Upstream Linux commit f90b68d6c8b0.
The CORE/BUS root slice has following design, simplied graph: The difference is core not have pre_div block. A composite core/bus clk has 8 inputs for mux to select, saying clk[0-7].
It support target(smart) interface and normal interface. Target interface is exported for programmer easy to configure ccm root. Normal interface is also exported, but we not use it in our driver, because it will introduce more complexity compared with target interface.
The normal interface simplified as below: SEL_A GA +--+ +-+ | +->+ +------+ CLK[0-7]--->+ | +-+ | | | | +----v---+ +----+ | +--+ |pre_diva+----> | +---------+ | +--------+ |mux +--+post_div | | +--+ |pre_divb+--->+ | +---------+ | | | +----^---+ +----+ +--->+ | +-+ | | +->+ +------+ +--+ +-+ SEL_B GB
The mux in the upper pic is not the target interface MUX, target interface MUX is hiding SEL_A and SEL_B. When you choose clk[0-7], you are actually writing SEL_A or SEL_B depends on the internal counter which will also control the internal "mux".
The target interface simplified as below which is used by Linux Kernel: CLK[0-7]--->MUX-->Gate-->pre_div-->post_div
A requirement of the Target Interface's software is that the target clock source is active, it means when setting SEL_A, the current input clk to SEL_A must be active, same to SEL_B.
We touch target interface, but hardware logic actually also need configure normal interface.
There will be system hang, when doing the following steps: The initial state: SEL_A/SEL_B are both sourcing from clk0, the internal counter choose SEL_A. 1. switch mux from clk0 to clk1 The hardware logic will choose SEL_B and configure SEL_B to clk1. SEL_A no changed. 2. gate off clk0 Disable clk0, then the input to SEL_A is off. 3. switch from clk1 to clk2 The hardware logic will choose SEL_A and configure SEL_A to clk2, however the current SEL_A input clk0 is off, the system hang.
The solution to fix the issue is in step 1, write twice to target interface MUX, it will make SEL_A/SEL_B both sources from clk1, then no need to care about the state of clk0. And finally system performs well.
Signed-off-by: Peng Fan peng.fan@nxp.com Reviewed-by: Dong Aisheng aisheng.dong@nxp.com Signed-off-by: Shawn Guo shawnguo@kernel.org Signed-off-by: Michael Trimarchi michael@amarulasolutions.com --- V1->V2: add original submitter in linux kernel, suggested by Fabio Estevam --- drivers/clk/imx/clk-composite-8m.c | 37 +++++++++++++++++++++++++++++- 1 file changed, 36 insertions(+), 1 deletion(-)
diff --git a/drivers/clk/imx/clk-composite-8m.c b/drivers/clk/imx/clk-composite-8m.c index 560d74aac80..2cb7d135000 100644 --- a/drivers/clk/imx/clk-composite-8m.c +++ b/drivers/clk/imx/clk-composite-8m.c @@ -117,6 +117,41 @@ static const struct clk_ops imx8m_clk_composite_divider_ops = { .set_rate = imx8m_clk_composite_divider_set_rate, };
+static int imx8m_clk_mux_set_parent(struct clk *clk, struct clk *parent) +{ + struct clk_mux *mux = to_clk_mux(clk); + int index; + u32 val; + u32 reg; + + index = clk_mux_fetch_parent_index(clk, parent); + if (index < 0) { + log_err("Could not fetch index\n"); + return index; + } + + val = clk_mux_index_to_val(mux->table, mux->flags, index); + + reg = readl(mux->reg); + reg &= ~(mux->mask << mux->shift); + val = val << mux->shift; + reg |= val; + + /* + * write twice to make sure non-target interface + * SEL_A/B point the same clk input. + */ + writel(reg, mux->reg); + writel(reg, mux->reg); + + return 0; +} + +const struct clk_ops imx8m_clk_mux_ops = { + .get_rate = clk_generic_get_rate, + .set_parent = imx8m_clk_mux_set_parent, +}; + struct clk *imx8m_clk_composite_flags(const char *name, const char * const *parent_names, int num_parents, void __iomem *reg, @@ -155,7 +190,7 @@ struct clk *imx8m_clk_composite_flags(const char *name,
clk = clk_register_composite(NULL, name, parent_names, num_parents, - &mux->clk, &clk_mux_ops, &div->clk, + &mux->clk, &imx8m_clk_mux_ops, &div->clk, &imx8m_clk_composite_divider_ops, &gate->clk, &clk_gate_ops, flags); if (IS_ERR(clk))