From patchwork Tue Oct 17 13:54:36 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Alexander Stein X-Patchwork-Id: 13425197 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 bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id E1E5CCDB474 for ; Tue, 17 Oct 2023 13:55:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:Message-Id:Date:Subject:Cc :To:From:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:In-Reply-To:References: List-Owner; bh=f2jj8ZJ77yVpPOSWq1/PDfpsRC+BCYBidIYwofl5jbA=; b=VXamNUwgdOOiko VcHxQraK0mmRr2AYJuZ/tvLKuU+yJRJe6hpJmRLdxZLqu2ojr7BzrfMFshRsoSthNlC+U2Jgg/GAp k+HE9+rcs2Yept+ChEh0PVBG/82uUzT6tMXb5xO5mox5uC3dbK1WS1qjzFJ2rmU27puK+23JF5NfS ds/90d+6RVCdRzgHCHT02gOSCqm8AyDJRI7tdjNEIY9zrzzHFqAFTMiRshZyAVNdOT9AhW1SNngsC Gg0/R4YHz3stH3XeQeAa/cGQUCEzKuC76XINu3lUhOURtF1pJZRtyVULX0l/QLkqK2qeoSF4EtM6P b8WxVLjRFbv1Pttb9pmw==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.96 #2 (Red Hat Linux)) id 1qskX2-00CSxw-0J; Tue, 17 Oct 2023 13:54:48 +0000 Received: from mx1.tq-group.com ([93.104.207.81]) by bombadil.infradead.org with esmtps (Exim 4.96 #2 (Red Hat Linux)) id 1qskWx-00CSxP-36 for linux-arm-kernel@lists.infradead.org; Tue, 17 Oct 2023 13:54:46 +0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=tq-group.com; i=@tq-group.com; q=dns/txt; s=key1; t=1697550884; x=1729086884; h=from:to:cc:subject:date:message-id:mime-version: content-transfer-encoding; bh=miZzG0khLIpx/7QCyEFXzuWod2hJR8wFw9VpN47ephA=; b=KyJp3lw3htG4NFNGczqnipmqfCwHicxpwxTlGQiPNogeqNu639w2ZM8Y CYY9svZZGQ27OHGmuOfYmy1bZ5aow+BeXHCGQ1YMObie/ssp4IX7UvbSY /DlU9Qtrin464iZWuBSN+eTLBvFD9CuXZS8F/LhjssOIFh4DKq5yguVgg 121aEo1RFykvOqRFiQHMPGcc9X5PLUO32fjUCCNHNzCt/YTGz3eS5BNeH M1b3LpPgXbspps46bc+BfF0Y3JVuEq5xkudIrr6CApdSymeVcctV4IXsy anAVyYLGGaKKU9O/xsq+YfbZFH9gaWOTv0tw1U2Mti92GEd+WByowQD9G Q==; X-IronPort-AV: E=Sophos;i="6.03,232,1694728800"; d="scan'208";a="33508839" Received: from vtuxmail01.tq-net.de ([10.115.0.20]) by mx1.tq-group.com with ESMTP; 17 Oct 2023 15:54:40 +0200 Received: from steina-w.tq-net.de (steina-w.tq-net.de [10.123.53.18]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by vtuxmail01.tq-net.de (Postfix) with ESMTPSA id 99D02280082; Tue, 17 Oct 2023 15:54:40 +0200 (CEST) From: Alexander Stein To: Dong Aisheng , Andi Shyti , Shawn Guo , Sascha Hauer , Fabio Estevam , Wolfram Sang Cc: Alexander Sverdlin , Pengutronix Kernel Team , NXP Linux Team , linux-i2c@vger.kernel.org, linux-arm-kernel@lists.infradead.org, Alexander Stein Subject: [PATCH v6 1/1] i2c: lpi2c: cache peripheral clock rate Date: Tue, 17 Oct 2023 15:54:36 +0200 Message-Id: <20231017135436.1241650-1-alexander.stein@ew.tq-group.com> X-Mailer: git-send-email 2.34.1 MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20231017_065444_385013_57D749E2 X-CRM114-Status: GOOD ( 18.66 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org From: Alexander Sverdlin One of the reasons to do it is to save some CPU cycles on cpu_freq_get() under