Skip to contentSkip to conversation

USB(u_serial.c): it has a NULL point

PATCH In review hui yang 3 participants 4 messages Nov 3, 2019 – Nov 13, 2019 USB

lore
Share & tools
lore

hui yang

author
#0 lore raw

PATCH 2/2 USB(u_serial.c): it has a NULL point

From: YangHui <[email protected]>

Unable to handle kernel NULL pointer dereference at virtual address 000002a0
Kernel BUG at tty_wakeup+0x18/0x8c
[<0000000095135e9d>] tty_wakeup+0x18/0x8c
[<00000000c35f19e4>] gs_start_io+0xc8/0x158
[<00000000b11b184e>] gserial_connect+0xd0/0x174
[<00000000642201de>] acm_set_alt+0x70/0x180
[<00000000a9d38e8b>] composite_setup+0x1848/0x1a7c
[<0000000072c88781>] android_setup+0x1ec/0x1f4
[<000000004c0127a5>] msm_udc_irq+0x8d0/0xc0c
[<00000000f8bb9acf>] __handle_irq_event_percpu+0x74/0x294
[<00000000afd57256>] handle_irq_event+0x60/0xc8
[<0000000076719b24>] handle_fasteoi_irq+0xc8/0x1c4
[<00000000421f3d87>] generic_handle_irq+0x34/0x4c
[<00000000b2f2d343>] __handle_domain_irq+0x6c/0xbc
[<00000000c74325e8>] gic_handle_irq+0x6c/0xe0

yes,we can see the port->port.tty==NULL, so system crashed

---
 drivers/usb/gadget/function/u_serial.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/usb/gadget/function/u_serial.c +2 -0
index 65f634e..a1c2ac2 100644
@@ -562,6 +562,8 @@ static int gs_start_io(struct gs_port *port)
 
 	/* unblock any pending writes into our circular buffer */
 	if (started) {
+		if (port->port.tty == NULL)
+			return -EIO;
 		tty_wakeup(port->port.tty);
 	} else {
 		gs_free_requests(ep, head, &port->read_allocated);
Signature · 1 line
2.7.4
maintainer
#1 lore raw
Quoted 29 lines · hui yang

From: YangHui <[email protected]>

Unable to handle kernel NULL pointer dereference at virtual address 000002a0
Kernel BUG at tty_wakeup+0x18/0x8c
[<0000000095135e9d>] tty_wakeup+0x18/0x8c
[<00000000c35f19e4>] gs_start_io+0xc8/0x158
[<00000000b11b184e>] gserial_connect+0xd0/0x174
[<00000000642201de>] acm_set_alt+0x70/0x180
[<00000000a9d38e8b>] composite_setup+0x1848/0x1a7c
[<0000000072c88781>] android_setup+0x1ec/0x1f4
[<000000004c0127a5>] msm_udc_irq+0x8d0/0xc0c
[<00000000f8bb9acf>] __handle_irq_event_percpu+0x74/0x294
[<00000000afd57256>] handle_irq_event+0x60/0xc8
[<0000000076719b24>] handle_fasteoi_irq+0xc8/0x1c4
[<00000000421f3d87>] generic_handle_irq+0x34/0x4c
[<00000000b2f2d343>] __handle_domain_irq+0x6c/0xbc
[<00000000c74325e8>] gic_handle_irq+0x6c/0xe0

yes,we can see the port->port.tty==NULL, so system crashed

---
 drivers/usb/gadget/function/u_serial.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/usb/gadget/function/u_serial.c +0 -0
index 65f634e..a1c2ac2 100644

drivers/usb/gadget/function/u_serial.c static int gs_start_io(struct gs_port *port) · line 566

Quoted 5 lines
drivers/usb/gadget/function/u_serial.c +2 -0
@@ -562,6 +562,8 @@ static int gs_start_io(struct gs_port *port)
 
 	/* unblock any pending writes into our circular buffer */
 	if (started) {
+		if (port->port.tty == NULL)
+			return -EIO;

How can this happen? Shouldn't this be handled earlier?

If not, can you document this how this can be true and what should be done about it?

thanks,

greg k-h

author
#2 lore raw

I'm sorry for my fault . Now I send to everyone .

When we use USB to transmit data
with PC, the phone will crashed Low probability .
We found port->port.tty became NULL.

