diff mbox series

[RFC] lsm,fs: fix vfs_getxattr_alloc() return type and caller error paths

Message ID 20221110043614.802364-1-paul@paul-moore.com (mailing list archive)
State New, archived
Headers show
Series [RFC] lsm,fs: fix vfs_getxattr_alloc() return type and caller error paths | expand

Commit Message

Paul Moore Nov. 10, 2022, 4:36 a.m. UTC
The vfs_getxattr_alloc() function currently returns a ssize_t value
despite the fact that it only uses int values internally for return
values.  Fix this by converting vfs_getxattr_alloc() to return an
int type and adjust the callers as necessary.  As part of these
caller modifications, some of the callers are fixed to properly free
the xattr value buffer on both success and failure to ensure that
memory is not leaked in the failure case.

Signed-off-by: Paul Moore <paul@paul-moore.com>
---
 fs/xattr.c                                |  5 +++--
 include/linux/xattr.h                     |  6 +++---
 security/apparmor/domain.c                |  3 +--
 security/commoncap.c                      | 22 ++++++++++------------
 security/integrity/evm/evm_crypto.c       |  5 +++--
 security/integrity/evm/evm_main.c         |  7 +++++--
 security/integrity/ima/ima.h              |  5 +++--
 security/integrity/ima/ima_appraise.c     |  6 +++---
 security/integrity/ima/ima_main.c         |  6 ++++--
 security/integrity/ima/ima_template_lib.c | 11 +++++------
 10 files changed, 40 insertions(+), 36 deletions(-)

Comments

Paul Moore Nov. 10, 2022, 4:51 a.m. UTC | #1
On Wed, Nov 9, 2022 at 11:36 PM Paul Moore <paul@paul-moore.com> wrote:
>
> The vfs_getxattr_alloc() function currently returns a ssize_t value
> despite the fact that it only uses int values internally for return
> values.  Fix this by converting vfs_getxattr_alloc() to return an
> int type and adjust the callers as necessary.  As part of these
> caller modifications, some of the callers are fixed to properly free
> the xattr value buffer on both success and failure to ensure that
> memory is not leaked in the failure case.
>
> Signed-off-by: Paul Moore <paul@paul-moore.com>
> ---
>  fs/xattr.c                                |  5 +++--
>  include/linux/xattr.h                     |  6 +++---
>  security/apparmor/domain.c                |  3 +--
>  security/commoncap.c                      | 22 ++++++++++------------
>  security/integrity/evm/evm_crypto.c       |  5 +++--
>  security/integrity/evm/evm_main.c         |  7 +++++--
>  security/integrity/ima/ima.h              |  5 +++--
>  security/integrity/ima/ima_appraise.c     |  6 +++---
>  security/integrity/ima/ima_main.c         |  6 ++++--
>  security/integrity/ima/ima_template_lib.c | 11 +++++------
>  10 files changed, 40 insertions(+), 36 deletions(-)

Mimi, I'm particularly interested in your thoughts on this patch as
there are a number of places in the IMA/EVM code that needed some
additional tweaks to prevent a memory leak like we fixed in the
capabilities code via 8cf0a1bc1287 ("capabilities: fix potential
memleak on error path from vfs_getxattr_alloc()").
Serge Hallyn Nov. 18, 2022, 1:54 a.m. UTC | #2
On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> The vfs_getxattr_alloc() function currently returns a ssize_t value
> despite the fact that it only uses int values internally for return
> values.  Fix this by converting vfs_getxattr_alloc() to return an
> int type and adjust the callers as necessary.  As part of these
> caller modifications, some of the callers are fixed to properly free
> the xattr value buffer on both success and failure to ensure that
> memory is not leaked in the failure case.
> 
> Signed-off-by: Paul Moore <paul@paul-moore.com>

Reviewed-by: Serge Hallyn <serge@hallyn.com>

Do I understand right that the change to process_measurement()
will avoid an unnecessary call to krealloc() if the xattr has
not changed size between the two calls to ima_read_xattr()?
If something more than that is going on there, it might be
worth pointing out in the commit message.

