From mboxrd@z Thu Jan  1 00:00:00 1970
From: sboyd@codeaurora.org (Stephen Boyd)
Date: Tue, 07 Oct 2014 18:09:22 -0700
Subject: [PATCH v2 4/4] clk: Use ww_mutexes for clk_prepare_{lock/unlock}
In-Reply-To: <5429F555.9030108@codeaurora.org>
References: <1409792466-5092-1-git-send-email-sboyd@codeaurora.org>
 <1409792466-5092-5-git-send-email-sboyd@codeaurora.org>
 <20140928024131.19023.40641@quantum> <5429F555.9030108@codeaurora.org>
Message-ID: <54348EC2.3040300@codeaurora.org>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On 09/29/2014 05:12 PM, Stephen Boyd wrote:
>   I'm also auditing clock
> drivers to find potential brokenness.
>

These are places where we re-enter the framework under drivers/clk/. It 
looks like sirf can be ported to use determine_rate() and something like 
my "safe parent" patch. Tegra is concerning given that they call 
clk_get_rate() in atomic context which is bad because that 
clk_get_rate() can sleep on the prepare mutex. Otherwise we can probably 
just convert that to use the unlocked variants. I'm aware of the qcom 
one, maybe we need a framework flag that indicates that all parent 
clocks must be enabled to switch this clock's parent. The last one is 
versatile which I hope we can convert to use assigned-rates?

drivers/clk/sirf/clk-common.c:409:        ret1 = clk_set_parent(hw->clk, 
clk_pll1.hw.clk);
drivers/clk/sirf/clk-common.c:414:        ret1 = clk_set_parent(hw->clk, 
clk_pll2.hw.clk);
drivers/clk/sirf/clk-common.c:419:        ret1 = clk_set_parent(hw->clk, 
clk_pll3.hw.clk);
drivers/clk/sirf/clk-common.c:427:        ret1 = clk_set_parent(hw->clk, 
clk_pll2.hw.clk);
drivers/clk/sirf/clk-common.c:433:    ret1 = clk_set_parent(hw->clk, 
clk_pll1.hw.clk);
drivers/clk/versatile/clk-sp810.c:103: clk_set_parent(hw->clk, new_parent);
drivers/clk/sirf/clk-common.c:431:    ret2 = 
clk_set_rate(clk_pll1.hw.clk, rate);
drivers/clk/qcom/mmcc-msm8960.c:520:        ret = 
clk_prepare_enable(clk_get_parent_by_index(clk, i));
drivers/clk/qcom/mmcc-msm8960.c:549: 
clk_disable_unprepare(clk_get_parent_by_index(clk, i));
Not necessary? drivers/clk/sirf/clk-common.c:168:    struct clk 
*parent_clk = clk_get_parent(hw->clk);
Not necessary? drivers/clk/sirf/clk-common.c:169:    struct clk 
*pll_parent_clk = clk_get_parent(parent_clk);
Not necessary! drivers/clk/sirf/clk-common.c:181:    struct clk 
*parent_clk = clk_get_parent(hw->clk);
drivers/clk/sirf/clk-common.c:423:    cur_parent = clk_get_parent(hw->clk);
drivers/clk/spear/clk-vco-pll.c:90: 
__clk_get_rate(__clk_get_parent(__clk_get_parent(hw->clk)));
drivers/clk/tegra/clk-pll.c:732:    unsigned long input_rate = 
clk_get_rate(clk_get_parent(hw->clk));
drivers/clk/tegra/clk-pll.c:1288:    unsigned long input_rate = 
clk_get_rate(clk_get_parent(hw->clk));
drivers/clk/versatile/clk-sp810.c:82:    struct clk *old_parent = 
__clk_get_parent(hw->clk);
drivers/clk/qcom/mmcc-msm8960.c:520:        ret = 
clk_prepare_enable(clk_get_parent_by_index(clk, i));
drivers/clk/versatile/clk-sp810.c:102: clk_prepare(new_parent);
drivers/clk/versatile/clk-sp810.c:104: clk_unprepare(old_parent);

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation