Re: [PATCH v3 3/3] Bluetooth: hci_h5: Add runtime suspend

From: Ondřej Jirman
Date: Thu Oct 28 2021 - 04:12:16 EST

Hell Archie,

On Thu, Oct 28, 2021 at 10:06:58AM +0800, Archie Pusaka wrote:
> Hi Ondrej,
>
> There's a mistake on my side with the WAKEUP_DISABLE flag, but it's
> fixed by Hans in this patch.
> https://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth-next.git/commit/drivers/bluetooth/hci_h5.c?id=9a9023f314873241a43b5a2b96e9c0caaa958433
> Could you try and see whether that fixes your issue?

Yes, that fixes the issue. Thanks. :)

kind regards,
o.

> Thanks,
> Archie
>
>
> On Thu, 28 Oct 2021 at 07:47, Ondřej Jirman <megi@xxxxxx> wrote:
> >
> > On Thu, Oct 28, 2021 at 12:23:26AM +0200, megi xff wrote:
> > > Hello Archie,
> > >
> > > On Fri, Jul 23, 2021 at 07:31:57PM +0800, Archie Pusaka wrote:
> > > > From: Archie Pusaka <apusaka@xxxxxxxxxxxx>
> > > >
> > > > This patch allows the controller to suspend after a short period of
> > > > inactivity.
> > >
> > > I see this pattern in dmesg after this patch: (I've added printks
> > > to many hci_h5 functions to see what's going on)
> > >
> > > [ 493.150325] h5_dequeue
> > > [ 493.150332] h5_dequeue
> > > [ 493.150336] h5_dequeue
> > > [ 493.150340] h5_dequeue
> > > [ 493.150370] h5_dequeue
> > > [ 493.150547] h5_recv
> > > [ 493.150863] h5_recv
> > > [ 493.150878] h5_dequeue
> > > [ 493.150885] h5_dequeue
> > > [ 493.150888] h5_dequeue
> > > [ 493.151315] h5_enqueue
> > > [ 493.151328] h5_dequeue
> > > [ 493.151350] h5_dequeue
> > > [ 493.151447] h5_dequeue
> > > [ 493.151612] h5_recv
> > > [ 493.151945] h5_recv
> > > [ 493.151961] h5_dequeue
> > > [ 493.151967] h5_dequeue
> > > [ 493.151970] h5_dequeue
> > > [ 495.171812] h5_flush
> > > [ 495.171845] h5_flush
> > > [ 499.267473] h5_serdev_suspend
> > > [ 499.267490] h5_btrtl_suspend
> > > [ 499.273784] h5_recv
> > > [ 499.273828] h5_serdev_resume
> > > [ 499.273833] h5_btrtl_resume
> > > [ 499.273837] h5_btrtl_resume / reprobe
> > > [ 499.273855] h5_btrtl_reprobe_worker
> > > [ 499.273913] h5_serdev_remove
> > > [ 499.274997] h5_close
> > > [ 499.275010] h5_btrtl_close
> > > [ 499.275624] h5_serdev_probe
> > > [ 499.276126] h5_open
> > > [ 499.276132] h5_btrtl_open
> > > [ 499.915820] h5_dequeue
> > > [ 499.915857] h5_dequeue
> > > [ 499.915863] h5_dequeue
> > > [ 499.916212] h5_dequeue
> > > [ 499.919643] h5_recv
> > > [ 499.919675] h5_dequeue
> > > [ 499.919682] h5_dequeue
> > > [ 499.919687] h5_dequeue
> > > [ 499.919692] h5_dequeue
> > >
> > > repeating ad nauseam every 6s.
> > >
> > > Basically bluetooth device reprobes every 6s. Looks like h5_recv call
> > > after h5_btrtl_suspend wakes the device immediately after suspend.
> > >
> > > (there are no users of bluetooth in my userspace) I'd expect the
> > > device to stay suspended after suspend.
> > >
> > > I have some extra patches to support 8723cs but nothing that
> > > would affect this codepath. https://megous.com/git/linux/log/?h=bt-5.15
> > >
> > > I assume it will have the same behavior with 8723bs which is already
> > > mainline. I guess this issue is specific to devices with H5_INFO_WAKEUP_DISABLE
> > > flag set.
> > >
> > > Do you have any ideas?
> >
> > I've added dump_stack() to first h5_recv call after suspend (the one
> > that's causing the immediate wakeup after runtime PM suspend), and it returns:
> >
> > [ 5.938258] recv
> > [ 13.377775] suspend
> > [ 13.384106] recv
> > [ 13.384120] CPU: 1 PID: 83 Comm: kworker/u8:1 Tainted: G C 5.15.0-rc7-00002-g64f2c49e8400 #23
> > [ 13.384141] Hardware name: Pine64 PinePhone (1.2) (DT)
> > [ 13.384151] Workqueue: events_unbound flush_to_ldisc
> > [ 13.384196] Call trace:
> > [ 13.384199] dump_backtrace+0x0/0x15c
> > [ 13.384224] show_stack+0x14/0x20
> > [ 13.384232] dump_stack_lvl+0x64/0x7c
> > [ 13.384250] dump_stack+0x14/0x2c
> > [ 13.384258] h5_recv+0x44/0xdbc [hci_uart]
> > [ 13.384288] hci_uart_receive_buf+0x6c/0x94 [hci_uart]
> > [ 13.384298] ttyport_receive_buf+0x60/0xf4
> > [ 13.384318] flush_to_ldisc+0xb0/0x160
