usb: dwc3: debugfs: do not queue work if try to change mode on non-drd
Subsystems (2)
PATCH 1/1 usb: dwc3: debugfs: do not queue work if try to change mode on non-drd
Do not try to queue a drd work for change mode if the port is not a drd, this is to avoid below kernel dump:
[ 60.115529] ------------[ cut here ]------------ [ 60.120166] WARNING: CPU: 1 PID: 627 at kernel/workqueue.c:1473 __queue_work+0x46c/0x520 [ 60.128254] Modules linked in: [ 60.131313] CPU: 1 PID: 627 Comm: sh Not tainted 5.7.0-rc4-00022-g914a586-dirty #135 [ 60.139054] Hardware name: NXP i.MX8MQ EVK (DT) [ 60.143585] pstate: a0000085 (NzCv daIf -PAN -UAO) [ 60.148376] pc : __queue_work+0x46c/0x520 [ 60.152385] lr : __queue_work+0x314/0x520 [ 60.156393] sp : ffff8000124ebc40 [ 60.159705] x29: ffff8000124ebc40 x28: ffff800011808018 [ 60.165018] x27: ffff800011819ef8 x26: ffff800011d39980 [ 60.170331] x25: ffff800011808018 x24: 0000000000000100 [ 60.175643] x23: 0000000000000013 x22: 0000000000000001 [ 60.180955] x21: ffff0000b7c08e00 x20: ffff0000b6c31080 [ 60.186267] x19: ffff0000bb99bc00 x18: 0000000000000000 [ 60.191579] x17: 0000000000000000 x16: 0000000000000000 [ 60.196891] x15: 0000000000000000 x14: 0000000000000000 [ 60.202202] x13: 0000000000000000 x12: 0000000000000000 [ 60.207515] x11: 0000000000000000 x10: 0000000000000040 [ 60.212827] x9 : ffff800011d55460 x8 : ffff800011d55458 [ 60.218138] x7 : ffff0000b7800028 x6 : 0000000000000000 [ 60.223450] x5 : ffff0000b7800000 x4 : 0000000000000000 [ 60.228762] x3 : ffff0000bb997cc0 x2 : 0000000000000001 [ 60.234074] x1 : 0000000000000000 x0 : ffff0000b6c31088 [ 60.239386] Call trace: [ 60.241834] __queue_work+0x46c/0x520 [ 60.245496] queue_work_on+0x6c/0x90 [ 60.249075] dwc3_set_mode+0x48/0x58 [ 60.252651] dwc3_mode_write+0xf8/0x150 [ 60.256489] full_proxy_write+0x5c/0xa8 [ 60.260327] __vfs_write+0x18/0x40 [ 60.263729] vfs_write+0xdc/0x1c8 [ 60.267045] ksys_write+0x68/0xf0 [ 60.270360] __arm64_sys_write+0x18/0x20 [ 60.274286] el0_svc_common.constprop.0+0x68/0x160 [ 60.279077] do_el0_svc+0x20/0x80 [ 60.282394] el0_sync_handler+0x10c/0x178 [ 60.286403] el0_sync+0x140/0x180 [ 60.289716] ---[ end trace 70b155582e2b7988 ]---
--- drivers/usb/dwc3/debugfs.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/usb/dwc3/debugfs.c +3 -0
@@ -428,6 +428,9 @@ static ssize_t dwc3_mode_write(struct file *file,if (copy_from_user (&buf, ubuf,min_t (size_t,sizeof (buf) -1 , count)))return -EFAULT ; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return count; +if (!strncmp (buf,"host" ,4 )) mode =DWC3_GCTL_PRTCAP_HOST ;
Signature · 1 line
2.7.4
Thinh Nguyen
maintainerHi,
Quoted 2 lines · Li Jun
Do not try to queue a drd work for change mode if the port is not a drd, this is to avoid below kernel dump:
Are you talking about OTG or DRD? This patch seems to be for OTG. If you need to debug and manually do role switch for DRD from userspace, enable and use role class sysfs attribute.
