Message-ID: <8023fa50a84817cc911a117db9bd3757c34fddfb.camel@linux.ibm.com>
Date: Thu, 20 Feb 2025 09:53:08 -0500
From: Mimi Zohar <zohar@...ux.ibm.com>
To: steven chen <chenste@...ux.microsoft.com>, stefanb@...ux.ibm.com,
        roberto.sassu@...weicloud.com, roberto.sassu@...wei.com,
        eric.snowberg@...cle.com, ebiederm@...ssion.com, paul@...l-moore.com,
        code@...icks.com, bauermann@...abnow.com,
        linux-integrity@...r.kernel.org, kexec@...ts.infradead.org,
        linux-security-module@...r.kernel.org, linux-kernel@...r.kernel.org
Cc: madvenka@...ux.microsoft.com, nramas@...ux.microsoft.com,
        James.Bottomley@...senPartnership.com, bhe@...hat.com,
        vgoyal@...hat.com, dyoung@...hat.com
Subject: Re: [PATCH v8 1/7] ima: define and call ima_alloc_kexec_file_buf

On Tue, 2025-02-18 at 14:54 -0800, steven chen wrote:
> Carrying the IMA measurement list across kexec requires allocating a
> buffer and copying the measurement records.  Separate allocating the
> buffer and copying the measurement records into separate functions in
> order to allocate the buffer at kexec 'load' and copy the measurements
> at kexec 'execute'.
> 
> This patch includes the following changes:
>  - Refactor ima_dump_measurement_list() to move the memory allocation
>    to a separate function ima_alloc_kexec_file_buf() which allocates
>    buffer of size 'kexec_segment_size' at kexec 'load'.
>  - Make the local variable ima_kexec_file in ima_dump_measurement_list()
>    a local static to the file, so that it can be accessed from 
>    ima_alloc_kexec_file_buf(). Compare actual memory required to ensure 
>    there is enough memory for the entire measurement record.
>  - Copy as many measurement events as possible.
>  - Make necessary changes to the function ima_add_kexec_buffer() to call
>    the above two functions.
>  - Compared the memory size allocated with memory size of the entire 
>    measurement record. If there is not enough memory, it will copy as many
>    IMA measurement records as possible, and this situation will result
>    in a failure of remote attestation.
> 
> Author: Tushar Sugandhi <tusharsu@...ux.microsoft.com>

Steven, thank you again for picking up this patch set.

As previously explained, there is no tag named "Author" in
https://www.kernel.org/doc/Documentation/process/submitting-patches.rst.  To give credit
to the original author use "Co-developed-by".  The "Co-developed-by:" tag is immediately
followed by the original author's "Signed-off-by:" tag.  Please refer to the document for
an example.

> Suggested-by: Mimi Zohar <zohar@...ux.ibm.com>

<--- "Co-developed-by:" would go here.
> Signed-off-by: Tushar Sugandhi <tusharsu@...ux.microsoft.com>
> Signed-off-by: steven chen <chenste@...ux.microsoft.com>
> ---
>  security/integrity/ima/ima.h       |   1 +
>  security/integrity/ima/ima_kexec.c | 102 +++++++++++++++++++++--------
>  security/integrity/ima/ima_queue.c |   4 +-
>  3 files changed, 77 insertions(+), 30 deletions(-)
> 
> diff --git a/security/integrity/ima/ima.h b/security/integrity/ima/ima.h
> index 24d09ea91b87..4428fcf42167 100644
> --- a/security/integrity/ima/ima.h
> +++ b/security/integrity/ima/ima.h
> @@ -274,6 +274,7 @@ bool ima_template_has_modsig(const struct ima_template_desc
> *ima_template);
>  int ima_restore_measurement_entry(struct ima_template_entry *entry);
>  int ima_restore_measurement_list(loff_t bufsize, void *buf);
>  int ima_measurements_show(struct seq_file *m, void *v);
> +int ima_get_binary_runtime_entry_size(struct ima_template_entry *entry);
>  unsigned long ima_get_binary_runtime_size(void);
>  int ima_init_template(void);
>  void ima_init_template_list(void);
> diff --git a/security/integrity/ima/ima_kexec.c b/security/integrity/ima/ima_kexec.c
> index 9d45f4d26f73..89088f1fa989 100644
> --- a/security/integrity/ima/ima_kexec.c
> +++ b/security/integrity/ima/ima_kexec.c
> @@ -15,63 +15,97 @@
>  #include "ima.h"
>  
>  #ifdef CONFIG_IMA_KEXEC
> +static struct seq_file ima_kexec_file;
> +
> +static void ima_reset_kexec_file(struct seq_file *sf)
> +{
> +	sf->buf = NULL;
> +	sf->size = 0;
> +	sf->read_pos = 0;
> +	sf->count = 0;
> +}
> +
> +static void ima_free_kexec_file_buf(struct seq_file *sf)
> +{
> +	vfree(sf->buf);
> +	ima_reset_kexec_file(sf);
> +}
> +
> +static int ima_alloc_kexec_file_buf(size_t segment_size)
> +{
> +	/*
> +	 * kexec 'load' may be called multiple times.
> +	 * Free and realloc the buffer only if the segment_size is
> +	 * changed from the previous kexec 'load' call.
> +	 */
> +	if (ima_kexec_file.buf && ima_kexec_file.size == segment_size)
> +		goto out;
> +
> +	ima_free_kexec_file_buf(&ima_kexec_file);
> +
> +	/* segment size can't change between kexec load and execute */
> +	ima_kexec_file.buf = vmalloc(segment_size);
> +	if (!ima_kexec_file.buf)
> +		return -ENOMEM;
> +
> +	ima_kexec_file.size = segment_size;
> +
> +out:
> +	ima_kexec_file.read_pos = 0;
> +	ima_kexec_file.count = sizeof(struct ima_kexec_hdr);	/* reserved space */
> +
> +	return 0;
> +}
> +

