Provide a method to filter out sockaddr and bind calls by network address family. Existing SOCKADDR records are listed for any network activity. Implement the AUDIT_SADDR_FAM field selector to be able to classify or limit records to specific network address families, such as AF_INET or AF_INET6. An example of a network record that is unlikely to be useful and flood the logs: type=SOCKADDR msg=audit(07/27/2017 12:18:27.019:845) : saddr={ fam=local path=/var/run/nscd/socket } type=SYSCALL msg=audit(07/27/2017 12:18:27.019:845) : arch=x86_64 syscall=connect success=no exit=ENOENT(No such file or directory) a0=0x3 a1=0x7fff229c4980 a2=0x6e a3=0x6 items=1 ppid=3301 pid=6145 auid=sgrubb uid=sgrubb gid=sgrubb euid=sgrubb suid=sgrubb fsuid=sgrubb egid=sgrubb sgid=sgrubb fsgid=sgrubb tty=pts3 ses=4 comm=bash exe=/usr/bin/bash subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 key=network-test Please see the audit-testsuite PR at https://github.com/linux-audit/audit-testsuite/pull/87 Please see the github issue https://github.com/linux-audit/audit-kernel/issues/64 Please see the github issue for the accompanying userspace support https://github.com/linux-audit/audit-userspace/issues/93 Signed-off-by: Richard Guy Briggs <rgb(a)redhat.com&gt; --- Changelog: v2: - rebase on ghak73 v2 - check for valid range of saddr_fam value include/uapi/linux/audit.h | 1 + kernel/auditfilter.c | 5 +++++ kernel/auditsc.c | 5 +++++ 3 files changed, 11 insertions(+) diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h index a1280af20336..c89c6495983d 100644 --- a/include/uapi/linux/audit.h +++ b/include/uapi/linux/audit.h @@ -281,6 +281,7 @@ #define AUDIT_OBJ_GID 110 #define AUDIT_FIELD_COMPARE 111 #define AUDIT_EXE 112 +#define AUDIT_SADDR_FAM 113 #define AUDIT_ARG0 200 #define AUDIT_ARG1 (AUDIT_ARG0+1) diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c index 5beb2244d5ba..4c897281beb8 100644 --- a/kernel/auditfilter.c +++ b/kernel/auditfilter.c @@ -387,6 +387,7 @@ static int audit_field_valid(struct audit_entry *entry, struct audit_field *f) case AUDIT_SUCCESS: case AUDIT_INODE: case AUDIT_SESSIONID: + case AUDIT_SADDR_FAM: /* bit ops are only useful on syscall args */ if (f->op == Audit_bitmask || f->op == Audit_bittest) return -EINVAL; @@ -438,6 +439,10 @@ static int audit_field_valid(struct audit_entry *entry, struct audit_field *f) if (f->val > AUDIT_MAX_FIELD_COMPARE) return -EINVAL; break; + case AUDIT_SADDR_FAM: + if (f->val <= AF_UNSPEC || f->val >= AF_MAX)

+ return -EINVAL; + break; default: break; } diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 4bd0ec60a0e8..aab364804b9b 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -623,6 +623,11 @@ static int audit_filter_rules(struct task_struct *tsk, case AUDIT_LOGINUID_SET: result = audit_comparator(audit_loginuid_set(tsk), f->op, f->val); break; + case AUDIT_SADDR_FAM: + if (ctx->sockaddr) + result = audit_comparator(ctx->sockaddr->ss_family, + f->op, f->val); + break; case AUDIT_SUBJ_USER: case AUDIT_SUBJ_ROLE: case AUDIT_SUBJ_TYPE: -- 1.8.3.1