BR,
Thinh
Quoted 61 lines
[ 60.115529] ------------[ cut here ]------------ [ 60.120166] WARNING: CPU: 1 PID: 627 at kernel/workqueue.c:1473 __queue_work+0x46c/0x520 [ 60.128254] Modules linked in: [ 60.131313] CPU: 1 PID: 627 Comm: sh Not tainted 5.7.0-rc4-00022-g914a586-dirty #135 [ 60.139054] Hardware name: NXP i.MX8MQ EVK (DT) [ 60.143585] pstate: a0000085 (NzCv daIf -PAN -UAO) [ 60.148376] pc : __queue_work+0x46c/0x520 [ 60.152385] lr : __queue_work+0x314/0x520 [ 60.156393] sp : ffff8000124ebc40 [ 60.159705] x29: ffff8000124ebc40 x28: ffff800011808018 [ 60.165018] x27: ffff800011819ef8 x26: ffff800011d39980 [ 60.170331] x25: ffff800011808018 x24: 0000000000000100 [ 60.175643] x23: 0000000000000013 x22: 0000000000000001 [ 60.180955] x21: ffff0000b7c08e00 x20: ffff0000b6c31080 [ 60.186267] x19: ffff0000bb99bc00 x18: 0000000000000000 [ 60.191579] x17: 0000000000000000 x16: 0000000000000000 [ 60.196891] x15: 0000000000000000 x14: 0000000000000000 [ 60.202202] x13: 0000000000000000 x12: 0000000000000000 [ 60.207515] x11: 0000000000000000 x10: 0000000000000040 [ 60.212827] x9 : ffff800011d55460 x8 : ffff800011d55458 [ 60.218138] x7 : ffff0000b7800028 x6 : 0000000000000000 [ 60.223450] x5 : ffff0000b7800000 x4 : 0000000000000000 [ 60.228762] x3 : ffff0000bb997cc0 x2 : 0000000000000001 [ 60.234074] x1 : 0000000000000000 x0 : ffff0000b6c31088 [ 60.239386] Call trace: [ 60.241834] __queue_work+0x46c/0x520 [ 60.245496] queue_work_on+0x6c/0x90 [ 60.249075] dwc3_set_mode+0x48/0x58 [ 60.252651] dwc3_mode_write+0xf8/0x150 [ 60.256489] full_proxy_write+0x5c/0xa8 [ 60.260327] __vfs_write+0x18/0x40 [ 60.263729] vfs_write+0xdc/0x1c8 [ 60.267045] ksys_write+0x68/0xf0 [ 60.270360] __arm64_sys_write+0x18/0x20 [ 60.274286] el0_svc_common.constprop.0+0x68/0x160 [ 60.279077] do_el0_svc+0x20/0x80 [ 60.282394] el0_sync_handler+0x10c/0x178 [ 60.286403] el0_sync+0x140/0x180 [ 60.289716] ---[ end trace 70b155582e2b7988 ]---Signed-off-by Li Jun <[email protected]>--- drivers/usb/dwc3/debugfs.c | 3 +++ 1 file changed, 3 insertions(+)drivers/usb/dwc3/debugfs.c +3
-0@@ -428,6 +428,9 @@ static ssize_t dwc3_mode_write(struct file *file,if (copy_from_user (&buf, ubuf,min_t (size_t,sizeof (buf) -1 , count)))return -EFAULT ; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return count; +if (!strncmp (buf,"host" ,4 )) mode =DWC3_GCTL_PRTCAP_HOST ;
Quoted 12 lines
-----Original Message-----
From: Thinh Nguyen <[email protected]>
Sent: Thursday, July 23, 2020 4:36 AM
To: Jun Li <[email protected]>; [email protected]
Cc: [email protected]; [email protected]
Subject: Re: [PATCH] usb: dwc3: debugfs: do not queue work if try to change mode on non-drdHi,
Quoted 7 lines
Quoted 2 lines · Li Jun
Do not try to queue a drd work for change mode if the port is not a drd, this is to avoid below kernel dump:
Are you talking about OTG or DRD? This patch seems to be for OTG. If you need to debug and manually do role switch for DRD from userspace, enable and use role class sysfs attribute.
There is no problem in normal cases, I triggered this warning when try to change the role for a single role port via this debugfs file, I think nothing should stop user to use this file for debug purpose.
