[PATCH] tpm: improve output in sysfs files when the TPM fails

[PATCH] tpm: improve output in sysfs files when the TPM fails

[Kylene Hall]

When the TPM is in a disabled or deactivated state the sysfs pcrs and 
pubek files will appear empty.  To remove any confusion this might cause, 
the files will instead contain the error the TPM returned (also indicative 
of what state the TPM is in and what actions might be needed to change 
that state).

Signed-off-by: Kylene Hall <kjhall@us.ibm.com>
---
diff -uprN linux-2.6.12-rc3/drivers/char/tpm/tpm.c /home/kylie/kernel/linux-2.6.12-rc3-tpmdd/drivers/char/tpm/tpm.c
--- linux-2.6.12-rc3/drivers/char/tpm/tpm.c	2005-05-12 18:03:43.000000000 -0500
+++ /home/kylie/kernel/linux-2.6.12-rc3-tpmdd/drivers/char/tpm/tpm.c	2005-05-12 17:40:26.000000000 -0500
@@ -212,8 +183,11 @@ ssize_t tpm_show_pcrs(struct device *dev
 
 	memcpy(data, cap_pcr, sizeof(cap_pcr));
 	if ((len = tpm_transmit(chip, data, sizeof(data)))
-	    < CAP_PCR_RESULT_SIZE)
-		return len;
+	    < CAP_PCR_RESULT_SIZE) {
+		str += sprintf( str, "TPM ERROR: %d\n", 
+			be32_to_cpu(*((__be32 *) (data + 6))));
+		goto out;
+	}
 
 	num_pcrs = be32_to_cpu(*((__be32 *) (data + 14)));
 
@@ -222,13 +196,17 @@ ssize_t tpm_show_pcrs(struct device *dev
 		index = cpu_to_be32(i);
 		memcpy(data + 10, &index, 4);
 		if ((len = tpm_transmit(chip, data, sizeof(data)))
-		    < READ_PCR_RESULT_SIZE)
-			return len;
+		    < READ_PCR_RESULT_SIZE){
+			str += sprintf( str, "TPM ERROR: %d\n", 
+				be32_to_cpu(*((__be32 *) (data + 6))));
+			goto out;
+		}
 		str += sprintf(str, "PCR-%02d: ", i);
 		for (j = 0; j < TPM_DIGEST_SIZE; j++)
 			str += sprintf(str, "%02X ", *(data + 10 + j));
 		str += sprintf(str, "\n");
 	}
+out:
 	return str - buf;
 }
 
@@ -262,7 +240,8 @@ ssize_t tpm_show_pubek(struct device *de
 
 	if ((len = tpm_transmit(chip, data, READ_PUBEK_RESULT_SIZE)) <
 	    READ_PUBEK_RESULT_SIZE) {
-		rc = len;
+		str += sprintf( str, "TPM ERROR: %d\n", 
+			be32_to_cpu(*((__be32 *) (data + 6))));
 		goto out;
 	}
 
@@ -294,8 +273,8 @@ ssize_t tpm_show_pubek(struct device *de
 		if ((i + 1) % 16 == 0)
 			str += sprintf(str, "\n");
 	}
-	rc = str - buf;
 out:
+	rc = str - buf;
 	kfree(data);
 	return rc;
 }

Re: [PATCH] tpm: improve output in sysfs files when the TPM fails

[Greg KH]

On Thu, May 12, 2005 at 05:20:22PM -0500, Kylene Hall wrote:
> When the TPM is in a disabled or deactivated state the sysfs pcrs and 
> pubek files will appear empty.  To remove any confusion this might cause, 
> the files will instead contain the error the TPM returned (also indicative 
> of what state the TPM is in and what actions might be needed to change 
> that state).

No, sysfs files are not error logs.  Please use the standard system wide
error log for this (syslog).

Why not just change the mode of the sysfs file instead, or delete it
entirely in this case?

thanks,

greg k-h

Re: [PATCH] tpm: improve output in sysfs files when the TPM fails

[Kylene Hall]

On Thu, 12 May 2005, Greg KH wrote:
> On Thu, May 12, 2005 at 05:20:22PM -0500, Kylene Hall wrote:
> > When the TPM is in a disabled or deactivated state the sysfs pcrs and 
> > pubek files will appear empty.  To remove any confusion this might cause, 
> > the files will instead contain the error the TPM returned (also indicative 
> > of what state the TPM is in and what actions might be needed to change 
> > that state).
> 
> No, sysfs files are not error logs.  Please use the standard system wide
> error log for this (syslog).
> 
> Why not just change the mode of the sysfs file instead, or delete it
> entirely in this case?