On 2019-05-08 18:05, Paul Moore wrote: > On Wed, May 8, 2019 at 12:46 PM Richard Guy Briggs <rgb(a)redhat.com&gt; wrote: > > > > Provide a method to filter out sockaddr and bind calls by network > > address family. > > > > Existing SOCKADDR records are listed for any network activity. > > Implement the AUDIT_SADDR_FAM field selector to be able to classify or > > limit records to specific network address families, such as AF_INET or > > AF_INET6. > > > > An example of a network record that is unlikely to be useful and flood > > the logs: > > > > type=SOCKADDR msg=audit(07/27/2017 12:18:27.019:845) : saddr={ fam=local > > path=/var/run/nscd/socket } > > type=SYSCALL msg=audit(07/27/2017 12:18:27.019:845) : arch=x86_64 > > syscall=connect success=no exit=ENOENT(No such file or directory) a0=0x3 > > a1=0x7fff229c4980 a2=0x6e a3=0x6 items=1 ppid=3301 pid=6145 auid=sgrubb > > uid=sgrubb gid=sgrubb euid=sgrubb suid=sgrubb fsuid=sgrubb egid=sgrubb > > sgid=sgrubb fsgid=sgrubb tty=pts3 ses=4 comm=bash exe=/usr/bin/bash > > subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 > > key=network-test > > > > Please see the audit-testsuite PR at > > https://github.com/linux-audit/audit-testsuite/pull/87 > > Please see the github issue > > https://github.com/linux-audit/audit-kernel/issues/64 > > Please see the github issue for the accompanying userspace support > > https://github.com/linux-audit/audit-userspace/issues/93 > > > > Signed-off-by: Richard Guy Briggs <rgb(a)redhat.com&gt; > > --- > > Changelog: > > v2: > > - rebase on ghak73 v2 > > - check for valid range of saddr_fam value > > > > include/uapi/linux/audit.h | 1 + > > kernel/auditfilter.c | 5 +++++ > > kernel/auditsc.c | 5 +++++ > > 3 files changed, 11 insertions(+) > > > > diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h > > index a1280af20336..c89c6495983d 100644 > > --- a/include/uapi/linux/audit.h > > +++ b/include/uapi/linux/audit.h > > @@ -281,6 +281,7 @@ > > #define AUDIT_OBJ_GID 110 > > #define AUDIT_FIELD_COMPARE 111 > > #define AUDIT_EXE 112 > > +#define AUDIT_SADDR_FAM 113 > > > > #define AUDIT_ARG0 200 > > #define AUDIT_ARG1 (AUDIT_ARG0+1) > > diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c > > index 5beb2244d5ba..4c897281beb8 100644 > > --- a/kernel/auditfilter.c > > +++ b/kernel/auditfilter.c > > @@ -387,6 +387,7 @@ static int audit_field_valid(struct audit_entry *entry, struct audit_field *f) > > case AUDIT_SUCCESS: > > case AUDIT_INODE: > > case AUDIT_SESSIONID: > > + case AUDIT_SADDR_FAM: > > /* bit ops are only useful on syscall args */ > > if (f->op == Audit_bitmask || f->op == Audit_bittest) > > return -EINVAL; > > @@ -438,6 +439,10 @@ static int audit_field_valid(struct audit_entry *entry, struct audit_field *f) > > if (f->val > AUDIT_MAX_FIELD_COMPARE) > > return -EINVAL; > > break; > > + case AUDIT_SADDR_FAM: > > + if (f->val <= AF_UNSPEC || f->val >= AF_MAX) > > AF_UNSPEC is a valid address family for some operations, and while I'm > not sure what value there is in an auditing these events, we should > allow it. It's also worth noting that f->val is an unsigned value so > we are never going to see a value less than AF_UNSPEC/0. This is why > on your earlier patch I only talked about AF_MAX and not AF_UNSPEC. On reflection, I agree about the unsigned nature of f->val. I'm also prepared to allow AF_UNSPEC/0. However, I'm having trouble reconciling your last sentence with the previous thread in which you explicitly mention AF_UNSPEC and sanity checks related to it: (Am I losing it, or misunderstanding?)