usb: chipidea: fixes for usb-linus
Subsystems (3)
Peter Chen
maintainercover usb: chipidea: fixes for usb-linus
From: Peter Chen <[email protected]>
Peter Chen (1):
usb: chipidea: udc: fix sleeping function called from invalid context
drivers/usb/chipidea/udc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Signature · 1 line
2.17.1
Peter Chen
maintainerPATCH 1/1 usb: chipidea: udc: fix sleeping function called from invalid context
From: Peter Chen <[email protected]>
The code calls pm_runtime_get_sync with irq disabled, it causes below warning:
BUG: sleeping function called from invalid context at wer/runtime.c:1075 in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: er/u8:1 CPU: 1 PID: 37 Comm: kworker/u8:1 Not tainted 20200304-00181-gbebfd2a5be98 #1588 Hardware name: NVIDIA Tegra SoC (Flattened Device Tree) Workqueue: ci_otg ci_otg_work
[<c010e8bd>] (unwind_backtrace) from [<c010a315>] 1/0x14)
[<c010a315>] (show_stack) from [<c0987d29>] 5/0x94)
[<c0987d29>] (dump_stack) from [<c013e77f>] +0xeb/0x118)
[<c013e77f>] (___might_sleep) from [<c052fa1d>] esume+0x75/0x78)
[<c052fa1d>] (__pm_runtime_resume) from [<c0627a33>] 0x23/0x74)
[<c0627a33>] (ci_udc_pullup) from [<c062fb93>] nect+0x2b/0xcc)
[<c062fb93>] (usb_gadget_connect) from [<c062769d>] _connect+0x59/0x104)
[<c062769d>] (ci_hdrc_gadget_connect) from [<c062778b>] ssion+0x43/0x48)
[<c062778b>] (ci_udc_vbus_session) from [<c062f997>] s_connect+0x17/0x9c)
[<c062f997>] (usb_gadget_vbus_connect) from [<c062634d>] bd/0x128)
[<c062634d>] (ci_otg_work) from [<c0134719>] rk+0x149/0x404)
[<c0134719>] (process_one_work) from [<c0134acb>] 0xf7/0x3bc)
[<c0134acb>] (worker_thread) from [<c0139433>] x118)
[<c0139433>] (kthread) from [<c01010bd>] (ret_from_fork+0x11/0x34)
--- drivers/usb/chipidea/udc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/usb/chipidea/udc.c +3 -1
@@ -1539,9 +1539,11 @@ static void ci_hdrc_gadget_connect(struct usb_gadget *_gadget, int is_active)if (ci->driver) {hw_device_state (ci, ci->ep0out->qh.dma);usb_gadget_set_state (_gadget,USB_STATE_POWERED ); +spin_unlock_irqrestore (&ci->lock, flags);usb_udc_vbus_handler (_gadget, true); + }else { +spin_unlock_irqrestore (&ci->lock, flags); } -spin_unlock_irqrestore (&ci->lock, flags); }else {usb_udc_vbus_handler (_gadget, false);if (ci->driver)
Signature · 1 line
2.17.1
Re: [PATCH 1/1] usb: chipidea: udc: fix sleeping function called from invalid context
Quoted 9 lines · Peter Chen
From: Peter Chen <[email protected]>
The code calls
pm_runtime_get_syncwith irq disabled, it causes below warning:BUG: sleeping function called from invalid context at
wer/runtime.c:1075in_atomic(): 1,irqs_disabled(): 128,non_block: 0, pid:
er/u8:1
...
Quoted 14 lines
Tested-by Dmitry Osipenko <[email protected]>Cc <[email protected]> #v5.5Fixes 72dc8df7920f ("usb: chipidea: udc: protect usb interrupt enable")Reported-by Dmitry Osipenko <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/chipidea/udc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/usb/chipidea/udc.c +0
-0
drivers/usb/chipidea/udc.c static void ci_hdrc_gadget_connect(struct usb_gadget *_gadget, int is_active) · line 1544
Quoted 9 lines
drivers/usb/chipidea/udc.c +3
-1@@ -1539,9 +1539,11 @@ static void ci_hdrc_gadget_connect(struct usb_gadget *_gadget, int is_active)if (ci->driver) {hw_device_state (ci, ci->ep0out->qh.dma);usb_gadget_set_state (_gadget,USB_STATE_POWERED ); +spin_unlock_irqrestore (&ci->lock, flags);usb_udc_vbus_handler (_gadget, true); + }else { +spin_unlock_irqrestore (&ci->lock, flags); } -spin_unlock_irqrestore (&ci->lock, flags);
There's something strange about this patch.
Do you really know that interrupts will be enabled following the spin_unlock_irqrestore()? In other words, do you know that interrupts were enabled upon entry to this routine?
If they were, then why use spin_lock_irqsave/spin_unlock_irqrestore? Why not use spin_lock_irq and spin_unlock_irq?
Alan Stern
Re: [PATCH 1/1] usb: chipidea: udc: fix sleeping function called from invalid context
Quoted 40 lines · Alan Stern
Quoted 9 lines · Peter Chen
From: Peter Chen <[email protected]>
The code calls
pm_runtime_get_syncwith irq disabled, it causes below warning:BUG: sleeping function called from invalid context at
wer/runtime.c:1075in_atomic(): 1,irqs_disabled(): 128,non_block: 0, pid:
er/u8:1...
Quoted 14 lines
Tested-by Dmitry Osipenko <[email protected]>Cc <[email protected]> #v5.5Fixes 72dc8df7920f ("usb: chipidea: udc: protect usb interrupt enable")Reported-by Dmitry Osipenko <[email protected]>Signed-off-by Peter Chen <[email protected]>--- drivers/usb/chipidea/udc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/usb/chipidea/udc.c +0
-0drivers/usb/chipidea/udc.c static void ci_hdrc_gadget_connect(struct usb_gadget *_gadget, int is_active) · line 1544
Quoted 9 lines
drivers/usb/chipidea/udc.c +3
-1@@ -1539,9 +1539,11 @@ static void ci_hdrc_gadget_connect(struct usb_gadget *_gadget, int is_active)if (ci->driver) {hw_device_state (ci, ci->ep0out->qh.dma);usb_gadget_set_state (_gadget,USB_STATE_POWERED ); +spin_unlock_irqrestore (&ci->lock, flags);usb_udc_vbus_handler (_gadget, true); + }else { +spin_unlock_irqrestore (&ci->lock, flags); } -spin_unlock_irqrestore (&ci->lock, flags);There's something strange about this patch.
Quoted 7 lines
Do you really know that interrupts will be enabled following the
spin_unlock_irqrestore()? In other words, do you know that interrupts were enabled upon entry to this routine?If they were, then why use spin_lock_irqsave/spin_unlock_irqrestore? Why not use
spin_lock_irqandspin_unlock_irq?
This function is called at process context, so the interrupt is enabled, I will use spin_lock_irq, thanks, Alan.
Signature · 2 lines
Thanks, Peter Chen