| Message ID | 20200724091520.880211-1-tweek@google.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | selinux: add tracepoint on denials | expand |
On Fri, Jul 24, 2020 at 5:15 AM Thiébaud Weksteen <tweek@google.com> wrote: > > The audit data currently captures which process and which target > is responsible for a denial. There is no data on where exactly in the > process that call occurred. Debugging can be made easier by being able to > reconstruct the unified kernel and userland stack traces [1]. Add a > tracepoint on the SELinux denials which can then be used by userland > (i.e. perf). > > Although this patch could manually be added by each OS developer to > trouble shoot a denial, adding it to the kernel streamlines the > developers workflow. > > [1] https://source.android.com/devices/tech/debug/native_stack_dump > > Signed-off-by: Thiébaud Weksteen <tweek@google.com> > Signed-off-by: Joel Fernandes <joelaf@google.com> > --- > MAINTAINERS | 1 + > include/trace/events/selinux.h | 35 ++++++++++++++++++++++++++++++++++ > security/selinux/avc.c | 6 ++++++ > 3 files changed, 42 insertions(+) > create mode 100644 include/trace/events/selinux.h > > diff --git a/MAINTAINERS b/MAINTAINERS > index e64cdde81851..6b6cd5e13537 100644 > --- a/MAINTAINERS > +++ b/MAINTAINERS > @@ -15358,6 +15358,7 @@ T: git git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git > F: Documentation/ABI/obsolete/sysfs-selinux-checkreqprot > F: Documentation/ABI/obsolete/sysfs-selinux-disable > F: Documentation/admin-guide/LSM/SELinux.rst > +F: include/trace/events/selinux.h > F: include/uapi/linux/selinux_netlink.h > F: scripts/selinux/ > F: security/selinux/ > diff --git a/include/trace/events/selinux.h b/include/trace/events/selinux.h > new file mode 100644 > index 000000000000..e247187a8135 > --- /dev/null > +++ b/include/trace/events/selinux.h > @@ -0,0 +1,35 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +#undef TRACE_SYSTEM > +#define TRACE_SYSTEM selinux > + > +#if !defined(_TRACE_SELINUX_H) || defined(TRACE_HEADER_MULTI_READ) > +#define _TRACE_SELINUX_H > + > +#include <linux/ktime.h> > +#include <linux/tracepoint.h> > + > +TRACE_EVENT(selinux_denied, > + > + TP_PROTO(int cls, int av), > + > + TP_ARGS(cls, av), > + > + TP_STRUCT__entry( > + __field(int, cls) > + __field(int, av) > + ), > + > + TP_fast_assign( > + __entry->cls = cls; > + __entry->av = av; > + ), > + > + TP_printk("denied %d %d", > + __entry->cls, > + __entry->av) > +); I would think you would want to log av as %x for easier interpretation especially when there are multiple permissions being checked at once (which can happen). Also both cls and av would properly be unsigned values. Only other question I have is whether it would be beneficial to include other information here to help uniquely identify/correlate the denial with the avc: message and whether any decoding of the class, av, or other information could/should be done here versus in some userland helper.