> ---
>  fs/xattr.c                                |  5 +++--
>  include/linux/xattr.h                     |  6 +++---
>  security/apparmor/domain.c                |  3 +--
>  security/commoncap.c                      | 22 ++++++++++------------
>  security/integrity/evm/evm_crypto.c       |  5 +++--
>  security/integrity/evm/evm_main.c         |  7 +++++--
>  security/integrity/ima/ima.h              |  5 +++--
>  security/integrity/ima/ima_appraise.c     |  6 +++---
>  security/integrity/ima/ima_main.c         |  6 ++++--
>  security/integrity/ima/ima_template_lib.c | 11 +++++------
>  10 files changed, 40 insertions(+), 36 deletions(-)
> 
> diff --git a/fs/xattr.c b/fs/xattr.c
> index 61107b6bbed29..f38fd969b5fcd 100644
> --- a/fs/xattr.c
> +++ b/fs/xattr.c
> @@ -354,11 +354,12 @@ xattr_getsecurity(struct user_namespace *mnt_userns, struct inode *inode,
>   * vfs_getxattr_alloc - allocate memory, if necessary, before calling getxattr
>   *
>   * Allocate memory, if not already allocated, or re-allocate correct size,
> - * before retrieving the extended attribute.
> + * before retrieving the extended attribute.  The xattr value buffer should
> + * always be freed by the caller, even on error.
>   *
>   * Returns the result of alloc, if failed, or the getxattr operation.
>   */
> -ssize_t
> +int
>  vfs_getxattr_alloc(struct user_namespace *mnt_userns, struct dentry *dentry,
>  		   const char *name, char **xattr_value, size_t xattr_size,
>  		   gfp_t flags)
> diff --git a/include/linux/xattr.h b/include/linux/xattr.h
> index 4c379d23ec6e7..2218a9645b89d 100644
> --- a/include/linux/xattr.h
> +++ b/include/linux/xattr.h
> @@ -68,9 +68,9 @@ int __vfs_removexattr_locked(struct user_namespace *, struct dentry *,
>  int vfs_removexattr(struct user_namespace *, struct dentry *, const char *);
>  
>  ssize_t generic_listxattr(struct dentry *dentry, char *buffer, size_t buffer_size);
> -ssize_t vfs_getxattr_alloc(struct user_namespace *mnt_userns,
> -			   struct dentry *dentry, const char *name,
> -			   char **xattr_value, size_t size, gfp_t flags);
> +int vfs_getxattr_alloc(struct user_namespace *mnt_userns,
> +		       struct dentry *dentry, const char *name,
> +		       char **xattr_value, size_t size, gfp_t flags);
>  
>  int xattr_supported_namespace(struct inode *inode, const char *prefix);
>  
> diff --git a/security/apparmor/domain.c b/security/apparmor/domain.c
> index 91689d34d2811..04a818d516047 100644
> --- a/security/apparmor/domain.c
> +++ b/security/apparmor/domain.c
> @@ -311,10 +311,9 @@ static int aa_xattrs_match(const struct linux_binprm *bprm,
>  			   struct aa_profile *profile, unsigned int state)
>  {
>  	int i;
> -	ssize_t size;
>  	struct dentry *d;
>  	char *value = NULL;
> -	int value_size = 0, ret = profile->xattr_count;
> +	int size, value_size = 0, ret = profile->xattr_count;
>  
>  	if (!bprm || !profile->xattr_count)
>  		return 0;
> diff --git a/security/commoncap.c b/security/commoncap.c
> index 5fc8986c3c77c..d4fc890955134 100644
> --- a/security/commoncap.c
> +++ b/security/commoncap.c
> @@ -350,14 +350,14 @@ static __u32 sansflags(__u32 m)
>  	return m & ~VFS_CAP_FLAGS_EFFECTIVE;
>  }
>  
> -static bool is_v2header(size_t size, const struct vfs_cap_data *cap)
> +static bool is_v2header(int size, const struct vfs_cap_data *cap)
>  {
>  	if (size != XATTR_CAPS_SZ_2)
>  		return false;
>  	return sansflags(le32_to_cpu(cap->magic_etc)) == VFS_CAP_REVISION_2;
>  }
>  
> -static bool is_v3header(size_t size, const struct vfs_cap_data *cap)
> +static bool is_v3header(int size, const struct vfs_cap_data *cap)
>  {
>  	if (size != XATTR_CAPS_SZ_3)
>  		return false;
> @@ -379,7 +379,7 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns,
>  			  struct inode *inode, const char *name, void **buffer,
>  			  bool alloc)
>  {
> -	int size, ret;
> +	int size;
>  	kuid_t kroot;
>  	u32 nsmagic, magic;
>  	uid_t root, mappedroot;
> @@ -395,20 +395,18 @@ int cap_inode_getsecurity(struct user_namespace *mnt_userns,
>  	dentry = d_find_any_alias(inode);
>  	if (!dentry)
>  		return -EINVAL;
> -
> -	size = sizeof(struct vfs_ns_cap_data);
> -	ret = (int)vfs_getxattr_alloc(mnt_userns, dentry, XATTR_NAME_CAPS,
> -				      &tmpbuf, size, GFP_NOFS);
> +	size = vfs_getxattr_alloc(mnt_userns, dentry, XATTR_NAME_CAPS, &tmpbuf,
> +				  sizeof(struct vfs_ns_cap_data), GFP_NOFS);
>  	dput(dentry);
> -
> -	if (ret < 0 || !tmpbuf)
> -		return ret;
> +	/* gcc11 complains if we don't check for !tmpbuf */
> +	if (size < 0 || !tmpbuf)
> +		goto out_free;
>  
>  	fs_ns = inode->i_sb->s_user_ns;
>  	cap = (struct vfs_cap_data *) tmpbuf;
> -	if (is_v2header((size_t) ret, cap)) {
> +	if (is_v2header(size, cap)) {
>  		root = 0;
> -	} else if (is_v3header((size_t) ret, cap)) {
> +	} else if (is_v3header(size, cap)) {
>  		nscap = (struct vfs_ns_cap_data *) tmpbuf;
>  		root = le32_to_cpu(nscap->rootid);
>  	} else {
> diff --git a/security/integrity/evm/evm_crypto.c b/security/integrity/evm/evm_crypto.c
> index 708de9656bbd2..fa5ff13fa8c97 100644
> --- a/security/integrity/evm/evm_crypto.c
> +++ b/security/integrity/evm/evm_crypto.c
> @@ -335,14 +335,15 @@ static int evm_is_immutable(struct dentry *dentry, struct inode *inode)
>  				(char **)&xattr_data, 0, GFP_NOFS);
>  	if (rc <= 0) {
>  		if (rc == -ENODATA)
> -			return 0;
> -		return rc;
> +			rc = 0;
> +		goto out;
>  	}
>  	if (xattr_data->type == EVM_XATTR_PORTABLE_DIGSIG)
>  		rc = 1;
>  	else
>  		rc = 0;
>  
> +out:
>  	kfree(xattr_data);
>  	return rc;
