usb: gadget: avoid using gadget after freed
PATCH 1/1 usb: gadget: avoid using gadget after freed
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
--- drivers/usb/gadget/function/f_ecm.c | 10 ++++++---- drivers/usb/gadget/function/f_eem.c | 10 ++++++---- drivers/usb/gadget/function/f_ncm.c | 11 +++++++---- drivers/usb/gadget/function/f_phonet.c | 2 +- drivers/usb/gadget/function/f_rndis.c | 2 +- drivers/usb/gadget/function/f_subset.c | 10 ++++++---- 6 files changed, 27 insertions(+), 18 deletions(-)
drivers/usb/gadget/function/f_ecm.c +6 -4
@@ -695,15 +695,17 @@ ecm_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to ecm_opts->bound access */ +mutex_lock (&ecm_opts->lock); +gether_set_gadget (ecm_opts->net, cdev->gadget);if (!ecm_opts->bound) { -mutex_lock (&ecm_opts->lock); -gether_set_gadget (ecm_opts->net, cdev->gadget); status =gether_register_netdev (ecm_opts->net); -mutex_unlock (&ecm_opts->lock); -if (status) +if (status) { +mutex_unlock (&ecm_opts->lock);return status; + } ecm_opts->bound = true; } +mutex_unlock (&ecm_opts->lock); ecm_string_defs[1 ].s = ecm->ethaddr;
drivers/usb/gadget/function/f_eem.c +6 -4
@@ -256,15 +256,17 @@ static int eem_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to eem_opts->bound access */ +mutex_lock (&eem_opts->lock); +gether_set_gadget (eem_opts->net, cdev->gadget);if (!eem_opts->bound) { -mutex_lock (&eem_opts->lock); -gether_set_gadget (eem_opts->net, cdev->gadget); status =gether_register_netdev (eem_opts->net); -mutex_unlock (&eem_opts->lock); -if (status) +if (status) { +mutex_unlock (&eem_opts->lock);return status; + } eem_opts->bound = true; } +mutex_unlock (&eem_opts->lock); us =usb_gstrings_attach (cdev, eem_strings,ARRAY_SIZE (eem_string_defs));
drivers/usb/gadget/function/f_ncm.c +7 -4
@@ -1409,15 +1409,18 @@ static int ncm_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to ncm_opts->bound access */ +mutex_lock (&ncm_opts->lock); +gether_set_gadget (ncm_opts->net, cdev->gadget);if (!ncm_opts->bound) { -mutex_lock (&ncm_opts->lock); -gether_set_gadget (ncm_opts->net, cdev->gadget); status =gether_register_netdev (ncm_opts->net); -mutex_unlock (&ncm_opts->lock); -if (status) +if (status) { +mutex_unlock (&ncm_opts->lock); goto fail; + } ncm_opts->bound = true; } +mutex_unlock (&ncm_opts->lock); + us =usb_gstrings_attach (cdev, ncm_strings,ARRAY_SIZE (ncm_string_defs));if (IS_ERR (us)) {
drivers/usb/gadget/function/f_phonet.c +1 -1
@@ -494,8 +494,8 @@ static int pn_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to phonet_opts->bound access */ +gphonet_set_gadget (phonet_opts->net, gadget);if (!phonet_opts->bound) { -gphonet_set_gadget (phonet_opts->net, gadget); status =gphonet_register_netdev (phonet_opts->net);if (status)return status;
drivers/usb/gadget/function/f_rndis.c +1 -1
@@ -698,8 +698,8 @@ rndis_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to rndis_opts->bound access */ +gether_set_gadget (rndis_opts->net, cdev->gadget);if (!rndis_opts->bound) { -gether_set_gadget (rndis_opts->net, cdev->gadget); status =gether_register_netdev (rndis_opts->net);if (status) goto fail;
drivers/usb/gadget/function/f_subset.c +6 -4
@@ -308,15 +308,17 @@ geth_bind(struct usb_configuration *c, struct usb_function *f)* with list_for_each_entry, so we assume no race condition * with regard to gether_opts->bound access */ +mutex_lock (&gether_opts->lock); +gether_set_gadget (gether_opts->net, cdev->gadget);if (!gether_opts->bound) { -mutex_lock (&gether_opts->lock); -gether_set_gadget (gether_opts->net, cdev->gadget); status =gether_register_netdev (gether_opts->net); -mutex_unlock (&gether_opts->lock); -if (status) +if (status) { +mutex_unlock (&gether_opts->lock);return status; + } gether_opts->bound = true; } +mutex_unlock (&gether_opts->lock); us =usb_gstrings_attach (cdev, geth_strings,ARRAY_SIZE (geth_string_defs));
Signature · 1 line
2.17.1
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>
I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Signature · 1 line
balbi
Quoted 18 lines · Felipe Balbi
Quoted 15 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Quoted 7 lines
Quoted 1 line
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
--
balbi
The issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4) [ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTED [ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deleted
The UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g. f_ncm) will not set the new gadget.