On Fri, 24 Jul 2020 11:15:03 +0200 "Thiébaud Weksteen" <tweek@google.com> wrote: > diff --git a/security/selinux/avc.c b/security/selinux/avc.c > index d18cb32a242a..85d2e22ab656 100644 > --- a/security/selinux/avc.c > +++ b/security/selinux/avc.c > @@ -31,6 +31,9 @@ > #include "avc_ss.h" > #include "classmap.h" > > +#define CREATE_TRACE_POINTS > +#include <trace/events/selinux.h> > + > #define AVC_CACHE_SLOTS 512 > #define AVC_DEF_CACHE_THRESHOLD 512 > #define AVC_CACHE_RECLAIM 16 > @@ -672,6 +675,9 @@ static void avc_audit_pre_callback(struct audit_buffer *ab, void *a) > return; > } > > + if (sad->denied) First, I would like to deny sadness as well ;-) Now, there is a way to add that branch within the "nop" area of the trace event, and remove the conditional branch from the main code. > + trace_selinux_denied(sad->tclass, av); > + Instead have this: trace_selinux_denied(sad, av); > perms = secclass_map[sad->tclass-1].perms; > > audit_log_format(ab, " {"); > --- /dev/null > +++ b/include/trace/events/selinux.h > @@ -0,0 +1,35 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +#undef TRACE_SYSTEM > +#define TRACE_SYSTEM selinux > + > +#if !defined(_TRACE_SELINUX_H) || defined(TRACE_HEADER_MULTI_READ) > +#define _TRACE_SELINUX_H > + > +#include <linux/ktime.h> > +#include <linux/tracepoint.h> > + > +TRACE_EVENT(selinux_denied, TRACE_EVENT_CONDITION(selinux_denied, > + > + TP_PROTO(int cls, int av), TP_PROTO(struct selinux_audit_data sad, int av) > + > + TP_ARGS(cls, av), > + TP_CONDITION(sad->denied), The above condition will be tested before calling the tracepoint. But only if the trace event is enabled. > + TP_STRUCT__entry( > + __field(int, cls) > + __field(int, av) > + ), > + > + TP_fast_assign( > + __entry->cls = cls; __entry->cls = sad->tclass; > + __entry->av = av; > + ), > + > + TP_printk("denied %d %d", > + __entry->cls, > + __entry->av) > +); > + > +#endif > + > +/* This part must be outside protection */ > +#include <trace/define_trace.h> -- Steve
On Fri, Jul 24, 2020 at 9:32 AM Stephen Smalley <stephen.smalley.work@gmail.com> wrote: > On Fri, Jul 24, 2020 at 5:15 AM Thiébaud Weksteen <tweek@google.com> wrote: > > The audit data currently captures which process and which target > > is responsible for a denial. There is no data on where exactly in the > > process that call occurred. Debugging can be made easier by being able to > > reconstruct the unified kernel and userland stack traces [1]. Add a > > tracepoint on the SELinux denials which can then be used by userland > > (i.e. perf). > > > > Although this patch could manually be added by each OS developer to > > trouble shoot a denial, adding it to the kernel streamlines the > > developers workflow. > > > > [1] https://source.android.com/devices/tech/debug/native_stack_dump > > > > Signed-off-by: Thiébaud Weksteen <tweek@google.com> > > Signed-off-by: Joel Fernandes <joelaf@google.com> > > --- > > MAINTAINERS | 1 + > > include/trace/events/selinux.h | 35 ++++++++++++++++++++++++++++++++++ > > security/selinux/avc.c | 6 ++++++ > > 3 files changed, 42 insertions(+) > > create mode 100644 include/trace/events/selinux.h > > > > diff --git a/MAINTAINERS b/MAINTAINERS > > index e64cdde81851..6b6cd5e13537 100644 > > --- a/MAINTAINERS > > +++ b/MAINTAINERS > > @@ -15358,6 +15358,7 @@ T: git git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git > > F: Documentation/ABI/obsolete/sysfs-selinux-checkreqprot > > F: Documentation/ABI/obsolete/sysfs-selinux-disable > > F: Documentation/admin-guide/LSM/SELinux.rst > > +F: include/trace/events/selinux.h > > F: include/uapi/linux/selinux_netlink.h > > F: scripts/selinux/ > > F: security/selinux/ > > diff --git a/include/trace/events/selinux.h b/include/trace/events/selinux.h > > new file mode 100644 > > index 000000000000..e247187a8135 > > --- /dev/null > > +++ b/include/trace/events/selinux.h > > @@ -0,0 +1,35 @@ > > +/* SPDX-License-Identifier: GPL-2.0 */ > > +#undef