>  }
> diff --git a/security/integrity/evm/evm_main.c b/security/integrity/evm/evm_main.c
> index 23d484e05e6f2..bce72e80fd123 100644
> --- a/security/integrity/evm/evm_main.c
> +++ b/security/integrity/evm/evm_main.c
> @@ -519,14 +519,17 @@ static int evm_xattr_change(struct user_namespace *mnt_userns,
>  
>  	rc = vfs_getxattr_alloc(&init_user_ns, dentry, xattr_name, &xattr_data,
>  				0, GFP_NOFS);
> -	if (rc < 0)
> -		return 1;
> +	if (rc < 0) {
> +		rc = 1;
> +		goto out;
> +	}
>  
>  	if (rc == xattr_value_len)
>  		rc = !!memcmp(xattr_value, xattr_data, rc);
>  	else
>  		rc = 1;
>  
> +out:
>  	kfree(xattr_data);
>  	return rc;
>  }
> diff --git a/security/integrity/ima/ima.h b/security/integrity/ima/ima.h
> index be965a8715e4e..03b440921e615 100644
> --- a/security/integrity/ima/ima.h
> +++ b/security/integrity/ima/ima.h
> @@ -326,7 +326,7 @@ enum integrity_status ima_get_cache_status(struct integrity_iint_cache *iint,
>  enum hash_algo ima_get_hash_algo(const struct evm_ima_xattr_data *xattr_value,
>  				 int xattr_len);
>  int ima_read_xattr(struct dentry *dentry,
> -		   struct evm_ima_xattr_data **xattr_value);
> +		   struct evm_ima_xattr_data **xattr_value, int xattr_len);
>  
>  #else
>  static inline int ima_check_blacklist(struct integrity_iint_cache *iint,
> @@ -372,7 +372,8 @@ ima_get_hash_algo(struct evm_ima_xattr_data *xattr_value, int xattr_len)
>  }
>  
>  static inline int ima_read_xattr(struct dentry *dentry,
> -				 struct evm_ima_xattr_data **xattr_value)
> +				 struct evm_ima_xattr_data **xattr_value,
> +				 int xattr_len)
>  {
>  	return 0;
>  }
> diff --git a/security/integrity/ima/ima_appraise.c b/security/integrity/ima/ima_appraise.c
> index 3e0fbbd995342..88ffb15ca0e2e 100644
> --- a/security/integrity/ima/ima_appraise.c
> +++ b/security/integrity/ima/ima_appraise.c
> @@ -221,12 +221,12 @@ enum hash_algo ima_get_hash_algo(const struct evm_ima_xattr_data *xattr_value,
>  }
>  
>  int ima_read_xattr(struct dentry *dentry,
> -		   struct evm_ima_xattr_data **xattr_value)
> +		   struct evm_ima_xattr_data **xattr_value, int xattr_len)
>  {
> -	ssize_t ret;
> +	int ret;
>  
>  	ret = vfs_getxattr_alloc(&init_user_ns, dentry, XATTR_NAME_IMA,
> -				 (char **)xattr_value, 0, GFP_NOFS);
> +				 (char **)xattr_value, xattr_len, GFP_NOFS);
>  	if (ret == -EOPNOTSUPP)
>  		ret = 0;
>  	return ret;
> diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c
> index 040b03ddc1c77..0226899947d6a 100644
> --- a/security/integrity/ima/ima_main.c
> +++ b/security/integrity/ima/ima_main.c
> @@ -293,7 +293,8 @@ static int process_measurement(struct file *file, const struct cred *cred,
>  	/* HASH sets the digital signature and update flags, nothing else */
>  	if ((action & IMA_HASH) &&
>  	    !(test_bit(IMA_DIGSIG, &iint->atomic_flags))) {
> -		xattr_len = ima_read_xattr(file_dentry(file), &xattr_value);
> +		xattr_len = ima_read_xattr(file_dentry(file),
> +					   &xattr_value, xattr_len);
>  		if ((xattr_value && xattr_len > 2) &&
>  		    (xattr_value->type == EVM_IMA_XATTR_DIGSIG))
>  			set_bit(IMA_DIGSIG, &iint->atomic_flags);
> @@ -316,7 +317,8 @@ static int process_measurement(struct file *file, const struct cred *cred,
>  	if ((action & IMA_APPRAISE_SUBMASK) ||
>  	    strcmp(template_desc->name, IMA_TEMPLATE_IMA_NAME) != 0) {
>  		/* read 'security.ima' */
> -		xattr_len = ima_read_xattr(file_dentry(file), &xattr_value);
> +		xattr_len = ima_read_xattr(file_dentry(file),
> +					   &xattr_value, xattr_len);
>  
>  		/*
>  		 * Read the appended modsig if allowed by the policy, and allow
> diff --git a/security/integrity/ima/ima_template_lib.c b/security/integrity/ima/ima_template_lib.c
> index 7bf9b15072202..4564faae7d673 100644
> --- a/security/integrity/ima/ima_template_lib.c
> +++ b/security/integrity/ima/ima_template_lib.c
> @@ -601,16 +601,15 @@ int ima_eventevmsig_init(struct ima_event_data *event_data,
>  	rc = vfs_getxattr_alloc(&init_user_ns, file_dentry(event_data->file),
>  				XATTR_NAME_EVM, (char **)&xattr_data, 0,
>  				GFP_NOFS);
> -	if (rc <= 0)
> -		return 0;
> -
> -	if (xattr_data->type != EVM_XATTR_PORTABLE_DIGSIG) {
> -		kfree(xattr_data);
> -		return 0;
> +	if (rc <= 0 || xattr_data->type != EVM_XATTR_PORTABLE_DIGSIG) {
> +		rc = 0;
> +		goto out;
>  	}
>  
>  	rc = ima_write_template_field_data((char *)xattr_data, rc, DATA_FMT_HEX,
>  					   field_data);
> +
> +out:
>  	kfree(xattr_data);
>  	return rc;
>  }
> -- 
> 2.38.1
Paul Moore Nov. 18, 2022, 1:44 p.m. UTC | #3
On Thu, Nov 17, 2022 at 8:54 PM Serge E. Hallyn <serge@hallyn.com> wrote:
> On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> > The vfs_getxattr_alloc() function currently returns a ssize_t value
> > despite the fact that it only uses int values internally for return
> > values.  Fix this by converting vfs_getxattr_alloc() to return an
> > int type and adjust the callers as necessary.  As part of these
> > caller modifications, some of the callers are fixed to properly free
> > the xattr value buffer on both success and failure to ensure that
> > memory is not leaked in the failure case.
> >
> > Signed-off-by: Paul Moore <paul@paul-moore.com>
>
> Reviewed-by: Serge Hallyn <serge@hallyn.com>
>
> Do I understand right that the change to process_measurement()
> will avoid an unnecessary call to krealloc() if the xattr has
> not changed size between the two calls to ima_read_xattr()?
> If something more than that is going on there, it might be
> worth pointing out in the commit message.

