[PATCH] filter: add filesystem filter with fstype
Richard Guy Briggs
rgb at redhat.com
Mon Jun 19 16:56:17 UTC 2017
On 2017-06-19 12:46, Steve Grubb wrote:
> On Friday, June 16, 2017 4:29:46 PM EDT Richard Guy Briggs wrote:
> > Tracefs or debugfs were causing hundreds to thousands of PATH records to
> > be associated with the init_module and finit_module SYSCALL records on a
> > few modules when the following rule was in place for startup:
> > -a always,exit -F arch=x86_64 -S init_module -F key=mod-load
> >
> > Add the new "filesystem" filter list anchored in __audit_inode_child() to
> > filter out PATH records from uninteresting filesystem types, "fstype",
> > keying on their kernel hexadecimal 4-octet magic identifier.
> >
> > An example rule would look like:
> > -a never,filesystem -F fstype=0x74726163 -F key=ignore_tracefs
> > -a never,filesystem -F fstype=0x64626720 -F key=ignore_debugfs
> >
> > Note: "always,filesystem" will log the PATH record anyways and add latency.
> >
> > See: https://github.com/linux-audit/audit-kernel/issues/8
> > Test case: https://github.com/linux-audit/audit-testsuite/issues/42
> >
> > Signed-off-by: Richard Guy Briggs <rgb at redhat.com>
> > ---
> > docs/audit_add_rule_data.3 | 3 +++
> > lib/errormsg.h | 5 +++++
> > lib/fieldtab.h | 2 ++
> > lib/flagtab.h | 10 ++++++----
> > lib/libaudit.c | 26 ++++++++++++++++++++++++--
> > lib/libaudit.h | 10 ++++++++++
> > lib/private.h | 1 +
> > src/auditctl-listing.c | 6 ++++--
> > src/auditctl.c | 16 ++++++++++++++--
> > 9 files changed, 69 insertions(+), 10 deletions(-)
> >
> > diff --git a/docs/audit_add_rule_data.3 b/docs/audit_add_rule_data.3
> > index a0802c0..1e7540c 100644
> > --- a/docs/audit_add_rule_data.3
> > +++ b/docs/audit_add_rule_data.3
> > @@ -22,6 +22,9 @@ AUDIT_FILTER_EXIT - Apply rule at syscall exit. This is
> > the main filter that is .TP
> > \(bu
> > AUDIT_FILTER_TYPE - Apply rule at audit_log_start. This is the exclude
> > filter which discards any records that match. +.TP
> > +\(bu
> > +AUDIT_FILTER_FS - Apply rule when adding PATH auxiliary records to SYSCALL
> > events. This is the filesystem filter. This is used to ignore PATH records
> > that are not of interest. .LP
> >
> > .PP
> > diff --git a/lib/errormsg.h b/lib/errormsg.h
> > index 91d8252..ef54589 100644
> > --- a/lib/errormsg.h
> > +++ b/lib/errormsg.h
> > @@ -20,6 +20,7 @@
> > * Authors:
> > * Zhang Xiliang <zhangxiliang at cn.fujitsu.com>
> > * Steve Grubb <sgrubb at redhat.com>
> > + * Richard Guy Briggs <rgb at redhat.com>
> > */
> >
> > struct msg_tab {
> > @@ -66,6 +67,8 @@ struct msg_tab {
> > #define EAU_FIELDNOFILTER 31
> > #define EAU_FILTERMISSING 32
> > #define EAU_COMPINCOMPAT 33
> > +#define EAU_FIELDUNAVAIL 34
> > +#define EAU_FILTERNOSUPPORT 35
> > static const struct msg_tab err_msgtab[] = {