TRACE_SYSTEM > > +#define TRACE_SYSTEM selinux > > + > > +#if !defined(_TRACE_SELINUX_H) || defined(TRACE_HEADER_MULTI_READ) > > +#define _TRACE_SELINUX_H > > + > > +#include <linux/ktime.h> > > +#include <linux/tracepoint.h> > > + > > +TRACE_EVENT(selinux_denied, > > + > > + TP_PROTO(int cls, int av), > > + > > + TP_ARGS(cls, av), > > + > > + TP_STRUCT__entry( > > + __field(int, cls) > > + __field(int, av) > > + ), > > + > > + TP_fast_assign( > > + __entry->cls = cls; > > + __entry->av = av; > > + ), > > + > > + TP_printk("denied %d %d", > > + __entry->cls, > > + __entry->av) > > +); > > I would think you would want to log av as %x for easier interpretation > especially when there are multiple permissions being checked at once > (which can happen). Also both cls and av would properly be unsigned > values. Only other question I have is whether it would be beneficial > to include other information here to help uniquely identify/correlate > the denial with the avc: message and whether any decoding of the > class, av, or other information could/should be done here versus in > some userland helper. It does seem like at the very least it would be nice to see the av as hex values instead of integers, e.g. "%x" in the TP_printk() call. Considering this patch is about making dev's lives easier, I tend to agree with Stephen questioning if you should go a step further and convert both the class and av values into string representations.
Thanks for the review! I'll send a new revision of the patch with the %x formatter and using the TP_CONDITION macro. On adding further information to the trace event, I would prefer adding the strict minimum to be able to correlate the event with the avc message. The reason is that tracevents have a fixed size (see https://www.kernel.org/doc/Documentation/trace/events.txt). For instance, we would need to decide on a maximum size for the string representation of the list of permissions. This would also duplicate the reporting done in the avc audit event. I'll simply add the pid as part of the printk, which should be sufficient for the correlation. On Fri, Jul 24, 2020 at 3:55 PM Paul Moore <paul@paul-moore.com> wrote: > > On Fri, Jul 24, 2020 at 9:32 AM Stephen Smalley > <stephen.smalley.work@gmail.com> wrote: > > On Fri, Jul 24, 2020 at 5:15 AM Thiébaud Weksteen <tweek@google.com> wrote: > > > The audit data currently captures which process and which target > > > is responsible for a denial. There is no data on where exactly in the > > > process that call occurred. Debugging can be made easier by being able to > > > reconstruct the unified kernel and userland stack traces [1]. Add a > > > tracepoint on the SELinux denials which can then be used by userland > > > (i.e. perf). > > > > > > Although this patch could manually be added by each OS developer to > > > trouble shoot a denial, adding it to the kernel streamlines the > > > developers workflow. > > > > > > [1] https://source.android.com/devices/tech/debug/native_stack_dump > > > > > > Signed-off-by: Thiébaud Weksteen <tweek@google.com> > > > Signed-off-by: Joel Fernandes <joelaf@google.com> > > > --- > > > MAINTAINERS | 1 + > > > include/trace/events/selinux.h | 35 ++++++++++++++++++++++++++++++++++ > > > security/selinux/avc.c | 6 ++++++ > > > 3 files changed, 42 insertions(+) > > > create mode 100644 include/trace/events/selinux.h > > > > > > diff --git a/MAINTAINERS b/MAINTAINERS > > > index e64cdde81851..6b6cd5e13537 100644 > > > --- a/MAINTAINERS > > > +++ b/MAINTAINERS > > > @@ -15358,6 +15358,7 @@ T: git git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git > > > F: Documentation/ABI/obsolete/sysfs-selinux-checkreqprot > > > F: Documentation/ABI/obsolete/sysfs-selinux-disable > > > F: Documentation/admin-guide/LSM/SELinux.rst > > > +F: include/trace/events/selinux.h > > > F: include/uapi/linux/selinux_netlink.h > > > F: scripts/selinux/ > > > F: security/selinux/ > > > diff --git