runtime: performance regression due to bad instruction used in morestack_noctxt for ppc64 in CL 425396 [1.18 backport]

[gopherbot]

@archanaravindar requested issue #57741 to be considered for backport to the next 1.18 minor release.

@gopherbot please backport this to previous releases Go1.19 and Go1.18
Since the fix 54332 that causes this regression on ppc64 has been likewise back ported to these two releases

[archanaravindar]

Since the fix removes the regression in Go1.18 for the following benchmarks, we request this issue to be back ported

StackCopyPtr-64 138ms           85ms%   -38.36% 
StackCopy-64    137ms           83ms%   -39.48% 
StackCopyNoCache-64     15.1ms  2.5ms   -83.49% 
StackCopyWithStkobj-64  39.9ms  14.4ms  -63.89% 
                                                                        
Hash65536-64    23.5µs          3.9µs   -83.28% 
MegEqMap-64     146µs           22µs    -84.63% 
RepeatedLookupStrMapKey1M-64    403µs   63µs    -84.45% 
MakeChan/Struct/40-64   197ns   146ns   -25.83% 
GoMapClear/NonReflexive/10000-64        116µs   18µs    -84.79% 
MapPop10000-64  9.44ms          4.09ms  -56.72% 

[gopherbot]

Change https://go.dev/cl/462336 mentions this issue: [release-branch.go1.18] runtime: fix performance regression in morestack_noctxt on ppc64

[cherrymui]

As Go 1.20 is about to be released and there will not be any more Go 1.18 minor release, closing this as unplanned. Thanks.

[laboger]

This regression was introduced in 1.18.6 and based on the testing we've done so far it affects p10 the most. If someone needs to use 1.18, using 1.18.5 or earlier avoids this problem on ppc64le.