[PATCH] Enable splitting the logs to both auditd and kernel simultaneously
Eric Paris
eparis at parisplace.org
Fri May 24 20:03:16 UTC 2013
A memset to 0 is adequate if you want to be able to turn features back
off. You have to know if the 0 or 1 is meaningful.
Without the mask userland would have to read the features from the
kernel, twiddle whichever bits they want to twiddle, then write the
features back to the kernel. With the mask userspace can specify
which bits they are twiddling and which bits the kernel just shouldn't
touch. So the read before write isn't necessary and removes that
little race.
I thought it made setting/unsetting bits easier on the userspace side.
I can get rid of the mask if you think the read/twiddle/write option
is better than set-mask/write
-Eric
On Fri, May 24, 2013 at 3:33 PM, William Roberts
<bill.c.roberts at gmail.com> wrote:
> Well when I was adding to the AUDIT MSG types, I felt pretty dirty :-P
>
> I think versioning and bitpacking the features is the right way to go.
>
> Whats the purpose of mask in audit_features struct?
>
> Looks like you use it to indicate which bits userspace intended to set,
> incase the structure wasn't memset to 0?
>
> Bill
>
>
> On Fri, May 24, 2013 at 9:34 AM, Eric Paris <eparis at parisplace.org> wrote:
>>
>> On Wed, 2013-05-22 at 19:18 -0700, William Roberts wrote:
>> > This need came about in Samsung's mobile products. Id like to see and
>> > attempt to get this mainstreamed. Any feedback would be greatly
>> > appreciated. Thanks.
>>
>> My first thought is that I just sent a patch to introduce a flexible
>> on/off flipping of bits for audit system features. Rather timely, so we
>> don't have to keep introducing message types. AUDIT_SET/AUDIT_GET
>> certainly weren't well thought out in that way. I'd like your thoughts
>> on those patches (really 1/7 does the heavy lifting 6/7 and 7/7 use
>> it)
>>
>> I thinking porting on top of my series will result in your patch being
>> about 10 lines.
>>
>> I also don't love the name. This isn't about splitting logs, it's about
>> also logging to kmsg.
>>
>> Do we need the Kconfig default? I'd think we should be able to get this
>> set early enough in userspace. Any messages before auditd are going to
>> go to kmsg. I envision that we should be able to set features in the
>> audit.rules so you should be able to turn this on before auditd starts
>> and things would stop going to kmsg.
>>
>> But I don't see a problem with it. If you port on top of this series
>> I'm ok with queuing for 3.11
>>
>> > On Wed, May 22, 2013 at 7:16 PM, William Roberts
>> > <bill.c.roberts at gmail.com> wrote:
>> > Allow the audit subsystem to send audit events to both the
>> > kernel
>> > message buffer and auditd at the same time.
>> >
>> > Signed-off-by: William Roberts <w.roberts at sta.samsung.com>
>> > ---
>> > include/uapi/linux/audit.h | 11 ++++++++++
>> > init/Kconfig | 8 +++++++
>> > kernel/audit.c | 52
>> > ++++++++++++++++++++++++++++++++++++++++------
>> > kernel/audit.h | 6 ++++++
>> > 4 files changed, 71 insertions(+), 6 deletions(-)
>> >
>> > diff --git a/include/uapi/linux/audit.h
>> > b/include/uapi/linux/audit.h
>> > index 75cef3f..0af5d78 100644
>> > --- a/include/uapi/linux/audit.h
>> > +++ b/include/uapi/linux/audit.h
>> > @@ -68,6 +68,8 @@
>> > #define AUDIT_MAKE_EQUIV 1015 /* Append to watched
>> > tree */
>> > #define AUDIT_TTY_GET 1016 /* Get TTY auditing
>> > status */
>> > #define AUDIT_TTY_SET 1017 /* Set TTY auditing
>> > status */
>> > +#define AUDIT_LOGSPLIT_GET 1018 /* Get logsplit
>> > status */
>> > +#define AUDIT_LOGSPLIT_SET 1019 /* Set logsplit
>> > status */
>> >
>> > #define AUDIT_FIRST_USER_MSG 1100 /* Userspace messages
>> > mostly uninteresting to kernel */
>> > #define AUDIT_USER_AVC 1107 /* We filter this
>> > differently */
>> > @@ -321,6 +323,10 @@ enum {
>> > #define AUDIT_FAIL_PRINTK 1
>> > #define AUDIT_FAIL_PANIC 2
>> >
>> > +/* Audit splitlog options */
>> > +#define AUDIT_LOGSPLIT_OFF 0
>> > +#define AUDIT_LOGSPLIT_ON 1
>> > +
>> > /* distinguish syscall tables */
>> > #define __AUDIT_ARCH_64BIT 0x80000000
>> > #define __AUDIT_ARCH_LE 0x40000000
>> > @@ -374,6 +380,11 @@ struct audit_tty_status {
>> > __u32 log_passwd; /* 1 = enabled, 0 =
>> > disabled */
>> > };
>> >
>> > +struct audit_logsplit_status {
>> > + __u32 enabled; /* AUDIT_LOGSPLIT_ON
>> > or
>> > +
>> > AUDIT_LOGSPLIT_OFF */
>> > +};
>> > +
>> > /* audit_rule_data supports filter rules with both integer
>> > and string
>> > * fields. It corresponds with AUDIT_ADD_RULE,
>> > AUDIT_DEL_RULE and
>> > * AUDIT_LIST_RULES requests.
