Re: [PATCH] accel/tcg: assert insn_idx will always be valid before plugin_inject_cb

[Richard Henderson]

On 9/13/21 7:06 AM, Alex Bennée wrote:
Hmm so we have a separate list for speedy access:
     /* list to quickly access the injected ops */
     QSIMPLEQ_HEAD(, TCGOp) plugin_ops;

I wonder if we should drop that and just scan QTAILQ_HEAD(, TCGOp) ops
so we can be properly aligned with the current instruction.
Alternatively we could just emit INDEX_op_insn_start to the plugin list
as well?

I suspect that just scanning the whole list would be easiest.  Then you don't need to 
maintain your own separate list.


r~