All of lore.kernel.org
 help / color / mirror / Atom feed
From: Mimi Zohar <zohar@linux.ibm.com>
To: Stephen Smalley <stephen.smalley.work@gmail.com>
Cc: linux-integrity@vger.kernel.org, bill.c.roberts@gmail.com
Subject: Re: [PATCH v3 ima-evm-utils] extend ima_measurement --pcrs option to support per-bank pcr files
Date: Fri, 24 Jul 2020 15:12:47 -0400	[thread overview]
Message-ID: <1595617967.5017.31.camel@linux.ibm.com> (raw)
In-Reply-To: <20200724145242.31178-1-stephen.smalley.work@gmail.com>

On Fri, 2020-07-24 at 10:52 -0400, Stephen Smalley wrote:
> Extend the ima_measurement --pcrs option to support per-bank pcr files.
> The extended syntax is "--pcrs algorithm,pathname".  If no algorithm
> is specified, it defaults to sha1 as before.  Multiple --pcrs options
> are now supported, one per bank of PCRs. The file format remains
> unchanged.  If --pcrs is specified, only try to read PCRs from the
> specified file(s); do not fall back to trying to read from sysfs
> or the TPM itself in this case since the user requested use of
> the files.
> 
> Create per-bank pcr files, depends on "tpm: add sysfs exports for all
> banks of PCR registers" kernel patch:
> $ cat tpm2pcrread.sh
> #!/bin/sh
> for alg in sha1 sha256
> do
>   rm -f pcr-$alg
>   pcr=0;
>   while [ $pcr -lt 24 ];
>   do
>     printf "PCR-%02d: " $pcr >> pcr-$alg;
>     cat /sys/class/tpm/tpm0/pcr-$alg/$pcr >> pcr-$alg;
>     pcr=$[$pcr+1];
>   done
> done
> $ sh ./tpm2pcrread.sh
> 
> Pass only the sha1 PCRs to evmctl defaulting to sha1:
> $ sudo evmctl ima_measurement --pcrs pcr-sha1 /sys/kernel/security/integrity/ima/binary_runtime_measurements
> 
> Pass only the sha1 PCRs to evmctl with explicit selection of sha1:
> $ sudo evmctl ima_measurement --pcrs sha1,pcr-sha1 /sys/kernel/security/integrity/ima/binary_runtime_measurements
> 
> Pass both sha1 and sha256 PCRs to evmctl:
> $ sudo evmctl ima_measurement --pcrs sha1,pcr-sha1 --pcrs sha256,pcr-sha256 /sys/kernel/security/integrity/ima/binary_runtime_measurements
> 
> Signed-off-by: Stephen Smalley <stephen.smalley.work@gmail.com>

I reviewed the patch and everything looks good, other than a few
comments below.  Weirdly the order in which the TPM bank files
containing the PCRs are supplied on the command line is affecting
being able to verify the measurement list - sha1, sha256 versus
sha256, sha1.  Perhaps during the time it takes to read the different
banks, the PCRs have changed?  Not quite sure what is happening.
 
> ---
> v3 splits reading the pcr files and the sysfs files into two separate functions
> each called from read_tpm_banks() with different error handling. Thus, if
> --pcrs is specified, then evmctl will only try to read from the specified
> file(s) and any errors will cause it to fail rather than trying to read
> directly from the TPM.  If --pcrs is not specified, then evmctl will try
> to read from the sysfs files and then fall back to reading directly
> from the TPM if that fails. v3 also reduces the limit on the number of
> PCRs read from the files to NUM_PCRS since otherwise it will overflow
> the pcrs[] array.  Further improvements to the file parser are deferred
> to future patches.
> 
> @@ -1829,20 +1783,108 @@ static void extend_tpm_banks(struct template_entry *entry, int num_banks,
>  #endif
>  }
>  
> -/* Read TPM 1.2 PCRs */
> -static int read_tpm_pcrs(int num_banks, struct tpm_bank_info *tpm_banks)
> +static int read_one_bank(struct tpm_bank_info *tpm_bank, FILE *fp)
>  {
> -	int i;
> +	char *p, pcr_str[8], buf[80];

As log as we're changing this from 70 to 80, might as well define it
as MAX_DITEST_SIZE * 2 + X.  I'm pretty sure support for additional,
larger digests will be added.  This can be deferred to the subsequent
patch. 

> +	int i = 0;
> +	int result = -1;
> +	for (;;) {
> +		p = fgets(buf, sizeof(buf), fp);
> +		if (!p || i >= NUM_PCRS)
> +			break;
> +		sprintf(pcr_str, "PCR-%2.2d", i);
> +		if (!strncmp(p, pcr_str, 6))
> +			hex2bin(tpm_bank->pcr[i++], p + 7, tpm_bank->digest_size);
> +		result = 0;
> +	}
> +	return result;
> +}
> +
> +static char *pcrs = "/sys/class/tpm/tpm0/device/pcrs";  /* Kernels >= 4.0 */
> +static char *misc_pcrs = "/sys/class/misc/tpm0/device/pcrs";
> +
> +/* Read one of the TPM 1.2 sysfs files if present */
> +static int read_sysfs_pcrs(int num_banks, struct tpm_bank_info *tpm_banks)
> +{
> +	FILE *fp;
> +	int i, result;
>  
> -	if (tpm_pcr_read(tpm_banks, SHA_DIGEST_LENGTH)) {
> -		log_debug("Failed to read TPM 1.2 PCRs.\n");
> +	fp = fopen(pcrs, "r");
> +	if (!fp)
> +		fp = fopen(misc_pcrs, "r");
> +	if (!fp)
>  		return -1;
> -	}
>  
> +	result = read_one_bank(&tpm_banks[0], fp);
> +	fclose(fp);
> +	if (result < 0)
> +		return result;
>  	tpm_banks[0].supported = 1;
>  	for (i = 1; i < num_banks; i++)
>  		tpm_banks[i].supported = 0;
>  	return 0;
> +
> +}
> +
> +/* Read PCRs from per-bank file(s) specified via --pcrs */
> +static int read_file_pcrs(int num_banks, struct tpm_bank_info *tpm_banks)
> +{
> +	struct stat s;
> +	FILE *fp;
> +	char *p;
> +	const char *alg, *path;
> +	int i, j, bank, result;
> +
> +	for (i = 0; i < num_banks; i++)
> +		tpm_banks[i].supported = 0;
> +
> +	for (i = 0; i < npcrfile; i++) {

Being able to provide more PCR files than banks is a bit strange.  I
guess the TPM could export additional banks, than what is currently
supported by ima-evm-utils, and programmatically supply them to ima-
evm-utils.

Mimi

> +		p = strchr(pcrfile[i], ',');
> +		if (p) {
> +			*p = 0;
> +			alg = pcrfile[i];
> +			path = ++p;
> +		} else {
> 

  reply	other threads:[~2020-07-24 19:13 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-24 14:52 [PATCH v3 ima-evm-utils] extend ima_measurement --pcrs option to support per-bank pcr files Stephen Smalley
2020-07-24 19:12 ` Mimi Zohar [this message]
2020-07-24 19:44   ` Mimi Zohar

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1595617967.5017.31.camel@linux.ibm.com \
    --to=zohar@linux.ibm.com \
    --cc=bill.c.roberts@gmail.com \
    --cc=linux-integrity@vger.kernel.org \
    --cc=stephen.smalley.work@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.