[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <D9QTEAUN0RNE.11G3ZW4IBGL5M@ventanamicro.com>
Date: Thu, 08 May 2025 15:31:07 +0200
From: Radim Krčmář <rkrcmar@...tanamicro.com>
To: "Atish Patra" <atishp@...osinc.com>, "Anup Patel" <anup@...infault.org>,
"Atish Patra" <atishp@...shpatra.org>, "Paul Walmsley"
<paul.walmsley@...ive.com>, "Palmer Dabbelt" <palmer@...belt.com>,
"Alexandre Ghiti" <alex@...ti.fr>
Cc: <kvm@...r.kernel.org>, <kvm-riscv@...ts.infradead.org>,
<linux-riscv@...ts.infradead.org>, <linux-kernel@...r.kernel.org>,
"linux-riscv" <linux-riscv-bounces@...ts.infradead.org>
Subject: Re: [PATCH 1/5] RISC-V: KVM: Lazy enable hstateen IMSIC & ISEL bit
2025-05-05T14:39:26-07:00, Atish Patra <atishp@...osinc.com>:
> Currently, we enable the smstateen bit at vcpu configure time by
> only checking the presence of required ISA extensions.
>
> These bits are not required to be enabled if the guest never uses
> the corresponding architectural state. Enable the smstaeen bits
> at runtime lazily upon first access.
What is the advantage of enabling them lazily?
To make the trap useful, we would have to lazily perform initialization
of the AIA. I think it would require notable changes to AIA, though...
Thanks.
Powered by blists - more mailing lists