mutex. The second reason if the (false-positive) lockdep splat caused by the recursive feature of the "prepare_lock" (one clock instance is I2C peripheral clock and another is pcf85063 RTC as clock provider): ====================================================== WARNING: possible circular locking dependency detected 5.15.71+... #1 Tainted: G O ------------------------------------------------------ .../2332 is trying to acquire lock: ffff8000096cae08 (prepare_lock){+.+.}-{3:3}, at: clk_prepare_lock+0x50/0xb0 but task is already holding lock: ffff000011021100 (i2c_register_adapter){+.+.}-{3:3}, at: i2c_adapter_lock_bus+0x2c/0x3c which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (i2c_register_adapter){+.+.}-{3:3}: lock_acquire+0x68/0x8c rt_mutex_lock_nested+0x88/0xe0 i2c_adapter_lock_bus+0x2c/0x3c i2c_transfer+0x58/0x130 regmap_i2c_read+0x64/0xb0 _regmap_raw_read+0x114/0x440 _regmap_bus_read+0x4c/0x84 _regmap_read+0x6c/0x270 regmap_read+0x54/0x80 pcf85063_probe+0xec/0x4cc i2c_device_probe+0x10c/0x350 really_probe+0xc4/0x470 __driver_probe_device+0x11c/0x190 driver_probe_device+0x48/0x110 __device_attach_driver+0xc4/0x160 bus_for_each_drv+0x80/0xe0 __device_attach+0xb0/0x1f0 device_initial_probe+0x1c/0x2c bus_probe_device+0xa4/0xb0 device_add+0x398/0x8ac device_register+0x28/0x40 i2c_new_client_device+0x144/0x290 of_i2c_register_devices+0x18c/0x230 i2c_register_adapter+0x1dc/0x6b0 __i2c_add_numbered_adapter+0x68/0xbc i2c_add_adapter+0xb0/0xe0 lpi2c_imx_probe+0x354/0x5e0 platform_probe+0x70/0xec really_probe+0xc4/0x470 __driver_probe_device+0x11c/0x190 driver_probe_device+0x48/0x110 __device_attach_driver+0xc4/0x160 bus_for_each_drv+0x80/0xe0 __device_attach+0xb0/0x1f0 device_initial_probe+0x1c/0x2c bus_probe_device+0xa4/0xb0 deferred_probe_work_func+0xa0/0xfc process_one_work+0x2ac/0x6f4 worker_thread+0x7c/0x47c kthread+0x150/0x16c ret_from_fork+0x10/0x20 -> #1 (rtc_pcf85063:560:(&config->regmap)->lock){+.+.}-{3:3}: lock_acquire+0x68/0x8c __mutex_lock+0x9c/0x4d0 mutex_lock_nested+0x48/0x5c regmap_lock_mutex+0x1c/0x30 regmap_read+0x44/0x80 pcf85063_clkout_recalc_rate+0x34/0x80 __clk_register+0x520/0x880 devm_clk_register+0x64/0xc4 pcf85063_probe+0x24c/0x4cc i2c_device_probe+0x10c/0x350 really_probe+0xc4/0x470 __driver_probe_device+0x11c/0x190 driver_probe_device+0x48/0x110 __device_attach_driver+0xc4/0x160 bus_for_each_drv+0x80/0xe0 __device_attach+0xb0/0x1f0 device_initial_probe+0x1c/0x2c bus_probe_device+0xa4/0xb0 device_add+0x398/0x8ac device_register+0x28/0x40 i2c_new_client_device+0x144/0x290 of_i2c_register_devices+0x18c/0x230 i2c_register_adapter+0x1dc/0x6b0 __i2c_add_numbered_adapter+0x68/0xbc i2c_add_adapter+0xb0/0xe0 lpi2c_imx_probe+0x354/0x5e0 platform_probe+0x70/0xec really_probe+0xc4/0x470 __driver_probe_device+0x11c/0x190 driver_probe_device+0x48/0x110 __device_attach_driver+0xc4/0x160 bus_for_each_drv+0x80/0xe0 __device_attach+0xb0/0x1f0 device_initial_probe+0x1c/0x2c bus_probe_device+0xa4/0xb0 deferred_probe_work_func+0xa0/0xfc process_one_work+0x2ac/0x6f4 worker_thread+0x7c/0x47c kthread+0x150/0x16c ret_from_fork+0x10/0x20 -> #0 (prepare_lock){+.