On Wed, 2021-04-28 at 15:35 +0000, Roberto Sassu wrote:
> > From: Roberto Sassu
> > Sent: Friday, March 5, 2021 4:19 PM
> > ima_inode_setxattr() and ima_inode_removexattr() hooks are called before
> > an
> > operation is performed. Thus, ima_reset_appraise_flags() should not be
> > called there, as flags might be unnecessarily reset if the operation is
> > denied.
> >
> > This patch introduces the post hooks ima_inode_post_setxattr() and
> > ima_inode_post_removexattr(), and adds the call to
> > ima_reset_appraise_flags() in the new functions.
> >
> > Cc: Casey Schaufler <casey@xxxxxxxxxxxxxxxx>
> > Signed-off-by: Roberto Sassu <roberto.sassu@xxxxxxxxxx>
> > ---
> > fs/xattr.c | 2 ++
> > include/linux/ima.h | 18 ++++++++++++++++++
> > security/integrity/ima/ima_appraise.c | 25 ++++++++++++++++++++++---
> > security/security.c | 1 +
> > 4 files changed, 43 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/xattr.c b/fs/xattr.c
> > index b3444e06cded..81847f132d26 100644
> > --- a/fs/xattr.c
> > +++ b/fs/xattr.c
> > @@ -16,6 +16,7 @@
> > #include <linux/namei.h>
> > #include <linux/security.h>
> > #include <linux/evm.h>
> > +#include <linux/ima.h>
> > #include <linux/syscalls.h>
> > #include <linux/export.h>
> > #include <linux/fsnotify.h>
> > @@ -502,6 +503,7 @@ __vfs_removexattr_locked(struct user_namespace
> > *mnt_userns,
> >
> > if (!error) {
> > fsnotify_xattr(dentry);
> > + ima_inode_post_removexattr(dentry, name);
> > evm_inode_post_removexattr(dentry, name);
> > }
> >
> > diff --git a/include/linux/ima.h b/include/linux/ima.h
> > index 61d5723ec303..5e059da43857 100644
> > --- a/include/linux/ima.h
> > +++ b/include/linux/ima.h
> > @@ -171,7 +171,13 @@ extern void ima_inode_post_setattr(struct
> > user_namespace *mnt_userns,
> > struct dentry *dentry);
> > extern int ima_inode_setxattr(struct dentry *dentry, const char *xattr_name,
> > const void *xattr_value, size_t xattr_value_len);
> > +extern void ima_inode_post_setxattr(struct dentry *dentry,
> > + const char *xattr_name,
> > + const void *xattr_value,
> > + size_t xattr_value_len);
> > extern int ima_inode_removexattr(struct dentry *dentry, const char
> > *xattr_name);
> > +extern void ima_inode_post_removexattr(struct dentry *dentry,
> > + const char *xattr_name);
> > #else
> > static inline bool is_ima_appraise_enabled(void)
> > {
> > @@ -192,11 +198,23 @@ static inline int ima_inode_setxattr(struct dentry
> > *dentry,
> > return 0;
> > }
> >
> > +static inline void ima_inode_post_setxattr(struct dentry *dentry,
> > + const char *xattr_name,
> > + const void *xattr_value,
> > + size_t xattr_value_len)
> > +{
> > +}
> > +
> > static inline int ima_inode_removexattr(struct dentry *dentry,
> > const char *xattr_name)
> > {
> > return 0;
> > }
> > +
> > +static inline void ima_inode_post_removexattr(struct dentry *dentry,
> > + const char *xattr_name)
> > +{
> > +}
> > #endif /* CONFIG_IMA_APPRAISE */
> >
> > #if defined(CONFIG_IMA_APPRAISE) &&
> > defined(CONFIG_INTEGRITY_TRUSTED_KEYRING)
> > diff --git a/security/integrity/ima/ima_appraise.c
> > b/security/integrity/ima/ima_appraise.c
> > index 565e33ff19d0..1f029e4c8d7f 100644
> > --- a/security/integrity/ima/ima_appraise.c
> > +++ b/security/integrity/ima/ima_appraise.c
> > @@ -577,21 +577,40 @@ int ima_inode_setxattr(struct dentry *dentry, const
> > char *xattr_name,
> > if (result == 1) {
> > if (!xattr_value_len || (xvalue->type >= IMA_XATTR_LAST))
> > return -EINVAL;
> > - ima_reset_appraise_flags(d_backing_inode(dentry),
> > - xvalue->type == EVM_IMA_XATTR_DIGSIG);
> > result = 0;
> > }
> > return result;
> > }
> >
> > +void ima_inode_post_setxattr(struct dentry *dentry, const char
> > *xattr_name,
> > + const void *xattr_value, size_t xattr_value_len)
> > +{
> > + const struct evm_ima_xattr_data *xvalue = xattr_value;
> > + int result;
> > +
> > + result = ima_protect_xattr(dentry, xattr_name, xattr_value,
> > + xattr_value_len);
> > + if (result == 1)
> > + ima_reset_appraise_flags(d_backing_inode(dentry),
>
> I found an issue in this patch.
>
> Moving ima_reset_appraise_flags() to the post hook causes this
> function to be executed also when __vfs_setxattr_noperm() is
> called.
>
> The problem is that at the end of a write IMA calls
> ima_collect_measurement() to recalculate the file digest and
> update security.ima. ima_collect_measurement() sets
> IMA_COLLECTED.
>
> However, after that __vfs_setxattr_noperm() causes
> IMA_COLLECTED to be reset, and to unnecessarily recalculate
> the file digest. This wouldn't happen if ima_reset_appraise_flags()
> is in the pre hook.
>
> I solved by replacing:
> iint->flags &= ~IMA_DONE_MASK;
> with:
> iint->flags &= ~(IMA_DONE_MASK & ~IMA_COLLECTED);
>
> just when the IMA_CHANGE_XATTR bit is set. It should
> not be a problem since setting an xattr does not influence
> the file content.
>
> Mimi, what do you think?
Thank yor for noticing this.
Without seeing the actual change it is hard to tell. The only place
that "iint->flags &= ~IMA_DONE_MASK;" occurs is in neither of the above
functions, but in process_measurement(). There it is a part of a
compound "if" statement. Perhaps it would be ok to change it for just
the IMA_CHANGE_XATTR test, but definitely not for the other conditions,
like untrusted mounts.
Moving ima_reset_appraise_flags() to the post hooks is to minimize
resetting the flags unnecessarily. That is really a performance fix,
not something necessary for making the EVM portable & immutable
signatures more usable. As much as possible, please minimize the
changes to facilitate review and testing.
thanks,
Mimi