>> > diff --git a/init/Kconfig b/init/Kconfig
>> > index 9d3a788..52efca6 100644
>> > --- a/init/Kconfig
>> > +++ b/init/Kconfig
>> > @@ -286,6 +286,14 @@ config AUDIT_LOGINUID_IMMUTABLE
>> > one to drop potentially dangerous capabilites from
>> > the login tasks,
>> > but may not be backwards compatible with older init
>> > systems.
>> >
>> > +config AUDIT_SPLITLOG
>> > + bool "Split audit messages to userspace and kernel by
>> > default"
>> > + depends on AUDIT
>> > + help
>> > + Setting this to true will cause the audit messages
>> > to be split
>> > + by default to both the kernel message log and a
>> > userspace audit
>> > + daemon if registered.
>> > +
>> > source "kernel/irq/Kconfig"
>> > source "kernel/time/Kconfig"
>> >
>> > diff --git a/kernel/audit.c b/kernel/audit.c
>> > index 21c7fa6..ccd2f60 100644
>> > --- a/kernel/audit.c
>> > +++ b/kernel/audit.c
>> > @@ -88,6 +88,9 @@ static int audit_default;
>> > /* If auditing cannot proceed, audit_failure selects what
>> > happens. */
>> > static int audit_failure = AUDIT_FAIL_PRINTK;
>> >
>> > +/* Whether or not logsplit is enabled */
>> > +static int audit_logsplit = AUDIT_SPLITLOG_INIT;
>> > +
>> > /*
>> > * If audit records are to be written to the netlink socket,
>> > audit_pid
>> > * contains the pid of the auditd process and
>> > audit_nlk_portid contains
>> > @@ -343,6 +346,17 @@ static int audit_set_failure(int state)
>> > return audit_do_config_change("audit_failure",
>> > &audit_failure, state);
>> > }
>> >
>> > +static int audit_set_logsplit(int state)
>> > +{
>> > + if (state != AUDIT_LOGSPLIT_OFF
>> > + && state != AUDIT_LOGSPLIT_ON)
>> > + return -EINVAL;
>> > +
>> > + return audit_do_config_change("audit_logsplit",
>> > &audit_logsplit, state);
>> > +}
>> > +
>> > +
>> > +
>> > /*
>> > * Queue skbs to be sent to auditd when/if it comes back.
>> > These skbs should
>> > * already have been sent via prink/syslog and so if these
>> > messages are dropped
>> > @@ -361,11 +375,8 @@ static void audit_hold_skb(struct sk_buff
>> > *skb)
>> > kfree_skb(skb);
>> > }
>> >
>> > -/*
>> > - * For one reason or another this nlh isn't getting delivered
>> > to the userspace
>> > - * audit daemon, just send it to printk.