Yes, that was the intent, trying to avoid extra calls to krealloc().

Mimi, have you had a chance to look at this patch yet?  In addition to
cleaning up the vfs_getxattr_alloc() function it resolves some issues
with IMA (memory leaks), but as you're the IMA expert I really need
your review on this ...
Mimi Zohar Nov. 18, 2022, 6:29 p.m. UTC | #4
On Fri, 2022-11-18 at 08:44 -0500, Paul Moore wrote:
> On Thu, Nov 17, 2022 at 8:54 PM Serge E. Hallyn <serge@hallyn.com> wrote:
> > On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> > > The vfs_getxattr_alloc() function currently returns a ssize_t value
> > > despite the fact that it only uses int values internally for return
> > > values.  Fix this by converting vfs_getxattr_alloc() to return an
> > > int type and adjust the callers as necessary.  As part of these
> > > caller modifications, some of the callers are fixed to properly free
> > > the xattr value buffer on both success and failure to ensure that
> > > memory is not leaked in the failure case.
> 
> > > Signed-off-by: Paul Moore <paul@paul-moore.com>
> >
> > Reviewed-by: Serge Hallyn <serge@hallyn.com>
> >
> > Do I understand right that the change to process_measurement()
> > will avoid an unnecessary call to krealloc() if the xattr has
> > not changed size between the two calls to ima_read_xattr()?
> > If something more than that is going on there, it might be
> > worth pointing out in the commit message.
> 
> Yes, that was the intent, trying to avoid extra calls to krealloc().
> 
> Mimi, have you had a chance to look at this patch yet?  In addition to
> cleaning up the vfs_getxattr_alloc() function it resolves some issues
> with IMA (memory leaks), but as you're the IMA expert I really need
> your review on this ...b

All the other vfs_{get/set/list}xattr functions return ssize_t.  Why
should vfs_getxattr_alloc() be any different?

The only time there could be a memory leak is when the
vfs_getxattr_alloc() caller provides a buffer which isn't large enough.
The one example in IMA/EVM is the call to evm_calc_hmac_or_hash(),
which is freeing the memory.

