Skip to contentSkip to conversation

ipmi:ssif: Exit early when there is a SMBus error

PATCH In review Ivan T. Ivanov 2 participants 13 messages Aug 16, 2024 – Sep 10, 2024 IPMI

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ipmi:ssif: Exit early when there is a SMBus error

It is pointless to continue module probing when communication with device is failing. This just fill logs with misleading messages like this:

[Fri Jul 26 18:32:34 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Error fetching SSIF: -121 180453376 62, your system probably doesn't support this command so using defaults
[Fri Jul 26 18:32:54 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to clear message flags: -121 180453376 62
[Fri Jul 26 18:33:14 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Error getting global enables: -121 180453376 62
[Fri Jul 26 18:33:49 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:33:50 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:07 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:07 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:25 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:25 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:43 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:43 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:35:01 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:35:01 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: Unable to get the device id: -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to register device: error -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to start IPMI SSIF: -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif: probe of i2c-IPI0001:00 failed with error -5

Also in some of these prints uninitialized variables are used. So just exit early when communication with device is flawed.

---
 drivers/char/ipmi/ipmi_ssif.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/char/ipmi/ipmi_ssif.c +20 -0
index 96ad571d041a..37516733e5c8 100644
@@ -1315,6 +1315,16 @@ static int read_response(struct i2c_client *client, unsigned char *resp)
 	return ret;
 }
 
+/* Filter SMBus communication errors from incorrect response errors */
+static bool is_smbus_error(struct device *dev, int err)
+{
+	if (!err || err == -EINVAL || err == -E2BIG)
+		return false;
+
+	dev_err(dev, "SMbus error: %d\n", err);
+	return true;
+}
+
 static int do_cmd(struct i2c_client *client, int len, unsigned char *msg,
 		  int *resp_len, unsigned char *resp)
 {
@@ -1709,6 +1719,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_GET_SYSTEM_INTERFACE_CAPABILITIES_CMD;
 	msg[2] = 0; /* SSIF */
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (!rv && (len >= 3) && (resp[2] == 0)) {
 		if (len < 7) {
 			if (ssif_dbg_probe)
@@ -1767,6 +1779,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_CLEAR_MSG_FLAGS_CMD;
 	msg[2] = WDT_PRE_TIMEOUT_INT;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 3) || (resp[2] != 0))
 		dev_warn(&ssif_info->client->dev,
 			 "Unable to clear message flags: %d %d %2.2x\n",
@@ -1776,6 +1790,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
 	msg[1] = IPMI_GET_BMC_GLOBAL_ENABLES_CMD;
 	rv = do_cmd(client, 2, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 4) || (resp[2] != 0)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error getting global enables: %d %d %2.2x\n",
@@ -1796,6 +1812,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
 	msg[2] = ssif_info->global_enables | IPMI_BMC_EVT_MSG_BUFF;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 2)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error setting global enables: %d %d %2.2x\n",
@@ -1818,6 +1836,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
 	msg[2] = ssif_info->global_enables | IPMI_BMC_RCV_MSG_INTR;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 2)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error setting global enables: %d %d %2.2x\n",
Signature · 1 line
2.43.0
maintainer
#1 lore raw
Quoted 3 lines · Ivan T. Ivanov

It is pointless to continue module probing when communication with device is failing. This just fill logs with misleading messages like this:

So the BMC (or whatever is there) responds to a GET_DEVICE_ID command, but then doesn't properly handle any other valid and mandatory commands? And this device was added via ACPI or SMBIOS or device tree, almost certainly.

My comments are:

1) This fix is wrong, because it may mask important things that need to be reported.

2) Fix the source of the problem. You can't expect software to compensate for all bad hardware and firmware. I'd guess the firmware tables are pointing to something that's not a BMC.

3) It appears the response to the GET_DEVICE_ID command, though a response is returned, is not valid. The right way to handle this would be to do more validation in the ssif_detect() function. It doesn't do any validation of the response data, and that's really what needs to be done.