>> > - */
>> > -static void audit_printk_skb(struct sk_buff *skb)
>> > +/* Just printks the skb, no audit_hold or free of any kind */
>> > +static void __audit_printk_skb(struct sk_buff *skb)
>> > {
>> > struct nlmsghdr *nlh = nlmsg_hdr(skb);
>> > char *data = nlmsg_data(nlh);
>> > @@ -376,7 +387,14 @@ static void audit_printk_skb(struct
>> > sk_buff *skb)
>> > else
>> > audit_log_lost("printk limit exceeded
>> > \n");
>> > }
>> > -
>> > +}
>> > +/*
>> > + * For one reason or another this nlh isn't getting delivered
>> > to the userspace
>> > + * audit daemon, just send it to printk.
>> > + */
>> > +static void audit_printk_skb(struct sk_buff *skb)
>> > +{
>> > + __audit_printk_skb(skb);
>> > audit_hold_skb(skb);
>> > }
>> >
>> > @@ -590,6 +608,8 @@ static int audit_netlink_ok(struct sk_buff
>> > *skb, u16 msg_type)
>> > case AUDIT_SIGNAL_INFO:
>> > case AUDIT_TTY_GET:
>> > case AUDIT_TTY_SET:
>> > + case AUDIT_LOGSPLIT_GET:
>> > + case AUDIT_LOGSPLIT_SET:
>> > case AUDIT_TRIM:
>> > case AUDIT_MAKE_EQUIV:
>> > if (!capable(CAP_AUDIT_CONTROL))
>> > @@ -843,7 +863,24 @@ static int audit_receive_msg(struct
>> > sk_buff *skb, struct nlmsghdr *nlh)
>> > spin_unlock(&tsk->sighand->siglock);
>> > break;
>> > }
>> > + case AUDIT_LOGSPLIT_GET: {
>> > + struct audit_logsplit_status s;
>> > + s.enabled = audit_logsplit;
>> > + audit_send_reply(NETLINK_CB(skb).portid, seq,
>> > + AUDIT_LOGSPLIT_GET, 0, 0, &s,
>> > sizeof(s));
>> > + break;
>> > + }
>> > + case AUDIT_LOGSPLIT_SET: {
>> > + struct audit_logsplit_status *s;
>> > + if (nlh->nlmsg_len < sizeof(struct
>> > audit_logsplit_status))
>> > + return -EINVAL;
>> > + s = data;
>> > + err = audit_set_logsplit(s->enabled);
>> > + break;
>> > + }
>> > +
>> > default:
>> > + printk(KERN_ERR "Unkown audit command");
>> > err = -EINVAL;
>> > break;
>> > }
>> > @@ -1670,6 +1707,9 @@ void audit_log_end(struct audit_buffer
>> > *ab)
>> > nlh->nlmsg_len = ab->skb->len - NLMSG_HDRLEN;
>> >
>> > if (audit_pid) {
>> > + if (audit_logsplit ==
>> > AUDIT_LOGSPLIT_ON)
>> > + __audit_printk_skb(ab->skb);
>> > +
>> > skb_queue_tail(&audit_skb_queue,
>> > ab->skb);
>> > wake_up_interruptible(&kauditd_wait);
>> > } else {
>> > diff --git a/kernel/audit.h b/kernel/audit.h
>> > index 1c95131..ef43bb1 100644
>> > --- a/kernel/audit.h
>> > +++ b/kernel/audit.h
>> > @@ -321,4 +321,10 @@ extern struct list_head
>> > *audit_killed_trees(void);
>> > #define audit_filter_inodes(t,c) AUDIT_DISABLED
>> > #endif
>> >
>> > +#ifdef CONFIG_AUDIT_SPLITLOG
>> > +#define AUDIT_SPLITLOG_INIT AUDIT_LOGSPLIT_ON
>> > +#else
>> > +#define AUDIT_SPLITLOG_INIT AUDIT_LOGSPLIT_OFF
>> > +#endif
>> > +
>> > extern struct mutex audit_cmd_mutex;
>> > --
>> > 1.8.2.2
>> >
>> >
>> >
>> >
>> >
>> > --
>> > Respectfully,
>> >
>> > William C Roberts
>> >
>> >
>> > --
>> > Linux-audit mailing list
>> > Linux-audit at redhat.com
>> > https://www.redhat.com/mailman/listinfo/linux-audit
>>
>>
>
>
>
> --
> Respectfully,
>
> William C Roberts
>
More information about the Linux-audit
mailing list