Chisel compile warning: random LFSR16

[ingallsj]

Type of change: other enhancement (paying off technical debt)
Impact: no functional change
Development Phase: implementation

fix these Chisel compile warnings:

[warn] /rocket-chip/src/main/scala/rocket/ICache.scala:191:14: method apply in object LFSR16 is deprecated (since 3.2): Use chisel3.util.random.LFSR(16) for a 16-bit LFSR
[warn]     val v0 = LFSR16(refill_fire)(log2Up(nWays)-1,0)
[warn]              ^
[warn] /rocket-chip/src/main/scala/tilelink/Arbiter.scala:100:14: method apply in object LFSR16 is deprecated (since 3.2): Use chisel3.util.random.LFSR(16) for a 16-bit LFSR
[warn]   val lfsr = LFSR16(Bool(true))
[warn]              ^
[warn] /rocket-chip/src/main/scala/util/ECC.scala:186:14: method apply in object LFSR16 is deprecated (since 3.2): Use chisel3.util.random.LFSR(16) for a 16-bit LFSR
[warn]     UIntToOH(LFSR16()(log2Up(width)+f-1,0))(width-1,0)
[warn]              ^
[warn] /rocket-chip/src/main/scala/util/Misc.scala:157:28: method apply in object LFSR16 is deprecated (since 3.2): Use chisel3.util.random.LFSR(16) for a 16-bit LFSR
[warn]   private def randomizer = LFSR16()
[warn]                            ^
[warn] /rocket-chip/src/main/scala/util/Replacement.scala:17:14: method apply in object LFSR16 is deprecated (since 3.2): Use chisel3.util.random.LFSR(16) for a 16-bit LFSR
[warn]   val lfsr = LFSR16(replace)
[warn]              ^

[seldridge]

Note: that the get the exact behavior of LFSR16 you need to use a special construction:

FibonacciLFSR
  .maxPeriod(16, increment, seed = Some(BigInt(1) << 15))
  .asBools
  .reverse

The new LFSR works in reverse of the old one.

[ingallsj]

Note: that the get the exact behavior of LFSR16 you need to use a special construction:
The new LFSR works in reverse of the old one.

Noted, and thanks, I didn't know that difference.
For all of these purposes, the new behavior (with minimal code bloat) should be fine too: we aren't trying for predictable behavior from the pseudo-random generators. :)

[aswaterman]

Yeah, the semantic difference is fine in this case