> > [ 13.384324] process_one_work+0x1d8/0x380
> > [ 13.384339] worker_thread+0x178/0x4e0
> > [ 13.384348] kthread+0x11c/0x130
> > [ 13.384359] ret_from_fork+0x10/0x20
> >
> > It's comming from here https://elixir.bootlin.com/linux/latest/source/drivers/tty/tty_buffer.c#L510
> >
> > Which can be scheduled from these places:
> >
> > - https://elixir.bootlin.com/linux/latest/source/drivers/tty/tty_buffer.c#L65
> > - https://elixir.bootlin.com/linux/latest/source/drivers/tty/tty_buffer.c#L413
> >
> > And that's where I lose a thread of what can be happening. :)
> >
> > Maybe h5_recv is not a good function to mark activity on the device,
> > due to tty_buffer code just calling it to check if some data are
> > available, even if none are? Even if nothing uses bluetooth from
> > userspace...
> >
> > kind regards,
> > o.
> >
> > > kind regards,
> > > o.
> > >
> > >
> > > > Signed-off-by: Archie Pusaka <apusaka@xxxxxxxxxxxx>
> > > > Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@xxxxxxxxxxxx>
> > > > Reviewed-by: Hilda Wu <hildawu@xxxxxxxxxxx>
> > > >
> > > > ---
> > > >
> > > > Changes in v3:
> > > > * Reordering #include
> > > >
> > > > drivers/bluetooth/hci_h5.c | 20 ++++++++++++++++++++
> > > > 1 file changed, 20 insertions(+)
> > > >
> > > > diff --git a/drivers/bluetooth/hci_h5.c b/drivers/bluetooth/hci_h5.c
> > > > index cbc63b057f33..0c0dedece59c 100644
> > > > --- a/drivers/bluetooth/hci_h5.c
> > > > +++ b/drivers/bluetooth/hci_h5.c
> > > > @@ -12,6 +12,7 @@
> > > > #include <linux/kernel.h>
> > > > #include <linux/mod_devicetable.h>
> > > > #include <linux/of_device.h>
> > > > +#include <linux/pm_runtime.h>
> > > > #include <linux/serdev.h>
> > > > #include <linux/skbuff.h>
> > > >
> > > > @@ -21,6 +22,8 @@
> > > > #include "btrtl.h"
> > > > #include "hci_uart.h"
> > > >
> > > > +#define SUSPEND_TIMEOUT_MS 6000
> > > > +
> > > > #define HCI_3WIRE_ACK_PKT 0
> > > > #define HCI_3WIRE_LINK_PKT 15
> > > >
> > > > @@ -584,6 +587,10 @@ static int h5_recv(struct hci_uart *hu, const void *data, int count)
> > > > count -= processed;
> > > > }
> > > >
> > > > + pm_runtime_get(&hu->serdev->dev);
> > > > + pm_runtime_mark_last_busy(&hu->serdev->dev);
> > > > + pm_runtime_put_autosuspend(&hu->serdev->dev);
> > > > +
> > > > return 0;
> > > > }
> > > >
> > > > @@ -620,6 +627,10 @@ static int h5_enqueue(struct hci_uart *hu, struct sk_buff *skb)
> > > > break;
> > > > }
> > > >
> > > > + pm_runtime_get_sync(&hu->serdev->dev);
> > > > + pm_runtime_mark_last_busy(&hu->serdev->dev);
> > > > + pm_runtime_put_autosuspend(&hu->serdev->dev);
> > > > +
> > > > return 0;
> > > > }
> > > >
> > > > @@ -951,6 +962,12 @@ static void h5_btrtl_open(struct h5 *h5)
> > > > serdev_device_set_parity(h5->hu->serdev, SERDEV_PARITY_EVEN);
> > > > serdev_device_set_baudrate(h5->hu->serdev, 115200);
> > > >
> > > > + pm_runtime_set_active(&h5->hu->serdev->dev);
> > > > + pm_runtime_use_autosuspend(&h5->hu->serdev->dev);
> > > > + pm_runtime_set_autosuspend_delay(&h5->hu->serdev->dev,
> > > > + SUSPEND_TIMEOUT_MS);
> > > > + pm_runtime_enable(&h5->hu->serdev->dev);
> > > > +
> > > > /* The controller needs up to 500ms to wakeup */
> > > > gpiod_set_value_cansleep(h5->enable_gpio, 1);
> > > > gpiod_set_value_cansleep(h5->device_wake_gpio, 1);
> > > > @@ -959,6 +976,8 @@ static void h5_btrtl_open(struct h5 *h5)
> > > >
> > > > static void h5_btrtl_close(struct h5 *h5)
> > > > {
> > > > + pm_runtime_disable(&h5->hu->serdev->dev);
> > > > +
> > > > gpiod_set_value_cansleep(h5->device_wake_gpio, 0);
> > > > gpiod_set_value_cansleep(h5->enable_gpio, 0);
> > > > }
> > > > @@ -1066,6 +1085,7 @@ MODULE_DEVICE_TABLE(acpi, h5_acpi_match);
> > > >
> > > > static const struct dev_pm_ops h5_serdev_pm_ops = {
> > > > SET_SYSTEM_SLEEP_PM_OPS(h5_serdev_suspend, h5_serdev_resume)
> > > > + SET_RUNTIME_PM_OPS(h5_serdev_suspend, h5_serdev_resume, NULL)
> > > > };
> > > >
> > > > static const struct of_device_id rtl_bluetooth_of_match[] = {
> > > > --
> > > > 2.32.0.432.gabb21c7263-goog
> > > >