For kernel 4.19+, the role switch command will be:
echo "device" > /sys/class/usb_role/intel_xhci_usb_sw-role-switch/role
The latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.c
Thanks,
Lianwei
Hi,
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)
oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTED
What is this android_work. That doesn't exist upstream.
Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.
Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.c
Right, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
Signature · 1 line
balbi
Quoted 72 lines · Felipe Balbi
Hi,
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTEDWhat is this
android_work. That doesn't exist upstream.Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.c
Quoted 6 lines
Right, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
--
balbi
The issue exist in main line kernel, but I can not test it with v5.2-rc5 kernel. I tested it with 4.19 kernel,
which, for the usb gadget part, has almost the same code as v5.2. It is 100% reproducible with dual role
USB controller or by removing UDC hardware. Take f_ncm for example, the use case as follows:
1. USB controller is in host mode, f_ncm and UDC is configured in configfs.
- The ncm is instanced and alloced when "functions/ncm.usb0" is
created and it will be freed
when those files are delted from configfs.
2. enable the gadget and bind it to this ncm function.
- For the first time running, ncm_opts->bound is 0 andgether_set_gadget is called to set the
gadget. The bound is set to 1 then.
3. If the UDC is disconnected from bus, then the UDC and its gadget is deleted. But because the
ncm.usb0 is still there, ncm object is not freed andncm_opts->bound is still set.
There are two ways to disconnect the UDC hardware. One is for dual
role host controller by switch
host controller role to device mode. Another way is by removing
the UDC hardware from bus, both
will generate an usb device disconnect event to UDC driver to
delete udc and gadget.
4. Now the bound of ncm is still set and gadget is deleted due to udc disconnected. And if we connect
the udc device again, then it will create new udc and gadget and
bind to ncm again. But because
bound is already set, the new gadget is not set to gether
(gether_register_netdev not called).
Not sure if this is clear to you. Please review the scenario and the patch.
Thanks,
Lianwei
Hi,
Quoted 72 lines · Lianwei Wang
Quoted 72 lines
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTEDWhat is this
android_work. That doesn't exist upstream.Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.cRight, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
Quoted 5 lines
Quoted 2 lines
--
balbiThe issue exist in main line kernel, but I can not test it with v5.2-rc5 kernel. I tested it with 4.19 kernel,
which of the v4.19?
Quoted 4 lines
which, for the usb gadget part, has almost the same code as v5.2. It is 100% reproducible with dual role
USB controller or by removing UDC hardware. Takef_ncmfor example, the use case as follows:
Keep in mind that the way android handles dual-role is completely different from what we have upstream.
Quoted 11 lines
1. USB controller is in host mode,
f_ncmand UDC is configured in configfs.
- The ncm is instanced and alloced when "functions/ncm.usb0" is
created and it will be freed
when those files are delted from configfs.2. enable the gadget and bind it to this ncm function.
- For the first time running,ncm_opts->boundis 0 andgether_set_gadgetis called to set the
gadget. The bound is set to 1 then.3. If the UDC is disconnected from bus, then the UDC and its gadget is
what do you mean by "disconnected from the bus"? Removing the cable (aka, disconnect) will only cause the ->disconnect() callback to be called. It will not result in the UDC being freed. Is this, perhaps, something specific to android?
Quoted 9 lines
deleted. But because the
ncm.usb0 is still there, ncm object is not freed andncm_opts->boundis still set.
There are two ways to disconnect the UDC hardware. One is for dual
role host controller by switch
host controller role to device mode. Another way is by removing
the UDC hardware from bus, both
will generate an usb device disconnect event to UDC driver to
delete udc and gadget.
not true, unless I misunderstand what you mean. Disconnect will generate a disconnect interrupt in most UDCs (except for dummy) and the ->disconnect() callback will be called. Nothing will be freed.
