RE: [PATCH] Hexagon (target/hexagon) Add overrides for cache/sync/barrier instructions

[Taylor Simpson]

> -----Original Message-----
> From: Richard Henderson <richard.henderson@linaro.org>
> Sent: Monday, April 10, 2023 8:30 PM
> To: Taylor Simpson <tsimpson@quicinc.com>; qemu-devel@nongnu.org
> Cc: philmd@linaro.org; ale@rev.ng; anjo@rev.ng; Brian Cain
> <bcain@quicinc.com>; Matheus Bernardino (QUIC)
> <quic_mathbern@quicinc.com>
> Subject: Re: [PATCH] Hexagon (target/hexagon) Add overrides for
> cache/sync/barrier instructions
> 
> On 4/10/23 13:24, Taylor Simpson wrote:
> > Most of these are not modelled in QEMU, so save the overhead of
> > calling a helper.
> >
> > The only exception is dczeroa.  It assigns to hex_dczero_addr, which
> > is handled during packet commit.
> >
> > Signed-off-by: Taylor Simpson<tsimpson@quicinc.com>
> > ---
> >   target/hexagon/gen_tcg.h | 24 ++++++++++++++++++++++++
> >   target/hexagon/macros.h  | 18 ++++--------------
> >   2 files changed, 28 insertions(+), 14 deletions(-)
> 
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
> 
> Something to look at in the future: I believe quite a lot of these variables 
> like
> dczero_addr are not "real" architectural state, in that they do not persist
> beyond the lifetime of the packet.  There are others, e.g. pkt_has_store_s1.

That's correct.

> These variables could be moved to DisasContext and allocated on demand.
> Even recently this was tedious, because of TCG temporary lifetime issues,
> but no longer.

I'll work on this.  The obvious advantage is to allow the TCG optimizer more 
opportunity to fold copies and propagate constants.

Any other advantage?

Thanks,
Taylor