* [RFC v4 0/2] Add support for time DT property in TCPM
@ 2024-09-25 3:11 Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 1/2] dt-bindings: connector: Add properties to define time values Amit Sunil Dhamne
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Amit Sunil Dhamne @ 2024-09-25 3:11 UTC (permalink / raw)
To: gregkh, robh, dmitry.baryshkov, heikki.krogerus
Cc: badhri, kyletso, rdbabiera, linux-kernel, linux-usb, devicetree,
Amit Sunil Dhamne
USB PD specification defines a bunch of timers that can have a range of
acceptable values instead of specific values. These values have to be
tuned based on the platform. However, TCPM currently sets them to a
default value without providing a mechanism to set platform specific
values.
This patchset adds new DT properties per timer to allow users to define
platform specific values.
Changes compared to v3:
- nit: removed an extra newline that got added in tcpm_register_port()
Changes compared to v2:
- Added min, max & default values to DT property in Documentation.
- Changed return type of tcpm_fw_get_timings to void instead of int.
Changes compared to v1:
- Defined new properties per timer that we are interested in rather
than defining a single pd-timers u32 array property.
- Better description of the timer properties.
- Since subject has changed, adding link for previous patchset for
posterity:
https://lore.kernel.org/all/20240911000715.554184-1-amitsd@google.com/
Amit Sunil Dhamne (2):
dt-bindings: connector: Add properties to define time values
usb: typec: tcpm: Add support for parsing time dt properties
.../bindings/connector/usb-connector.yaml | 35 ++++++++-
drivers/usb/typec/tcpm/tcpm.c | 73 +++++++++++++++----
2 files changed, 91 insertions(+), 17 deletions(-)
base-commit: 68d4209158f43a558c5553ea95ab0c8975eab18c
--
2.46.0.792.g87dc391469-goog
^ permalink raw reply [flat|nested] 5+ messages in thread
* [RFC v4 1/2] dt-bindings: connector: Add properties to define time values
2024-09-25 3:11 [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
@ 2024-09-25 3:11 ` Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 2/2] usb: typec: tcpm: Add support for parsing time dt properties Amit Sunil Dhamne
2024-10-21 23:01 ` [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2 siblings, 0 replies; 5+ messages in thread
From: Amit Sunil Dhamne @ 2024-09-25 3:11 UTC (permalink / raw)
To: gregkh, robh, dmitry.baryshkov, heikki.krogerus
Cc: badhri, kyletso, rdbabiera, linux-kernel, linux-usb, devicetree,
Amit Sunil Dhamne
This commit adds the following properties:
* sink-wait-cap-time-ms
* ps-source-off-time-ms
* cc-debounce-time-ms
This is to enable setting of platform/board specific timer values as
these timers have a range of acceptable values.
Signed-off-by: Amit Sunil Dhamne <amitsd@google.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
---
.../bindings/connector/usb-connector.yaml | 35 ++++++++++++++++++-
1 file changed, 34 insertions(+), 1 deletion(-)
diff --git a/Documentation/devicetree/bindings/connector/usb-connector.yaml b/Documentation/devicetree/bindings/connector/usb-connector.yaml
index fb216ce68bb3..21a0c58c65cd 100644
--- a/Documentation/devicetree/bindings/connector/usb-connector.yaml
+++ b/Documentation/devicetree/bindings/connector/usb-connector.yaml
@@ -253,6 +253,36 @@ properties:
additionalProperties: false
+ sink-wait-cap-time-ms:
+ description: Represents the max time in ms that USB Type-C port (in sink
+ role) should wait for the port partner (source role) to send source caps.
+ SinkWaitCap timer starts when port in sink role attaches to the source.
+ This timer will stop when sink receives PD source cap advertisement before
+ timeout in which case it'll move to capability negotiation stage. A
+ timeout leads to a hard reset message by the port.
