| Message ID | 20240523232216.3148367-1-dianders@chromium.org |
|---|---|
| Headers | show |
| Series | serial: Fix problems when serial transfer is happening at suspend time | expand |
On Thu, May 23, 2024 at 04:22:12PM -0700, Douglas Anderson wrote: > Recently, suspend testing on sc7180-trogdor based devices has started > to sometimes fail with messages like this: > > port a88000.serial:0.0: PM: calling pm_runtime_force_suspend+0x0/0xf8 @ 28934, parent: a88000.serial:0 > port a88000.serial:0.0: PM: dpm_run_callback(): pm_runtime_force_suspend+0x0/0xf8 returns -16 > port a88000.serial:0.0: PM: pm_runtime_force_suspend+0x0/0xf8 returned -16 after 33 usecs > port a88000.serial:0.0: PM: failed to suspend: error -16 > > I could reproduce these problem by logging in via an agetty on the > debug serial port (which was _not_ used for kernel console) and > running: > cat /var/log/messages > ...and then (via an SSH session) forcing a few suspend/resume cycles. > > Tracing through the code and doing some printf debugging shows that printf() ...or... printf()-based > the -16 (-EBUSY) comes from the recently added > serial_port_runtime_suspend(). > > The idea of the serial_port_runtime_suspend() function is to prevent > the port from being _runtime_ suspended if it still has bytes left to > transmit. Having bytes left to transmit isn't a reason to block > _system_ suspend, though. The DEFINE_RUNTIME_DEV_PM_OPS() used by the > serial_port code means that the system suspend function will be > pm_runtime_force_suspend(). In pm_runtime_force_suspend() we can see > that before calling the runtime suspend function we'll call > pm_runtime_disable(). This should be a reliable way to detect that > we're called from system suspend and that we shouldn't look for > busyness. ... > + /* > + * We only want to check the busyness of the port if PM Runtime is > + * enabled. Specifically PM Runtime will be disabled by > + * pm_runtime_force_suspend() during system suspend and we don't want > + * to block system suspend even if there is data still left to > + * transmit. We only want to block regulator PM Runtime transitions. regular > + */ > + if (!pm_runtime_enabled(dev)) > + return 0;
This is a set of two patches that fix problems related to suspending while a serial transfer is going on. The two patches are independent from each other and can land in any order. The only thing tying them together is that I used the same test to reproduce both of them. Specifically, I could reproduce my problemes by logging in via an agetty on the debug serial port (which was _not_ used for kernel console) and running: cat /var/log/messages ...and then (via an SSH session) forcing a few suspend/resume cycles. The first patch solves a problem that is probably more major. It was introduced recently and has even shown up in stable trees. Suspend/resume testing in ChromeOS test labs are hitting the problem fixed by this patch. The fix hasn't been tested in labs, but when I reproduced the problem locally I could see that the fix worked. IMO it should land ASAP. The second patch fixes an ancient problem that I only found because I was trying to reproduce the first problem. Given how long it's been around it's probably not urgent but it would be nice to get fixed. Douglas Anderson (2): serial: port: Don't block system suspend even if bytes are left to xmit serial: qcom-geni: Fix qcom_geni_serial_stop_tx_fifo() while xfer drivers/tty/serial/qcom_geni_serial.c | 45 +++++++++++++++++++++++++-- drivers/tty/serial/serial_port.c | 10 ++++++ 2 files changed, 52 insertions(+), 3 deletions(-)