-corey

Quoted 97 lines
[Fri Jul 26 18:32:34 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Error fetching SSIF: -121 180453376 62, your system probably doesn't support this command so using defaults
[Fri Jul 26 18:32:54 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to clear message flags: -121 180453376 62
[Fri Jul 26 18:33:14 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Error getting global enables: -121 180453376 62
[Fri Jul 26 18:33:49 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:33:50 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:07 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:07 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:25 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:25 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:34:43 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:34:43 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:35:01 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:35:01 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: BMC returned 0xff, retry get bmc device id
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: device id demangle failed: -22
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: IPMI message handler: Unable to get the device id: -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to register device: error -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif i2c-IPI0001:00: ipmi_ssif: Unable to start IPMI SSIF: -5
[Fri Jul 26 18:35:19 2024] ipmi_ssif: probe of i2c-IPI0001:00 failed with error -5

Also in some of these prints uninitialized variables are used. So just exit early when communication with device is flawed.

---
 drivers/char/ipmi/ipmi_ssif.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/char/ipmi/ipmi_ssif.c +20 -0
index 96ad571d041a..37516733e5c8 100644
@@ -1315,6 +1315,16 @@ static int read_response(struct i2c_client *client, unsigned char *resp)
 	return ret;
 }
 
+/* Filter SMBus communication errors from incorrect response errors */
+static bool is_smbus_error(struct device *dev, int err)
+{
+	if (!err || err == -EINVAL || err == -E2BIG)
+		return false;
+
+	dev_err(dev, "SMbus error: %d\n", err);
+	return true;
+}
+
 static int do_cmd(struct i2c_client *client, int len, unsigned char *msg,
 		  int *resp_len, unsigned char *resp)
 {
@@ -1709,6 +1719,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_GET_SYSTEM_INTERFACE_CAPABILITIES_CMD;
 	msg[2] = 0; /* SSIF */
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (!rv && (len >= 3) && (resp[2] == 0)) {
 		if (len < 7) {
 			if (ssif_dbg_probe)
@@ -1767,6 +1779,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_CLEAR_MSG_FLAGS_CMD;
 	msg[2] = WDT_PRE_TIMEOUT_INT;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 3) || (resp[2] != 0))
 		dev_warn(&ssif_info->client->dev,
 			 "Unable to clear message flags: %d %d %2.2x\n",
@@ -1776,6 +1790,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
 	msg[1] = IPMI_GET_BMC_GLOBAL_ENABLES_CMD;
 	rv = do_cmd(client, 2, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 4) || (resp[2] != 0)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error getting global enables: %d %d %2.2x\n",
@@ -1796,6 +1812,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
 	msg[2] = ssif_info->global_enables | IPMI_BMC_EVT_MSG_BUFF;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 2)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error setting global enables: %d %d %2.2x\n",
@@ -1818,6 +1836,8 @@ static int ssif_probe(struct i2c_client *client)
 	msg[1] = IPMI_SET_BMC_GLOBAL_ENABLES_CMD;
 	msg[2] = ssif_info->global_enables | IPMI_BMC_RCV_MSG_INTR;
 	rv = do_cmd(client, 3, msg, &len, resp);
+	if (is_smbus_error(&client->dev, rv))
+		goto out;
 	if (rv || (len < 2)) {
 		dev_warn(&ssif_info->client->dev,
 			 "Error setting global enables: %d %d %2.2x\n",
Signature · 1 line
2.43.0
#2 lore raw

Hi,

Quoted 6 lines · Corey Minyard
Quoted 3 lines · Ivan T. Ivanov

It is pointless to continue module probing when communication with device is failing. This just fill logs with misleading messages like this:

So the BMC (or whatever is there) responds to a GET_DEVICE_ID command,

Well, not really. In cases where ssif_detect() returns -ENODEV, i2c core i2c_detect_address() threat it as "We catch it here as this isn't an error”. See i2c_detect_address().