<--- ima_dump_measurement_list() function comment goes here.

>  static int ima_dump_measurement_list(unsigned long *buffer_size, void **buffer,
>  				     unsigned long segment_size)
>  {
>  	struct ima_queue_entry *qe;
> -	struct seq_file file;
>  	struct ima_kexec_hdr khdr;
>  	int ret = 0;
> +	size_t entry_size = 0;
>  
> -	/* segment size can't change between kexec load and execute */
> -	file.buf = vmalloc(segment_size);
> -	if (!file.buf) {
> -		ret = -ENOMEM;
> -		goto out;
> +	if (!ima_kexec_file.buf) {
> +		pr_err("Kexec file buf not allocated\n");
> +		return -EINVAL;
>  	}
>  
> -	file.file = NULL;
> -	file.size = segment_size;
> -	file.read_pos = 0;
> -	file.count = sizeof(khdr);	/* reserved space */
> -
>  	memset(&khdr, 0, sizeof(khdr));
>  	khdr.version = 1;
>  	/* This is an append-only list, no need to hold the RCU read lock */
> +	/* Copy as many IMA measurements list records as possible */

Having two consecutive comments like this looks weird.  Please refer to section "8)
Commenting" of https://www.kernel.org/doc/Documentation/process/coding-style.rst. 

The first comment is particular to list_for_each_entry_rcu() and should remain here.  The
latter comment is more generic and should be included as part of a function comment.

>  	list_for_each_entry_rcu(qe, &ima_measurements, later, true) {
> -		if (file.count < file.size) {
> +		entry_size += ima_get_binary_runtime_entry_size(qe->entry);
> +		if (entry_size <= segment_size) {

As much as possible splitting a function shouldn't change the existing code.  It makes it
harder to review.  This sort of change should be a separate patch with the Subject topic
line something like "ima: copy only complete measurement records across kexec".  Making
this change as the first patch in the patch set will also allow it to be backported.

>  			khdr.count++;
> -			ima_measurements_show(&file, qe);
> +			ima_measurements_show(&ima_kexec_file, qe);
>  		} else {
>  			ret = -EINVAL;
> +			pr_err("IMA log file is too big for Kexec buf\n");
>  			break;

We really DO want to "Copy as many IMA measurements list records as possible" as possible.
However, the code doesn't match the comment, since the caller of
ima_dump_measurement_list() treats -EINVAL as an error and bails.