a/include/trace/events/selinux.h b/include/trace/events/selinux.h > > > new file mode 100644 > > > index 000000000000..e247187a8135 > > > --- /dev/null > > > +++ b/include/trace/events/selinux.h > > > @@ -0,0 +1,35 @@ > > > +/* SPDX-License-Identifier: GPL-2.0 */ > > > +#undef TRACE_SYSTEM > > > +#define TRACE_SYSTEM selinux > > > + > > > +#if !defined(_TRACE_SELINUX_H) || defined(TRACE_HEADER_MULTI_READ) > > > +#define _TRACE_SELINUX_H > > > + > > > +#include <linux/ktime.h> > > > +#include <linux/tracepoint.h> > > > + > > > +TRACE_EVENT(selinux_denied, > > > + > > > + TP_PROTO(int cls, int av), > > > + > > > + TP_ARGS(cls, av), > > > + > > > + TP_STRUCT__entry( > > > + __field(int, cls) > > > + __field(int, av) > > > + ), > > > + > > > + TP_fast_assign( > > > + __entry->cls = cls; > > > + __entry->av = av; > > > + ), > > > + > > > + TP_printk("denied %d %d", > > > + __entry->cls, > > > + __entry->av) > > > +); > > > > I would think you would want to log av as %x for easier interpretation > > especially when there are multiple permissions being checked at once > > (which can happen). Also both cls and av would properly be unsigned > > values. Only other question I have is whether it would be beneficial > > to include other information here to help uniquely identify/correlate > > the denial with the avc: message and whether any decoding of the > > class, av, or other information could/should be done here versus in > > some userland helper. > > It does seem like at the very least it would be nice to see the av as > hex values instead of integers, e.g. "%x" in the TP_printk() call. > Considering this patch is about making dev's lives easier, I tend to > agree with Stephen questioning if you should go a step further and > convert both the class and av values into string representations. > > -- > paul moore > www.paul-moore.com
On 7/28/20 8:49 AM, Thiébaud Weksteen wrote: > Thanks for the review! I'll send a new revision of the patch with the > %x formatter and using the TP_CONDITION macro. > > On adding further information to the trace event, I would prefer > adding the strict minimum to be able to correlate the event with the > avc message. The reason is that tracevents have a fixed size (see > https://www.kernel.org/doc/Documentation/trace/events.txt). For > instance, we would need to decide on a maximum size for the string > representation of the list of permissions. This would also duplicate > the reporting done in the avc audit event. I'll simply add the pid as > part of the printk, which should be sufficient for the correlation. Ok, also please use unsigned int for the fields and %u for the cls value. (btw top-posting is discouraged for mailing list discussions, see http://vger.kernel.org/lkml/#s3-9)
On Tue, 28 Jul 2020 14:49:24 +0200 Thiébaud Weksteen <tweek@google.com> wrote: > Thanks for the review! I'll send a new revision of the patch with the > %x formatter and using the TP_CONDITION macro. > > On adding further information to the trace event, I would prefer > adding the strict minimum to be able to correlate the event with the > avc message. The reason is that tracevents have a fixed size (see > https://www.kernel.org/doc/Documentation/trace/events.txt). For Wait! What? Where in that document does it say that trace events have a fixed size. We have a lot of dynamically sized trace events. > instance, we would need to decide on a maximum size for the string > representation of the list of permissions. This would also duplicate > the reporting done in the avc audit event. I'll simply add the pid as > part of the printk, which should be sufficient for the correlation. > Please take a look at samples/trace_events/trace_events_sample.h and read the example on __print_symbolic(). I think that's what you are looking for. -- Steve
On Tue, Jul 28, 2020 at 3:04 PM Stephen Smalley
<stephen.smalley.work@gmail.com> wrote:
> Ok, also please use unsigned int for the fields and %u for the cls value.
