Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 14 Oct 2019 11:35:44 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     "Joel Fernandes (Google)" <joel@joelfernandes.org>
Cc:     linux-kernel@vger.kernel.org, rostedt@goodmis.org,
        primiano@google.com, rsavitski@google.com, jeffv@google.com,
        kernel-team@android.com, Alexei Starovoitov <ast@kernel.org>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        bpf@vger.kernel.org, Daniel Borkmann <daniel@iogearbox.net>,
        Ingo Molnar <mingo@redhat.com>,
        James Morris <jmorris@namei.org>, Jiri Olsa <jolsa@redhat.com>,
        Kees Cook <keescook@chromium.org>,
        linux-security-module@vger.kernel.org,
        Matthew Garrett <matthewgarrett@google.com>,
        Namhyung Kim <namhyung@kernel.org>, selinux@vger.kernel.org,
        Song Liu <songliubraving@fb.com>,
        "maintainer:X86 ARCHITECTURE (32-BIT AND 64-BIT)" <x86@kernel.org>,
        Yonghong Song <yhs@fb.com>
Subject: Re: [PATCH] perf_event: Add support for LSM and SELinux checks
Message-ID: <20191014093544.GB2328@hirez.programming.kicks-ass.net>
References: <20191011160330.199604-1-joel@joelfernandes.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20191011160330.199604-1-joel@joelfernandes.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Oct 11, 2019 at 12:03:30PM -0400, Joel Fernandes (Google) wrote:

> @@ -4761,6 +4762,7 @@ int perf_event_release_kernel(struct perf_event *event)
>  	}
>  
>  no_ctx:
> +	security_perf_event_free(event);
>  	put_event(event); /* Must be the 'last' reference */
>  	return 0;
>  }

> @@ -10553,11 +10568,16 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu,
>  		}
>  	}
>  
> +	err = security_perf_event_alloc(event);
> +	if (err)
> +		goto err_security;
> +
>  	/* symmetric to unaccount_event() in _free_event() */
>  	account_event(event);
>  
>  	return event;
>  
> +err_security:
>  err_addr_filters:
>  	kfree(event->addr_filter_ranges);
>  

There's a bunch of problems here I think:

 - err_security is named wrong; the naming scheme is to name the label
   after the last thing that succeeded / first thing that needs to be
   undone.

 - per that, you're forgetting to undo 'get_callchain_buffers()'

 - perf_event_release_kernel() is not a full match to
   perf_event_alloc(), inherited events get created by
   perf_event_alloc() but never pass through
   perf_event_release_kernel().


I'm thinking the below patch on top should ammend these issues; please
verify.

---
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4540,6 +4540,8 @@ static void _free_event(struct perf_even
 
 	unaccount_event(event);
 
+	security_perf_event_free(event);
+
 	if (event->rb) {
 		/*
 		 * Can happen when we close an event with re-directed output.
@@ -4774,7 +4776,6 @@ int perf_event_release_kernel(struct per
 	}
 
 no_ctx:
-	security_perf_event_free(event);
 	put_event(event); /* Must be the 'last' reference */
 	return 0;
 }
@@ -10595,14 +10596,18 @@ perf_event_alloc(struct perf_event_attr
 
 	err = security_perf_event_alloc(event);
 	if (err)
-		goto err_security;
+		goto err_callchain_buffer;
 
 	/* symmetric to unaccount_event() in _free_event() */
 	account_event(event);
 
 	return event;
 
-err_security:
+err_callchain_buffer:
+	if (!event->parent) {
+		if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
+			put_callchain_buffers();
+	}
 err_addr_filters:
 	kfree(event->addr_filter_ranges);