Quoted 4 lines

but then doesn't properly handle any other valid and mandatory commands? And this device was added via ACPI or SMBIOS or device tree, almost certainly.

Quoted 8 lines

My comments are:

1) This fix is wrong, because it may mask important things that need to be reported.

2) Fix the source of the problem. You can't expect software to compensate for all bad hardware and firmware. I'd guess the firmware tables are pointing to something that's not a BMC.

I am not saying that hardware or firmware do not have its issues in this case. But just as it is written now ssif_probe() is fragile. It continue asking device for features ignoring that there is no valid SMBus communication.

Quoted 7 lines

3) It appears the response to the GET_DEVICE_ID command, though a response is returned, is not valid. The right way to handle this would be to do more validation in the ssif_detect() function. It doesn't do any validation of the response data, and that's really what needs to be done.

do_cmd() in ssif_detect() already do validation. Perhaps, ssif_probe() should just not return ENODEV in case of error.

Thank you for your comments!

Regards,
Ivan

#3 lore raw
Quoted 9 lines · Ivan T. Ivanov
Quoted 8 lines

3) It appears the response to the GET_DEVICE_ID command, though a response is returned, is not valid. The right way to handle this would be to do more validation in the ssif_detect() function. It doesn't do any validation of the response data, and that's really what needs to be done.

Quoted 3 lines

do_cmd() in ssif_detect() already do validation. Perhaps, ssif_probe() should just not return ENODEV in case of error.

Oh, I wanted to say ssif_detect, not ssif_probe.

Regards,
Ivan

maintainer
#4 lore raw
Quoted 11 lines · Ivan T. Ivanov
Quoted 9 lines · Ivan T. Ivanov
Quoted 8 lines

3) It appears the response to the GET_DEVICE_ID command, though a response is returned, is not valid. The right way to handle this would be to do more validation in the ssif_detect() function. It doesn't do any validation of the response data, and that's really what needs to be done.

Quoted 5 lines
Quoted 3 lines

do_cmd() in ssif_detect() already do validation. Perhaps, ssif_probe() should just not return ENODEV in case of error.

Oh, I wanted to say ssif_detect, not ssif_probe.

Yeah, that's probably the right solution. I'll look at this a bit. But I see the problem. Do you want to do a patch, or do you want me to?

-corey

#5 lore raw

Hi,

Quoted 13 lines · Corey Minyard
Quoted 13 lines
Quoted 12 lines
Quoted 8 lines

3) It appears the response to the GET_DEVICE_ID command, though a response is returned, is not valid. The right way to handle this would be to do more validation in the ssif_detect() function. It doesn't do any validation of the response data, and that's really what needs to be done.

do_cmd() in ssif_detect() already do validation. Perhaps, ssif_probe() should just not return ENODEV in case of error.

Quoted 5 lines
Quoted 1 line

Oh, I wanted to say ssif_detect, not ssif_probe.

Yeah, that's probably the right solution. I'll look at this a bit. But I see the problem. Do you want to do a patch, or do you want me to?

If you want to do it, please go ahead. I don't mind.

Thanks,
Ivan

maintainer
#6 lore raw

PATCH 1/1 ipmi:ssif: Improve detecting during probing

In reply to Ivan T. Ivanov

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Ivan, is it possible for you to test this patch on the broken system? It should work based on what you reported, but it's nice to be sure.

Also, I discovered that the detect function is kind of bogus, it only works on an address list that isn't present (any more). However, I re-used it for my purposes in the probe function.