> > { -EAU_OPMISSING, 2, "-F missing operation for" },
> > { -EAU_FIELDUNKNOWN, 2, "-F unknown field:" },
> > @@ -100,5 +103,7 @@ static const struct msg_tab err_msgtab[] = {
> > { -EAU_FIELDNOFILTER, 1, "must be used with exclude, user, or exit
> > filter" }, { -EAU_FILTERMISSING, 0, "filter is missing from rule" },
> > { -EAU_COMPINCOMPAT, 2, "-C incompatible comparison" },
> > + { -EAU_FIELDUNAVAIL, 1, "field is not valid for the filter" },
> > + { -EAU_FILTERNOSUPPORT, 1, "filter is not supported ty kernel" },
> > };
> > #endif
> > diff --git a/lib/fieldtab.h b/lib/fieldtab.h
> > index 0c5e39d..c425d5b 100644
> > --- a/lib/fieldtab.h
> > +++ b/lib/fieldtab.h
> > @@ -18,6 +18,7 @@
> > *
> > * Authors:
> > * Steve Grubb <sgrubb at redhat.com>
> > + * Richard Guy Briggs <rgb at redhat.com>
> > */
> >
> > _S(AUDIT_PID, "pid" )
> > @@ -56,6 +57,7 @@ _S(AUDIT_WATCH, "path" )
> > _S(AUDIT_PERM, "perm" )
> > _S(AUDIT_DIR, "dir" )
> > _S(AUDIT_FILETYPE, "filetype" )
> > +_S(AUDIT_FSTYPE, "fstype" )
> > _S(AUDIT_OBJ_UID, "obj_uid" )
> > _S(AUDIT_OBJ_GID, "obj_gid" )
> > _S(AUDIT_FIELD_COMPARE, "field_compare" )
> > diff --git a/lib/flagtab.h b/lib/flagtab.h
> > index 4b04692..7a618e0 100644
> > --- a/lib/flagtab.h
> > +++ b/lib/flagtab.h
> > @@ -18,8 +18,10 @@
> > *
> > * Authors:
> > * Steve Grubb <sgrubb at redhat.com>
> > + * Richard Guy Briggs <rgb at redhat.com>
> > */
> > -_S(AUDIT_FILTER_TASK, "task" )
> > -_S(AUDIT_FILTER_EXIT, "exit" )
> > -_S(AUDIT_FILTER_USER, "user" )
> > -_S(AUDIT_FILTER_EXCLUDE, "exclude" )
> > +_S(AUDIT_FILTER_TASK, "task" )
> > +_S(AUDIT_FILTER_EXIT, "exit" )
> > +_S(AUDIT_FILTER_USER, "user" )
> > +_S(AUDIT_FILTER_EXCLUDE, "exclude" )
>
> Why are these ^^^ in the patch?
To re-align the closing round bracket with the longer "filesystem" line
that follows.
> > +_S(AUDIT_FILTER_FS, "filesystem")
> > diff --git a/lib/libaudit.c b/lib/libaudit.c
> > index 18cd384..34677b9 100644
> > --- a/lib/libaudit.c
> > +++ b/lib/libaudit.c
> > @@ -19,6 +19,7 @@
> > * Authors:
> > * Steve Grubb <sgrubb at redhat.com>
> > * Rickard E. (Rik) Faith <faith at redhat.com>
> > + * Richard Guy Briggs <rgb at redhat.com>
> > */
> >
> > #include "config.h"
> > @@ -85,6 +86,7 @@ int _audit_permadded = 0;
> > int _audit_archadded = 0;
> > int _audit_syscalladded = 0;
> > int _audit_exeadded = 0;
> > +int _audit_filterfsadded = 0;
> > unsigned int _audit_elf = 0U;
> > static struct libaudit_conf config;
> >
> > @@ -1466,6 +1468,23 @@ int audit_rule_fieldpair_data(struct audit_rule_data
> > **rulep, const char *pair, }
> > }
> >
> > + /* FS filter can be used only with FSTYPE field */
> > + if (flags == AUDIT_FILTER_FS) {
> > + uint32_t features = audit_get_features();
> > + if ((features & AUDIT_FEATURE_BITMAP_FILTER_PATH) == 0) {
>
> Doesn't the kernel feature bitmap names need updating?
Yes, I missed this one.
