Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from phobos.denx.de (phobos.denx.de [85.214.62.61]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id D2162C25B08 for ; Wed, 17 Aug 2022 10:43:21 +0000 (UTC) Received: from h2850616.stratoserver.net (localhost [IPv6:::1]) by phobos.denx.de (Postfix) with ESMTP id 737D384221; Wed, 17 Aug 2022 12:43:19 +0200 (CEST) Authentication-Results: phobos.denx.de; dmarc=none (p=none dis=none) header.from=denx.de Authentication-Results: phobos.denx.de; spf=pass smtp.mailfrom=u-boot-bounces@lists.denx.de DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=denx.de; s=phobos-20191101; t=1660732999; bh=/GYQuO2XOuRIvyEfBaqNgZYxfSewyMpITDG0DebZQ2s=; h=Date:Subject:To:Cc:References:From:In-Reply-To:List-Id: List-Unsubscribe:List-Archive:List-Post:List-Help:List-Subscribe: From; b=Qk3XlCjSAQElOa1c3GTYMaK3IcEyZM/DVO0/4Uyj7DXQZYR52WbqjE5/DiqCbBxBu JXwtt1c++5Vh1t7DhVkTHo632CGgR7ZDwTSHGSmhcXeGMWf4XFk6VBSJl4842Zed7t +Q+y9gtExd2Kz0shoZs+YjwgFRISPEHPDZTI9IXwzMNQFb6sYkw4ngamEBeNjRaYxO vq6ESIpJ4qF53uYnjqGzvIinGtrmpLU+3QtJ6eGH0yP/FredgRjkgs5vVJkb82C0ET db8uO9aPYgoNJevav+wDDZcIqOKwlqVHPeqeM5+KkOEHeBokmm7nKCnMbCKCPzckHZ iWvNmGbvU/yug== Received: by phobos.denx.de (Postfix, from userid 109) id 3F116847FC; Wed, 17 Aug 2022 12:43:17 +0200 (CEST) Received: from mout-u-107.mailbox.org (mout-u-107.mailbox.org [IPv6:2001:67c:2050:101:465::107]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)) (No client certificate requested) by phobos.denx.de (Postfix) with ESMTPS id E6FA783FA1 for ; Wed, 17 Aug 2022 12:43:13 +0200 (CEST) Authentication-Results: phobos.denx.de; dmarc=none (p=none dis=none) header.from=denx.de Authentication-Results: phobos.denx.de; spf=fail smtp.mailfrom=sr@denx.de Received: from smtp2.mailbox.org (smtp2.mailbox.org [IPv6:2001:67c:2050:b231:465::2]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange ECDHE (P-384) server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by mout-u-107.mailbox.org (Postfix) with ESMTPS id 4M74Nf5x0hz9sSR; Wed, 17 Aug 2022 12:42:58 +0200 (CEST) Message-ID: <1ecc8f9d-e6cd-8afa-2964-c85b3913eab9@denx.de> Date: Wed, 17 Aug 2022 12:42:57 +0200 MIME-Version: 1.0 Subject: Re: [PATCH] arm: kirkwood: 88f6281: Detect CONFIG_SYS_TCLK from SAR register Content-Language: en-US To: =?UTF-8?Q?Pali_Roh=c3=a1r?= , Michael Walle Cc: u-boot@lists.denx.de References: <20220816200016.18288-1-pali@kernel.org> <56f57c8a19d7d7f30fcbe442cecc4cc6@walle.cc> <20220816221420.2k5en3zvadaxsiqa@pali> <20220816223942.3acvgv5r7wfwhood@pali> <20220817083659.jlopv77svqmj25sg@pali> From: Stefan Roese In-Reply-To: <20220817083659.jlopv77svqmj25sg@pali> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-BeenThere: u-boot@lists.denx.de X-Mailman-Version: 2.1.39 Precedence: list List-Id: U-Boot discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: u-boot-bounces@lists.denx.de Sender: "U-Boot" X-Virus-Scanned: clamav-milter 0.103.6 at phobos.denx.de X-Virus-Status: Clean Hi Pali, On 17.08.22 10:36, Pali Rohár wrote: > On Wednesday 17 August 2022 01:23:34 Michael Walle wrote: >> Am 2022-08-17 00:39, schrieb Pali Rohár: >>> On Wednesday 17 August 2022 00:14:20 