Perhaps I'm missing something, but from an IMA/EVM perspective, I see a
style change (common exit), but not any memory leak fixes.  I'm fine
with the style change.
Paul Moore Nov. 18, 2022, 6:44 p.m. UTC | #5
On Fri, Nov 18, 2022 at 1:30 PM Mimi Zohar <zohar@linux.ibm.com> wrote:
> On Fri, 2022-11-18 at 08:44 -0500, Paul Moore wrote:
> > On Thu, Nov 17, 2022 at 8:54 PM Serge E. Hallyn <serge@hallyn.com> wrote:
> > > On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> > > > The vfs_getxattr_alloc() function currently returns a ssize_t value
> > > > despite the fact that it only uses int values internally for return
> > > > values.  Fix this by converting vfs_getxattr_alloc() to return an
> > > > int type and adjust the callers as necessary.  As part of these
> > > > caller modifications, some of the callers are fixed to properly free
> > > > the xattr value buffer on both success and failure to ensure that
> > > > memory is not leaked in the failure case.
> >
> > > > Signed-off-by: Paul Moore <paul@paul-moore.com>
> > >
> > > Reviewed-by: Serge Hallyn <serge@hallyn.com>
> > >
> > > Do I understand right that the change to process_measurement()
> > > will avoid an unnecessary call to krealloc() if the xattr has
> > > not changed size between the two calls to ima_read_xattr()?
> > > If something more than that is going on there, it might be
> > > worth pointing out in the commit message.
> >
> > Yes, that was the intent, trying to avoid extra calls to krealloc().
> >
> > Mimi, have you had a chance to look at this patch yet?  In addition to
> > cleaning up the vfs_getxattr_alloc() function it resolves some issues
> > with IMA (memory leaks), but as you're the IMA expert I really need
> > your review on this ...b
>
> All the other vfs_{get/set/list}xattr functions return ssize_t.  Why
> should vfs_getxattr_alloc() be any different?

The xattr_handler::get() function, the main engine behind
vfs_getxattr_alloc() and the source of the non-error return values,
returns an int.  The error return values returned by
vfs_getxattr_alloc() are the usual -E* integer values.

> The only time there could be a memory leak is when the
> vfs_getxattr_alloc() caller provides a buffer which isn't large enough.
> The one example in IMA/EVM is the call to evm_calc_hmac_or_hash(),
> which is freeing the memory.
>
> Perhaps I'm missing something, but from an IMA/EVM perspective, I see a
> style change (common exit), but not any memory leak fixes.  I'm fine
> with the style change.

Picking one at random, what about the change in
ima_eventevmsig_init()?  The current code does not free @xattr_data on
error which has the potential to leak memory if vfs_getxattr_alloc()'s
second call to the xattr get'er function fails.  Granted, the
likelihood of this, if it is even possible, is an open question, but I
don't think that is an excuse for the callers to not do The Right
Thing.
Mimi Zohar Nov. 18, 2022, 7:09 p.m. UTC | #6
On Fri, 2022-11-18 at 13:44 -0500, Paul Moore wrote:
> On Fri, Nov 18, 2022 at 1:30 PM Mimi Zohar <zohar@linux.ibm.com> wrote:
> > On Fri, 2022-11-18 at 08:44 -0500, Paul Moore wrote:
> > > On Thu, Nov 17, 2022 at 8:54 PM Serge E. Hallyn <serge@hallyn.com> wrote:
> > > > On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> > > > > The vfs_getxattr_alloc() function currently returns a ssize_t value
> > > > > despite the fact that it only uses int values internally for return
> > > > > values.  Fix this by converting vfs_getxattr_alloc() to return an
> > > > > int type and adjust the callers as necessary.  As part of these
> > > > > caller modifications, some of the callers are fixed to properly free
> > > > > the xattr value buffer on both success and failure to ensure that
> > > > > memory is not leaked in the failure case.
> > >
> > > > > Signed-off-by: Paul Moore <paul@paul-moore.com>
> > > >
> > > > Reviewed-by: Serge Hallyn <serge@hallyn.com>
> > > >
> > > > Do I understand right that the change to process_measurement()
> > > > will avoid an unnecessary call to krealloc() if the xattr has
> > > > not changed size between the two calls to ima_read_xattr()?
> > > > If something more than that is going on there, it might be
> > > > worth pointing out in the commit message.
> > >
> > > Yes, that was the intent, trying to avoid extra calls to krealloc().
> > >
> > > Mimi, have you had a chance to look at this patch yet?  In addition to
> > > cleaning up the vfs_getxattr_alloc() function it resolves some issues
> > > with IMA (memory leaks), but as you're the IMA expert I really need
> > > your review on this ...b
> >
> > All the other vfs_{get/set/list}xattr functions return ssize_t.  Why
> > should vfs_getxattr_alloc() be any different?
> 
> The xattr_handler::get() function, the main engine behind
> vfs_getxattr_alloc() and the source of the non-error return values,
> returns an int.  The error return values returned by
> vfs_getxattr_alloc() are the usual -E* integer values.
> 
> > The only time there could be a memory leak is when the
> > vfs_getxattr_alloc() caller provides a buffer which isn't large enough.
> > The one example in IMA/EVM is the call to evm_calc_hmac_or_hash(),
> > which is freeing the memory.
> >
> > Perhaps I'm missing something, but from an IMA/EVM perspective, I see a
> > style change (common exit), but not any memory leak fixes.  I'm fine
> > with the style change.
> 
> Picking one at random, what about the change in
> ima_eventevmsig_init()?  The current code does not free @xattr_data on
> error which has the potential to leak memory if vfs_getxattr_alloc()'s
> second call to the xattr get'er function fails.  Granted, the
> likelihood of this, if it is even possible, is an open question, but I
> don't think that is an excuse for the callers to not do The Right
> Thing.