Thanks.

drivers/char/ipmi/ipmi_ssif.c +23 -1
index e8e7b832c060..4c403e7a9fc8 100644
@@ -1368,8 +1368,20 @@ static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info)
 	rv = do_cmd(client, 2, msg, &len, resp);
 	if (rv)
 		rv = -ENODEV;
-	else
+	else {
+	    if (len < 3) {
+		rv = -ENODEV;
+	    } else {
+		struct ipmi_device_id id;
+
+		rv = ipmi_demangle_device_id(resp[0] >> 2, resp[1],
+					     resp + 2, len - 2, &id);
+		if (rv)
+		    rv = -ENODEV; /* Error means a BMC probably isn't there. */
+	    }
+	    if (!rv && info)
 		strscpy(info->type, DEVICE_NAME, I2C_NAME_SIZE);
+	}
 	kfree(resp);
 	return rv;
 }
@@ -1704,6 +1716,16 @@ static int ssif_probe(struct i2c_client *client)
 		ipmi_addr_src_to_str(ssif_info->addr_source),
 		client->addr, client->adapter->name, slave_addr);
 
+	/*
+	 * Send a get device id command and validate its response to
+	 * make sure a valid BMC is there.
+	 */
+	rv = ssif_detect(client, NULL);
+	if (rv) {
+		dev_err(&client->dev, "Not present\n");
+		goto out;
+	}
+
 	/* Now check for system interface capabilities */
 	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
 	msg[1] = IPMI_GET_SYSTEM_INTERFACE_CAPABILITIES_CMD;
Signature · 1 line
2.34.1
#7 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

Hi Corey,

Quoted 11 lines · Corey Minyard

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

Quoted 7 lines
---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Ivan, is it possible for you to test this patch on the broken system?

This exact system is not available to me at the moment. I have few other machines on which I could test this.

Quoted 13 lines

It should work based on what you reported, but it's nice to be sure.

Also, I discovered that the detect function is kind of bogus, it only works on an address list that isn't present (any more). However, I re-used it for my purposes in the probe function.

Thanks.

drivers/char/ipmi/ipmi_ssif.c +0 -0
index e8e7b832c060..4c403e7a9fc8 100644

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1370

Quoted 3 lines
drivers/char/ipmi/ipmi_ssif.c +0 -0
@@ -1368,8 +1368,20 @@ static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info)
 	rv = do_cmd(client, 2, msg, &len, resp);
 	if (rv)
 		rv = -ENODEV;

What is my worry is that in case of SMBus errors, device is there but for some reason it got stuck/crashed or whatever, so will get out of detect function from here and with ENODEV return code probe function will be called for no reason.

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1373

Quoted 4 lines
drivers/char/ipmi/ipmi_ssif.c +3 -1
-	else
+	else {
+	    if (len < 3) {
+		rv = -ENODEV;

No point to call probe(), right?

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1380

Quoted 7 lines
drivers/char/ipmi/ipmi_ssif.c +7 -0
+	    } else {
+		struct ipmi_device_id id;
+
+		rv = ipmi_demangle_device_id(resp[0] >> 2, resp[1],
+					     resp + 2, len - 2, &id);
+		if (rv)
+		    rv = -ENODEV; /* Error means a BMC probably isn't there. */

Same.

Quoted 11 lines
drivers/char/ipmi/ipmi_ssif.c +3 -0
+	    }
+	    if (!rv && info)
 		strscpy(info->type, DEVICE_NAME, I2C_NAME_SIZE);
+	}
 	kfree(resp);
 	return rv;
 }
@@ -1704,6 +1716,16 @@ static int ssif_probe(struct i2c_client *client)
 		ipmi_addr_src_to_str(ssif_info->addr_source),
 		client->addr, client->adapter->name, slave_addr);
 

drivers/char/ipmi/ipmi_ssif.c static int ssif_probe(struct i2c_client *client) · line 1728

Quoted 10 lines
drivers/char/ipmi/ipmi_ssif.c +10 -0
+	/*
+	 * Send a get device id command and validate its response to
+	 * make sure a valid BMC is there.
+	 */
+	rv = ssif_detect(client, NULL);
+	if (rv) {
+		dev_err(&client->dev, "Not present\n");
+		goto out;
+	}
+

The point is that even after this point IPMI device can start failing to properly communicate with the OS, real SMBus errors, like EREMOTEIO in my case, but unfortunately code bellow do not handle this very well, I think.

