Patch "perf/core: Fix perf_event_open() vs. execve() race" has been added to the 4.5-stable tree
Subsystems (2)
This is a note to let you know that I've just added the patch titled
perf/core: Fix perf_event_open() vs. execve() race
to the 4.5-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
perf-core-fix-perf_event_open-vs.-execve-race.patch
and it can be found in the queue-4.5 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.
Quoted 1 line
From
79c9ce57eb2d5f1497546a3946b4ae21b6fdc438Mon Sep 17 00:00:00 2001
From: Peter Zijlstra <[email protected]>
Date: Tue, 26 Apr 2016 11:36:53 +0200
Subject: perf/core: Fix perf_event_open() vs. execve() race
From: Peter Zijlstra <[email protected]>
commit 79c9ce57eb2d5f1497546a3946b4ae21b6fdc438 upstream.
Jann reported that the ptrace_may_access() check in find_lively_task_by_vpid() is racy against exec().
Specifically:
perf_event_open() execve()
ptrace_may_access()
commit_creds()
... if (get_dumpable() != SUID_DUMP_USER)
perf_event_exit_task();
perf_install_in_context()
would result in installing a counter across the creds boundary.
Fix this by wrapping lots of perf_event_open() in cred_guard_mutex. This should be fine as perf_event_exit_task() is already called with cred_guard_mutex held, so all perf locks already nest inside it.
--- kernel/events/core.c | 52 +++++++++++++++++++++++++++++++++++---------------- 1 file changed, 36 insertions(+), 16 deletions(-)
kernel/events/core.c +36 -16
@@ -1090,6 +1090,7 @@ static void put_ctx(struct perf_event_co* function. * * Lock order: +* cred_guard_mutex * task_struct::perf_event_mutex * perf_event_context::mutex * perf_event::child_mutex;
@@ -3415,7 +3416,6 @@ static struct task_struct * find_lively_task_by_vpid(pid_t vpid) {struct task_struct *task; -int err;rcu_read_lock ();if (!vpid)
@@ -3429,16 +3429,7 @@ find_lively_task_by_vpid(pid_t vpid)if (!task)return ERR_PTR (-ESRCH ); -/* Reuse ptrace permission checks for now. */ - err = -EACCES ; -if (!ptrace_may_access (task,PTRACE_MODE_READ_REALCREDS )) - goto errout; -return task; -errout: -put_task_struct (task); -return ERR_PTR (err); - }/*
@@ -8360,6 +8351,24 @@ SYSCALL_DEFINE5(perf_event_open,get_online_cpus (); +if (task) { + err =mutex_lock_interruptible (&task->signal->cred_guard_mutex); +if (err) + goto err_cpus; + +/* +* Reuse ptrace permission checks for now. +* +* We must hold cred_guard_mutex across this and any potential +* perf_install_in_context() call for this new event to +* serialize against exec() altering our credentials (and the +* perf_event_exit_task() that could imply). +*/ + err = -EACCES ; +if (!ptrace_may_access (task,PTRACE_MODE_READ_REALCREDS )) + goto err_cred; + } +if (flags &PERF_FLAG_PID_CGROUP ) cgroup_fd = pid;
@@ -8367,7 +8376,7 @@ SYSCALL_DEFINE5(perf_event_open, NULL, NULL, cgroup_fd);if (IS_ERR (event)) { err =PTR_ERR (event); - goto err_cpus; + goto err_cred; }if (is_sampling_event (event)) {
@@ -8426,11 +8435,6 @@ SYSCALL_DEFINE5(perf_event_open, goto err_context; } -if (task) { -put_task_struct (task); - task =NULL ; - } -/* * Look up the group leader (we will attach this event to it): */
@@ -8528,6 +8532,11 @@ SYSCALL_DEFINE5(perf_event_open,WARN_ON_ONCE (ctx->parent_ctx); +/* +* This is the point on no return; we cannot fail hereafter. This is +* where we start modifying current state. +*/ +if (move_group) {/* * See perf_event_ctx_lock() for comments on the details
@@ -8599,6 +8608,11 @@ SYSCALL_DEFINE5(perf_event_open,mutex_unlock (&gctx->mutex);mutex_unlock (&ctx->mutex); +if (task) { +mutex_unlock (&task->signal->cred_guard_mutex); +put_task_struct (task); + } +put_online_cpus ();mutex_lock (¤t->perf_event_mutex);
@@ -8631,6 +8645,9 @@ err_alloc:*/ if (!event_file)free_event (event); +err_cred: +if (task) +mutex_unlock (&task->signal->cred_guard_mutex); err_cpus:put_online_cpus (); err_task:
@@ -8915,6 +8932,9 @@ static void perf_event_exit_task_context/* * When a child task exits, feed back event values to parent events. +* +* Can be called with cred_guard_mutex held when called from +* install_exec_creds(). */ void perf_event_exit_task (struct task_struct *child) {
Patches currently in stable-queue which might be from [email protected] are
queue-4.5/perf-x86-intel-pt-generate-pmi-in-the-stop-region-as-well.patch queue-4.5/perf-core-fix-perf_event_open-vs.-execve-race.patch