > barrier; and the release barrier is a combo of load-store + store-store > barriers. >
All the above are two-way barriers. Acquire/Release are one-way barriers. So we cannot combine the above to get acquire/release semantics without being conservative. Thanks, -- Pranith