Pali Rohár wrote: >>>> On Tuesday 16 August 2022 23:34:13 Michael Walle wrote: >>>>> Am 2022-08-16 22:00, schrieb Pali Rohár: >>>>>> Bit 21 in SAR register specifies if TCLK is running at 166 MHz or 200 >>>>>> MHz. >>>>>> This information is undocumented in public Marvell Kirkwood Functional >>>>>> Specifications [2], but is available in Linux v3.15 kirkwood code [1]. >>>>>> >>>>>> Commit 8ac303d49f89 ("arm: kirkwood: Do not overwrite CONFIG_SYS_TCLK") >>>>>> broke support for Marvell 88F6281 SoCs because it was expected that all >>>>>> those SoCs have TCLK running at 200 MHz as specified in Marvell 88F6281 >>>>>> Hardware Specifications [3]. >>>>>> >>>>>> Fix broken support for 88F6281 by detecting CONFIG_SYS_TCLK from SAR >>>>>> register, like it was doing Linux v3.15. >>>>>> >>>>>> [1] - >>>>>> https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/arch/arm/mach-kirkwood/common.c?h=v3.15#n542 >>>>>> [2] - >>>>>> https://web.archive.org/web/20130730091033/http://www.marvell.com/embedded-processors/kirkwood/assets/FS_88F6180_9x_6281_OpenSource.pdf >>>>>> [3] - >>>>>> https://web.archive.org/web/20120620073511/http://www.marvell.com/embedded-processors/kirkwood/assets/HW_88F6281_OpenSource.pdf >>>>>> >>>>>> Fixes: 8ac303d49f89 ("arm: kirkwood: Do not overwrite CONFIG_SYS_TCLK") >>>>>> Signed-off-by: Pali Rohár >>>>>> --- >>>>>> Michael, please test this patch, if it fixes your boards. >>>>> >>>>> Thanks, but this doesn't compile: >>>>> >>>>> In file included from lib/time.c:19: >>>>> lib/time.c: In function ‘timer_get_boot_us’: >>>>> ./arch/arm/include/asm/io.h:108:19: error: token "{" is not valid in >>>>> preprocessor expressions >>>>> 108 | #define readl(c) ({ u32 __v = __arch_getl(c); __iormb(); __v; }) >>>>> | ^ >>>>> ./arch/arm/include/asm/arch/kw88f6281.h:18:29: note: in expansion of macro >>>>> ‘readl’ >>>>> 18 | #define CONFIG_SYS_TCLK ((readl(CONFIG_SAR_REG) & BIT(21)) ? \ >>>>> | ^~~~~ >>>>> ./arch/arm/include/asm/arch/config.h:56:32: note: in expansion of macro >>>>> ‘CONFIG_SYS_TCLK’ >>>>> 56 | #define CONFIG_SYS_TIMER_RATE CONFIG_SYS_TCLK >>>>> | ^~~~~~~~~~~~~~~ >>>>> lib/time.c:50:5: note: in expansion of macro ‘CONFIG_SYS_TIMER_RATE’ >>>>> 50 | #if CONFIG_SYS_TIMER_RATE == 1000000 >>>>> | ^~~~~~~~~~~~~~~~~~~~~ >>>>> ./arch/arm/include/asm/io.h:108:19: error: token "{" is not valid in >>>>> preprocessor expressions >>>>> 108 | #define readl(c) ({ u32 __v = __arch_getl(c); __iormb(); __v; }) >>>>> | ^ >>>>> ./arch/arm/include/asm/arch/kw88f6281.h:18:29: note: in expansion of macro >>>>> ‘readl’ >>>>> 18 | #define CONFIG_SYS_TCLK ((readl(CONFIG_SAR_REG) & BIT(21)) ? \ >>>>> | ^~~~~ >>>>> ./arch/arm/include/asm/arch/config.h:56:32: note: in expansion of macro >>>>> ‘CONFIG_SYS_TCLK’ >>>>> 56 | #define CONFIG_SYS_TIMER_RATE CONFIG_SYS_TCLK >>>>> | ^~~~~~~~~~~~~~~ >>>>> lib/time.c:52:7: note: in expansion of macro ‘CONFIG_SYS_TIMER_RATE’ >>>>> 52 | #elif CONFIG_SYS_TIMER_RATE > 1000000 >>>>> | ^~~~~~~~~~~~~~~~~~~~~ >>>>> make[1]: *** [scripts/Makefile.build:257: lib/time.o] Fehler 1 >>>>> >>>>> -michael >>>> >>>> Ah :-( And why it works on other Armada mvebu boards? Because of this: >>>> >>>> $ git grep CONFIG_SYS_TIMER_RATE arch/arm/mach-mvebu >>>> arch/arm/mach-mvebu/include/mach/config.h:#define >>>> CONFIG_SYS_TIMER_RATE 25000000 >>>> >>>> I have looked how is this option used and it is unit for >>>> CONFIG_SYS_TIMER_COUNTER option. On all kirkwood and mvebu platforms >>>> is >>>> that register set to SoC Global Timer 0 which on Armada 38x uses >>>> either >>>> fixed 25MHz clock or NBCLK clock, together with SoC Global Timer 0 >>>> Ratio. When NBCLK then the frequency of that timer register is: >>>> NBCLK / 2 / (2 ^ ratio). Default ratio is 0 (but can be configured >>>> up to >>>> 7). By default fixed 25MHz clock is used. >>>> >>>> Hence for mvebu it is OK as U-Boot does not change default values of >>>> SoC >>>> Global Timer 0. >>>> >>>> For kirkwood it looks like that those timers are based on TCLK (page >>>> 283 >>>> in Functional Specifications). >>>> >>>> How to fix it? I do not know right now. I have just quirk & dirty >>>> solution by moving code from preprocessor to compiler: >>>> >>>> diff --git a/lib/time.c b/lib/time.c >>>> index 96074b84af69..f2ffc0498d39 100644 >>>> --- a/lib/time.c >>>> +++ b/lib/time.c >>>> @@ -46,13 +46,15 @@ unsigned long notrace timer_read_counter(void) >>>> ulong timer_get_boot_us(void) >>>> { >>>> ulong count = timer_read_counter(); >>>> - >>>> -#if CONFIG_SYS_TIMER_RATE == 1000000 >>>> - return count; >>>> -#elif CONFIG_SYS_TIMER_RATE > 1000000 >>>> - return lldiv(count, CONFIG_SYS_TIMER_RATE / 1000000); >>>> -#elif defined(CONFIG_SYS_TIMER_RATE) >>>> - return (unsigned long long)count * 1000000 / CONFIG_SYS_TIMER_RATE; >>>> +#ifdef CONFIG_SYS_TIMER_RATE >>>> + const ulong timer_rate = CONFIG_SYS_TIMER_RATE; >>>> + >>>> + if (timer_rate) >>> ~~~~~~~~~~ >>> Here needs to be if (timer_rate == 1000000) >>> >>>> + return count; >>>> + else if (timer_rate > 1000000) >>>> + return lldiv(count, timer_rate / 1000000); >>>> + else >>>> + return (unsigned long long)count * 1000000 / timer_rate; >>>> #else >>>> /* Assume the counter is in microseconds */ >>>> return count; >>>> >>>> At least it could be used for verifying if CONFIG_SYS_TCLK is working. >> >> FWIW, this is working. >> >>> Anyway, it looks like that this timer code is deprecated and boards >>> should migrate to use DM. >>> >>> So maybe the proper fix for this would be to add kirkwood dm driver? >> >> Eventually sure, but for now I'd like to have a working board without >> spending too much time on fixing this. I'd propose to revert the >> broken commit for now and then, we can work on a proper timer driver. >> >> But I can already say that the u-boot binary is almost at its size >> limit. I've looked into CONFIG_TIMER and it seems that we'd also >> need the whole clock infrastructure (and DM_CLK). No? > > So, for now I see only two options: > > 1) Either apply that change which moves preprocessor macros to C code and apply this one fix. > > 2) Or revert that my kirkwood commit 8ac303d49f89 and do not apply this one fix. > > Stefan, what do you prefer? I prefer option 1. Could you please work on a new patch for this and include some reasoning for this change in the commit message? Thanks, Stefan