Delete GTF_BLK_VOLATILE and GTF_BLK_UNALIGNED

[SingleAccretion]

These struct ASG-specific flags are as legacy leftover from GT_COPYBLK days, inconsistent with how non-struct stores are handled.

Some diffs due to the better (more correct) handling of some volatile operations.

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch, @kunalspathak
See info in area-owners.md if you want to be subscribed.

Issue Details

---

These struct ASG-specific flags are as legacy leftover from GT_COPYBLK days, inconsistent with how non-struct stores are handled.

Expecting some diffs due to the better (more correct) handling of some volatile operations.

Author:	SingleAccretion
Assignees:	-
Labels:	area-CodeGen-coreclr, community-contribution
Milestone:	-

[SingleAccretion]

@dotnet/jit-contrib

[EgorBo]

Any idea what happened with e.g. TestLclFldAddrIntrinsicsSSE41_BlendVariable:

@@ -7,33 +7,42 @@
 ; Final local variable assignments
 ;
 ;* V00 arg0         [V00    ] (  0,  0   )  double  ->  zero-ref    single-def
-;* V01 loc0         [V01,T00] (  0,  0   )  struct (32) zero-ref    do-not-enreg[SF] ld-addr-op
+;  V01 loc0         [V01    ] (  2,  2   )  struct (32) [rsp+08H]   do-not-enreg[XSF] must-init addr-exposed ld-addr-op
 ;* V02 loc1         [V02    ] (  0,  0   )   byref  ->  zero-ref   
 ;* V03 loc2         [V03    ] (  0,  0   )  simd16  ->  zero-ref   
 ;# V04 OutArgs      [V04    ] (  1,  1   )  struct ( 0) [rsp+00H]   do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
 ;
-; Lcl frame size = 0
+; Lcl frame size = 40
 
 G_M20604_IG01:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, nogc <-- Prolog IG
+       sub      rsp, 40
        vzeroupper 
-						;; size=3 bbWeight=1 PerfScore 1.00
+       xor      eax, eax
+       mov      qword ptr [rsp+08H], rax
+       vxorps   xmm4, xmm4
+       vmovdqa  xmmword ptr [rsp+10H], xmm4
+       mov      qword ptr [rsp+20H], rax
+						;; size=29 bbWeight=1 PerfScore 5.83
 G_M20604_IG02:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
        vmovups  xmm0, xmmword ptr [reloc @RWD00]
-       vmovups  xmm1, xmmword ptr [reloc @RWD16]
-       vpblendvb xmm0, xmm0, [reloc @RWD32], xmm1
+       vmovups  xmmword ptr [rsp+18H], xmm0
+       vmovups  xmm0, xmmword ptr [reloc @RWD16]
+       vmovups  xmm1, xmmword ptr [reloc @RWD32]
+       vpblendvb xmm0, xmm0, [rsp+18H], xmm1
        vmovd    eax, xmm0
        vxorps   xmm0, xmm0
        vcvtsi2sd  xmm0, eax
-						;; size=38 bbWeight=1 PerfScore 17.33
+						;; size=50 bbWeight=1 PerfScore 21.33
 G_M20604_IG03:        ; bbWeight=1, epilog, nogc, extend
+       add      rsp, 40
        ret      
-						;; size=1 bbWeight=1 PerfScore 1.00
-RWD00  	dq	0000000100000001h, 0000000100000001h
-RWD16  	dq	0000000300000003h, 0000000300000003h
-RWD32  	dq	0000000200000002h, 0000000200000002h
+						;; size=5 bbWeight=1 PerfScore 1.25
+RWD00  	dq	0000000200000002h, 0000000200000002h
+RWD16  	dq	0000000100000001h, 0000000100000001h
+RWD32  	dq	0000000300000003h, 0000000300000003h
 
 
-; Total bytes of code 42, prolog size 3, PerfScore 23.53, instruction count 8, allocated bytes for code 42 (MethodHash=ab2aaf83) for method Runtime_39424:TestLclFldAddrIntrinsicsSSE41_BlendVariable(double):double
+; Total bytes of code 84, prolog size 29, PerfScore 36.82, instruction count 17, allocated bytes for code 84 (MethodHash=ab2aaf83) for method Runtime_39424:TestLclFldAddrIntrinsicsSSE41_BlendVariable(double):double
 ; ============================================================

[SingleAccretion]

Any idea what happened with e.g. TestLclFldAddrIntrinsicsSSE41_BlendVariable

Yes - all the diffs are from cases where we set the flags correctly but used not to. This particular one is from ldobj handling.

[EgorBo]

Any idea what happened with e.g. TestLclFldAddrIntrinsicsSSE41_BlendVariable

Yes - all the diffs are from cases where we set the flags correctly but used not to. This particular one is from ldobj handling.

Thanks, didn't notice volatile prefix in that IL test