| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20240830031709.134-1-shawn.shao@jaguarmicro.com>
Date: Fri, 30 Aug 2024 11:17:09 +0800
From: "Shawn.Shao" <shawn.shao@...uarmicro.com>
To: hminas@...opsys.com,
gregkh@...uxfoundation.org,
linux-usb@...r.kernel.org,
linux-kernel@...r.kernel.org
Cc: xiaowu.ding@...uarmicro.com,
angus.chen@...uarmicro.com,
Shawn Shao <shawn.shao@...uarmicro.com>
Subject: [PATCH] usb: dwc2: Adjust the timing of USB Driver Interrupt Registration in the Crashkernel Scenario
From: Shawn Shao <shawn.shao@...uarmicro.com>
The issue is that before entering the crash kernel, the DWC USB controller
did not perform operations such as resetting the interrupt mask bits.
After entering the crash kernel,before the USB interrupt handler
registration was completed while loading the DWC USB driver,an GINTSTS_SOF
interrupt was received.This triggered the misroute_irq process within the
GIC handling framework,ultimately leading to the misrouting of the
interrupt,causing it to be handled by the wrong interrupt handler
and resulting in the issue.
Summary:In a scenario where the kernel triggers a panic and enters
the crash kernel,it is necessary to ensure that the interrupt mask
bit is not enabled before the interrupt registration is complete.
If an interrupt reaches the CPU at this moment,it will certainly
not be handled correctly,especially in cases where this interrupt
is reported frequently.
Please refer to the Crashkernel dmesg information as follows
(the message on line 3 was added before devm_request_irq is
called by the dwc2_driver_probe function):
[ 5.866837][ T1] dwc2 JMIC0010:01: supply vusb_d not found, using dummy regulator
[ 5.874588][ T1] dwc2 JMIC0010:01: supply vusb_a not found, using dummy regulator
[ 5.882335][ T1] dwc2 JMIC0010:01: before devm_request_irq irq: [71], gintmsk[0xf300080e], gintsts[0x04200009]
[ 5.892686][ C0] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.10.0-jmnd1.2_RC #18
[ 5.900327][ C0] Hardware name: CMSS HyperCard4-25G/HyperCard4-25G, BIOS 1.6.4 Jul 8 2024
[ 5.908836][ C0] Call trace:
[ 5.911965][ C0] dump_backtrace+0x0/0x1f0
[ 5.916308][ C0] show_stack+0x20/0x30
[ 5.920304][ C0] dump_stack+0xd8/0x140
[ 5.924387][ C0] pcie_xxx_handler+0x3c/0x1d8
[ 5.930121][ C0] __handle_irq_event_percpu+0x64/0x1e0
[ 5.935506][ C0] handle_irq_event+0x80/0x1d0
[ 5.940109][ C0] try_one_irq+0x138/0x174
[ 5.944365][ C0] misrouted_irq+0x134/0x140
[ 5.948795][ C0] note_interrupt+0x1d0/0x30c
[ 5.953311][ C0] handle_irq_event+0x13c/0x1d0
[ 5.958001][ C0] handle_fasteoi_irq+0xd4/0x260
[ 5.962779][ C0] __handle_domain_irq+0x88/0xf0
[ 5.967555][ C0] gic_handle_irq+0x9c/0x2f0
[ 5.971985][ C0] el1_irq+0xb8/0x140
[ 5.975807][ C0] __setup_irq+0x3dc/0x7cc
[ 5.980064][ C0] request_threaded_irq+0xf4/0x1b4
[ 5.985015][ C0] devm_request_threaded_irq+0x80/0x100
[ 5.990400][ C0] dwc2_driver_probe+0x1b8/0x6b0
[ 5.995178][ C0] platform_drv_probe+0x5c/0xb0
[ 5.999868][ C0] really_probe+0xf8/0x51c
[ 6.004125][ C0] driver_probe_device+0xfc/0x170
[ 6.008989][ C0] device_driver_attach+0xc8/0xd0
[ 6.013853][ C0] __driver_attach+0xe8/0x1b0
[ 6.018369][ C0] bus_for_each_dev+0x7c/0xdc
[ 6.022886][ C0] driver_attach+0x2c/0x3c
[ 6.027143][ C0] bus_add_driver+0xdc/0x240
[ 6.031573][ C0] driver_register+0x80/0x13c
[ 6.036090][ C0] __platform_driver_register+0x50/0x5c
[ 6.041476][ C0] dwc2_platform_driver_init+0x24/0x30
[ 6.046774][ C0] do_one_initcall+0x50/0x25c
[ 6.051291][ C0] do_initcall_level+0xe4/0xfc
[ 6.055894][ C0] do_initcalls+0x80/0xa4
[ 6.060064][ C0] kernel_init_freeable+0x198/0x240
[ 6.065102][ C0] kernel_init+0x1c/0x12c
Signed-off-by: Shawn Shao <shawn.shao@...uarmicro.com>
---
drivers/usb/dwc2/platform.c | 26 ++++++++++++++------------
1 file changed, 14 insertions(+), 12 deletions(-)
diff --git a/drivers/usb/dwc2/platform.c b/drivers/usb/dwc2/platform.c
index 7b84416dfc2b..c1b7209b9483 100644
--- a/drivers/usb/dwc2/platform.c
+++ b/drivers/usb/dwc2/platform.c
@@ -469,18 +469,6 @@ static int dwc2_driver_probe(struct platform_device *dev)
spin_lock_init(&hsotg->lock);
- hsotg->irq = platform_get_irq(dev, 0);
- if (hsotg->irq < 0)
- return hsotg->irq;
-
- dev_dbg(hsotg->dev, "registering common handler for irq%d\n",
- hsotg->irq);
- retval = devm_request_irq(hsotg->dev, hsotg->irq,
- dwc2_handle_common_intr, IRQF_SHARED,
- dev_name(hsotg->dev), hsotg);
- if (retval)
- return retval;
-
hsotg->vbus_supply = devm_regulator_get_optional(hsotg->dev, "vbus");
if (IS_ERR(hsotg->vbus_supply)) {
retval = PTR_ERR(hsotg->vbus_supply);
@@ -524,6 +512,20 @@ static int dwc2_driver_probe(struct platform_device *dev)
if (retval)
goto error;
+ hsotg->irq = platform_get_irq(dev, 0);
+ if (hsotg->irq < 0) {
+ retval = hsotg->irq;
+ goto error;
+ }
+
+ dev_dbg(hsotg->dev, "registering common handler for irq%d\n",
+ hsotg->irq);
+ retval = devm_request_irq(hsotg->dev, hsotg->irq,
+ dwc2_handle_common_intr, IRQF_SHARED,
+ dev_name(hsotg->dev), hsotg);
+ if (retval)
+ goto error;
+
/*
* For OTG cores, set the force mode bits to reflect the value
* of dr_mode. Force mode bits should not be touched at any
--
2.34.1
Powered by blists - more mailing lists