There is condition check of USB_DR_MODE_OTG when init drd_work, but no this check when queue work, so maybe a better change is to move the condition check from __dwc3_set_mode() into dwc3_set_mode()
drivers/usb/dwc3/core.c +3 -3
@@ -120,9 +120,6 @@ static void __dwc3_set_mode(struct work_struct *work)unsigned long flags;int ret; -if (dwc->dr_mode !=USB_DR_MODE_OTG ) -return ; -pm_runtime_get_sync (dwc->dev);if (dwc->current_dr_role ==DWC3_GCTL_PRTCAP_OTG )
@@ -205,6 +202,9 @@ void dwc3_set_mode(struct dwc3 *dwc, u32 mode) {unsigned long flags; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return ; +spin_lock_irqsave (&dwc->lock, flags); dwc->desired_dr_role = mode;spin_unlock_irqrestore (&dwc->lock, flags);
Thanks
Li Jun
Quoted 65 lines
BR,
ThinhQuoted 61 lines
[ 60.115529] ------------[ cut here ]------------ [ 60.120166] WARNING: CPU: 1 PID: 627 at kernel/workqueue.c:1473 __queue_work+0x46c/0x520 [ 60.128254] Modules linked in: [ 60.131313] CPU: 1 PID: 627 Comm: sh Not tainted 5.7.0-rc4-00022-g914a586-dirty #135 [ 60.139054] Hardware name: NXP i.MX8MQ EVK (DT) [ 60.143585] pstate: a0000085 (NzCv daIf -PAN -UAO) [ 60.148376] pc : __queue_work+0x46c/0x520 [ 60.152385] lr : __queue_work+0x314/0x520 [ 60.156393] sp : ffff8000124ebc40 [ 60.159705] x29: ffff8000124ebc40 x28: ffff800011808018 [ 60.165018] x27: ffff800011819ef8 x26: ffff800011d39980 [ 60.170331] x25: ffff800011808018 x24: 0000000000000100 [ 60.175643] x23: 0000000000000013 x22: 0000000000000001 [ 60.180955] x21: ffff0000b7c08e00 x20: ffff0000b6c31080 [ 60.186267] x19: ffff0000bb99bc00 x18: 0000000000000000 [ 60.191579] x17: 0000000000000000 x16: 0000000000000000 [ 60.196891] x15: 0000000000000000 x14: 0000000000000000 [ 60.202202] x13: 0000000000000000 x12: 0000000000000000 [ 60.207515] x11: 0000000000000000 x10: 0000000000000040 [ 60.212827] x9 : ffff800011d55460 x8 : ffff800011d55458 [ 60.218138] x7 : ffff0000b7800028 x6 : 0000000000000000 [ 60.223450] x5 : ffff0000b7800000 x4 : 0000000000000000 [ 60.228762] x3 : ffff0000bb997cc0 x2 : 0000000000000001 [ 60.234074] x1 : 0000000000000000 x0 : ffff0000b6c31088 [ 60.239386] Call trace: [ 60.241834] __queue_work+0x46c/0x520 [ 60.245496] queue_work_on+0x6c/0x90 [ 60.249075] dwc3_set_mode+0x48/0x58 [ 60.252651] dwc3_mode_write+0xf8/0x150 [ 60.256489] full_proxy_write+0x5c/0xa8 [ 60.260327] __vfs_write+0x18/0x40 [ 60.263729] vfs_write+0xdc/0x1c8 [ 60.267045] ksys_write+0x68/0xf0 [ 60.270360] __arm64_sys_write+0x18/0x20 [ 60.274286] el0_svc_common.constprop.0+0x68/0x160 [ 60.279077] do_el0_svc+0x20/0x80 [ 60.282394] el0_sync_handler+0x10c/0x178 [ 60.286403] el0_sync+0x140/0x180 [ 60.289716] ---[ end trace 70b155582e2b7988 ]---Signed-off-by Li Jun <[email protected]>--- drivers/usb/dwc3/debugfs.c | 3 +++ 1 file changed, 3 insertions(+)drivers/usb/dwc3/debugfs.c +3
-0@@ -428,6 +428,9 @@ static ssize_t dwc3_mode_write(struct file *file,if (copy_from_user (&buf, ubuf,min_t (size_t,sizeof (buf) -1 , count)))return -EFAULT ; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return count; +if (!strncmp (buf,"host" ,4 )) mode =DWC3_GCTL_PRTCAP_HOST ;
Thinh Nguyen
maintainerQuoted 14 lines · Jun Li
Quoted 14 lines
-----Original Message-----
From: Thinh Nguyen <[email protected]>
Sent: Thursday, July 23, 2020 4:36 AM
To: Jun Li <[email protected]>; [email protected]
Cc: [email protected]; [email protected]
Subject: Re: [PATCH] usb: dwc3: debugfs: do not queue work if try to change mode on non-drdHi,
Quoted 2 lines · Li Jun
Do not try to queue a drd work for change mode if the port is not a drd, this is to avoid below kernel dump:
Quoted 7 lines
Quoted 3 lines
Are you talking about OTG or DRD? This patch seems to be for OTG. If you need to debug and manually do role switch for DRD from userspace, enable and use role class sysfs attribute.
There is no problem in normal cases, I triggered this warning when try to change the role for a single role port via this debugfs file, I think nothing should stop user to use this file for debug purpose.
Ok.
Quoted 4 lines
There is condition check of
USB_DR_MODE_OTGwhen initdrd_work, but no this check when queue work, so maybe a better change is to move the condition check from__dwc3_set_mode()intodwc3_set_mode()
Yeah, this looks better.