Quoted 8 lines
4. Now the bound of ncm is still set and gadget is deleted due to udc disconnected. And if we connect
the udc device again, then it will create new udc and gadget and
bind to ncm again. But because
bound is already set, the new gadget is not set to gether
(gether_register_netdevnot called).Not sure if this is clear to you. Please review the scenario and the patch.
This sounds a little like it's android-specific. Is your platform using dwc3? Can you capture tracepoints of the failure? ftrace_dump_on_oops will help getting the actual tracepoints in this case.
cheers
Signature · 1 line
balbi
Quoted 85 lines · Felipe Balbi
Hi,
Quoted 72 lines · Lianwei Wang
Quoted 72 lines
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTEDWhat is this
android_work. That doesn't exist upstream.Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.cRight, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
Quoted 5 lines
Quoted 2 lines
--
balbiThe issue exist in main line kernel, but I can not test it with v5.2-rc5 kernel. I tested it with 4.19 kernel,
which of the v4.19?
Quoted 8 lines
Quoted 4 lines
which, for the usb gadget part, has almost the same code as v5.2. It is 100% reproducible with dual role
USB controller or by removing UDC hardware. Takef_ncmfor example, the use case as follows:Keep in mind that the way android handles dual-role is completely different from what we have upstream.
Right, Our xchi controller support dual role and normally it works in host mode and the other device, e.g. phone can connect to our system as a gadget device. It works in the same way as PC.
Quoted 10 lines
Quoted 10 lines
1. USB controller is in host mode,
f_ncmand UDC is configured in configfs.
- The ncm is instanced and alloced when "functions/ncm.usb0" is
created and it will be freed
when those files are delted from configfs.2. enable the gadget and bind it to this ncm function.
- For the first time running,ncm_opts->boundis 0 andgether_set_gadgetis called to set the
gadget. The bound is set to 1 then.
Quoted 7 lines
Quoted 1 line
3. If the UDC is disconnected from bus, then the UDC and its gadget is
what do you mean by "disconnected from the bus"? Removing the cable (aka, disconnect) will only cause the ->disconnect() callback to be called. It will not result in the UDC being freed. Is this, perhaps, something specific to android?
Our UDC is removable so we can remove the UDC from our system and the udc and gadget will be deleted after that.
And our xhci support dual role mode and we can switch role between host and device mode. In host mode, the UDC will be connected to xhci host controller and the other device can connect to UDC. In device mode, it will work as a device
and can connect to other host computer.
Quoted 10 lines
Quoted 9 lines
deleted. But because the
ncm.usb0 is still there, ncm object is not freed andncm_opts->boundis still set.
There are two ways to disconnect the UDC hardware. One is for dual
role host controller by switch
host controller role to device mode. Another way is by removing
the UDC hardware from bus, both
will generate an usb device disconnect event to UDC driver to
delete udc and gadget.
Quoted 4 lines
not true, unless I misunderstand what you mean. Disconnect will generate a disconnect interrupt in most UDCs (except for dummy) and the ->disconnect() callback will be called. Nothing will be freed.
As I mentioned above, the disconnect means the UDC is disconnected.
Quoted 7 lines
Quoted 7 lines
4. Now the bound of ncm is still set and gadget is deleted due to udc disconnected. And if we connect
the udc device again, then it will create new udc and gadget and
bind to ncm again. But because
bound is already set, the new gadget is not set to gether
(gether_register_netdevnot called).
Quoted 8 lines
Quoted 1 line
Not sure if this is clear to you. Please review the scenario and the patch.
This sounds a little like it's android-specific. Is your platform using dwc3? Can you capture tracepoints of the failure?
ftrace_dump_on_oopswill help getting the actual tracepoints in this case.cheers
It is not android specific. For the dual role controller, in device mode the dwc3 will
work as gadget device and can connect to PC as device. In host mode, the dwc3 does not work, instead the removable UDC will be connected to our system as gadget device.
Quoted 2 lines
--
balbi
And the key point is that if the UDC can be disconnected as deleted? The answer is
yes for below two cases:
- The UDC is removable
- It is a dual role controller and can switch between host mode and
device mode. In host
mode it will work as a xhci host controller. And in device mode it works as dwc3 gadget
device.