Will do in v3. Thanks.
On Tue, Jul 28, 2020 at 3:12 PM Steven Rostedt <rostedt@goodmis.org> wrote: > Where in that document does it say that trace events have a fixed size. > We have a lot of dynamically sized trace events. My mistake. From the "format" pseudo-file, I assumed the offset and size were fixed. > Please take a look at samples/trace_events/trace_events_sample.h > and read the example on __print_symbolic(). > I think that's what you are looking for. Ack, thanks for pointing these out. I still think that my other argument (i.e. duplication of avc message) holds.
On Tue, Jul 28, 2020 at 8:49 AM Thiébaud Weksteen <tweek@google.com> wrote: > > Thanks for the review! I'll send a new revision of the patch with the > %x formatter and using the TP_CONDITION macro. > > On adding further information to the trace event, I would prefer > adding the strict minimum to be able to correlate the event with the > avc message. The reason is that tracevents have a fixed size (see > https://www.kernel.org/doc/Documentation/trace/events.txt). For > instance, we would need to decide on a maximum size for the string > representation of the list of permissions. It sounds like this is no longer an issue, hopefully this changes your thinking as I'm not sure how usable it would be in practice for users not overly familiar with SELinux. Perhaps it would be helpful if you provided an example of how one would be expected to use this new tracepoint? That would help put things in the proper perspective. > This would also duplicate > the reporting done in the avc audit event. I'll simply add the pid as > part of the printk, which should be sufficient for the correlation. Well, to be honest, the very nature of this tracepoint is duplicating the AVC audit record with a focus on using perf to establish a full backtrace at the expense of reduced information. At least that is how it appears to me.
On Fri, Jul 24, 2020 at 5:15 AM Thiébaud Weksteen <tweek@google.com> wrote: > > The audit data currently captures which process and which target > is responsible for a denial. There is no data on where exactly in the > process that call occurred. Debugging can be made easier by being able to > reconstruct the unified kernel and userland stack traces [1]. Add a > tracepoint on the SELinux denials which can then be used by userland > (i.e. perf). > > Although this patch could manually be added by each OS developer to > trouble shoot a denial, adding it to the kernel streamlines the > developers workflow. > > [1] https://source.android.com/devices/tech/debug/native_stack_dump > > Signed-off-by: Thiébaud Weksteen <tweek@google.com> > Signed-off-by: Joel Fernandes <joelaf@google.com> While I am in support of the general idea, could you change my SOB to something like Inspired-by? This is really your patch, but I did demonstrate the idea in an article where the intention was to apply a patch out of tree to do stack dumps / tracing. SOB on the other hand is supposed to track the flow of a patch (the people who the patch goes through) when it is sent upstream. Thanks, - Joel
On Tue, Jul 28, 2020 at 5:12 PM Paul Moore <paul@paul-moore.com> wrote: > Perhaps it would be helpful if you provided an example of how one > would be expected to use this new tracepoint? That would help put > things in the proper perspective. The best example is the one I provided in the commit message, that is using perf (or a perf equivalent), to hook onto that tracepoint. > Well, to be honest, the very nature of this tracepoint is duplicating > the AVC audit record with a focus on using perf to establish a full > backtrace at the expense of reduced information. At least that is how > it appears to me. I see both methods as complementary. By default, the kernel itself can do some reporting (i.e avc message) on which process triggered the denial, what was the context, etc. This is useful even in production and doesn't require any extra tooling. The case for adding this tracepoint can be seen as advanced debugging. That is, once an avc denial has been confirmed, a developer can use this tracepoint to surface the userland stacktrace. It requires more userland tools and symbols on the userland binaries.