+ minimum: 310
+ maximum: 620
+ default: 310
+
+ ps-source-off-time-ms:
+ description: Represents the max time in ms that a DRP in source role should
+ take to turn off power after the PsSourceOff timer starts. PsSourceOff
+ timer starts when a sink's PHY layer receives EOP of the GoodCRC message
+ (corresponding to an Accept message sent in response to a PR_Swap or a
+ FR_Swap request). This timer stops when last bit of GoodCRC EOP
+ corresponding to the received PS_RDY message is transmitted by the PHY
+ layer. A timeout shall lead to error recovery in the type-c port.
+ minimum: 750
+ maximum: 920
+ default: 920
+
+ cc-debounce-time-ms:
+ description: Represents the max time in ms that a port shall wait to
+ determine if it's attached to a partner.
+ minimum: 100
+ maximum: 200
+ default: 200
+
dependencies:
sink-vdos-v1: [ sink-vdos ]
sink-vdos: [ sink-vdos-v1 ]
@@ -380,7 +410,7 @@ examples:
};
# USB-C connector attached to a typec port controller(ptn5110), which has
- # power delivery support and enables drp.
+ # power delivery support, explicitly defines time properties and enables drp.
- |
#include <dt-bindings/usb/pd.h>
typec: ptn5110 {
@@ -393,6 +423,9 @@ examples:
sink-pdos = <PDO_FIXED(5000, 2000, PDO_FIXED_USB_COMM)
PDO_VAR(5000, 12000, 2000)>;
op-sink-microwatt = <10000000>;
+ sink-wait-cap-time-ms = <465>;
+ ps-source-off-time-ms = <835>;
+ cc-debounce-time-ms = <101>;
};
};
--
2.46.0.792.g87dc391469-goog
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [RFC v4 2/2] usb: typec: tcpm: Add support for parsing time dt properties
2024-09-25 3:11 [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 1/2] dt-bindings: connector: Add properties to define time values Amit Sunil Dhamne
@ 2024-09-25 3:11 ` Amit Sunil Dhamne
2024-10-21 23:01 ` [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2 siblings, 0 replies; 5+ messages in thread
From: Amit Sunil Dhamne @ 2024-09-25 3:11 UTC (permalink / raw)
To: gregkh, robh, dmitry.baryshkov, heikki.krogerus
Cc: badhri, kyletso, rdbabiera, linux-kernel, linux-usb, devicetree,
Amit Sunil Dhamne
Add support for DT time properties to allow users to define platform
specific timing deadlines of certain timers rather than using hardcoded
ones. For values that have not been explicitly defined in DT using this
property, default values will be set therefore, making this change
backward compatible.
Signed-off-by: Amit Sunil Dhamne <amitsd@google.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
---
drivers/usb/typec/tcpm/tcpm.c | 73 +++++++++++++++++++++++++++--------
1 file changed, 57 insertions(+), 16 deletions(-)
diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
index 4b02d6474259..5a661b925eba 100644
--- a/drivers/usb/typec/tcpm/tcpm.c
+++ b/drivers/usb/typec/tcpm/tcpm.c
@@ -310,6 +310,17 @@ struct pd_data {
unsigned int operating_snk_mw;
};
+/*
+ * @sink_wait_cap_time: Deadline (in ms) for tTypeCSinkWaitCap timer
+ * @ps_src_wait_off_time: Deadline (in ms) for tPSSourceOff timer
+ * @cc_debounce_time: Deadline (in ms) for tCCDebounce timer
+ */
+struct pd_timings {
+ u32 sink_wait_cap_time;
+ u32 ps_src_off_time;
+ u32 cc_debounce_time;
+};
+
struct tcpm_port {
struct device *dev;
@@ -552,6 +563,9 @@ struct tcpm_port {
*/
unsigned int message_id_prime;
unsigned int rx_msgid_prime;
+
+ /* Timer deadline values configured at runtime */