Quoted 17 lines
drivers/usb/dwc3/core.c +0
-3@@ -120,9 +120,6 @@ static void __dwc3_set_mode(struct work_struct *work)unsigned long flags;int ret; -if (dwc->dr_mode !=USB_DR_MODE_OTG ) -return ; -pm_runtime_get_sync (dwc->dev);if (dwc->current_dr_role ==DWC3_GCTL_PRTCAP_OTG )
drivers/usb/dwc3/core.c void dwc3_set_mode(struct dwc3 *dwc, u32 mode) · line 210
Quoted 9 lines
drivers/usb/dwc3/core.c +3
-0@@ -205,6 +202,9 @@ void dwc3_set_mode(struct dwc3 *dwc, u32 mode) {unsigned long flags; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return ; +spin_lock_irqsave (&dwc->lock, flags); dwc->desired_dr_role = mode;spin_unlock_irqrestore (&dwc->lock, flags);
BR,
Thinh
Hi
Quoted 72 lines
-----Original Message-----
From: Jun Li <[email protected]>
Sent: Wednesday, July 22, 2020 5:07 PM
To: [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH] usb: dwc3: debugfs: do not queue work if try to change mode on non-drdDo not try to queue a drd work for change mode if the port is not a drd, this is to avoid below kernel dump:
[ 60.115529] ------------[ cut here ]------------ [ 60.120166] WARNING: CPU: 1 PID: 627 at kernel/workqueue.c:1473 __queue_work+0x46c/0x520 [ 60.128254] Modules linked in: [ 60.131313] CPU: 1 PID: 627 Comm: sh Not tainted 5.7.0-rc4-00022-g914a586-dirty #135 [ 60.139054] Hardware name: NXP i.MX8MQ EVK (DT) [ 60.143585] pstate: a0000085 (NzCv daIf -PAN -UAO) [ 60.148376] pc : __queue_work+0x46c/0x520 [ 60.152385] lr : __queue_work+0x314/0x520 [ 60.156393] sp : ffff8000124ebc40 [ 60.159705] x29: ffff8000124ebc40 x28: ffff800011808018 [ 60.165018] x27: ffff800011819ef8 x26: ffff800011d39980 [ 60.170331] x25: ffff800011808018 x24: 0000000000000100 [ 60.175643] x23: 0000000000000013 x22: 0000000000000001 [ 60.180955] x21: ffff0000b7c08e00 x20: ffff0000b6c31080 [ 60.186267] x19: ffff0000bb99bc00 x18: 0000000000000000 [ 60.191579] x17: 0000000000000000 x16: 0000000000000000 [ 60.196891] x15: 0000000000000000 x14: 0000000000000000 [ 60.202202] x13: 0000000000000000 x12: 0000000000000000 [ 60.207515] x11: 0000000000000000 x10: 0000000000000040 [ 60.212827] x9 : ffff800011d55460 x8 : ffff800011d55458 [ 60.218138] x7 : ffff0000b7800028 x6 : 0000000000000000 [ 60.223450] x5 : ffff0000b7800000 x4 : 0000000000000000 [ 60.228762] x3 : ffff0000bb997cc0 x2 : 0000000000000001 [ 60.234074] x1 : 0000000000000000 x0 : ffff0000b6c31088 [ 60.239386] Call trace: [ 60.241834] __queue_work+0x46c/0x520 [ 60.245496] queue_work_on+0x6c/0x90 [ 60.249075] dwc3_set_mode+0x48/0x58 [ 60.252651] dwc3_mode_write+0xf8/0x150 [ 60.256489] full_proxy_write+0x5c/0xa8 [ 60.260327] __vfs_write+0x18/0x40 [ 60.263729] vfs_write+0xdc/0x1c8 [ 60.267045] ksys_write+0x68/0xf0 [ 60.270360] __arm64_sys_write+0x18/0x20 [ 60.274286] el0_svc_common.constprop.0+0x68/0x160 [ 60.279077] do_el0_svc+0x20/0x80 [ 60.282394] el0_sync_handler+0x10c/0x178 [ 60.286403] el0_sync+0x140/0x180 [ 60.289716] ---[ end trace 70b155582e2b7988 ]---Signed-off-by Li Jun <[email protected]>--- drivers/usb/dwc3/debugfs.c | 3 +++ 1 file changed, 3 insertions(+)drivers/usb/dwc3/debugfs.c index renamed from drivers/usb/dwc3/debugfs.c +0
-06d9de33..99e7f53 100644
drivers/usb/dwc3/debugfs.c +3
-1@@ -428,6 +428,9 @@ static ssize_t dwc3_mode_write(struct file *file,if (copy_from_user (&buf, ubuf,min_t (size_t,sizeof (buf) -1 , count)))return -EFAULT ; +if (dwc->dr_mode !=USB_DR_MODE_OTG ) +return count; +if (!strncmp (buf,"host" ,4 )) mode =DWC3_GCTL_PRTCAP_HOST ; --
Quoted 1 line
2.7.4
A gentle ping...
Li Jun