From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions When using eBPF to probe CS CALL instructions within a function, a crash can be triggered. The eBPF tool probes offset 257 of the __hrtimer_run_queues() function: <__hrtimer_run_queues+249>: nopl 0x0(%rax,%rax,1) <__hrtimer_run_queues+254>: mov %r14,%rdi <__hrtimer_run_queues+257>: cs call <__x86_indirect_thunk_r12> <__hrtimer_run_queues+263>: mov %eax,%r12d <__hrtimer_run_queues+266>: xchg %ax,%ax <__hrtimer_run_queues+268>: mov %r13,%rdi Which triggers this crash: BUG: unable to handle page fault for address: 00000000000f41c9 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 0 P4D 0 Oops: 0002 [#1] SMP NOPTI CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P RIP: 0010:__hrtimer_run_queues+0x106/0x230 Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is at the 6th byte of the above CS CALL instruction. Since the CS CALL instruction occupies 6 bytes, the exception occurred in the middle of that call instruction. The root cause is that when using eBPF tools to probe in the middle of a function, a kprobe with INT3 is used as the underlying implementation. During single-step emulation of the original CALL instruction, int3_emulate_call() assumes that the probed CALL instruction is 5 bytes long. However, the actual CS-prefixed CALL instruction occupies 6 bytes, so it constructs an incorrect exception return address. When the CPU returns from the kprobe handler, the next instruction to be executed is at the address of the last byte of that CS CALL instruction. Coincidentally, starting from that address, the CPU fetches and decodes a completely different instruction, which ultimately triggers a kernel crash. Fix the issue by using the actual instruction length obtained from the instruction decoder when constructing the exception return address, rather than relying on the hardcoded CALL_INSN_SIZE macro. [ mingo: Refined the changelog ] Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step") Suggested-by: Masami Hiramatsu (Google) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- drivers/soc/microchip/mpfs-sys-controller.c | 282 ++++++++++++++++++++++++++++ 1 file changed, 282 insertions(+) create mode 100644 drivers/soc/microchip/mpfs-sys-controller.c (limited to 'drivers/soc/microchip/mpfs-sys-controller.c') diff --git a/drivers/soc/microchip/mpfs-sys-controller.c b/drivers/soc/microchip/mpfs-sys-controller.c new file mode 100644 index 000000000..0400a01b2 --- /dev/null +++ b/drivers/soc/microchip/mpfs-sys-controller.c @@ -0,0 +1,282 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Microchip PolarFire SoC (MPFS) system controller driver + * + * Copyright (c) 2020-2021 Microchip Corporation. All rights reserved. + * + * Author: Conor Dooley + * + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* + * This timeout must be long, as some services (example: image authentication) + * take significant time to complete + */ +#define MPFS_SYS_CTRL_TIMEOUT_MS 30000 + +static DEFINE_MUTEX(transaction_lock); + +struct mpfs_sys_controller { + struct mbox_client client; + struct mbox_chan *chan; + struct completion c; + struct mtd_info *flash; + struct kref consumers; +}; + +struct mpfs_syscon_config { + unsigned int nb_subdevs; + struct platform_device *subdevs; +}; + +int mpfs_blocking_transaction(struct mpfs_sys_controller *sys_controller, struct mpfs_mss_msg *msg) +{ + unsigned long timeout = msecs_to_jiffies(MPFS_SYS_CTRL_TIMEOUT_MS); + int ret; + + ret = mutex_lock_interruptible(&transaction_lock); + if (ret) + return ret; + + reinit_completion(&sys_controller->c); + + ret = mbox_send_message(sys_controller->chan, msg); + if (ret < 0) { + dev_warn(sys_controller->client.dev, "MPFS sys controller service timeout\n"); + goto out; + } + + /* + * Unfortunately, the system controller will only deliver an interrupt + * if a service succeeds. mbox_send_message() will block until the busy + * flag is gone. If the busy flag is gone but no interrupt has arrived + * to trigger the rx callback then the service can be deemed to have + * failed. + * The caller can then interrogate msg::response::resp_status to + * determine the cause of the failure. + * mbox_send_message() returns positive integers in the success path, so + * ret needs to be cleared if we do get an interrupt. + */ + if (!wait_for_completion_timeout(&sys_controller->c, timeout)) { + ret = -EBADMSG; + dev_warn(sys_controller->client.dev, + "MPFS sys controller service failed with status: %d\n", + msg->response->resp_status); + } else { + ret = 0; + } + +out: + mutex_unlock(&transaction_lock); + + return ret; +} +EXPORT_SYMBOL(mpfs_blocking_transaction); + +static void mpfs_sys_controller_rx_callback(struct mbox_client *client, void *msg) +{ + struct mpfs_sys_controller *sys_controller = + container_of(client, struct mpfs_sys_controller, client); + + complete(&sys_controller->c); +} + +static void mpfs_sys_controller_delete(struct