Ok, instead of putting error information in the sysfs file this new patch 
will put an error entry in syslog.  The sysfs files can't easily be 
removed in these cases as the driver does not know this information and it 
can be changed by commands sent to the TPM from userspace.

Signed-off-by: Kylene Hall <kjhall@us.ibm.com>
---
diff -uprN linux-2.6.12-rc3/drivers/char/tpm/tpm.c /home/kylie/kernel/linux-2.6.12-rc3-tpmdd/drivers/char/tpm/tpm.c
--- linux-2.6.12-rc3/drivers/char/tpm/tpm.c	2005-05-12 18:03:43.000000000 -0500
+++ /home/kylie/kernel/linux-2.6.12-rc3-tpmdd/drivers/char/tpm/tpm.c	2005-05-13 13:43:52.000000000 -0500
@@ -212,8 +183,11 @@ ssize_t tpm_show_pcrs(struct device *dev
 
 	memcpy(data, cap_pcr, sizeof(cap_pcr));
 	if ((len = tpm_transmit(chip, data, sizeof(data)))
-	    < CAP_PCR_RESULT_SIZE)
-		return len;
+	    < CAP_PCR_RESULT_SIZE) {
+		dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred attempting to determine the number of PCRS\n", 
+			be32_to_cpu(*((__be32 *) (data + 6))));
+		return 0;
+	}
 
 	num_pcrs = be32_to_cpu(*((__be32 *) (data + 14)));
 
@@ -222,13 +196,17 @@ ssize_t tpm_show_pcrs(struct device *dev
 		index = cpu_to_be32(i);
 		memcpy(data + 10, &index, 4);
 		if ((len = tpm_transmit(chip, data, sizeof(data)))
-		    < READ_PCR_RESULT_SIZE)
-			return len;
+		    < READ_PCR_RESULT_SIZE){
+			dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred attempting to read PCR %d of %d\n", 
+				be32_to_cpu(*((__be32 *) (data + 6))), i, num_pcrs);
+			goto out;
+		}
 		str += sprintf(str, "PCR-%02d: ", i);
 		for (j = 0; j < TPM_DIGEST_SIZE; j++)
 			str += sprintf(str, "%02X ", *(data + 10 + j));
 		str += sprintf(str, "\n");
 	}
+out:
 	return str - buf;
 }
 
@@ -262,8 +240,9 @@ ssize_t tpm_show_pubek(struct device *de
 
 	if ((len = tpm_transmit(chip, data, READ_PUBEK_RESULT_SIZE)) <
 	    READ_PUBEK_RESULT_SIZE) {
-		rc = len;
-		goto out;
+		dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred attempting to read the PUBEK\n", 
+			    be32_to_cpu(*((__be32 *) (data + 6))));
+		return 0;
 	}
 
 	/* 
@@ -295,7 +274,6 @@ ssize_t tpm_show_pubek(struct device *de
 			str += sprintf(str, "\n");
 	}
 	rc = str - buf;
-out:
 	kfree(data);
 	return rc;
 }

Re: [PATCH] tpm: improve output in sysfs files when the TPM fails

[Andrew Morton]

Kylene Hall <kjhall@us.ibm.com> wrote:
>
>  On Thu, 12 May 2005, Greg KH wrote:
>  > On Thu, May 12, 2005 at 05:20:22PM -0500, Kylene Hall wrote:
>  > > When the TPM is in a disabled or deactivated state the sysfs pcrs and 
>  > > pubek files will appear empty.  To remove any confusion this might cause, 
>  > > the files will instead contain the error the TPM returned (also indicative 
>  > > of what state the TPM is in and what actions might be needed to change 
>  > > that state).
>  > 
>  > No, sysfs files are not error logs.  Please use the standard system wide
>  > error log for this (syslog).
>  > 
>  > Why not just change the mode of the sysfs file instead, or delete it
>  > entirely in this case?
> 
>  Ok, instead of putting error information in the sysfs file this new patch 
>  will put an error entry in syslog.  The sysfs files can't easily be 
>  removed in these cases as the driver does not know this information and it 
>  can be changed by commands sent to the TPM from userspace.

Will this change permit unprivileged users to create large amounts of
syslog output?  If so, this is considered poor form.

IOW: please confirm that the relevant sysfs files are root-read-only?