In both way we see that the udc is disconnected and deleted in our system.
config USB_ROLES_INTEL_XHCI
tristate "Intel XHCI USB Role Switch"
depends on ACPI && X86
help
Driver for the internal USB role switch for switching the USB data lines between the xHCI host controller and the dwc3 gadget controller found on various Intel SoCs.
To compile the driver as a module, choose M here: the module will be called intel-xhci-usb-role-switch.
The issue should be 100% reproducible with that support Intel XHCI USB role switch, or
the udc can be removabled.
Thanks,
Lianwei
Hi,
Quoted 184 lines · Lianwei Wang
Quoted 85 lines · Felipe Balbi
Hi,
Quoted 72 lines · Lianwei Wang
Quoted 72 lines
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTEDWhat is this
android_work. That doesn't exist upstream.Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.cRight, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
Quoted 5 lines
Quoted 2 lines
--
balbiThe issue exist in main line kernel, but I can not test it with v5.2-rc5 kernel. I tested it with 4.19 kernel,
which of the v4.19?
Quoted 8 lines
Quoted 4 lines
which, for the usb gadget part, has almost the same code as v5.2. It is 100% reproducible with dual role
USB controller or by removing UDC hardware. Takef_ncmfor example, the use case as follows:Keep in mind that the way android handles dual-role is completely different from what we have upstream.
Right, Our xchi controller support dual role and normally it works in host mode and the other device, e.g. phone can connect to our system as a gadget device. It works in the same way as PC.
Quoted 10 lines
Quoted 10 lines
1. USB controller is in host mode,
f_ncmand UDC is configured in configfs.
- The ncm is instanced and alloced when "functions/ncm.usb0" is
created and it will be freed
when those files are delted from configfs.2. enable the gadget and bind it to this ncm function.
- For the first time running,ncm_opts->boundis 0 andgether_set_gadgetis called to set the
gadget. The bound is set to 1 then.Quoted 7 lines
Quoted 1 line
3. If the UDC is disconnected from bus, then the UDC and its gadget is
what do you mean by "disconnected from the bus"? Removing the cable (aka, disconnect) will only cause the ->disconnect() callback to be called. It will not result in the UDC being freed. Is this, perhaps, something specific to android?
Our UDC is removable so we can remove the UDC from our system and the udc and gadget will be deleted after that.
And our xhci support dual role mode and we can switch role between host and device mode. In host mode, the UDC will be connected to xhci host controller and the other device can connect to UDC. In device mode, it will work as a device
and can connect to other host computer.Quoted 10 lines
Quoted 9 lines
deleted. But because the
ncm.usb0 is still there, ncm object is not freed andncm_opts->boundis still set.
There are two ways to disconnect the UDC hardware. One is for dual
role host controller by switch
host controller role to device mode. Another way is by removing
the UDC hardware from bus, both
will generate an usb device disconnect event to UDC driver to
delete udc and gadget.Quoted 4 lines
not true, unless I misunderstand what you mean. Disconnect will generate a disconnect interrupt in most UDCs (except for dummy) and the ->disconnect() callback will be called. Nothing will be freed.
As I mentioned above, the disconnect means the UDC is disconnected.
Quoted 7 lines
Quoted 7 lines
4. Now the bound of ncm is still set and gadget is deleted due to udc disconnected. And if we connect
the udc device again, then it will create new udc and gadget and
bind to ncm again. But because
bound is already set, the new gadget is not set to gether
(gether_register_netdevnot called).Quoted 8 lines
Quoted 1 line
Not sure if this is clear to you. Please review the scenario and the patch.
This sounds a little like it's android-specific. Is your platform using dwc3? Can you capture tracepoints of the failure?
ftrace_dump_on_oopswill help getting the actual tracepoints in this case.cheers
It is not android specific. For the dual role controller, in device mode the dwc3 will
work as gadget device and can connect to PC as device. In host mode, the dwc3 does not work, instead the removable UDC will be connected to our system as gadget device.Quoted 2 lines
--
balbiAnd the key point is that if the UDC can be disconnected as deleted? The answer is
yes for below two cases:
- The UDC is removable
- It is a dual role controller and can switch between host mode and
device mode. In host
mode it will work as a xhci host controller. And in device mode it works as dwc3 gadget
device.In both way we see that the udc is disconnected and deleted in our system.
config
USB_ROLES_INTEL_XHCI
tristate "Intel XHCI USB Role Switch"
depends on ACPI && X86
help
Driver for the internal USB role switch for switching the USB data lines between the xHCI host controller and the dwc3 gadget controller found on various Intel SoCs.