Oh!  This is about the 2nd handler call failing.

Reviewed-by: Mimi Zohar <zohar@linux.ibm.com>b
Paul Moore Nov. 18, 2022, 10:20 p.m. UTC | #7
On Fri, Nov 18, 2022 at 2:09 PM Mimi Zohar <zohar@linux.ibm.com> wrote:
> On Fri, 2022-11-18 at 13:44 -0500, Paul Moore wrote:
> > On Fri, Nov 18, 2022 at 1:30 PM Mimi Zohar <zohar@linux.ibm.com> wrote:
> > > On Fri, 2022-11-18 at 08:44 -0500, Paul Moore wrote:
> > > > On Thu, Nov 17, 2022 at 8:54 PM Serge E. Hallyn <serge@hallyn.com> wrote:
> > > > > On Wed, Nov 09, 2022 at 11:36:14PM -0500, Paul Moore wrote:
> > > > > > The vfs_getxattr_alloc() function currently returns a ssize_t value
> > > > > > despite the fact that it only uses int values internally for return
> > > > > > values.  Fix this by converting vfs_getxattr_alloc() to return an
> > > > > > int type and adjust the callers as necessary.  As part of these
> > > > > > caller modifications, some of the callers are fixed to properly free
> > > > > > the xattr value buffer on both success and failure to ensure that
> > > > > > memory is not leaked in the failure case.
> > > >
> > > > > > Signed-off-by: Paul Moore <paul@paul-moore.com>
> > > > >
> > > > > Reviewed-by: Serge Hallyn <serge@hallyn.com>
> > > > >
> > > > > Do I understand right that the change to process_measurement()
> > > > > will avoid an unnecessary call to krealloc() if the xattr has
> > > > > not changed size between the two calls to ima_read_xattr()?
> > > > > If something more than that is going on there, it might be
> > > > > worth pointing out in the commit message.
> > > >
> > > > Yes, that was the intent, trying to avoid extra calls to krealloc().
> > > >
> > > > Mimi, have you had a chance to look at this patch yet?  In addition to
> > > > cleaning up the vfs_getxattr_alloc() function it resolves some issues
> > > > with IMA (memory leaks), but as you're the IMA expert I really need
> > > > your review on this ...b
> > >
> > > All the other vfs_{get/set/list}xattr functions return ssize_t.  Why
> > > should vfs_getxattr_alloc() be any different?
> >
> > The xattr_handler::get() function, the main engine behind
> > vfs_getxattr_alloc() and the source of the non-error return values,
> > returns an int.  The error return values returned by
> > vfs_getxattr_alloc() are the usual -E* integer values.
> >
> > > The only time there could be a memory leak is when the
> > > vfs_getxattr_alloc() caller provides a buffer which isn't large enough.
> > > The one example in IMA/EVM is the call to evm_calc_hmac_or_hash(),
> > > which is freeing the memory.
> > >
> > > Perhaps I'm missing something, but from an IMA/EVM perspective, I see a
> > > style change (common exit), but not any memory leak fixes.  I'm fine
> > > with the style change.
> >
> > Picking one at random, what about the change in
> > ima_eventevmsig_init()?  The current code does not free @xattr_data on
> > error which has the potential to leak memory if vfs_getxattr_alloc()'s
> > second call to the xattr get'er function fails.  Granted, the
> > likelihood of this, if it is even possible, is an open question, but I
> > don't think that is an excuse for the callers to not do The Right
> > Thing.
>
> Oh!  This is about the 2nd handler call failing.
>
> Reviewed-by: Mimi Zohar <zohar@linux.ibm.com>b

Merged into lsm/next, thanks all.
diff mbox series

Patch

diff --git a/fs/xattr.c b/fs/xattr.c
index 61107b6bbed29..f38fd969b5fcd 100644
--- a/fs/xattr.c
+++ b/fs/xattr.c
@@ -354,11 +354,12 @@  xattr_getsecurity(struct user_namespace *mnt_userns, struct inode *inode,
  * vfs_getxattr_alloc - allocate memory, if necessary, before calling getxattr
  *
  * Allocate memory, if not already allocated, or re-allocate correct size,
- * before retrieving the extended attribute.
+ * before retrieving the extended attribute.  The xattr value buffer should
+ * always be freed by the caller, even on error.
  *
  * Returns the result of alloc, if failed, or the getxattr operation.
  */
-ssize_t
+int
 vfs_getxattr_alloc(struct user_namespace *mnt_userns, struct dentry *dentry,
 		   const char *name, char **xattr_value, size_t xattr_size,
 		   gfp_t flags)
diff --git a/include/linux/xattr.h b/include/linux/xattr.h
index 4c379d23ec6e7..2218a9645b89d 100644
--- a/include/linux/xattr.h
+++ b/include/linux/xattr.h
@@ -68,9 +68,9 @@  int __vfs_removexattr_locked(struct user_namespace *, struct dentry *,
 int vfs_removexattr(struct user_namespace *, struct dentry *, const char *);
 