Re: [PATCH] tpm: improve output in sysfs files when the TPM fails

[Kylene Hall]

On Fri, 13 May 2005, Andrew Morton wrote:

> Kylene Hall <kjhall@us.ibm.com> wrote:
> >
> >  On Thu, 12 May 2005, Greg KH wrote:
> >  > On Thu, May 12, 2005 at 05:20:22PM -0500, Kylene Hall wrote:
> >  > > When the TPM is in a disabled or deactivated state the sysfs pcrs and 
> >  > > pubek files will appear empty.  To remove any confusion this might cause, 
> >  > > the files will instead contain the error the TPM returned (also indicative 
> >  > > of what state the TPM is in and what actions might be needed to change 
> >  > > that state).
> >  > 
> >  > No, sysfs files are not error logs.  Please use the standard system wide
> >  > error log for this (syslog).
> >  > 
> >  > Why not just change the mode of the sysfs file instead, or delete it
> >  > entirely in this case?
> > 
> >  Ok, instead of putting error information in the sysfs file this new patch 
> >  will put an error entry in syslog.  The sysfs files can't easily be 
> >  removed in these cases as the driver does not know this information and it 
> >  can be changed by commands sent to the TPM from userspace.
> 
> Will this change permit unprivileged users to create large amounts of
> syslog output?  If so, this is considered poor form.
> 
> IOW: please confirm that the relevant sysfs files are root-read-only?
> 
> 

Please back this patch out since it is undesirable to use the sysfs 
file or syslog for the error reporting for various reasons and is 
desirable to keep the files not root-read-only.  The status of the TPM can 
be gleaned from other sources.

Thanks,
Kylie  

Re: [PATCH] tpm: improve output in sysfs files when the TPM fails

[Kylene Jo Hall]

On Fri, 2005-05-13 at 22:30 -0700, Andrew Morton wrote:
> Will this change permit unprivileged users to create large amounts of
> syslog output?  If so, this is considered poor form.
> 
> IOW: please confirm that the relevant sysfs files are root-read-only?
> 
Since after reconsideration this is more debug output than an error (the
TPM is operating correctly given the current state) I have changed the
statements to dbg rather than err.  Is this sufficient?  Also this patch
corrects a memory leak if the error path is taken in the tpm_show_pubek
function.

Thanks,
Kylie Hall

Signed-off-by: Kylene Hall <kjhall@us.ibm.com>
---
--- linux-2.6.12-rc6-mm1/drivers/char/tpm/tpm.c.orig	2005-06-14 09:14:37.000000000 -0500
+++ linux-2.6.12-rc6-mm1/drivers/char/tpm/tpm.c	2005-06-14 09:24:23.000000000 -0500
@@ -147,7 +147,7 @@ ssize_t tpm_show_pcrs(struct device *dev
 	memcpy(data, cap_pcr, sizeof(cap_pcr));
 	if ((len = tpm_transmit(chip, data, sizeof(data)))
 	    < CAP_PCR_RESULT_SIZE) {
-		dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred "
+		dev_dbg(&chip->pci_dev->dev, "A TPM error (%d) occurred "
 				"attempting to determine the number of PCRS\n",
 			be32_to_cpu(*((__be32 *) (data + 6))));
 		return 0;
@@ -161,7 +161,7 @@ ssize_t tpm_show_pcrs(struct device *dev
 		memcpy(data + 10, &index, 4);
 		if ((len = tpm_transmit(chip, data, sizeof(data)))
 		    < READ_PCR_RESULT_SIZE){
-			dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred"
+			dev_dbg(&chip->pci_dev->dev, "A TPM error (%d) occurred"
 				" attempting to read PCR %d of %d\n",
 				be32_to_cpu(*((__be32 *) (data + 6))), i, num_pcrs);
 			goto out;
@@ -205,10 +205,11 @@ ssize_t tpm_show_pubek(struct device *de
 
 	if ((len = tpm_transmit(chip, data, READ_PUBEK_RESULT_SIZE)) <
 	    READ_PUBEK_RESULT_SIZE) {
-		dev_err(&chip->pci_dev->dev, "A TPM error (%d) occurred "
+		dev_dbg(&chip->pci_dev->dev, "A TPM error (%d) occurred "
 				"attempting to read the PUBEK\n",
 			    be32_to_cpu(*((__be32 *) (data + 6))));
-		return 0;
+		rc = 0;
+		goto out;
 	}
 
 	/* 
@@ -240,6 +241,7 @@ ssize_t tpm_show_pubek(struct device *de
 			str += sprintf(str, "\n");
 	}
 	rc = str - buf;
+out:
 	kfree(data);
 	return rc;
 }