Hi Mike,
On 23/03/26 10:58 AM, Michael Meissner wrote: > On Wed, Mar 11, 2026 at 04:34:21PM +0530, Surya Kumari Jangala wrote: >> Hi Kishan, >> >> On 04/03/26 12:17 pm, Kishan Parmar wrote: >>> Hello, >>> >>> This patch depends on the -mcpu=future patch and will be upstreamed >>> after that patch is upstreamed. >>> >>> Thanks and regards, >>> Kishan Parmar >>> >>> Add support for gating rs6000 built-ins on a new target predicate >>> "future", corresponding to -mcpu=future. >>> >>> Extend rs6000-gen-builtins.cc and rs6000-builtin.cc to recognize >>> [future] as a valid built-in gating predicate to enable defining new >>> built-ins in .bif files. >>> >>> 2025-03-03 Kishan Parmar <[email protected]> >>> Michael Meissner <[email protected]> >>> >>> gcc/ChangeLog: >>> >>> * config/rs6000/rs6000-builtin.cc (rs6000_invalid_builtin): Handle >>> ENB_FUTURE >>> and issue diagnostic requiring -mcpu=future. >>> (rs6000_builtin_is_supported): Return TARGET_FUTURE for ENB_FUTURE >>> built-ins. >>> * config/rs6000/rs6000-gen-builtins.cc (enum bif_stanza): Add >>> BSTZ_FUTURE. >>> (struct attrinfo): Add isfuture member. >> >> I don't see why we need 'future' as an attribute. We will be using it as a >> stanza >> header, but I don't foresee any need of 'future' as an attribute. > > One place that it is useful is if you compile rs6000-gen-builtins.cc > with DEBUG, the diag at line 1468 wants to print out all of the > attributes. Yes, the diag line prints out the attributes. But we don't need 'future' attribute. Attributes are special conditions that apply to the built-in function. We don't have P8, P10 etc attributes, so I don't see why we need 'future' attribute. -Surya >
