Skip to contentSkip to conversation

platform/chrome: cros_ec_debugfs: Support pd_info v2 format

PATCH In review Stephen Boyd 3 participants 5 messages Sep 9, 2020 – Sep 16, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 platform/chrome: cros_ec_debugfs: Support pd_info v2 format

Let's try to read more information out of more modern cros_ec devices by using the v2 format first and then fall back to the v1 format. This gives us more information about things such as DP mode of the typec pins and the CC state, along with some more things.

Cc Gwendal Grignou <[email protected]>
Cc Prashant Malani <[email protected]>
Cc Guenter Roeck <[email protected]>

---

Should we move read_buf to the heap?

 drivers/platform/chrome/cros_ec_debugfs.c | 40 +++++++++++++++++------
 1 file changed, 30 insertions(+), 10 deletions(-)
drivers/platform/chrome/cros_ec_debugfs.c +30 -10
index 272c89837d74..4f8c902c0de6 100644
@@ -195,28 +195,31 @@ static ssize_t cros_ec_pdinfo_read(struct file *file,
 				   size_t count,
 				   loff_t *ppos)
 {
-	char read_buf[EC_USB_PD_MAX_PORTS * 40], *p = read_buf;
+	char read_buf[EC_USB_PD_MAX_PORTS * 64], *p = read_buf;
 	struct cros_ec_debugfs *debug_info = file->private_data;
 	struct cros_ec_device *ec_dev = debug_info->ec->ec_dev;
 	struct {
 		struct cros_ec_command msg;
 		union {
-			struct ec_response_usb_pd_control_v1 resp;
+			struct ec_response_usb_pd_control_v2 resp_v2;
+			struct ec_response_usb_pd_control_v1 resp_v1;
 			struct ec_params_usb_pd_control params;
 		};
 	} __packed ec_buf;
 	struct cros_ec_command *msg;
-	struct ec_response_usb_pd_control_v1 *resp;
+	struct ec_response_usb_pd_control_v1 *resp_v1;
+	struct ec_response_usb_pd_control_v2 *resp_v2;
 	struct ec_params_usb_pd_control *params;
 	int i;
 
 	msg = &ec_buf.msg;
 	params = (struct ec_params_usb_pd_control *)msg->data;
-	resp = (struct ec_response_usb_pd_control_v1 *)msg->data;
+	resp_v1 = (struct ec_response_usb_pd_control_v1 *)msg->data;
+	resp_v2 = (struct ec_response_usb_pd_control_v2 *)msg->data;
 
 	msg->command = EC_CMD_USB_PD_CONTROL;
-	msg->version = 1;
-	msg->insize = sizeof(*resp);
+	msg->version = 2;
+	msg->insize = sizeof(*resp_v2);
 	msg->outsize = sizeof(*params);
 
 	/*
@@ -229,13 +232,30 @@ static ssize_t cros_ec_pdinfo_read(struct file *file,
 		params->mux = 0;
 		params->swap = 0;
 
-		if (cros_ec_cmd_xfer_status(ec_dev, msg) < 0)
+		if (cros_ec_cmd_xfer_status(ec_dev, msg) < 0) {
+			if (i == 0 && msg->version == 2) {
+				/* Try again with version 1 */
+				msg->version = 1;
+				msg->insize = sizeof(*resp_v1);
+				i = 0;
+				continue;
+			}
+
 			break;
+		}
 
 		p += scnprintf(p, sizeof(read_buf) + read_buf - p,
-			       "p%d: %s en:%.2x role:%.2x pol:%.2x\n", i,
-			       resp->state, resp->enabled, resp->role,
-			       resp->polarity);
+			       "p%d: %s en:%.2x role:%.2x pol:%.2x", i,
+			       resp_v1->state, resp_v1->enabled, resp_v1->role,
+			       resp_v1->polarity);
+		if (msg->version == 2) {
+			p += scnprintf(p, sizeof(read_buf) + read_buf - p,
+				       " cc:%.2x dp:%.2x ctrl:%.2x cs:%.2x gen:%.2x",
+				       resp_v2->cc_state, resp_v2->dp_mode,
+				       resp_v2->control_flags, resp_v2->cable_speed,
+				       resp_v2->cable_gen);
+		}
+		p += scnprintf(p, sizeof(read_buf) + read_buf - p, "\n");
 	}
 
 	return simple_read_from_buffer(user_buf, count, ppos,
Signature · 1 line
Sent by a computer, using git, on the internet
#1 lore raw