>  		}
>  	}
>  
> -	if (ret < 0)
> -		goto out;
> -
>  	/*
>  	 * fill in reserved space with some buffer details
>  	 * (eg. version, buffer size, number of measurements)
>  	 */
> -	khdr.buffer_size = file.count;
> +	khdr.buffer_size = ima_kexec_file.count;
>  	if (ima_canonical_fmt) {
>  		khdr.version = cpu_to_le16(khdr.version);
>  		khdr.count = cpu_to_le64(khdr.count);
>  		khdr.buffer_size = cpu_to_le64(khdr.buffer_size);
>  	}
> -	memcpy(file.buf, &khdr, sizeof(khdr));
> +	memcpy(ima_kexec_file.buf, &khdr, sizeof(khdr));
>  
>  	print_hex_dump_debug("ima dump: ", DUMP_PREFIX_NONE, 16, 1,
> -			     file.buf, file.count < 100 ? file.count : 100,
> +			     ima_kexec_file.buf, ima_kexec_file.count < 100 ?
> +			     ima_kexec_file.count : 100,
>  			     true);
>  
> -	*buffer_size = file.count;
> -	*buffer = file.buf;
> -out:
> -	if (ret == -EINVAL)
> -		vfree(file.buf);
> +	*buffer_size = ima_kexec_file.count;
> +	*buffer = ima_kexec_file.buf;
> +
>  	return ret;
>  }
>  
> @@ -90,7 +124,7 @@ void ima_add_kexec_buffer(struct kimage *image)
>  
>  	/* use more understandable variable names than defined in kbuf */
>  	void *kexec_buffer = NULL;
> -	size_t kexec_buffer_size;
> +	size_t kexec_buffer_size = 0;
>  	size_t kexec_segment_size;
>  	int ret;
>  
> @@ -110,13 +144,19 @@ void ima_add_kexec_buffer(struct kimage *image)
>  		return;
>  	}
>  
> -	ima_dump_measurement_list(&kexec_buffer_size, &kexec_buffer,
> -				  kexec_segment_size);
> -	if (!kexec_buffer) {
> +	ret = ima_alloc_kexec_file_buf(kexec_segment_size);
> +	if (ret < 0) {
>  		pr_err("Not enough memory for the kexec measurement buffer.\n");
>  		return;
>  	}
>  
> +	ret = ima_dump_measurement_list(&kexec_buffer_size, &kexec_buffer,
> +					kexec_segment_size);
> +	if (ret < 0) {
> +		pr_err("Failed to dump IMA measurements. Error:%d.\n", ret);
> +		return;
> +	}
> +

As mentioned above, we really do want to copy as many measurement records as possible
across kexec.

>  	kbuf.buffer = kexec_buffer;
>  	kbuf.bufsz = kexec_buffer_size;
>  	kbuf.memsz = kexec_segment_size;
> @@ -131,6 +171,12 @@ void ima_add_kexec_buffer(struct kimage *image)
>  	image->ima_buffer_size = kexec_segment_size;
>  	image->ima_buffer = kexec_buffer;
>  
> +	/*
> +	 * kexec owns kexec_buffer after kexec_add_buffer() is called
> +	 * and it will vfree() that buffer.
> +	 */
> +	ima_reset_kexec_file(&ima_kexec_file);
> +
>  	kexec_dprintk("kexec measurement buffer for the loaded kernel at 0x%lx.\n",
>  		      kbuf.mem);
>  }
> diff --git a/security/integrity/ima/ima_queue.c b/security/integrity/ima/ima_queue.c
> index 83d53824aa98..3dfd178d4292 100644
> --- a/security/integrity/ima/ima_queue.c
> +++ b/security/integrity/ima/ima_queue.c
> @@ -78,7 +78,7 @@ static struct ima_queue_entry *ima_lookup_digest_entry(u8
> *digest_value,
>   * binary_runtime_measurement list entry, which contains a
>   * couple of variable length fields (e.g template name and data).
>   */
> -static int get_binary_runtime_size(struct ima_template_entry *entry)
> +int ima_get_binary_runtime_entry_size(struct ima_template_entry *entry)
>  {
>  	int size = 0;
>  
> @@ -122,7 +122,7 @@ static int ima_add_digest_entry(struct ima_template_entry *entry,
>  	if (binary_runtime_size != ULONG_MAX) {
>  		int size;
>  
> -		size = get_binary_runtime_size(entry);
> +		size = ima_get_binary_runtime_entry_size(entry);
>  		binary_runtime_size = (binary_runtime_size < ULONG_MAX - size) ?
>  		     binary_runtime_size + size : ULONG_MAX;
>  	}

This change would be included in the new first patch named something like "ima: copy only
complete measurement records across kexec".

Thanks,

Mimi

Powered by blists - more mailing lists