+.}-{3:3}: __lock_acquire+0x1298/0x20d0 lock_acquire.part.0+0xf0/0x250 lock_acquire+0x68/0x8c __mutex_lock+0x9c/0x4d0 mutex_lock_nested+0x48/0x5c clk_prepare_lock+0x50/0xb0 clk_get_rate+0x28/0x80 lpi2c_imx_xfer+0xb0/0xa9c __i2c_transfer+0x174/0xa80 i2c_transfer+0x68/0x130 regmap_i2c_read+0x64/0xb0 _regmap_raw_read+0x114/0x440 regmap_raw_read+0x19c/0x28c regmap_bulk_read+0x1b8/0x244 at24_read+0x14c/0x2c4 nvmem_reg_read+0x2c/0x54 bin_attr_nvmem_read+0x8c/0xbc sysfs_kf_bin_read+0x74/0x94 kernfs_fop_read_iter+0xb0/0x1d0 new_sync_read+0xf0/0x184 vfs_read+0x154/0x1f0 ksys_read+0x70/0x100 __arm64_sys_read+0x24/0x30 invoke_syscall+0x50/0x120 el0_svc_common.constprop.0+0x68/0x124 do_el0_svc+0x30/0x9c el0_svc+0x54/0x110 el0t_64_sync_handler+0xa4/0x130 el0t_64_sync+0x1a0/0x1a4 other info that might help us debug this: Chain exists of: prepare_lock --> rtc_pcf85063:560:(&config->regmap)->lock --> i2c_register_adapter Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(i2c_register_adapter); lock(rtc_pcf85063:560:(&config->regmap)->lock); lock(i2c_register_adapter); lock(prepare_lock); *** DEADLOCK *** 4 locks held by .../2332: #0: ffff0000146eb288 (&of->mutex){+.+.}-{3:3}, at: kernfs_fop_read_iter+0x74/0x1d0 #1: ffff000010fe4400 (kn->active#72){.+.+}-{0:0}, at: kernfs_fop_read_iter+0x7c/0x1d0 #2: ffff0000110168e8 (&at24->lock){+.+.}-{3:3}, at: at24_read+0x8c/0x2c4 #3: ffff000011021100 (i2c_register_adapter){+.+.}-{3:3}, at: i2c_adapter_lock_bus+0x2c/0x3c stack backtrace: CPU: 1 PID: 2332 Comm: ... Tainted: G O 5.15.71+... #1 Hardware name: ... (DT) Call trace: dump_backtrace+0x0/0x1d4 show_stack+0x20/0x2c dump_stack_lvl+0x8c/0xb8 dump_stack+0x18/0x34 print_circular_bug+0x1f8/0x200 check_noncircular+0x130/0x144 __lock_acquire+0x1298/0x20d0 lock_acquire.part.0+0xf0/0x250 lock_acquire+0x68/0x8c __mutex_lock+0x9c/0x4d0 mutex_lock_nested+0x48/0x5c clk_prepare_lock+0x50/0xb0 clk_get_rate+0x28/0x80 lpi2c_imx_xfer+0xb0/0xa9c __i2c_transfer+0x174/0xa80 i2c_transfer+0x68/0x130 regmap_i2c_read+0x64/0xb0 _regmap_raw_read+0x114/0x440 regmap_raw_read+0x19c/0x28c regmap_bulk_read+0x1b8/0x244 at24_read+0x14c/0x2c4 nvmem_reg_read+0x2c/0x54 bin_attr_nvmem_read+0x8c/0xbc sysfs_kf_bin_read+0x74/0x94 kernfs_fop_read_iter+0xb0/0x1d0 new_sync_read+0xf0/0x184 vfs_read+0x154/0x1f0 ksys_read+0x70/0x100 __arm64_sys_read+0x24/0x30 invoke_syscall+0x50/0x120 el0_svc_common.constprop.0+0x68/0x124 do_el0_svc+0x30/0x9c el0_svc+0x54/0x110 el0t_64_sync_handler+0xa4/0x130 el0t_64_sync+0x1a0/0x1a4 Fixes: a55fa9d0e42e ("i2c: imx-lpi2c: add low power i2c bus driver") Signed-off-by: Alexander Sverdlin Reviewed-by: Alexander Stein Signed-off-by: Alexander Stein --- This is a v6 picking up Alexander Sverdlin's patch. Up to v5 is done by Alexander Sverdlin. I'm not sure if in this case both