possible x86-64 assembler bug

[Douglas Katzman]

Enabling both #+allocator-metrics and #+cons-profiling causes backpatching to fail:

#<THREAD "main thread" RUNNING {1001548143}>:
  The value
    137
  is not of type
    (SIGNED-BYTE 8)

Type HELP for debugger help, or (SB-EXT:EXIT) to exit from SBCL.

restarts (invokable by number or by possibly-abbreviated name):
  0: [ABORT] Exit debugger, returning to top level.

((LAMBDA (SB-X86-64-ASM::SEGMENT SB-X86-64-ASM::POSN) :IN SB-X86-64-ASM::EMIT-BYTE-DISPLACEMENT-BACKPATCH) #<SB-ASSEM:SEGMENT {10015D4FF3}> 136)

Minimal input program:
(defun lll (&rest x) (declare (optimize (sb-c::instrument-consing 2))) x)

If I had to guess, the long-nop in the instrument-consing code is the problem, but that code has been in use for a very long time, so I don't see why it would choose now to go awry.

[Douglas Katzman (dougk)]

the cause is extra alignment, and use of JRCXZ in listify-rest-args. There is no 32-bit displacement form of JRCXZ. So this isn't an assembler bug, it's a misuse of the instruction.