gcn: Fix __builtin_gcn_first_call_this_thread_p

  It turned out that cprop cleverly propagated the unspec_volatile
to the preceding (pseudo)register, permitting to remove the
'set (s0) (pseudoregister)' at -O2.  Unfortunately, it does
matter whether the assignment is done to 's2' (previously: pseudoregister)
or to s1. – Just having a hard register is not enough ...

Solution: Use USE (alias gen_rtx_USE) instead.

Additionally, I removed the s0 modification (that should lead to the unchanged result)
by adding 'gcn_operand_part (DImode, reg, 1)' and then working with SImode. Result:

   if (__builtin_gcn_first_call_this_thread_p())
     x = 42;

becomes now (with -O2) the following; the builtin code is up to to (and including)
'.L2', the rest is the 'if' and 'x=42':

         s_lshr_b32      s2, s1, 16
         s_cmpk_lg_u32   s2, 12345
         s_mov_b32       s12, scc
         s_mov_b32       vcc_lo, scc
         s_mov_b32       vcc_hi, 0
         s_cbranch_vccz  .L2
         s_and_b32       s2, s1, 65535   (= 0xFFFF)
         s_or_b32        s1, s2, 809041920 (= 0x30390000 = (12345 << 16))
.L2:
         s_getpc_b64     s[2:3]
         s_add_u32       s2, s2, x@rel32@lo+4
         s_addc_u32      s3, s3, x@rel32@hi+4
         s_mov_b32       vcc_lo, s12
         s_mov_b32       vcc_hi, 0
         s_cbranch_vccz  .L3
         s_mov_b32       s12, 42
         v_writelane_b32 v0, s12, 0
         s_mov_b64       exec, 1
         global_store_dword      v1, v0, s[2:3]
.L3:


OK for mainline?

Tobias
-----------------
Siemens Electronic Design Automation GmbH; Anschrift: Arnulfstraße 201, 80634 München; Gesellschaft mit beschränkter Haftung; Geschäftsführer: Thomas Heurung, Frank Thürauf; Sitz der Gesellschaft: München; Registergericht München, HRB 106955
  

On 28/11/2022 07:40, Tobias Burnus wrote:
> It turned out that cprop cleverly propagated the unspec_volatile
> to the preceding (pseudo)register, permitting to remove the
> 'set (s0) (pseudoregister)' at -O2.  Unfortunately, it does
> matter whether the assignment is done to 's2' (previously: pseudoregister)
> or to s1. – Just having a hard register is not enough ...
> 
> Solution: Use USE (alias gen_rtx_USE) instead.
> 
> Additionally, I removed the s0 modification (that should lead to the 
> unchanged result)
> by adding 'gcn_operand_part (DImode, reg, 1)' and then working with 
> SImode. Result:
> 
>    if (__builtin_gcn_first_call_this_thread_p())
>      x = 42;
> 
> becomes now (with -O2) the following; the builtin code is up to to (and 
> including)
> '.L2', the rest is the 'if' and 'x=42':
> 
>          s_lshr_b32      s2, s1, 16
>          s_cmpk_lg_u32   s2, 12345
>          s_mov_b32       s12, scc
>          s_mov_b32       vcc_lo, scc
>          s_mov_b32       vcc_hi, 0
>          s_cbranch_vccz  .L2
>          s_and_b32       s2, s1, 65535   (= 0xFFFF)
>          s_or_b32        s1, s2, 809041920 (= 0x30390000 = (12345 << 16))
> .L2:
>          s_getpc_b64     s[2:3]
>          s_add_u32       s2, s2, x@rel32@lo+4
>          s_addc_u32      s3, s3, x@rel32@hi+4
>          s_mov_b32       vcc_lo, s12
>          s_mov_b32       vcc_hi, 0
>          s_cbranch_vccz  .L3
>          s_mov_b32       s12, 42
>          v_writelane_b32 v0, s12, 0
>          s_mov_b64       exec, 1
>          global_store_dword      v1, v0, s[2:3]
> .L3:
> 
> 
> OK for mainline?

OK.

Andrew