[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <f4adcf94-8a06-4a24-8359-917016b2c4d6@broadcom.com>
Date: Tue, 4 Feb 2025 16:42:20 -0800
From: Florian Fainelli <florian.fainelli@...adcom.com>
To: Conor Dooley <conor@...nel.org>, daniel.lezcano@...aro.org,
tglx@...utronix.de, linux-kernel@...r.kernel.org, etienne.buira@...e.fr
Subject: Re: Kconfig condition on BCM2835_TIMER (and others)
On 2/4/25 15:31, Conor Dooley wrote:
> On Mon, Feb 03, 2025 at 01:44:05PM +0100, Etienne Buira wrote:
>> Resend with corrected email address.
>>
>> On Mon, Feb 03, 2025 at 01:39:33PM +0100, Etienne Buira wrote:
>>> Hi all,
>>>
>>> I want to build a kernel for Raspberry pi 4B, which, according to the
>>> dtb, has a bc2835-system-timer. This device is handled by
>>> drivers/clocksource/bcm2835_timer.c, which is built depending on
>>> CONFIG_BCM2835_TIMER.
>>>
>>> However, the config option is guarded by "if COMPILE_TEST" condition.
>>>
>>> Shouldn't that condition be ARM64 || COMPILE_TEST (ie be selectable on
>>> arm64 builds)?
>
> Perhaps, but I'd wonder why arm64 and arm32 are being treated differently.
> Florian, why's there no select on arm64 when the BCM2835 family option
> is enabled?
The Raspberry Pi 4B (Cortex-A72 CPU) primarily makes use of the ARM CPU
local timer under drivers/clocksource/arm_arch_timer.c, therefore using
bcm2835_timer.c is not necessary at all as a functional boot requirement.
There is no harm in enabling bcm2835_timer, and the arch_arm_timer.c
rating being greater (400) than the bcm2835_timer (300), you will get
the former to be selected anyway.
Feel free to submit a patch to enable the driver for ARM64.
--
Florian
Powered by blists - more mailing lists