Quoted 6 lines
To compile the driver as a module, choose M here: the module will be called intel-xhci-usb-role-switch.
The issue should be 100% reproducible with that support Intel XHCI USB role switch, or
the udc can be removabled.
what's specific to your case is *how* you implemented dual-role. None of that is upstream and I have no means of verifying that what you claim is true.
Sorry
Signature · 1 line
balbi
Quoted 197 lines · Felipe Balbi
Hi,
Quoted 184 lines · Lianwei Wang
Quoted 85 lines · Felipe Balbi
Hi,
Quoted 72 lines · Lianwei Wang
Quoted 72 lines
Quoted 24 lines · Lianwei Wang
Quoted 23 lines · Felipe Balbi
Quoted 12 lines · Lianwei Wang
The udc and gadget device will be deleted when udc device is disconnected and the related function will be unbind with it.
But if the configfs is not deleted, then the function object will be kept and the bound status is kept true.
Then after udc device is connected again and a new udc and gadget objects will be created and passed to bind interface. But because the bound is still true, the new gadget is not updated to netdev and a previous freed gadget will be used in netdev after bind.
Quoted 4 lines
To fix this using after freed issue, always set the gadget object to netdev in bind interface.
Signed-off-by Lianwei Wang <[email protected]>I can't actually understand what's the problem here. The gadget is not deleted when we disconnect the cable.
Quoted 8 lines
Quoted 2 lines
--
balbiThe issue was observed with a dual-role capable USB controller (e.g. Intel XHCI controller), which has the ability to switch role between host and device mode. The gadget is deleted when we switch role to device mode from host mode. See below log:
# echo p > /sys/devices/pci0000:00/0000:00:15.1/intel-cht-otg.0/mux_state #(4.4)oh, so you're using a modified tree :-) Then we can't really help.
Quoted 8 lines
[ 41.170891] intel-cht-otg intel-cht-otg.0: p: set PERIPHERAL mode [ 41.171895] dwc3 dwc3.0.auto: DWC3 OTG Notify USB_EVENT_VBUS [ 41.187420] dwc3 dwc3.0.auto: dwc3_resume_common [ 41.191192] usb 1-1: USB disconnect, device number 3 [ 41.191284] usb 1-1.1: USB disconnect, device number 4 [ 41.218958] usb 1-1.5: USB disconnect, device number 5 [ 41.238117] android_work: sent uevent USB_STATE=CONFIGURED [ 41.240572] android_work: sent uevent USB_STATE=DISCONNECTEDWhat is this
android_work. That doesn't exist upstream.Quoted 15 lines
[ 41.263285] platform dabr_udc.0: unregister gadget driver 'configfs-gadget' [ 41.263413] configfs-gadget gadget: unbind function 'Function FS Gadget'/ffff8801db049e38 [ 41.263969] configfs-gadget gadget: unbind function 'cdc_network'/ffff8801d8897400 [ 41.325943] dabridge 1-1.5:1.0: Port 3 VBUS OFF [ 41.720957] dabr_udc deleted [ 41.721097] dabridge 1-5 deletedThe UDC and gadget will be deleted after switch role to device mode. And they will be
created as new object when switching back to host mode. At this time the bind in function
driver (e.g.f_ncm) will not set the new gadget.Quoted 5 lines
For kernel 4.19+, the role switch command will be:
echo "device" >/sys/class/usb_role/intel_xhci_usb_sw-role-switch/roleThe latest Intel role switch kernel driver can be found here:
https://elixir.bootlin.com/linux/v5.2-rc5/source/drivers/usb/roles/intel-xhci-usb-role-switch.cRight, please test against v5.2-rc5 and show me the problem on that kernel. I can't apply patches for problems that may not even exist in upstream, sorry.
Quoted 5 lines
Quoted 2 lines
--
balbiThe issue exist in main line kernel, but I can not test it with v5.2-rc5 kernel. I tested it with 4.19 kernel,
which of the v4.19?
Quoted 8 lines
Quoted 4 lines
which, for the usb gadget part, has almost the same code as v5.2. It is 100% reproducible with dual role
USB controller or by removing UDC hardware. Takef_ncmfor example, the use case as follows:Keep in mind that the way android handles dual-role is completely different from what we have upstream.