Thanks

Quoted 45 lines · Greg KH <[email protected]> 于2019年11月3日周日 下午8:42写道:
Quoted 29 lines · hui yang

From: YangHui <[email protected]>

Unable to handle kernel NULL pointer dereference at virtual address 000002a0
Kernel BUG at tty_wakeup+0x18/0x8c
[<0000000095135e9d>] tty_wakeup+0x18/0x8c
[<00000000c35f19e4>] gs_start_io+0xc8/0x158
[<00000000b11b184e>] gserial_connect+0xd0/0x174
[<00000000642201de>] acm_set_alt+0x70/0x180
[<00000000a9d38e8b>] composite_setup+0x1848/0x1a7c
[<0000000072c88781>] android_setup+0x1ec/0x1f4
[<000000004c0127a5>] msm_udc_irq+0x8d0/0xc0c
[<00000000f8bb9acf>] __handle_irq_event_percpu+0x74/0x294
[<00000000afd57256>] handle_irq_event+0x60/0xc8
[<0000000076719b24>] handle_fasteoi_irq+0xc8/0x1c4
[<00000000421f3d87>] generic_handle_irq+0x34/0x4c
[<00000000b2f2d343>] __handle_domain_irq+0x6c/0xbc
[<00000000c74325e8>] gic_handle_irq+0x6c/0xe0

yes,we can see the port->port.tty==NULL, so system crashed

---
 drivers/usb/gadget/function/u_serial.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/usb/gadget/function/u_serial.c +0 -0
index 65f634e..a1c2ac2 100644

drivers/usb/gadget/function/u_serial.c static int gs_start_io(struct gs_port *port) · line 566

Quoted 5 lines
drivers/usb/gadget/function/u_serial.c +2 -0
@@ -562,6 +562,8 @@ static int gs_start_io(struct gs_port *port)
 
      /* unblock any pending writes into our circular buffer */
      if (started) {
+             if (port->port.tty == NULL)
+                     return -EIO;

How can this happen? Shouldn't this be handled earlier?

If not, can you document this how this can be true and what should be done about it?

thanks,

greg k-h

#3 lore raw
Quoted 5 lines · hui yang

I'm sorry for my fault . Now I send to everyone .

When we use USB to transmit data
with PC, the phone will crashed Low probability .
We found port->port.tty became NULL.

Would you please debug more, it seems gs open/close (gs_open/gs_close) are something wrong. The value port->port.tty and port->port.count are inconsistent.

Peter

Quoted 49 lines

Thanks

Quoted 45 lines · Greg KH <[email protected]> 于2019年11月3日周日 下午8:42写道:
Quoted 29 lines · hui yang

From: YangHui <[email protected]>

Unable to handle kernel NULL pointer dereference at virtual address 000002a0
Kernel BUG at tty_wakeup+0x18/0x8c
[<0000000095135e9d>] tty_wakeup+0x18/0x8c
[<00000000c35f19e4>] gs_start_io+0xc8/0x158
[<00000000b11b184e>] gserial_connect+0xd0/0x174
[<00000000642201de>] acm_set_alt+0x70/0x180
[<00000000a9d38e8b>] composite_setup+0x1848/0x1a7c
[<0000000072c88781>] android_setup+0x1ec/0x1f4
[<000000004c0127a5>] msm_udc_irq+0x8d0/0xc0c
[<00000000f8bb9acf>] __handle_irq_event_percpu+0x74/0x294
[<00000000afd57256>] handle_irq_event+0x60/0xc8
[<0000000076719b24>] handle_fasteoi_irq+0xc8/0x1c4
[<00000000421f3d87>] generic_handle_irq+0x34/0x4c
[<00000000b2f2d343>] __handle_domain_irq+0x6c/0xbc
[<00000000c74325e8>] gic_handle_irq+0x6c/0xe0

yes,we can see the port->port.tty==NULL, so system crashed

---
 drivers/usb/gadget/function/u_serial.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/usb/gadget/function/u_serial.c +0 -0
index 65f634e..a1c2ac2 100644

drivers/usb/gadget/function/u_serial.c static int gs_start_io(struct gs_port *port) · line 566

Quoted 5 lines
drivers/usb/gadget/function/u_serial.c +2 -0
@@ -562,6 +562,8 @@ static int gs_start_io(struct gs_port *port)
 
      /* unblock any pending writes into our circular buffer */
      if (started) {
+             if (port->port.tty == NULL)
+                     return -EIO;

How can this happen? Shouldn't this be handled earlier?

If not, can you document this how this can be true and what should be done about it?

thanks,

greg k-h

Signature · 2 lines
Thanks,
Peter Chen