| Message ID | 83796b73-cc62-5112-0838-8c4d541dbe75@linux.vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
| Series | Rename __builtin_rs6000_speculation_barrier | expand |
Hi! On Sun, Mar 04, 2018 at 01:52:20PM -0600, Bill Schmidt wrote: > We realized recently that the use of "rs6000" in a builtin name doesn't agree > with our normal naming conventions. Thus this patch changes such a builtin > to __builtin_powerpc_speculation_barrier instead. > > Bootstrapped and tested on powerpc64le-linux-gnu with no regressions. Is > this okay for trunk? Most of our builtins have *no* powerpc or similar in the name, and the two that do have "ppc". Should we use __builtin_speculation_barrier? It sounds more likely that other archs will want that as well, than that it will conflict. For the .md patterns we already have various that are named rs6000_* (and none ppc* or power*). Segher
> On Mar 5, 2018, at 9:04 AM, Segher Boessenkool <segher@kernel.crashing.org> wrote: > > Hi! > > On Sun, Mar 04, 2018 at 01:52:20PM -0600, Bill Schmidt wrote: >> We realized recently that the use of "rs6000" in a builtin name doesn't agree >> with our normal naming conventions. Thus this patch changes such a builtin >> to __builtin_powerpc_speculation_barrier instead. >> >> Bootstrapped and tested on powerpc64le-linux-gnu with no regressions. Is >> this okay for trunk? > > Most of our builtins have *no* powerpc or similar in the name, and the > two that do have "ppc". > > Should we use __builtin_speculation_barrier? It sounds more likely > that other archs will want that as well, than that it will conflict. That was my initial proposal. Richard asked me to insert the "rs6000_" to keep this in a separate namespace for the time being. There is still discussion about a more general speculation barrier builtin for all targets, though that has languished for over a month now. > > For the .md patterns we already have various that are named rs6000_* > (and none ppc* or power*). > Would it be reasonable to go with __builtin_ppc_speculation_barrier for the builtin name and revert to rs6000_* in the .md patterns, etc.? Thanks, Bill > > Segher >
On Mon, Mar 05, 2018 at 09:58:50AM -0600, Bill Schmidt wrote: > > > On Mar 5, 2018, at 9:04 AM, Segher Boessenkool <segher@kernel.crashing.org> wrote: > > > > Hi! > > > > On Sun, Mar 04, 2018 at 01:52:20PM -0600, Bill Schmidt wrote: > >> We realized recently that the use of "rs6000" in a builtin name doesn't agree > >> with our normal naming conventions. Thus this patch changes such a builtin > >> to __builtin_powerpc_speculation_barrier instead. > >> > >> Bootstrapped and tested on powerpc64le-linux-gnu with no regressions. Is > >> this okay for trunk? > > > > Most of our builtins have *no* powerpc or similar in the name, and the > > two that do have "ppc". > > > > Should we use __builtin_speculation_barrier? It sounds more likely > > that other archs will want that as well, than that it will conflict. > > That was my initial proposal. Richard asked me to insert the "rs6000_" to keep this > in a separate namespace for the time being. There is still discussion about a > more general speculation barrier builtin for all targets, though that has languished > for over a month now. > > > > For the .md patterns we already have various that are named rs6000_* > > (and none ppc* or power*). > > > Would it be reasonable to go with __builtin_ppc_speculation_barrier for the > builtin name and revert to rs6000_* in the .md patterns, etc.? Yeah that sounds fine. Okay for trunk with such changes. Thanks! Segher
Index: gcc/config/rs6000/rs6000-builtin.def =================================================================== --- gcc/config/rs6000/rs6000-builtin.def (revision 258157) +++ gcc/config/rs6000/rs6000-builtin.def (working copy) @@ -2348,7 +2348,7 @@ BU_DFP_MISC_2 (DSCRIQ, "dscriq", CONST, dfp_dscri /* 0 argument void function that we pretend was added in ISA 2.06. It's a special nop recognized by 2018+ firmware for P7 and up, with speculation barrier semantics. */ -BU_P7_MISC_X (SPEC_BARRIER, "rs6000_speculation_barrier", MISC) +BU_P7_MISC_X (SPEC_BARRIER, "powerpc_speculation_barrier", MISC) /* 1 argument BCD functions added in ISA 2.06. */ BU_P7_MISC_1 (CDTBCD, "cdtbcd", CONST, cdtbcd) Index: gcc/config/rs6000/rs6000.c =================================================================== --- gcc/config/rs6000/rs6000.c (revision 258157) +++ gcc/config/rs6000/rs6000.c (working copy) @@ -16874,7 +16874,7 @@ rs6000_expand_builtin (tree exp, rtx target, rtx s case MISC_BUILTIN_SPEC_BARRIER: { - emit_insn (gen_rs6000_speculation_barrier ()); + emit_insn (gen_powerpc_speculation_barrier ()); return NULL_RTX; } @@ -17250,7 +17250,7 @@ rs6000_init_builtins (void) ftype = build_function_type_list (void_type_node, NULL_TREE); def_builtin ("__builtin_cpu_init", ftype, RS6000_BUILTIN_CPU_INIT); - def_builtin ("__builtin_rs6000_speculation_barrier", ftype, + def_builtin ("__builtin_powerpc_speculation_barrier", ftype, MISC_BUILTIN_SPEC_BARRIER); ftype = build_function_type_list (bool_int_type_node, const_ptr_type_node, Index: gcc/config/rs6000/rs6000.md =================================================================== --- gcc/config/rs6000/rs6000.md (revision 258157) +++ gcc/config/rs6000/rs6000.md (working copy) @@ -12574,7 +12574,7 @@ return "ori 2,2,0"; }) -(define_insn "rs6000_speculation_barrier" +(define_insn "powerpc_speculation_barrier" [(unspec_volatile:BLK [(const_int 0)] UNSPECV_SPEC_BARRIER)] "" "ori 31,31,0") Index: gcc/testsuite/gcc.target/powerpc/spec-barr-1.c =================================================================== --- gcc/testsuite/gcc.target/powerpc/spec-barr-1.c (revision 258157) +++ gcc/testsuite/gcc.target/powerpc/spec-barr-1.c (working copy) @@ -4,7 +4,7 @@ void foo () { - __builtin_rs6000_speculation_barrier (); + __builtin_powerpc_speculation_barrier (); } /* { dg-final { scan-assembler "ori 31,31,0" } } */