Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220902100057.404817-1-elver@google.com>
In-Reply-To: <20220902100057.404817-1-elver@google.com>
From:   Dmitry Vyukov <dvyukov@google.com>
Date:   Mon, 5 Sep 2022 17:53:50 +0200
Message-ID: <CACT4Y+ZCzoK9+3qKOzXSHvZ5cfaXpYXbBL+R9bicapOuRpUNFQ@mail.gmail.com>
Subject: Re: [PATCH] perf: Allow restricted kernel breakpoints on user addresses
To:     Marco Elver <elver@google.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Jiri Olsa <jolsa@kernel.org>,
        Namhyung Kim <namhyung@kernel.org>,
        linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org,
        kasan-dev@googlegroups.com, Jann Horn <jannh@google.com>,
        Thomas Gleixner <tglx@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 2 Sept 2022 at 12:01, Marco Elver <elver@google.com> wrote:
>
> Allow the creation of restricted breakpoint perf events that also fire
> in the kernel (!exclude_kernel), if:
>
>   1. No sample information is requested; samples may contain IPs,
>      registers, or other information that may disclose kernel addresses.
>
>   2. The breakpoint (viz. data watchpoint) is on a user address.
>
> The rules constrain the allowable perf events such that no sensitive
> kernel information can be disclosed.
>
> Despite no explicit kernel information disclosure, the following
> questions may need answers:
>
>  1. Is obtaining information that the kernel accessed a particular
>     user's known memory location revealing new information?
>     Given the kernel's user space ABI, there should be no "surprise
>     accesses" to user space memory in the first place.
>
>  2. Does causing breakpoints on user memory accesses by the kernel
>     potentially impact timing in a sensitive way?
>     Since hardware breakpoints trigger regardless of the state of
>     perf_event_attr::exclude_kernel, but are filtered in the perf
>     subsystem, this possibility already exists independent of the
>     proposed change.

I don't see how this gives userspace any new information.
As you noted userspace already should know what userspace addresses
kernel accesses. Additionally since the breakpoint fires anyway (just
filtered out), the fact of it firing should be easily recoverable from
the timing side-channel already. So:

Acked-by: Dmitry Vyukov <dvyukov@google.com>


> Signed-off-by: Marco Elver <elver@google.com>
> ---
>
> Changelog
> ~~~~~~~~~
>
> v1:
> * Rebase.
>
> RFC: https://lkml.kernel.org/r/20220601093502.364142-1-elver@google.com
> ---
>  include/linux/perf_event.h |  8 +-------
>  kernel/events/core.c       | 38 ++++++++++++++++++++++++++++++++++++++
>  2 files changed, 39 insertions(+), 7 deletions(-)
>
> diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
> index a784e055002e..907b0e3f1318 100644
> --- a/include/linux/perf_event.h
> +++ b/include/linux/perf_event.h
> @@ -1367,13 +1367,7 @@ static inline int perf_is_paranoid(void)
>         return sysctl_perf_event_paranoid > -1;
>  }
>
> -static inline int perf_allow_kernel(struct perf_event_attr *attr)
> -{
> -       if (sysctl_perf_event_paranoid > 1 && !perfmon_capable())
> -               return -EACCES;
> -
> -       return security_perf_event_open(attr, PERF_SECURITY_KERNEL);
> -}
> +extern int perf_allow_kernel(struct perf_event_attr *attr);
>
>  static inline int perf_allow_cpu(struct perf_event_attr *attr)
>  {
> diff --git a/kernel/events/core.c b/kernel/events/core.c
> index 2621fd24ad26..75f5705b6892 100644
> --- a/kernel/events/core.c
> +++ b/kernel/events/core.c
> @@ -3176,6 +3176,12 @@ static int perf_event_modify_attr(struct perf_event *event,
>                 return -EOPNOTSUPP;
>         }
>
> +       if (!event->attr.exclude_kernel) {
> +               err = perf_allow_kernel(attr);
> +               if (err)
> +                       return err;
> +       }
> +
>         WARN_ON_ONCE(event->ctx->parent_ctx);
>
>         mutex_lock(&event->child_mutex);
> @@ -12037,6 +12043,38 @@ perf_check_permission(struct perf_event_attr *attr, struct task_struct *task)
>         return is_capable || ptrace_may_access(task, ptrace_mode);
>  }
>
> +/*
> + * Check if unprivileged users are allowed to set up breakpoints on user
> + * addresses that also count when the kernel accesses them.
> + */
> +static bool perf_allow_kernel_breakpoint(struct perf_event_attr *attr)
> +{
> +       if (attr->type != PERF_TYPE_BREAKPOINT)
> +               return false;
> +
> +       /*
> +        * The sample may contain IPs, registers, or other information that may
> +        * disclose kernel addresses or timing information. Disallow any kind of
> +        * additional sample information.
> +        */
> +       if (attr->sample_type)
> +               return false;
> +
> +       /*
> +        * Only allow kernel breakpoints on user addresses.
> +        */
> +       return access_ok((void __user *)(unsigned long)attr->bp_addr, attr->bp_len);
> +}
> +
> +int perf_allow_kernel(struct perf_event_attr *attr)
> +{
> +       if (sysctl_perf_event_paranoid > 1 && !perfmon_capable() &&
> +           !perf_allow_kernel_breakpoint(attr))
> +               return -EACCES;
> +
> +       return security_perf_event_open(attr, PERF_SECURITY_KERNEL);
> +}
> +
>  /**
>   * sys_perf_event_open - open a performance event, associate it to a task/cpu
>   *
> --
> 2.37.2.789.g6183377224-goog
>