+ struct pd_timings timings;
#ifdef CONFIG_DEBUG_FS
struct dentry *dentry;
struct mutex logbuffer_lock; /* log buffer access lock */
@@ -4639,15 +4653,15 @@ static void run_state_machine(struct tcpm_port *port)
case SRC_ATTACH_WAIT:
if (tcpm_port_is_debug(port))
tcpm_set_state(port, DEBUG_ACC_ATTACHED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
else if (tcpm_port_is_audio(port))
tcpm_set_state(port, AUDIO_ACC_ATTACHED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
else if (tcpm_port_is_source(port) && port->vbus_vsafe0v)
tcpm_set_state(port,
tcpm_try_snk(port) ? SNK_TRY
: SRC_ATTACHED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
break;
case SNK_TRY:
@@ -4698,7 +4712,7 @@ static void run_state_machine(struct tcpm_port *port)
}
break;
case SRC_TRYWAIT_DEBOUNCE:
- tcpm_set_state(port, SRC_ATTACHED, PD_T_CC_DEBOUNCE);
+ tcpm_set_state(port, SRC_ATTACHED, port->timings.cc_debounce_time);
break;
case SRC_TRYWAIT_UNATTACHED:
tcpm_set_state(port, SNK_UNATTACHED, 0);
@@ -4901,7 +4915,7 @@ static void run_state_machine(struct tcpm_port *port)
(port->cc1 != TYPEC_CC_OPEN &&
port->cc2 == TYPEC_CC_OPEN))
tcpm_set_state(port, SNK_DEBOUNCED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
else if (tcpm_port_is_disconnected(port))
tcpm_set_state(port, SNK_UNATTACHED,
PD_T_PD_DEBOUNCE);
@@ -4941,7 +4955,7 @@ static void run_state_machine(struct tcpm_port *port)
break;
case SNK_TRYWAIT:
tcpm_set_cc(port, TYPEC_CC_RD);
- tcpm_set_state(port, SNK_TRYWAIT_VBUS, PD_T_CC_DEBOUNCE);
+ tcpm_set_state(port, SNK_TRYWAIT_VBUS, port->timings.cc_debounce_time);
break;
case SNK_TRYWAIT_VBUS:
/*
@@ -5014,7 +5028,7 @@ static void run_state_machine(struct tcpm_port *port)
break;
case SNK_DISCOVERY_DEBOUNCE:
tcpm_set_state(port, SNK_DISCOVERY_DEBOUNCE_DONE,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
break;
case SNK_DISCOVERY_DEBOUNCE_DONE:
if (!tcpm_port_is_disconnected(port) &&
@@ -5041,10 +5055,10 @@ static void run_state_machine(struct tcpm_port *port)
if (port->vbus_never_low) {
port->vbus_never_low = false;
tcpm_set_state(port, SNK_SOFT_RESET,
- PD_T_SINK_WAIT_CAP);
+ port->timings.sink_wait_cap_time);
} else {
tcpm_set_state(port, SNK_WAIT_CAPABILITIES_TIMEOUT,
- PD_T_SINK_WAIT_CAP);
+ port->timings.sink_wait_cap_time);
}
break;
case SNK_WAIT_CAPABILITIES_TIMEOUT:
@@ -5066,7 +5080,8 @@ static void run_state_machine(struct tcpm_port *port)
if (tcpm_pd_send_control(port, PD_CTRL_GET_SOURCE_CAP, TCPC_TX_SOP))
tcpm_set_state_cond(port, hard_reset_state(port), 0);
else
- tcpm_set_state(port, hard_reset_state(port), PD_T_SINK_WAIT_CAP);
+ tcpm_set_state(port, hard_reset_state(port),
+ port->timings.sink_wait_cap_time);
break;
case SNK_NEGOTIATE_CAPABILITIES:
port->pd_capable = true;
@@ -5203,7 +5218,7 @@ static void run_state_machine(struct tcpm_port *port)
tcpm_set_state(port, ACC_UNATTACHED, 0);
break;
case AUDIO_ACC_DEBOUNCE:
- tcpm_set_state(port, ACC_UNATTACHED, PD_T_CC_DEBOUNCE);
+ tcpm_set_state(port, ACC_UNATTACHED, port->timings.cc_debounce_time);
break;
/* Hard_Reset states */
@@ -5420,7 +5435,7 @@ static void run_state_machine(struct tcpm_port *port)
tcpm_set_state(port, ERROR_RECOVERY, 0);
break;
case FR_SWAP_SNK_SRC_TRANSITION_TO_OFF:
- tcpm_set_state(port, ERROR_RECOVERY, PD_T_PS_SOURCE_OFF);
+ tcpm_set_state(port, ERROR_RECOVERY, port->timings.ps_src_off_time);
break;
case FR_SWAP_SNK_SRC_NEW_SINK_READY:
if (port->vbus_source)
@@ -5475,7 +5490,7 @@ static void run_state_machine(struct tcpm_port *port)
tcpm_set_cc(port, TYPEC_CC_RD);
/* allow CC debounce */
tcpm_set_state(port, PR_SWAP_SRC_SNK_SOURCE_OFF_CC_DEBOUNCED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
break;
case PR_SWAP_SRC_SNK_SOURCE_OFF_CC_DEBOUNCED:
/*
@@ -5510,7 +5525,7 @@ static void run_state_machine(struct tcpm_port *port)
port->pps_data.active, 0);
tcpm_set_charge(port, false);
tcpm_set_state(port, hard_reset_state(port),
- PD_T_PS_SOURCE_OFF);
+ port->timings.ps_src_off_time);
break;
case PR_SWAP_SNK_SRC_SOURCE_ON:
tcpm_enable_auto_vbus_discharge(port, true);
@@ -5666,7 +5681,7 @@ static void run_state_machine(struct tcpm_port *port)
case PORT_RESET_WAIT_OFF:
tcpm_set_state(port,
tcpm_default_state(port),
- port->vbus_present ? PD_T_PS_SOURCE_OFF : 0);
+ port->vbus_present ? port->timings.ps_src_off_time : 0);
break;
/* AMS intermediate state */
@@ -6157,7 +6172,7 @@ static void _tcpm_pd_vbus_vsafe0v(struct tcpm_port *port)
case SRC_ATTACH_WAIT:
if (tcpm_port_is_source(port))
tcpm_set_state(port, tcpm_try_snk(port) ? SNK_TRY : SRC_ATTACHED,
- PD_T_CC_DEBOUNCE);
+ port->timings.cc_debounce_time);
break;
case SRC_STARTUP:
case SRC_SEND_CAPABILITIES:
@@ -7053,6 +7068,30 @@ static int tcpm_port_register_pd(struct tcpm_port *port)
return ret;
}
+static void tcpm_fw_get_timings(struct tcpm_port *port, struct fwnode_handle *fwnode)
+{
+ int ret;
+ u32 val;
+
+ ret = fwnode_property_read_u32(fwnode, "sink-wait-cap-time-ms", &val);
+ if (!ret)
+ port->timings.sink_wait_cap_time = val;
+ else
+ port->timings.sink_wait_cap_time = PD_T_SINK_WAIT_CAP;
+
+ ret = fwnode_property_read_u32(fwnode, "ps-source-off-time-ms", &val);
+ if (!ret)
+ port->timings.ps_src_off_time = val;
+ else
+ port->timings.ps_src_off_time = PD_T_PS_SOURCE_OFF;
+
+ ret = fwnode_property_read_u32(fwnode, "cc-debounce-time-ms", &val);
+ if (!ret)
+ port->timings.cc_debounce_time = val;
+ else
+ port->timings.cc_debounce_time = PD_T_CC_DEBOUNCE;
+}
+
static int tcpm_fw_get_caps(struct tcpm_port *port, struct fwnode_handle *fwnode)
{
struct fwnode_handle *capabilities, *child, *caps = NULL;
@@ -7615,6 +7654,8 @@ struct tcpm_port *tcpm_register_port(struct device *dev, struct tcpc_dev *tcpc)
if (err < 0)
goto out_destroy_wq;
+ tcpm_fw_get_timings(port, tcpc->fwnode);
+
port->try_role = port->typec_caps.prefer_role;
port->typec_caps.revision = 0x0120; /* Type-C spec release 1.2 */
--
2.46.0.792.g87dc391469-goog
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [RFC v4 0/2] Add support for time DT property in TCPM
2024-09-25 3:11 [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 1/2] dt-bindings: connector: Add properties to define time values Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 2/2] usb: typec: tcpm: Add support for parsing time dt properties Amit Sunil Dhamne
@ 2024-10-21 23:01 ` Amit Sunil Dhamne
2024-10-22 5:14 ` Greg KH
2 siblings, 1 reply; 5+ messages in thread
From: Amit Sunil Dhamne @ 2024-10-21 23:01 UTC (permalink / raw)
To: gregkh, robh, heikki.krogerus, krzk+dt, conor+dt
Cc: badhri, kyletso, rdbabiera, linux-kernel, linux-usb, devicetree,
dmitry.baryshkov
Hi,
I had a process related question. Once an RFC patchset gets a
Reviewed-by tag, do I need to send a formal [PATCH] or is an RFC patch
sufficient for being accepted?