Hi Stephen, Prashant,

Quoted 11 lines · Stephen Boyd

Let's try to read more information out of more modern cros_ec devices by using the v2 format first and then fall back to the v1 format. This gives us more information about things such as DP mode of the typec pins and the CC state, along with some more things.

Cc Gwendal Grignou <[email protected]>
Cc Prashant Malani <[email protected]>
Cc Guenter Roeck <[email protected]>

---

I saw some discussion going on in gerrit (a pity the discussion didn't happen in mainline) Did you end with a conclusion? Can I remove this patch from my backlog?

Thanks,
Enric

Quoted 83 lines

Should we move read_buf to the heap?

 drivers/platform/chrome/cros_ec_debugfs.c | 40 +++++++++++++++++------
 1 file changed, 30 insertions(+), 10 deletions(-)
drivers/platform/chrome/cros_ec_debugfs.c +30 -10
index 272c89837d74..4f8c902c0de6 100644
@@ -195,28 +195,31 @@ static ssize_t cros_ec_pdinfo_read(struct file *file,
 				   size_t count,
 				   loff_t *ppos)
 {
-	char read_buf[EC_USB_PD_MAX_PORTS * 40], *p = read_buf;
+	char read_buf[EC_USB_PD_MAX_PORTS * 64], *p = read_buf;
 	struct cros_ec_debugfs *debug_info = file->private_data;
 	struct cros_ec_device *ec_dev = debug_info->ec->ec_dev;
 	struct {
 		struct cros_ec_command msg;
 		union {
-			struct ec_response_usb_pd_control_v1 resp;
+			struct ec_response_usb_pd_control_v2 resp_v2;
+			struct ec_response_usb_pd_control_v1 resp_v1;
 			struct ec_params_usb_pd_control params;
 		};
 	} __packed ec_buf;
 	struct cros_ec_command *msg;
-	struct ec_response_usb_pd_control_v1 *resp;
+	struct ec_response_usb_pd_control_v1 *resp_v1;
+	struct ec_response_usb_pd_control_v2 *resp_v2;
 	struct ec_params_usb_pd_control *params;
 	int i;
 
 	msg = &ec_buf.msg;
 	params = (struct ec_params_usb_pd_control *)msg->data;
-	resp = (struct ec_response_usb_pd_control_v1 *)msg->data;
+	resp_v1 = (struct ec_response_usb_pd_control_v1 *)msg->data;
+	resp_v2 = (struct ec_response_usb_pd_control_v2 *)msg->data;
 
 	msg->command = EC_CMD_USB_PD_CONTROL;
-	msg->version = 1;
-	msg->insize = sizeof(*resp);
+	msg->version = 2;
+	msg->insize = sizeof(*resp_v2);
 	msg->outsize = sizeof(*params);
 
 	/*
@@ -229,13 +232,30 @@ static ssize_t cros_ec_pdinfo_read(struct file *file,
 		params->mux = 0;
 		params->swap = 0;
 
