| Message ID | 20250526-topic-tcpm-tx-poll-cond-v1-1-38a81560ffff@linaro.org |
|---|---|
| State | New |
| Headers | show |
| Series | tcpm: fix pd_transmit poll condition | expand |
Hi Neil Good catch. Thanks! On Mon, 2025-05-26 at 09:42 +0200, Neil Armstrong wrote: > The read_poll_timeout() exit "cond" parameter wait for a complete > transmit state. > > The "!tx_complete" cond means that on the first tcpm_transmit_helper() > call, if the tx_complete returns to false, the poll loop would exit. > But since at probe the tx_status is left at 0, the tx is considered > as successful. And then later the proper status is returned and > ignored. > > But if on the first tcpm_transmit_helper() call the TX status is > returned the read_poll_timeout() would return in timeout and the > negociation would fail. > > Fixes: 1db4c0ac77e ("usb: tcpm: add core framework") > Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org> Tested-by: Marcel Ziswiler <marcel.ziswiler@codethink.co.uk> # rock5b > --- > drivers/usb/tcpm/tcpm.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/usb/tcpm/tcpm.c b/drivers/usb/tcpm/tcpm.c > index 0aee57cb2f4a44af7b483e0e609e39ba584a06c0..c8679615cd7701982b64b1905c5822ac98ed6f79 100644 > --- a/drivers/usb/tcpm/tcpm.c > +++ b/drivers/usb/tcpm/tcpm.c > @@ -209,7 +209,7 @@ static int tcpm_pd_transmit(struct udevice *dev, > */ > timeout_us *= 5; > ret = read_poll_timeout(tcpm_transmit_helper, tx_complete, > - !tx_complete, false, timeout_us, dev); > + tx_complete, false, timeout_us, dev); > if (ret < 0) { > dev_err(dev, "TCPM: PD transmit data failed: %d\n", ret); > return ret; > > --- > base-commit: bab54f5942c428be698216224fd10b91d974d4da > change-id: 20250526-topic-tcpm-tx-poll-cond-8dd3ebbbeddc > > Best regards, > -- > Neil Armstrong <neil.armstrong@linaro.org> Cheers Marcel
diff --git a/drivers/usb/tcpm/tcpm.c b/drivers/usb/tcpm/tcpm.c index 0aee57cb2f4a44af7b483e0e609e39ba584a06c0..c8679615cd7701982b64b1905c5822ac98ed6f79 100644 --- a/drivers/usb/tcpm/tcpm.c +++ b/drivers/usb/tcpm/tcpm.c @@ -209,7 +209,7 @@ static int tcpm_pd_transmit(struct udevice *dev, */ timeout_us *= 5; ret = read_poll_timeout(tcpm_transmit_helper, tx_complete, - !tx_complete, false, timeout_us, dev); + tx_complete, false, timeout_us, dev); if (ret < 0) { dev_err(dev, "TCPM: PD transmit data failed: %d\n", ret); return ret;
The read_poll_timeout() exit "cond" parameter wait for a complete transmit state. The "!tx_complete" cond means that on the first tcpm_transmit_helper() call, if the tx_complete returns to false, the poll loop would exit. But since at probe the tx_status is left at 0, the tx is considered as successful. And then later the proper status is returned and ignored. But if on the first tcpm_transmit_helper() call the TX status is returned the read_poll_timeout() would return in timeout and the negociation would fail. Fixes: 1db4c0ac77e ("usb: tcpm: add core framework") Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org> --- drivers/usb/tcpm/tcpm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- base-commit: bab54f5942c428be698216224fd10b91d974d4da change-id: 20250526-topic-tcpm-tx-poll-cond-8dd3ebbbeddc Best regards,