Right, Our xchi controller support dual role and normally it works in host mode and the other device, e.g. phone can connect to our system as a gadget device. It works in the same way as PC.
Quoted 10 lines
Quoted 10 lines
1. USB controller is in host mode,
f_ncmand UDC is configured in configfs.
- The ncm is instanced and alloced when "functions/ncm.usb0" is
created and it will be freed
when those files are delted from configfs.2. enable the gadget and bind it to this ncm function.
- For the first time running,ncm_opts->boundis 0 andgether_set_gadgetis called to set the
gadget. The bound is set to 1 then.Quoted 7 lines
Quoted 1 line
3. If the UDC is disconnected from bus, then the UDC and its gadget is
what do you mean by "disconnected from the bus"? Removing the cable (aka, disconnect) will only cause the ->disconnect() callback to be called. It will not result in the UDC being freed. Is this, perhaps, something specific to android?
Our UDC is removable so we can remove the UDC from our system and the udc and gadget will be deleted after that.
And our xhci support dual role mode and we can switch role between host and device mode. In host mode, the UDC will be connected to xhci host controller and the other device can connect to UDC. In device mode, it will work as a device
and can connect to other host computer.Quoted 10 lines
Quoted 9 lines
deleted. But because the
ncm.usb0 is still there, ncm object is not freed andncm_opts->boundis still set.
There are two ways to disconnect the UDC hardware. One is for dual
role host controller by switch
host controller role to device mode. Another way is by removing
the UDC hardware from bus, both
will generate an usb device disconnect event to UDC driver to
delete udc and gadget.Quoted 4 lines
not true, unless I misunderstand what you mean. Disconnect will generate a disconnect interrupt in most UDCs (except for dummy) and the ->disconnect() callback will be called. Nothing will be freed.
As I mentioned above, the disconnect means the UDC is disconnected.
Quoted 7 lines
Quoted 7 lines
4. Now the bound of ncm is still set and gadget is deleted due to udc disconnected. And if we connect
the udc device again, then it will create new udc and gadget and
bind to ncm again. But because
bound is already set, the new gadget is not set to gether
(gether_register_netdevnot called).Quoted 8 lines
Quoted 1 line
Not sure if this is clear to you. Please review the scenario and the patch.
This sounds a little like it's android-specific. Is your platform using dwc3? Can you capture tracepoints of the failure?
ftrace_dump_on_oopswill help getting the actual tracepoints in this case.cheers
It is not android specific. For the dual role controller, in device mode the dwc3 will
work as gadget device and can connect to PC as device. In host mode, the dwc3 does not work, instead the removable UDC will be connected to our system as gadget device.Quoted 2 lines
--
balbiAnd the key point is that if the UDC can be disconnected as deleted? The answer is
yes for below two cases:
- The UDC is removable
- It is a dual role controller and can switch between host mode and
device mode. In host
mode it will work as a xhci host controller. And in device mode it works as dwc3 gadget
device.In both way we see that the udc is disconnected and deleted in our system.
config
USB_ROLES_INTEL_XHCI
tristate "Intel XHCI USB Role Switch"
depends on ACPI && X86
help
Driver for the internal USB role switch for switching the USB data lines between the xHCI host controller and the dwc3 gadget controller found on various Intel SoCs.Quoted 6 lines
To compile the driver as a module, choose M here: the module will be called intel-xhci-usb-role-switch.
The issue should be 100% reproducible with that support Intel XHCI USB role switch, or
the udc can be removabled.
Quoted 8 lines
what's specific to your case is *how* you implemented dual-role. None of that is upstream and I have no means of verifying that what you claim is true.
Sorry
--
balbi
That's Ok, Maybe I'm misunderstanding how upstream kernel works.
The issue is happened on Intel hardware which should be support by upstream kernel.
I did not implement any dual role, it is from upstream. And the UDC driver is working as the same mechanism as upstream too. The upstream kernel support remove/disconnect UDC device.
I try to explain it clearly but unfortunately... my failure. If we look at the code,
we have remove function defined in udc driver, and which will finally delete UDC (usb_del_gadget_udc).
Could you tell me how the upstream kernel guarantee that the remove and usb_del_gadget_udc function is never called?
Thanks,
Lianwei