From 06a8566bcf5cf7db9843a82cde7a33c7bf3947d9 Mon Sep 17 00:00:00 2001 From: Lv Zheng Date: Fri, 13 Sep 2013 13:13:23 +0800 Subject: ACPI / IPMI: Fix atomic context requirement of ipmi_msg_handler() This patch fixes the issues indicated by the test results that ipmi_msg_handler() is invoked in atomic context. BUG: scheduling while atomic: kipmi0/18933/0x10000100 Modules linked in: ipmi_si acpi_ipmi ... CPU: 3 PID: 18933 Comm: kipmi0 Tainted: G AW 3.10.0-rc7+ #2 Hardware name: QCI QSSC-S4R/QSSC-S4R, BIOS QSSC-S4R.QCI.01.00.0027.070120100606 07/01/2010 ffff8838245eea00 ffff88103fc63c98 ffffffff814c4a1e ffff88103fc63ca8 ffffffff814bfbab ffff88103fc63d28 ffffffff814c73e0 ffff88103933cbd4 0000000000000096 ffff88103fc63ce8 ffff88102f618000 ffff881035c01fd8 Call Trace: [] dump_stack+0x19/0x1b [] __schedule_bug+0x46/0x54 [] __schedule+0x83/0x59c [] __cond_resched+0x22/0x2d [] _cond_resched+0x14/0x1d [] mutex_lock+0x11/0x32 [] ? __default_send_IPI_dest_field.constprop.0+0x53/0x58 [] ipmi_msg_handler+0x23/0x166 [ipmi_si] [] deliver_response+0x55/0x5a [] handle_new_recv_msgs+0xb67/0xc65 [] ? read_tsc+0x9/0x19 [] ? _raw_spin_lock_irq+0xa/0xc [] ipmi_thread+0x5c/0x146 [ipmi_si] ... Also Tony Camuso says: We were getting occasional "Scheduling while atomic" call traces during boot on some systems. Problem was first seen on a Cisco C210 but we were able to reproduce it on a Cisco c220m3. Setting CONFIG_LOCKDEP and LOCKDEP_SUPPORT to 'y' exposed a lockdep around tx_msg_lock in acpi_ipmi.c struct acpi_ipmi_device. ================================= [ INFO: inconsistent lock state ] 2.6.32-415.el6.x86_64-debug-splck #1 --------------------------------- inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. ksoftirqd/3/17 [HC0[0]:SC1[1]:HE1:SE0] takes: (&ipmi_device->tx_msg_lock){+.?...}, at: [] ipmi_msg_handler+0x71/0x126 {SOFTIRQ-ON-W} state was registered at: [] __lock_acquire+0x63c/0x1570 [] lock_acquire+0xa4/0x120 [] __mutex_lock_common+0x4c/0x400 [] mutex_lock_nested+0x4a/0x60 [] acpi_ipmi_space_handler+0x11b/0x234 [] acpi_ev_address_space_dispatch+0x170/0x1be The fix implemented by this change has been tested by Tony: Tested the patch in a boot loop with lockdep debug enabled and never saw the problem in over 400 reboots. Reported-and-tested-by: Tony Camuso Signed-off-by: Lv Zheng Reviewed-by: Huang Ying Signed-off-by: Rafael J. Wysocki diff --git a/drivers/acpi/acpi_ipmi.c b/drivers/acpi/acpi_ipmi.c index f40acef..a6977e1 100644 --- a/drivers/acpi/acpi_ipmi.c +++ b/drivers/acpi/acpi_ipmi.c @@ -39,6 +39,7 @@ #include #include #include +#include MODULE_AUTHOR("Zhao Yakui"); MODULE_DESCRIPTION("ACPI IPMI Opregion driver"); @@ -57,7 +58,7 @@ struct acpi_ipmi_device { struct list_head head; /* the IPMI request message list */ struct list_head tx_msg_list; - struct mutex tx_msg_lock; + spinlock_t tx_msg_lock; acpi_handle handle; struct pnp_dev *pnp_dev; ipmi_user_t user_interface; @@ -147,6 +148,7 @@ static void acpi_format_ipmi_msg(struct acpi_ipmi_msg *tx_msg, struct kernel_ipmi_msg *msg; struct acpi_ipmi_buffer *buffer; struct acpi_ipmi_device *device; + unsigned long flags; msg = &tx_msg->tx_message; /* @@ -177,10 +179,10 @@ static void acpi_format_ipmi_msg(struct acpi_ipmi_msg *tx_msg, /* Get the msgid */ device = tx_msg->device; - mutex_lock(&device->tx_msg_lock); + spin_lock_irqsave(&device->tx_msg_lock, flags); device->curr_msgid++; tx_msg->tx_msgid = device->curr_msgid; - mutex_unlock(&device->tx_msg_lock); + spin_unlock_irqrestore(&device->tx_msg_lock, flags); } static void acpi_format_ipmi_response(struct acpi_ipmi_msg *msg, @@ -242,6 +244,7 @@ static void ipmi_msg_handler(struct ipmi_recv_msg *msg, void *user_msg_data) int msg_found = 0; struct acpi_ipmi_msg *tx_msg; struct pnp_dev *pnp_dev = ipmi_device->pnp_dev; + unsigned long flags; if (msg->user != ipmi_device->user_interface) { dev_warn(&pnp_dev->dev, "Unexpected response is returned. " @@ -250,7 +253,7 @@ static void ipmi_msg_handler(struct ipmi_recv_msg *msg, void *user_msg_data) ipmi_free_recv_msg(msg); return; } - mutex_lock(&ipmi_device->tx_msg_lock); + spin_lock_irqsave(&ipmi_device->tx_msg_lock, flags); list_for_each_entry(tx_msg, &ipmi_device->tx_msg_list, head) { if (msg->msgid == tx_msg->tx_msgid) { msg_found = 1; @@ -258,7 +261,7 @@ static void ipmi_msg_handler(struct ipmi_recv_msg *msg, void *user_msg_data) } } - mutex_unlock(&ipmi_device->tx_msg_lock); + spin_unlock_irqrestore(&ipmi_device->tx_msg_lock, flags); if (!msg_found) { dev_warn(&pnp_dev->dev, "Unexpected response (msg id %ld) is " "returned.\n", msg->msgid); @@ -378,6 +381,7 @@ acpi_ipmi_space_handler(u32 function, acpi_physical_address address, struct acpi_ipmi_device *ipmi_device = handler_context; int err, rem_time; acpi_status status; + unsigned long flags; /* * IPMI opregion message. * IPMI message is firstly written to the BMC and system software @@ -395,9 +399,9 @@ acpi_ipmi_space_handler(u32 function, acpi_physical_address address, return AE_NO_MEMORY; acpi_format_ipmi_msg(tx_msg, address, value); - mutex_lock(&ipmi_device->tx_msg_lock); + spin_lock_irqsave(&ipmi_device->tx_msg_lock, flags); list_add_tail(&tx_msg->head, &ipmi_device->tx_msg_list); - mutex_unlock(&ipmi_device->tx_msg_lock); + spin_unlock_irqrestore(&ipmi_device->tx_msg_lock, flags); err = ipmi_request_settime(ipmi_device->user_interface, &tx_msg->addr, tx_msg->tx_msgid, @@ -413,9 +417,9 @@ acpi_ipmi_space_handler(u32 function, acpi_physical_address address, status = AE_OK; end_label: - mutex_lock(&ipmi_device->tx_msg_lock); + spin_lock_irqsave(&ipmi_device->tx_msg_lock, flags); list_del(&tx_msg->head); - mutex_unlock(&ipmi_device->tx_msg_lock); + spin_unlock_irqrestore(&ipmi_device->tx_msg_lock, flags); kfree(tx_msg); return status; } @@ -457,7 +461,7 @@ static void acpi_add_ipmi_device(struct acpi_ipmi_device *ipmi_device) INIT_LIST_HEAD(&ipmi_device->head); - mutex_init(&ipmi_device->tx_msg_lock); + spin_lock_init(&ipmi_device->tx_msg_lock); INIT_LIST_HEAD(&ipmi_device->tx_msg_list); ipmi_install_space_handler(ipmi_device); -- cgit v0.10.2