 ssize_t generic_listxattr(struct dentry *dentry, char *buffer, size_t buffer_size);
-ssize_t vfs_getxattr_alloc(struct user_namespace *mnt_userns,
-			   struct dentry *dentry, const char *name,
-			   char **xattr_value, size_t size, gfp_t flags);
+int vfs_getxattr_alloc(struct user_namespace *mnt_userns,
+		       struct dentry *dentry, const char *name,
+		       char **xattr_value, size_t size, gfp_t flags);
 
 int xattr_supported_namespace(struct inode *inode, const char *prefix);
 
diff --git a/security/apparmor/domain.c b/security/apparmor/domain.c
index 91689d34d2811..04a818d516047 100644
--- a/security/apparmor/domain.c
+++ b/security/apparmor/domain.c
@@ -311,10 +311,9 @@  static int aa_xattrs_match(const struct linux_binprm *bprm,
 			   struct aa_profile *profile, unsigned int state)
 {
 	int i;
-	ssize_t size;
 	struct dentry *d;
 	char *value = NULL;
-	int value_size = 0, ret = profile->xattr_count;
+	int size, value_size = 0, ret = profile->xattr_count;
 
 	if (!bprm || !profile->xattr_count)
 		return 0;
diff --git a/security/commoncap.c b/security/commoncap.c
index 5fc8986c3c77c..d4fc890955134 100644
--- a/security/commoncap.c
+++ b/security/commoncap.c
@@ -350,14 +350,14 @@  static __u32 sansflags(__u32 m)
 	return m & ~VFS_CAP_FLAGS_EFFECTIVE;
 }
 
-static bool is_v2header(size_t size, const struct vfs_cap_data *cap)
+static bool is_v2header(int size, const struct vfs_cap_data *cap)
 {
 	if (size != XATTR_CAPS_SZ_2)
 		return false;
 	return sansflags(le32_to_cpu(cap->magic_etc)) == VFS_CAP_REVISION_2;
 }
 
-static bool is_v3header(size_t size, const struct vfs_cap_data *cap)
+static bool is_v3header(int size, const struct vfs_cap_data *cap)
 {
 	if (size != XATTR_CAPS_SZ_3)
 		return false;
@@ -379,7 +379,7 @@  int cap_inode_getsecurity(struct user_namespace *mnt_userns,
 			  struct inode *inode, const char *name, void **buffer,
 			  bool alloc)
 {
-	int size, ret;
+	int size;
 	kuid_t kroot;
 	u32 nsmagic, magic;
 	uid_t root, mappedroot;
@@ -395,20 +395,18 @@  int cap_inode_getsecurity(struct user_namespace *mnt_userns,
 	dentry = d_find_any_alias(inode);
 	if (!dentry)
 		return -EINVAL;
-
-	size = sizeof(struct vfs_ns_cap_data);
-	ret = (int)vfs_getxattr_alloc(mnt_userns, dentry, XATTR_NAME_CAPS,
-				      &tmpbuf, size, GFP_NOFS);
+	size = vfs_getxattr_alloc(mnt_userns, dentry, XATTR_NAME_CAPS, &tmpbuf,
+				  sizeof(struct vfs_ns_cap_data), GFP_NOFS);
 	dput(dentry);
-
-	if (ret < 0 || !tmpbuf)
-		return ret;
+	/* gcc11 complains if we don't check for !tmpbuf */
+	if (size < 0 || !tmpbuf)
+		goto out_free;
 
 	fs_ns = inode->i_sb->s_user_ns;
 	cap = (struct vfs_cap_data *) tmpbuf;
-	if (is_v2header((size_t) ret, cap)) {
+	if (is_v2header(size, cap)) {
 		root = 0;
-	} else if (is_v3header((size_t) ret, cap)) {
+	} else if (is_v3header(size, cap)) {
 		nscap = (struct vfs_ns_cap_data *) tmpbuf;
 		root = le32_to_cpu(nscap->rootid);
 	} else {
diff --git a/security/integrity/evm/evm_crypto.c b/security/integrity/evm/evm_crypto.c
index 708de9656bbd2..fa5ff13fa8c97 100644
--- a/security/integrity/evm/evm_crypto.c
+++ b/security/integrity/evm/evm_crypto.c
@@ -335,14 +335,15 @@  static int evm_is_immutable(struct dentry *dentry, struct inode *inode)
 				(char **)&xattr_data, 0, GFP_NOFS);
 	if (rc <= 0) {
 		if (rc == -ENODATA)
-			return 0;
-		return rc;
+			rc = 0;
+		goto out;
 	}
 	if (xattr_data->type == EVM_XATTR_PORTABLE_DIGSIG)
 		rc = 1;
 	else
 		rc = 0;
 
+out:
 	kfree(xattr_data);
 	return rc;
 }
diff --git a/security/integrity/evm/evm_main.c b/security/integrity/evm/evm_main.c
index 23d484e05e6f2..bce72e80fd123 100644
--- a/security/integrity/evm/evm_main.c
+++ b/security/integrity/evm/evm_main.c
@@ -519,14 +519,17 @@  static int evm_xattr_change(struct user_namespace *mnt_userns,
 
 	rc = vfs_getxattr_alloc(&init_user_ns, dentry, xattr_name, &xattr_data,
 				0, GFP_NOFS);
-	if (rc < 0)
-		return 1;
+	if (rc < 0) {
+		rc = 1;
+		goto out;
+	}
 
 	if (rc == xattr_value_len)
 		rc = !!memcmp(xattr_value, xattr_data, rc);
 	else
 		rc = 1;
 