> > + return -EAU_FILTERNOSUPPORT;
> > + } else {
> > + switch(field) {
> > + case AUDIT_FSTYPE:
> > + _audit_filterfsadded = 1;
> > + case AUDIT_FILTERKEY:
> > + break;
> > + default:
> > + return -EAU_FIELDUNAVAIL;
> > + }
> > + }
> > + }
> > +
> > rule->fields[rule->field_count] = field;
> > rule->fieldflags[rule->field_count] = op;
> > switch (field)
> > @@ -1580,7 +1599,8 @@ int audit_rule_fieldpair_data(struct audit_rule_data
> > **rulep, const char *pair, }
> > if (field == AUDIT_FILTERKEY &&
> > !(_audit_syscalladded || _audit_permadded ||
> > - _audit_exeadded))
> > + _audit_exeadded ||
> > + _audit_filterfsadded))
> > return -EAU_KEYDEP;
> > vlen = strlen(v);
> > if (field == AUDIT_FILTERKEY &&
> > @@ -1715,7 +1735,7 @@ int audit_rule_fieldpair_data(struct audit_rule_data
> > **rulep, const char *pair, return -EAU_EXITONLY;
> > /* fallthrough */
> > default:
> > - if (field == AUDIT_INODE) {
> > + if (field == AUDIT_INODE || field == AUDIT_FSTYPE) {
> > if (!(op == AUDIT_NOT_EQUAL ||
> > op == AUDIT_EQUAL))
> > return -EAU_OPEQNOTEQ;
> > @@ -1727,6 +1747,8 @@ int audit_rule_fieldpair_data(struct audit_rule_data
> > **rulep, const char *pair, if (!isdigit((char)*(v)))
> > return -EAU_FIELDVALNUM;
> >
> > + if (field == AUDIT_FSTYPE && flags != AUDIT_FILTER_FS)
> > + return -EAU_FIELDUNAVAIL;
> > if (field == AUDIT_INODE)
> > rule->values[rule->field_count] =
> > strtoul(v, NULL, 0);
> > diff --git a/lib/libaudit.h b/lib/libaudit.h
> > index e5c7a4d..d7336fd 100644
> > --- a/lib/libaudit.h
> > +++ b/lib/libaudit.h
> > @@ -277,6 +277,9 @@ extern "C" {
> > #define AUDIT_KEY_SEPARATOR 0x01
> >
> > /* These are used in filter control */
> > +#ifndef AUDIT_FILTER_FS
> > +#define AUDIT_FILTER_FS 0x06 /* FS record filter in __audit_inode_child
> */
> > +#endif
> > #define AUDIT_FILTER_EXCLUDE AUDIT_FILTER_TYPE
> > #define AUDIT_FILTER_MASK 0x07 /* Mask to get actual filter */
> > #define AUDIT_FILTER_UNSET 0x80 /* This value means filter is unset */
> > @@ -305,6 +308,9 @@ extern "C" {
> > #ifndef AUDIT_FEATURE_BITMAP_LOST_RESET
> > #define AUDIT_FEATURE_BITMAP_LOST_RESET 0x00000020
> > #endif
> > +#ifndef AUDIT_FEATURE_BITMAP_FILTER_PATH
> > +#define AUDIT_FEATURE_BITMAP_FILTER_PATH 0x00000040
> > +#endif
>
> Same comment. Otherwise, it looks OK.
Ok, thanks. I'll go ahead and re-spin kernel patches as well.
> -Steve
>
> > /* Defines for interfield comparison update */
> > #ifndef AUDIT_OBJ_UID
> > @@ -324,6 +330,10 @@ extern "C" {
> > #define AUDIT_SESSIONID 25
> > #endif
> >
> > +#ifndef AUDIT_FSTYPE
> > +#define AUDIT_FSTYPE 26
> > +#endif
> > +
> > #ifndef AUDIT_COMPARE_UID_TO_OBJ_UID
> > #define AUDIT_COMPARE_UID_TO_OBJ_UID 1
> > #endif
> > diff --git a/lib/private.h b/lib/private.h
> > index cde1906..bd5e8b3 100644
> > --- a/lib/private.h
> > +++ b/lib/private.h
> > @@ -139,6 +139,7 @@ extern int _audit_permadded;
> > extern int _audit_archadded;
> > extern int _audit_syscalladded;
> > extern int _audit_exeadded;
> > +extern int _audit_filterfsadded;
> > extern unsigned int _audit_elf;
> >
> > #ifdef __cplusplus
> > diff --git a/src/auditctl-listing.c b/src/auditctl-listing.c
> > index 3bc8e71..50bc0b8 100644
> > --- a/src/auditctl-listing.c
> > +++ b/src/auditctl-listing.c
