[Bug middle-end/104195] Fails to optimize nested array indexing p[i/N].data[i%N]

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104195

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2022-01-23

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Looks like this could be done at the rtl level 
(set (reg:SI 104 [ _3->data[_4] ])
    (mem:SI (plus:DI (and:DI (mult:DI (reg/v:DI 98 [ i ])
                    (const_int 4 [0x4]))
                (const_int 12 [0xc]))
            (plus:DI (and:DI (mult:DI (reg/v:DI 98 [ i ])
                        (const_int 4 [0x4]))
                    (const_int -16 [0xfffffffffffffff0]))
                (reg:DI 105))) [1 _3->data[_4]+0 S4 A32]))


((i*4)&0xc) + (((i*4)&~0xf) + r105)
->
(((i*4)&0xc) + ((i*4)&~0xf)) + r105
->
(((i*4)&0xc) | ((i*4)&~0xf)) + r105
->
(i*4)&~0x4 + r105
->
i*4 + r105


GCC does optimize the address case though:
    struct CChunk { int data[4]; };
    int *f(struct CChunk *p, __SIZE_TYPE__ i) { return &p[i/4].data[i%4]; }