Quoted 6 lines
drivers/char/ipmi/ipmi_ssif.c +0 -0
 	/* Now check for system interface capabilities */
 	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
 	msg[1] = IPMI_GET_SYSTEM_INTERFACE_CAPABILITIES_CMD;
Signature · 1 line
2.34.1

Regards,
Ivan

maintainer
#8 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

Quoted 33 lines · Ivan T. Ivanov

Hi Corey,

Quoted 11 lines · Corey Minyard

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

Quoted 7 lines
---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Ivan, is it possible for you to test this patch on the broken system?

This exact system is not available to me at the moment. I have few other machines on which I could test this.

Quoted 8 lines

It should work based on what you reported, but it's nice to be sure.

Also, I discovered that the detect function is kind of bogus, it only works on an address list that isn't present (any more). However, I re-used it for my purposes in the probe function.

Thanks.

Quoted 13 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1370

Quoted 8 lines
drivers/char/ipmi/ipmi_ssif.c +0 -0
index e8e7b832c060..4c403e7a9fc8 100644
@@ -1368,8 +1368,20 @@ static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info)
 	rv = do_cmd(client, 2, msg, &len, resp);
 	if (rv)
 		rv = -ENODEV;

What is my worry is that in case of SMBus errors, device is there but for some reason it got stuck/crashed or whatever, so will get out of detect function from here and with ENODEV return code probe function will be called for no reason.

That's not how the i2c code works. See my next comment.

Quoted 7 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1373

Quoted 5 lines
drivers/char/ipmi/ipmi_ssif.c +3 -1
-	else
+	else {
+	    if (len < 3) {
+		rv = -ENODEV;

No point to call probe(), right?

Originally (before I add the call from ssif_probe()), this is not involved in the probe() call. Instead, the detect function is involved in calling a table of addresses in driver->address_list. So in this case this function is never called at all from the i2c code, since there is no address list.

Quoted 22 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1380

Quoted 8 lines
drivers/char/ipmi/ipmi_ssif.c +7 -0
+	    } else {
+		struct ipmi_device_id id;
+
+		rv = ipmi_demangle_device_id(resp[0] >> 2, resp[1],
+					     resp + 2, len - 2, &id);
+		if (rv)
+		    rv = -ENODEV; /* Error means a BMC probably isn't there. */

Same.

Quoted 11 lines
drivers/char/ipmi/ipmi_ssif.c +3 -0
+	    }
+	    if (!rv && info)
 		strscpy(info->type, DEVICE_NAME, I2C_NAME_SIZE);
+	}
 	kfree(resp);
 	return rv;
 }
@@ -1704,6 +1716,16 @@ static int ssif_probe(struct i2c_client *client)
 		ipmi_addr_src_to_str(ssif_info->addr_source),
 		client->addr, client->adapter->name, slave_addr);
 
Quoted 15 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_probe(struct i2c_client *client) · line 1728

Quoted 10 lines
drivers/char/ipmi/ipmi_ssif.c +10 -0
+	/*
+	 * Send a get device id command and validate its response to
+	 * make sure a valid BMC is there.
+	 */
+	rv = ssif_detect(client, NULL);
+	if (rv) {
+		dev_err(&client->dev, "Not present\n");
+		goto out;
+	}
+

The point is that even after this point IPMI device can start failing to properly communicate with the OS, real SMBus errors, like EREMOTEIO in my case, but unfortunately code bellow do not handle this very well, I think.

It is possible that the BMC gets rebooted or something between the call to ssif_detect() and the code below, but the probability is really low. If it answers a detect, the rest of the things should work.