-		if (cros_ec_cmd_xfer_status(ec_dev, msg) < 0)
+		if (cros_ec_cmd_xfer_status(ec_dev, msg) < 0) {
+			if (i == 0 && msg->version == 2) {
+				/* Try again with version 1 */
+				msg->version = 1;
+				msg->insize = sizeof(*resp_v1);
+				i = 0;
+				continue;
+			}
+
 			break;
+		}
 
 		p += scnprintf(p, sizeof(read_buf) + read_buf - p,
-			       "p%d: %s en:%.2x role:%.2x pol:%.2x\n", i,
-			       resp->state, resp->enabled, resp->role,
-			       resp->polarity);
+			       "p%d: %s en:%.2x role:%.2x pol:%.2x", i,
+			       resp_v1->state, resp_v1->enabled, resp_v1->role,
+			       resp_v1->polarity);
+		if (msg->version == 2) {
+			p += scnprintf(p, sizeof(read_buf) + read_buf - p,
+				       " cc:%.2x dp:%.2x ctrl:%.2x cs:%.2x gen:%.2x",
+				       resp_v2->cc_state, resp_v2->dp_mode,
+				       resp_v2->control_flags, resp_v2->cable_speed,
+				       resp_v2->cable_gen);
+		}
+		p += scnprintf(p, sizeof(read_buf) + read_buf - p, "\n");
 	}
 
 	return simple_read_from_buffer(user_buf, count, ppos,
#2 lore raw

HI Enric,

Quoted 16 lines · Enric Balletbo i Serra

Hi Stephen, Prashant,

Quoted 11 lines · Stephen Boyd

Let's try to read more information out of more modern cros_ec devices by using the v2 format first and then fall back to the v1 format. This gives us more information about things such as DP mode of the typec pins and the CC state, along with some more things.

Cc Gwendal Grignou <[email protected]>
Cc Prashant Malani <[email protected]>
Cc Guenter Roeck <[email protected]>

---

Quoted 2 lines

I saw some discussion going on in gerrit (a pity the discussion didn't happen in mainline) Did you end with a conclusion? Can I remove this patch from my backlog?

My apologies for not posting the comment here.
To summarize: the userspace EC utility ectool [1] can offer the equivalent output, but with better formatting. So I believe the decision is to use that instead of this patch.
I also posed a counter-question: can we remove this debugfs pdinfo file entirely, since we can pull this information using ectool?

[1]: https://chromium.googlesource.com/chromiumos/platform/ec/+/refs/heads/master/util/ectool.c

Best regards,

#3 lore raw

Hi Prashant,

Quoted 23 lines · Prashant Malani

HI Enric,

Quoted 16 lines · Enric Balletbo i Serra

Hi Stephen, Prashant,

Quoted 11 lines · Stephen Boyd

Let's try to read more information out of more modern cros_ec devices by using the v2 format first and then fall back to the v1 format. This gives us more information about things such as DP mode of the typec pins and the CC state, along with some more things.

Cc Gwendal Grignou <[email protected]>
Cc Prashant Malani <[email protected]>
Cc Guenter Roeck <[email protected]>

---

Quoted 2 lines

I saw some discussion going on in gerrit (a pity the discussion didn't happen in mainline) Did you end with a conclusion? Can I remove this patch from my backlog?

Quoted 7 lines

My apologies for not posting the comment here.
To summarize: the userspace EC utility ectool [1] can offer the equivalent output, but with better formatting. So I believe the decision is to use that instead of this patch.
I also posed a counter-question: can we remove this debugfs pdinfo file entirely, since we can pull this information using ectool?

If I am not mistaken pdinfo is used by your userspace, so, assuming we don't break (or we have a plan to not break) things I'm fine with it. In general, delegating things to userspace and get rid of kernel code is good for everyone.

Thanks,
Enric

Quoted 4 lines

[1]: https://chromium.googlesource.com/chromiumos/platform/ec/+/refs/heads/master/util/ectool.c

Best regards,

#4 lore raw

Hi Enric,

Quoted 7 lines · Enric Balletbo i Serra

Hi Prashant,

If I am not mistaken pdinfo is used by your userspace, so, assuming we don't break (or we have a plan to not break) things I'm fine with it. In general, delegating things to userspace and get rid of kernel code is good for everyone.

Sounds good, thanks! I will look into this and if it isn't breaking anything I'll send a patch which does away with pdinfo.

Best regards,