my R-b and S-o-b are necessary/desired. Changelog: v6: - rebased to next-20231017 - Just replaced clk_get_rate with atomic_read to shrink the diff v5: rebased onto Wolfram's i2c/for-next v4: switched to atomic_t included clk_rate_exclusive_get()/clk_rate_exclusive_put() v3: fixed build error reported by kernel test robot https://lore.kernel.org/oe-kbuild-all/202303102010.pAv56wKs-lkp@intel.com/ v2: added clk_notifier as Alexander suggested drivers/i2c/busses/i2c-imx-lpi2c.c | 41 +++++++++++++++++++++++++++++- 1 file changed, 40 insertions(+), 1 deletion(-) diff --git a/drivers/i2c/busses/i2c-imx-lpi2c.c b/drivers/i2c/busses/i2c-imx-lpi2c.c index 678b30e90492a..e04faa261e7b1 100644 --- a/drivers/i2c/busses/i2c-imx-lpi2c.c +++ b/drivers/i2c/busses/i2c-imx-lpi2c.c @@ -5,6 +5,7 @@ * Copyright 2016 Freescale Semiconductor, Inc. */ +#include #include #include #include @@ -99,6 +100,8 @@ struct lpi2c_imx_struct { __u8 *rx_buf; __u8 *tx_buf; struct completion complete; + struct notifier_block clk_change_nb; + atomic_t rate_per; unsigned int msglen; unsigned int delivered; unsigned int block_data; @@ -197,6 +200,20 @@ static void lpi2c_imx_stop(struct lpi2c_imx_struct *lpi2c_imx) } while (1); } +static int lpi2c_imx_clk_change_cb(struct notifier_block *nb, + unsigned long action, void *data) +{ + struct clk_notifier_data *ndata = data; + struct lpi2c_imx_struct *lpi2c_imx = container_of(nb, + struct lpi2c_imx_struct, + clk_change_nb); + + if (action & POST_RATE_CHANGE) + atomic_set(&lpi2c_imx->rate_per, ndata->new_rate); + + return NOTIFY_OK; +} + /* CLKLO = I2C_CLK_RATIO * CLKHI, SETHOLD = CLKHI, DATAVD = CLKHI/2 */ static int lpi2c_imx_config(struct lpi2c_imx_struct *lpi2c_imx) { @@ -207,7 +224,7 @@ static int lpi2c_imx_config(struct lpi2c_imx_struct *lpi2c_imx) lpi2c_imx_set_mode(lpi2c_imx); - clk_rate = clk_get_rate(lpi2c_imx->clks[0].clk); + clk_rate = atomic_read(&lpi2c_imx->rate_per); if (!clk_rate) return -EINVAL; @@ -590,6 +607,28 @@ static int lpi2c_imx_probe(struct platform_device *pdev) if (ret) return ret; + lpi2c_imx->clk_change_nb.notifier_call = lpi2c_imx_clk_change_cb; + ret = devm_clk_notifier_register(&pdev->dev, lpi2c_imx->clks[0].clk, + &lpi2c_imx->clk_change_nb); + if (ret) + return dev_err_probe(&pdev->dev, ret, + "can't register peripheral clock notifier\n"); + /* + * Lock the clock rate to avoid rate change between clk_get_rate() and + * atomic_set() + */ + ret = clk_rate_exclusive_get(lpi2c_imx->clks[0].clk); + if (ret) { + dev_err(&pdev->dev, "can't lock I2C peripheral clock rate\n"); + return ret; + } + atomic_set(&lpi2c_imx->rate_per, clk_get_rate(lpi2c_imx->clks[0].clk)); + clk_rate_exclusive_put(lpi2c_imx->clks[0].clk); + if (!atomic_read(&lpi2c_imx->rate_per)) { + dev_err(&pdev->dev, "can't get I2C peripheral clock rate\n"); + return -EINVAL; + } + pm_runtime_set_autosuspend_delay(&pdev->dev, I2C_PM_TIMEOUT); pm_runtime_use_autosuspend(&pdev->dev); pm_runtime_get_noresume(&pdev->dev);