On 7/28/20 12:02 PM, Thiébaud Weksteen wrote: > On Tue, Jul 28, 2020 at 5:12 PM Paul Moore <paul@paul-moore.com> wrote: >> Perhaps it would be helpful if you provided an example of how one >> would be expected to use this new tracepoint? That would help put >> things in the proper perspective. > The best example is the one I provided in the commit message, that is > using perf (or a perf equivalent), to hook onto that tracepoint. > >> Well, to be honest, the very nature of this tracepoint is duplicating >> the AVC audit record with a focus on using perf to establish a full >> backtrace at the expense of reduced information. At least that is how >> it appears to me. > I see both methods as complementary. By default, the kernel itself can > do some reporting (i.e avc message) on which process triggered the > denial, what was the context, etc. This is useful even in production > and doesn't require any extra tooling. > The case for adding this tracepoint can be seen as advanced debugging. > That is, once an avc denial has been confirmed, a developer can use > this tracepoint to surface the userland stacktrace. It requires more > userland tools and symbols on the userland binaries. Providing an example of the tracepoint output in the patch description would be helpful IMHO.
On Tue, Jul 28, 2020 at 12:02 PM Thiébaud Weksteen <tweek@google.com> wrote: > On Tue, Jul 28, 2020 at 5:12 PM Paul Moore <paul@paul-moore.com> wrote: > > Perhaps it would be helpful if you provided an example of how one > > would be expected to use this new tracepoint? That would help put > > things in the proper perspective. > > The best example is the one I provided in the commit message, that is > using perf (or a perf equivalent), to hook onto that tracepoint. I probably wasn't as clear as I should have been. I think it would be helpful if you demonstrated how one would take the SELinux data in the perf event and translated that into something meaningful.
On 7/28/20 6:02 PM, Thiébaud Weksteen wrote: > On Tue, Jul 28, 2020 at 5:12 PM Paul Moore <paul@paul-moore.com> wrote: >> Perhaps it would be helpful if you provided an example of how one >> would be expected to use this new tracepoint? That would help put >> things in the proper perspective. > The best example is the one I provided in the commit message, that is > using perf (or a perf equivalent), to hook onto that tracepoint. > >> Well, to be honest, the very nature of this tracepoint is duplicating >> the AVC audit record with a focus on using perf to establish a full >> backtrace at the expense of reduced information. At least that is how >> it appears to me. > I see both methods as complementary. By default, the kernel itself can > do some reporting (i.e avc message) on which process triggered the > denial, what was the context, etc. This is useful even in production > and doesn't require any extra tooling. > The case for adding this tracepoint can be seen as advanced debugging. > That is, once an avc denial has been confirmed, a developer can use > this tracepoint to surface the userland stacktrace. It requires more > userland tools and symbols on the userland binaries. I think from development view you would like to have a better way to trap this events in userspace. One idea that I have is is to have more outcomes from a rule. We have today allow, dontaudit, auditallow i think it would be good to have signal sent too. "signal-xxx-allow" for some set of signals. SIGBUS, SIGSEGV, SIGABRT maybe. That will be a good way to pickup the problem with a debugger or generate a a core file. I have also done some selinux trace functions. I think they collide with this set, but I think I can rebase them upon yours and see if they give some more functionality. I see this functionality very much needed in some form.
