[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <mhng-2dc5552b-3c35-4469-b511-3c6dc257b610@palmer-ri-x1c9>
Date: Wed, 05 Feb 2025 07:31:37 -0800 (PST)
From: Palmer Dabbelt <palmer@...belt.com>
To: Conor Dooley <conor@...nel.org>
CC: arikalo@...il.com, linux-riscv@...ts.infradead.org,
Paul Walmsley <paul.walmsley@...ive.com>, aou@...s.berkeley.edu, daniel.lezcano@...aro.org, tglx@...utronix.de,
djordje.todorovic@...cgroup.com, Dragan.Mladjenovic@...mia.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] riscv: Access time mmio instead of rdtime.
On Sun, 29 Dec 2024 07:09:13 PST (-0800), Conor Dooley wrote:
> On Fri, Dec 27, 2024 at 04:00:56PM +0100, Aleksandar Rikalo wrote:
>> +#if defined(CONFIG_RISCV_TIME_MMIO)
>> + if (!of_property_read_u64(n, "clock-reg", &mmio_addr)) {
>
> This is a non-standard, undocumented property. reg & reg-names are what
> you should be using here.
We were talking some in the patchwork meeting. If this system has a
CLINT then it's probably best to refactor the timers such that MMIO
CLINT access is independent from M-mode (as there's some patches
floating around to make M-mode use rdtime on systems that implement
rdtime).
>> + riscv_time_val = ioremap((long)mmio_addr, 8);
>> + if (riscv_time_val) {
>> + pr_info("Using mmio time register at 0x%llx\n", mmio_addr);
>> + static_branch_enable(&riscv_time_mmio_available);
>> + } else {
>> + pr_warn("Unable to use mmio time at 0x%llx\n", mmio_addr);
>> + }
>> + }
>> +#endif
Powered by blists - more mailing lists