[vectorIntrinsics+mask] RFR: 8266287: Basic mask IR implementation for the Vector API masking feature support

[Xiaohong Gong]

On Wed, 12 May 2021 10:33:50 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>> Based on [1], this patch adds the C2 compiler mid-end changes for the masking feature support. It mainly contains:
>>  1) Generation of the mask IRs for vector mask, including:
>>     - Mask generations, e.g. load/compare/maskAll
>>     - Mask operations, e.g. and/or/xor
>>  2) Conversions between vector and mask after loading mask values from memory and before storing mask values into memory
>>  3) Generation of the vector IRs which need the mask value as the control
>>     - The mask value is appended to the original vector node's input list
>> 
>> With this change, the bottom type of the vector mask will be set to `"TypeVectMask"` if the platform supports the masking feature and the backend implementations are added.
>> 
>> Note that this patch only contains the compiler mid-end changes. The backend implementations for SVE/AVX-512 will be in the
>> followed-up patches.
>> 
>> [1] https://github.com/openjdk/panama-vector/pull/57
>
> src/hotspot/share/opto/matcher.cpp line 2332:
> 
>> 2330:     return;
>> 2331:   }
>> 2332: 
> 
> If we add a new flag for masked nodes, it may simplify the checking logic for masking operations.  e.g. 
> node.hpp:
> class Node {
>   // Flags are sorted by usage frequency.
>   enum NodeFlags {
>     Flag_is_Copy                     = 1 << 0, // should be first bit to avoid shift
>     Flag_rematerialize               = 1 << 1,
>     Flag_needs_anti_dependence_check = 1 << 2,
>     Flag_is_macro                    = 1 << 3,
>     Flag_is_Con                      = 1 << 4,
>     Flag_is_cisc_alternate           = 1 << 5,
>     Flag_is_dead_loop_safe           = 1 << 6,
>     Flag_may_be_short_branch         = 1 << 7,
>     Flag_avoid_back_to_back_before   = 1 << 8,
>     Flag_avoid_back_to_back_after    = 1 << 9,
>     Flag_has_call                    = 1 << 10,
>     Flag_is_reduction                = 1 << 11,
>     Flag_is_scheduled                = 1 << 12,
>     **Flag_has_vector_mask_set**         = 1 << 13,

Good suggestion! Thanks! So do you mean using the new flag instead of `"n->in(3)->is_VectorMask()" ` ? I'm not very clarified about what the checking logic for masking operations you mean. Could you please elaborate more about this? Thanks so much!

-------------

PR: https://git.openjdk.java.net/panama-vector/pull/78