On Tue, Jul 28, 2020 at 6:20 PM Paul Moore <paul@paul-moore.com> wrote: > I probably wasn't as clear as I should have been. I think it would be > helpful if you demonstrated how one would take the SELinux data in the > perf event and translated that into something meaningful. So the data itself is not that relevant. What is important is the ability to hook the kernel at the right location, at the right time. Here is an example on how this patch can be used on Android (simpleperf is the Android equivalent of perf), running dmesg as the shell user which is not permitted: # simpleperf record -e selinux:selinux_denied -a -g --duration 10 # simpleperf report -g --full-callgraph Cmdline: /system/bin/simpleperf record -e selinux:selinux_denied -a -g --duration 10 Arch: arm64 Event: selinux:selinux_denied (type 2, config 493) Samples: 1 Event count: 1 Children Self Command Pid Tid Shared Object Symbol 100.00% 0.00% dmesg 3511 3511 /apex/com.android.runtime/lib64/bionic/libc.so __libc_init | -- __libc_init | -- main toybox_main toy_exec_which dmesg_main klogctl el0_svc_naked sys_syslog do_syslog security_syslog selinux_syslog avc_has_perm slow_avc_audit common_lsm_audit avc_audit_pre_callback You can see the combined user and kernel stacks which is useful to understand where and why the denial happened. The key point is that simpleperf is doing the heavy work (i.e names resolution), while the kernel only shares the strict minimum for that to happen. This can be correlated with the pid of the avc denial message (I'm assuming we are trouble shooting one specific denial). It is also possible to manually use ftrace. For instance, after enabling and triggering the denial: bonito:/sys/kernel/debug/tracing # cat trace # tracer: nop # # entries-in-buffer/entries-written: 1/1 #P:8 # # _-----=> irqs-off # / _----=> need-resched # | / _---=> hardirq/softirq # || / _--=> preempt-depth # ||| / delay # TASK-PID CPU# |||| TIMESTAMP FUNCTION # | | | |||| | | dmesg-3624 [001] .... 13072.325358: selinux_denied: denied pid=3624 tclass=4 audited=2 This can be correlated with the following avc denial: [ 2180.183062] type=1400 audit(1596111144.026:27): avc: denied { syslog_read } for comm="dmesg" scontext=u:r:shell:s0 tcontext=u:r:kernel:s0 tclass=system permissive=0 Here, there is limited value of having that tracepoint as we are only duplicating the avc message content. Nevertheless, the filtering part of Peter's patch would be useful to be more precise on which denial we are targeting (I'll reply to the other thread as well). I hope this clarifies the usage. Thanks.
diff --git a/MAINTAINERS b/MAINTAINERS index e64cdde81851..6b6cd5e13537 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -15358,6 +15358,7 @@ T: git git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git F: Documentation/ABI/obsolete/sysfs-selinux-checkreqprot F: Documentation/ABI/obsolete/sysfs-selinux-disable F: Documentation/admin-guide/LSM/SELinux.rst +F: include/trace/events/selinux.h F: include/uapi/linux/selinux_netlink.h F: scripts/selinux/ F: security/selinux/ diff --git a/include/trace/events/selinux.h b/include/trace/events/selinux.h new file mode 100644 index 000000000000..e247187a8135 --- /dev/null +++ b/include/trace/events/selinux.h @@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM selinux + +#if !defined(_TRACE_SELINUX_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_SELINUX_H + +#include <linux/ktime.h> +#include <linux/tracepoint.h> + +TRACE_EVENT(selinux_denied, + + TP_PROTO(int cls, int av), + + TP_ARGS(cls, av), + + TP_STRUCT__entry( + __field(int, cls) + __field(int, av) + ), + + TP_fast_assign( + __entry->cls = cls; + __entry->av = av; + ), + + TP_printk("denied %d %d", + __entry->cls, + __entry->av) +); + +#endif + +/* This part must be outside protection */ +#include <trace/define_trace.h> diff --git a/security/selinux/avc.c b/security/selinux/avc.c index d18cb32a242a..85d2e22ab656 100644 --- a/security/selinux/avc.c +++ b/security/selinux/avc.c @@ -31,6 +31,9 @@ #include "avc_ss.h" #include "classmap.h" +#define CREATE_TRACE_POINTS +#include <trace/events/selinux.h> + #define AVC_CACHE_SLOTS 512 #define AVC_DEF_CACHE_THRESHOLD 512 #define AVC_CACHE_RECLAIM 16 @@ -672,6 +675,9 @@ static void avc_audit_pre_callback(struct audit_buffer *ab, void *a) return; } + if (sad->denied) + trace_selinux_denied(sad->tclass, av); + perms = secclass_map[sad->tclass-1].perms; audit_log_format(ab, " {");