On Tue, 2024-11-05 at 18:33 +0300, Michael Tokarev wrote:
> 04.11.2024 03:17, Nicholas Piggin wrote:
> > From: Ilya Leoshkevich <i...@linux.ibm.com>
> >
> > This is like commit 86e6202a57b1 ("target/ppc: Make divw[u] handler
> > method decodetree compatible."), but for gen_op_arith_divd().
>
> 86e6202a57b1 is v9.0.0-880-g86e6202a57b1, ie, it is in 9.1 only.
> So I wonder what should we do for older stable series (namely for
> 8.2 lts and 9.0), -- should we pick this one together with the
> mentioned 86e6202a57b1, or neither?
>
> A separate question is whenever it makes sense to pick these for 7.2?
>
> Thanks,
>
> /mjt
Come to think of it, I would recommend only the previous commit
("target/ppc: Set ctx->opcode for decode_insn32()") for stable. It
solves the whole class of potential problems like this one, and it
should apply to 7.2 as well.
The reason we also have this fix in the pull request is that long-term
the ctx->opcode field will be removed.
[...]