I tried to look for it in the Kernel documentation but couldn't find.
Please can you help me on the next steps.
Thanks,
Amit
On 9/24/24 8:11 PM, Amit Sunil Dhamne wrote:
> USB PD specification defines a bunch of timers that can have a range of
> acceptable values instead of specific values. These values have to be
> tuned based on the platform. However, TCPM currently sets them to a
> default value without providing a mechanism to set platform specific
> values.
>
> This patchset adds new DT properties per timer to allow users to define
> platform specific values.
>
> Changes compared to v3:
> - nit: removed an extra newline that got added in tcpm_register_port()
>
> Changes compared to v2:
> - Added min, max & default values to DT property in Documentation.
> - Changed return type of tcpm_fw_get_timings to void instead of int.
>
> Changes compared to v1:
> - Defined new properties per timer that we are interested in rather
> than defining a single pd-timers u32 array property.
> - Better description of the timer properties.
> - Since subject has changed, adding link for previous patchset for
> posterity:
> https://lore.kernel.org/all/20240911000715.554184-1-amitsd@google.com/
>
> Amit Sunil Dhamne (2):
> dt-bindings: connector: Add properties to define time values
> usb: typec: tcpm: Add support for parsing time dt properties
>
> .../bindings/connector/usb-connector.yaml | 35 ++++++++-
> drivers/usb/typec/tcpm/tcpm.c | 73 +++++++++++++++----
> 2 files changed, 91 insertions(+), 17 deletions(-)
>
>
> base-commit: 68d4209158f43a558c5553ea95ab0c8975eab18c
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [RFC v4 0/2] Add support for time DT property in TCPM
2024-10-21 23:01 ` [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
@ 2024-10-22 5:14 ` Greg KH
0 siblings, 0 replies; 5+ messages in thread
From: Greg KH @ 2024-10-22 5:14 UTC (permalink / raw)
To: Amit Sunil Dhamne
Cc: robh, heikki.krogerus, krzk+dt, conor+dt, badhri, kyletso,
rdbabiera, linux-kernel, linux-usb, devicetree, dmitry.baryshkov
On Mon, Oct 21, 2024 at 04:01:10PM -0700, Amit Sunil Dhamne wrote:
> Hi,
>
> I had a process related question. Once an RFC patchset gets a Reviewed-by
> tag, do I need to send a formal [PATCH] or is an RFC patch sufficient for
> being accepted?
You need to send a real series (and please do not top-post), for it to
be considered for being accepted.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2024-10-22 5:14 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-09-25 3:11 [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 1/2] dt-bindings: connector: Add properties to define time values Amit Sunil Dhamne
2024-09-25 3:11 ` [RFC v4 2/2] usb: typec: tcpm: Add support for parsing time dt properties Amit Sunil Dhamne
2024-10-21 23:01 ` [RFC v4 0/2] Add support for time DT property in TCPM Amit Sunil Dhamne
2024-10-22 5:14 ` Greg KH
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).