-corey

Quoted 12 lines
Quoted 8 lines
drivers/char/ipmi/ipmi_ssif.c +0 -0
 	/* Now check for system interface capabilities */
 	msg[0] = IPMI_NETFN_APP_REQUEST << 2;
 	msg[1] = IPMI_GET_SYSTEM_INTERFACE_CAPABILITIES_CMD;
Signature · 1 line
2.34.1

Regards,
Ivan

#9 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

Hi,

Quoted 21 lines · Corey Minyard
Quoted 10 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1370

Quoted 5 lines

+++ b/drivers/char/ipmi/ipmi_ssif.c

drivers/char/ipmi/ipmi_ssif.c +0 -0
@@ -1368,8 +1368,20 @@ static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info)
 	rv = do_cmd(client, 2, msg, &len, resp);
 	if (rv)
 		rv = -ENODEV;

What is my worry is that in case of SMBus errors, device is there but for some reason it got stuck/crashed or whatever, so will get out of detect function from here and with ENODEV return code probe function will be called for no reason.

That's not how the i2c code works. See my next comment.

Quoted 7 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_detect(struct i2c_client *client, struct i2c_board_info *info) · line 1373

Quoted 5 lines
drivers/char/ipmi/ipmi_ssif.c +3 -1
-	else
+	else {
+	    if (len < 3) {
+		rv = -ENODEV;

No point to call probe(), right?

Quoted 5 lines

Originally (before I add the call from ssif_probe()), this is not involved in the probe() call. Instead, the detect function is involved in calling a table of addresses in driver->address_list. So in this case this function is never called at all from the i2c code, since there is no address list.

I see, thank you for explanation.

Quoted 17 lines
Quoted 16 lines

drivers/char/ipmi/ipmi_ssif.c static int ssif_probe(struct i2c_client *client) · line 1728

Quoted 11 lines
drivers/char/ipmi/ipmi_ssif.c +10 -0
+	/*
+	 * Send a get device id command and validate its response to
+	 * make sure a valid BMC is there.
+	 */
+	rv = ssif_detect(client, NULL);
+	if (rv) {
+		dev_err(&client->dev, "Not present\n");
+		goto out;
+	}
+

The point is that even after this point IPMI device can start failing to properly communicate with the OS, real SMBus errors, like EREMOTEIO in my case, but unfortunately code bellow do not handle this very well, I think.

Quoted 4 lines

It is possible that the BMC gets rebooted or something between the call to ssif_detect() and the code below, but the probability is really low. If it answers a detect, the rest of the things should work.

I have my daubs, but patch proposed will fix the issue that I see.

Thank you,
Ivan

#10 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

In reply to Corey Minyard

Hi Corey,

Quoted 11 lines · Corey Minyard

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

Quoted 5 lines
---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Do you plan to merge this fix?

Regards,
Ivan

maintainer
#11 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

Quoted 15 lines · Ivan T. Ivanov

Hi Corey,

Quoted 11 lines · Corey Minyard

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

Quoted 7 lines
Quoted 5 lines
---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Do you plan to merge this fix?

Yes, it's queued for the next release window. I should have responded with that.

-corey

Quoted 4 lines

Regards,
Ivan

#12 lore raw

Re: [PATCH] ipmi:ssif: Improve detecting during probing

Quoted 17 lines · Corey Minyard
Quoted 17 lines
Quoted 16 lines

If an IPMI SSIF device is probed and there is something there, but probably not an actual BMC, the code would just issue a lot of errors before it failed. We kind of need these errors to help with certain issues, and some of the failure reports are non-fatal.

However, a get device id command should alway work. If that fails, nothing else is going to work and it's a pretty good indication that there's no valid BMC there. So issue and check that command and bail if it fails.

---
 drivers/char/ipmi/ipmi_ssif.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)
Quoted 5 lines
Quoted 1 line

Do you plan to merge this fix?

Yes, it's queued for the next release window. I should have responded with that.

Thank you! No worries.

Regards,
Ivan