[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20220125144441.17750-1-ceggers@arri.de>
Date: Tue, 25 Jan 2022 15:44:35 +0100
From: Christian Eggers <ceggers@...i.de>
To: Abel Vesa <abel.vesa@....com>
CC: Michael Turquette <mturquette@...libre.com>,
Stephen Boyd <sboyd@...nel.org>,
Shawn Guo <shawnguo@...nel.org>,
Sascha Hauer <s.hauer@...gutronix.de>,
Pengutronix Kernel Team <kernel@...gutronix.de>,
Fabio Estevam <festevam@...il.com>,
NXP Linux Team <linux-imx@....com>,
<linux-clk@...r.kernel.org>,
<linux-arm-kernel@...ts.infradead.org>,
<linux-kernel@...r.kernel.org>,
Stefan Riedmueller <s.riedmueller@...tec.de>, <ceggers@...i.de>
Subject: [PATCH RESEND 0/6] clk: imx6*: avoid GPMI clock glitches on reparenting/divider change
RESEND because the series got corrupted on first tranmission
On the i.MX6 series (but not on i.MX7/8), most clock multiplexers and
switchable dividers are not "glitch safe", so switching them while a
consumer is connected can cause glitches with a higher frequency than
supported by the consuming peripheral.
Without fixes, peripherals can fail occasionally. One example is
that system boot fails due to lockup of the GPMI NAND controller:
f53d4c109a66 ("mtd: rawnand: gpmi: Add ERR007117 protection for nfc_apply_timings")
The conditions where this can appear also depend on the device
from which the system is booted and whether the bootloader has already
performed all reparenting of the clocks.
To avoid these problems, the clock subsystem must:
- Gate consumer clocks during reparenting
- Enforce that all consumer clocks are gated before client drivers can change the divider
Powered by blists - more mailing lists