kref *kref) +{ + struct mpfs_sys_controller *sys_controller = + container_of(kref, struct mpfs_sys_controller, consumers); + + mbox_free_channel(sys_controller->chan); + kfree(sys_controller); +} + +static void mpfs_sys_controller_put(void *data) +{ + struct mpfs_sys_controller *sys_controller = data; + + kref_put(&sys_controller->consumers, mpfs_sys_controller_delete); +} + +struct mtd_info *mpfs_sys_controller_get_flash(struct mpfs_sys_controller *mpfs_client) +{ + return mpfs_client->flash; +} +EXPORT_SYMBOL(mpfs_sys_controller_get_flash); + +static int mpfs_sys_controller_probe(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + struct mpfs_sys_controller *sys_controller; + struct mpfs_syscon_config *of_data; + struct device_node *np; + int i, ret; + + sys_controller = kzalloc_obj(*sys_controller); + if (!sys_controller) + return -ENOMEM; + + np = of_parse_phandle(dev->of_node, "microchip,bitstream-flash", 0); + if (!np) + goto no_flash; + + sys_controller->flash = of_get_mtd_device_by_node(np); + of_node_put(np); + if (IS_ERR(sys_controller->flash)) { + ret = dev_err_probe(dev, PTR_ERR(sys_controller->flash), "Failed to get flash\n"); + goto out_free; + } + +no_flash: + sys_controller->client.dev = dev; + sys_controller->client.rx_callback = mpfs_sys_controller_rx_callback; + sys_controller->client.tx_block = 1U; + sys_controller->client.tx_tout = msecs_to_jiffies(MPFS_SYS_CTRL_TIMEOUT_MS); + + sys_controller->chan = mbox_request_channel(&sys_controller->client, 0); + if (IS_ERR(sys_controller->chan)) { + ret = dev_err_probe(dev, PTR_ERR(sys_controller->chan), + "Failed to get mbox channel\n"); + goto out_free; + } + + init_completion(&sys_controller->c); + kref_init(&sys_controller->consumers); + + platform_set_drvdata(pdev, sys_controller); + + of_data = (struct mpfs_syscon_config *) device_get_match_data(dev); + if (!of_data) { + ret = dev_err_probe(dev, -EINVAL, "Error getting match data\n"); + goto out_free_channel; + } + + for (i = 0; i < of_data->nb_subdevs; i++) { + of_data->subdevs[i].dev.parent = dev; + if (platform_device_register(&of_data->subdevs[i])) + dev_warn(dev, "Error registering sub device %s\n", + of_data->subdevs[i].name); + } + + dev_info(&pdev->dev, "Registered MPFS system controller\n"); + + return 0; + +out_free_channel: + mbox_free_channel(sys_controller->chan); +out_free: + kfree(sys_controller); + return ret; +} + +static void mpfs_sys_controller_remove(struct platform_device *pdev) +{ + struct mpfs_sys_controller *sys_controller = platform_get_drvdata(pdev); + + mpfs_sys_controller_put(sys_controller); +} + +static struct platform_device mpfs_subdevs[] = { + { + .name = "mpfs-rng", + .id = -1, + }, + { + .name = "mpfs-generic-service", + .id = -1, + }, + { + .name = "mpfs-auto-update", + .id = -1, + }, +}; + +static struct platform_device pic64gx_subdevs[] = { + { + .name = "mpfs-rng", + .id = -1, + }, + { + .name = "mpfs-generic-service", + .id = -1, + }, +}; + +static const struct mpfs_syscon_config mpfs_config = { + .nb_subdevs = ARRAY_SIZE(mpfs_subdevs), + .subdevs = mpfs_subdevs, +}; + +static const struct mpfs_syscon_config pic64gx_config = { + .nb_subdevs = ARRAY_SIZE(pic64gx_subdevs), + .subdevs = pic64gx_subdevs, +}; + +static const struct of_device_id mpfs_sys_controller_of_match[] = { + {.compatible = "microchip,mpfs-sys-controller", .data = &mpfs_config}, + {.compatible = "microchip,pic64gx-sys-controller", .data = &pic64gx_config}, + {}, +}; +MODULE_DEVICE_TABLE(of, mpfs_sys_controller_of_match); + +struct mpfs_sys_controller *mpfs_sys_controller_get(struct device *dev) +{ + const struct of_device_id *match; + struct mpfs_sys_controller *sys_controller; + int ret; + + if (!dev->parent) + goto err_no_device; + + match = of_match_node(mpfs_sys_controller_of_match, dev->parent->of_node); + of_node_put(dev->parent->of_node); + if (!match) + goto err_no_device; + + sys_controller = dev_get_drvdata(dev->parent); + if (!sys_controller) + goto err_bad_device; + + if (!kref_get_unless_zero(&sys_controller->consumers)) + goto err_bad_device; + + ret = devm_add_action_or_reset(dev, mpfs_sys_controller_put, sys_controller); + if (ret) + return ERR_PTR(ret); + + return sys_controller; + +err_no_device: + dev_dbg(dev, "Parent device was not an MPFS system controller\n"); + return ERR_PTR(-ENODEV); + +err_bad_device: + dev_dbg(dev, "MPFS system controller found but could not register as a sub device\n"); + return ERR_PTR(-EPROBE_DEFER); +} +EXPORT_SYMBOL(mpfs_sys_controller_get); + +static struct platform_driver mpfs_sys_controller_driver = { + .driver = { + .name = "mpfs-sys-controller", + .of_match_table = mpfs_sys_controller_of_match, + }, + .probe = mpfs_sys_controller_probe, + .remove = mpfs_sys_controller_remove, +}; +module_platform_driver(mpfs_sys_controller_driver); + +MODULE_LICENSE("GPL v2"); +MODULE_AUTHOR("Conor Dooley "); +MODULE_DESCRIPTION("MPFS system controller driver"); -- cgit v1.3.1