> > @@ -91,7 +91,8 @@ static int is_watch(const struct audit_rule_data *r)
> >
> > if (((r->flags & AUDIT_FILTER_MASK) != AUDIT_FILTER_USER) &&
> > ((r->flags & AUDIT_FILTER_MASK) != AUDIT_FILTER_TASK) &&
> > - ((r->flags & AUDIT_FILTER_MASK) != AUDIT_FILTER_EXCLUDE)) {
> > + ((r->flags & AUDIT_FILTER_MASK) != AUDIT_FILTER_EXCLUDE) &&
> > + ((r->flags & AUDIT_FILTER_MASK) != AUDIT_FILTER_FS)) {
> > for (i = 0; i < (AUDIT_BITMASK_SIZE-1); i++) {
> > if (r->mask[i] != (uint32_t)~0) {
> > all = 0;
> > @@ -139,7 +140,8 @@ static int print_syscall(const struct audit_rule_data
> > *r, unsigned int *sc) /* Rules on the following filters do not take a
> > syscall */
> > if (((r->flags & AUDIT_FILTER_MASK) == AUDIT_FILTER_USER) ||
> > ((r->flags & AUDIT_FILTER_MASK) == AUDIT_FILTER_TASK) ||
> > - ((r->flags &AUDIT_FILTER_MASK) == AUDIT_FILTER_EXCLUDE))
> > + ((r->flags &AUDIT_FILTER_MASK) == AUDIT_FILTER_EXCLUDE) ||
> > + ((r->flags &AUDIT_FILTER_MASK) == AUDIT_FILTER_FS))
> > return 0;
> >
> > /* See if its all or specific syscalls */
> > diff --git a/src/auditctl.c b/src/auditctl.c
> > index 04765f4..b99c957 100644
> > --- a/src/auditctl.c
> > +++ b/src/auditctl.c
> > @@ -19,6 +19,7 @@
> > * Authors:
> > * Steve Grubb <sgrubb at redhat.com>
> > * Rickard E. (Rik) Faith <faith at redhat.com>
> > + * Richard Guy Briggs <rgb at redhat.com>
> > */
> >
> > #include "config.h"
> > @@ -74,6 +75,7 @@ static int reset_vars(void)
> > _audit_permadded = 0;
> > _audit_archadded = 0;
> > _audit_exeadded = 0;
> > + _audit_filterfsadded = 0;
> > _audit_elf = 0;
> > add = AUDIT_FILTER_UNSET;
> > del = AUDIT_FILTER_UNSET;
> > @@ -151,6 +153,8 @@ static int lookup_filter(const char *str, int *filter)
> > *filter = AUDIT_FILTER_EXIT;
> > else if (strcmp(str, "user") == 0)
> > *filter = AUDIT_FILTER_USER;
> > + else if (strcmp(str, "filesystem") == 0)
> > + *filter = AUDIT_FILTER_FS;
> > else if (strcmp(str, "exclude") == 0) {
> > *filter = AUDIT_FILTER_EXCLUDE;
> > exclude = 1;
> > @@ -760,6 +764,13 @@ static int setopt(int count, int lineno, char *vars[])
> > audit_msg(LOG_ERR,
> > "Error: syscall auditing being added to user list");
> > return -1;
> > + } else if (((add & (AUDIT_FILTER_MASK|AUDIT_FILTER_UNSET)) ==
> > + AUDIT_FILTER_FS || (del &
> > + (AUDIT_FILTER_MASK|AUDIT_FILTER_UNSET)) ==
> > + AUDIT_FILTER_FS)) {
> > + audit_msg(LOG_ERR,
> > + "Error: syscall auditing being added to filesystem list");
> > + return -1;
> > } else if (exclude) {
> > audit_msg(LOG_ERR,
> > "Error: syscall auditing cannot be put on exclude list");
> > @@ -936,8 +947,9 @@ static int setopt(int count, int lineno, char *vars[])
> > break;
> > case 'k':
> > if (!(_audit_syscalladded || _audit_permadded ||
> > - _audit_exeadded) || (add==AUDIT_FILTER_UNSET &&
> > - del==AUDIT_FILTER_UNSET)) {
> > + _audit_exeadded ||
> > + _audit_filterfsadded) ||
> > + (add==AUDIT_FILTER_UNSET && del==AUDIT_FILTER_UNSET)) {
> > audit_msg(LOG_ERR,
> > "key option needs a watch or syscall given prior to it");
> > retval = -1;
>
>
- RGB
--
Richard Guy Briggs <rgb at redhat.com>
Sr. S/W Engineer, Kernel Security, Base Operating Systems
Remote, Ottawa, Red Hat Canada
IRC: rgb, SunRaycer
Voice: +1.647.777.2635, Internal: (81) 32635
More information about the Linux-audit
mailing list