+out:
 	kfree(xattr_data);
 	return rc;
 }
diff --git a/security/integrity/ima/ima.h b/security/integrity/ima/ima.h
index be965a8715e4e..03b440921e615 100644
--- a/security/integrity/ima/ima.h
+++ b/security/integrity/ima/ima.h
@@ -326,7 +326,7 @@  enum integrity_status ima_get_cache_status(struct integrity_iint_cache *iint,
 enum hash_algo ima_get_hash_algo(const struct evm_ima_xattr_data *xattr_value,
 				 int xattr_len);
 int ima_read_xattr(struct dentry *dentry,
-		   struct evm_ima_xattr_data **xattr_value);
+		   struct evm_ima_xattr_data **xattr_value, int xattr_len);
 
 #else
 static inline int ima_check_blacklist(struct integrity_iint_cache *iint,
@@ -372,7 +372,8 @@  ima_get_hash_algo(struct evm_ima_xattr_data *xattr_value, int xattr_len)
 }
 
 static inline int ima_read_xattr(struct dentry *dentry,
-				 struct evm_ima_xattr_data **xattr_value)
+				 struct evm_ima_xattr_data **xattr_value,
+				 int xattr_len)
 {
 	return 0;
 }
diff --git a/security/integrity/ima/ima_appraise.c b/security/integrity/ima/ima_appraise.c
index 3e0fbbd995342..88ffb15ca0e2e 100644
--- a/security/integrity/ima/ima_appraise.c
+++ b/security/integrity/ima/ima_appraise.c
@@ -221,12 +221,12 @@  enum hash_algo ima_get_hash_algo(const struct evm_ima_xattr_data *xattr_value,
 }
 
 int ima_read_xattr(struct dentry *dentry,
-		   struct evm_ima_xattr_data **xattr_value)
+		   struct evm_ima_xattr_data **xattr_value, int xattr_len)
 {
-	ssize_t ret;
+	int ret;
 
 	ret = vfs_getxattr_alloc(&init_user_ns, dentry, XATTR_NAME_IMA,
-				 (char **)xattr_value, 0, GFP_NOFS);
+				 (char **)xattr_value, xattr_len, GFP_NOFS);
 	if (ret == -EOPNOTSUPP)
 		ret = 0;
 	return ret;
diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c
index 040b03ddc1c77..0226899947d6a 100644
--- a/security/integrity/ima/ima_main.c
+++ b/security/integrity/ima/ima_main.c
@@ -293,7 +293,8 @@  static int process_measurement(struct file *file, const struct cred *cred,
 	/* HASH sets the digital signature and update flags, nothing else */
 	if ((action & IMA_HASH) &&
 	    !(test_bit(IMA_DIGSIG, &iint->atomic_flags))) {
-		xattr_len = ima_read_xattr(file_dentry(file), &xattr_value);
+		xattr_len = ima_read_xattr(file_dentry(file),
+					   &xattr_value, xattr_len);
 		if ((xattr_value && xattr_len > 2) &&
 		    (xattr_value->type == EVM_IMA_XATTR_DIGSIG))
 			set_bit(IMA_DIGSIG, &iint->atomic_flags);
@@ -316,7 +317,8 @@  static int process_measurement(struct file *file, const struct cred *cred,
 	if ((action & IMA_APPRAISE_SUBMASK) ||
 	    strcmp(template_desc->name, IMA_TEMPLATE_IMA_NAME) != 0) {
 		/* read 'security.ima' */
-		xattr_len = ima_read_xattr(file_dentry(file), &xattr_value);
+		xattr_len = ima_read_xattr(file_dentry(file),
+					   &xattr_value, xattr_len);
 
 		/*
 		 * Read the appended modsig if allowed by the policy, and allow
diff --git a/security/integrity/ima/ima_template_lib.c b/security/integrity/ima/ima_template_lib.c
index 7bf9b15072202..4564faae7d673 100644
--- a/security/integrity/ima/ima_template_lib.c
+++ b/security/integrity/ima/ima_template_lib.c
@@ -601,16 +601,15 @@  int ima_eventevmsig_init(struct ima_event_data *event_data,
 	rc = vfs_getxattr_alloc(&init_user_ns, file_dentry(event_data->file),
 				XATTR_NAME_EVM, (char **)&xattr_data, 0,
 				GFP_NOFS);
-	if (rc <= 0)
-		return 0;
-
-	if (xattr_data->type != EVM_XATTR_PORTABLE_DIGSIG) {
-		kfree(xattr_data);
-		return 0;
+	if (rc <= 0 || xattr_data->type != EVM_XATTR_PORTABLE_DIGSIG) {
+		rc = 0;
+		goto out;
 	}
 
 	rc = ima_write_template_field_data((char *)xattr_data, rc, DATA_FMT_HEX,
 					   field_data);
+
+out:
 	kfree(xattr_data);
 	return rc;
 }