| Message ID | 20190325225636.11635-2-aarcange@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | userfaultfd: use RCU to free the task struct when fork fails | expand |
On 26.03.2019 01:56, Andrea Arcangeli wrote: > MEMCG depends on the task structure not to be freed under > rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences > mm->owner. > > An alternate possible fix would be to defer the delivery of the > userfaultfd contexts to the monitor until after fork() is guaranteed > to succeed. Such a change would require more changes because it would > create a strict ordering dependency where the uffd methods would need > to be called beyond the last potentially failing branch in order to be > safe. This solution as opposed only adds the dependency to common code > to set mm->owner to NULL and to free the task struct that was pointed > by mm->owner with RCU, if fork ends up failing. The userfaultfd > methods can still be called anywhere during the fork runtime and the > monitor will keep discarding orphaned "mm" coming from failed forks in > userland. > > This race condition couldn't trigger if CONFIG_MEMCG was set =n at > build time. > > Fixes: 893e26e61d04 ("userfaultfd: non-cooperative: Add fork() event") > Cc: stable@kernel.org > Tested-by: zhong jiang <zhongjiang@huawei.com> > Reported-by: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com > Signed-off-by: Andrea Arcangeli <aarcange@redhat.com> > --- > kernel/fork.c | 34 ++++++++++++++++++++++++++++++++-- > 1 file changed, 32 insertions(+), 2 deletions(-) > > diff --git a/kernel/fork.c b/kernel/fork.c > index 9dcd18aa210b..a19790e27afd 100644 > --- a/kernel/fork.c > +++ b/kernel/fork.c > @@ -952,6 +952,15 @@ static void mm_init_aio(struct mm_struct *mm) > #endif > } > > +static __always_inline void mm_clear_owner(struct mm_struct *mm, > + struct task_struct *p) > +{ > +#ifdef CONFIG_MEMCG > + if (mm->owner == p) > + WRITE_ONCE(mm->owner, NULL); > +#endif > +} > + > static void mm_init_owner(struct mm_struct *mm, struct task_struct *p) > { > #ifdef CONFIG_MEMCG > @@ -1331,6 +1340,7 @@ static struct mm_struct *dup_mm(struct task_struct *tsk) > free_pt: > /* don't put binfmt in mmput, we haven't got module yet */ > mm->binfmt = NULL; > + mm_init_owner(mm, NULL); > mmput(mm); > > fail_nomem: > @@ -1662,6 +1672,24 @@ static inline void rcu_copy_process(struct task_struct *p) > #endif /* #ifdef CONFIG_TASKS_RCU */ > } > > +#ifdef CONFIG_MEMCG > +static void __delayed_free_task(struct rcu_head *rhp) > +{ > + struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); > + > + free_task(tsk); > +} > +#endif /* CONFIG_MEMCG */ > + > +static __always_inline void delayed_free_task(struct task_struct *tsk) > +{ > +#ifdef CONFIG_MEMCG > + call_rcu(&tsk->rcu, __delayed_free_task); > +#else /* CONFIG_MEMCG */ > + free_task(tsk); > +#endif /* CONFIG_MEMCG */ > +} > + > /* > * This creates a new process as a copy of the old one, > * but does not actually start it yet. > @@ -2123,8 +2151,10 @@ static __latent_entropy struct task_struct *copy_process( > bad_fork_cleanup_namespaces: > exit_task_namespaces(p); > bad_fork_cleanup_mm: > - if (p->mm) > + if (p->mm) { > + mm_clear_owner(p->mm, p); > mmput(p->mm); > + } > bad_fork_cleanup_signal: > if (!(clone_flags & CLONE_THREAD)) > free_signal_struct(p->signal); > @@ -2155,7 +2185,7 @@ static __latent_entropy struct task_struct *copy_process( > bad_fork_free: > p->state = TASK_DEAD; > put_task_stack(p); > - free_task(p); > + delayed_free_task(p); Can't call_rcu(&p->rcu, delayed_put_task_struct) be used instead this? Kirill
On 26.03.2019 11:07, Kirill Tkhai wrote: > On 26.03.2019 01:56, Andrea Arcangeli wrote: >> MEMCG depends on the task structure not to be freed under >> rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences >> mm->owner. >> >> An alternate possible fix would be to defer the delivery of the >> userfaultfd contexts to the monitor until after fork() is guaranteed >> to succeed. Such a change would require more changes because it would >> create a strict ordering dependency where the uffd methods would need >> to be called beyond the last potentially failing branch in order to be >> safe. This solution as opposed only adds the dependency to common code >> to set mm->owner to NULL and to free the task struct that was pointed >> by mm->owner with RCU, if fork ends up failing. The userfaultfd >> methods can still be called anywhere during the fork runtime and the >> monitor will keep discarding orphaned "mm" coming from failed forks in >> userland. >> >> This race condition couldn't trigger if CONFIG_MEMCG was set =n at >> build time. >> >> Fixes: 893e26e61d04 ("userfaultfd: non-cooperative: Add fork() event") >> Cc: stable@kernel.org >> Tested-by: zhong jiang <zhongjiang@huawei.com> >> Reported-by: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com >> Signed-off-by: Andrea Arcangeli <aarcange@redhat.com> >> --- >> kernel/fork.c | 34 ++++++++++++++++++++++++++++++++-- >> 1 file changed, 32 insertions(+), 2 deletions(-) >> >> diff --git a/kernel/fork.c b/kernel/fork.c >> index 9dcd18aa210b..a19790e27afd 100644 >> --- a/kernel/fork.c >> +++ b/kernel/fork.c >> @@ -952,6 +952,15 @@ static void mm_init_aio(struct mm_struct *mm) >> #endif >> } >> >> +static __always_inline void mm_clear_owner(struct mm_struct *mm, >> + struct task_struct *p) >> +{ >> +#ifdef CONFIG_MEMCG >> + if (mm->owner == p) >> + WRITE_ONCE(mm->owner, NULL); >> +#endif >> +} >> + >> static void mm_init_owner(struct mm_struct *mm, struct task_struct *p) >> { >> #ifdef CONFIG_MEMCG >> @@ -1331,6 +1340,7 @@ static struct mm_struct *dup_mm(struct task_struct *tsk) >> free_pt: >> /* don't put binfmt in mmput, we haven't got module yet */ >> mm->binfmt = NULL; >> + mm_init_owner(mm, NULL); >> mmput(mm); >> >> fail_nomem: >> @@ -1662,6 +1672,24 @@ static inline void rcu_copy_process(struct task_struct *p) >> #endif /* #ifdef CONFIG_TASKS_RCU */ >> } >> >> +#ifdef CONFIG_MEMCG >> +static void __delayed_free_task(struct rcu_head *rhp) >> +{ >> + struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); >> + >> + free_task(tsk); >> +} >> +#endif /* CONFIG_MEMCG */ >> + >> +static __always_inline void delayed_free_task(struct task_struct *tsk) >> +{ >> +#ifdef CONFIG_MEMCG >> + call_rcu(&tsk->rcu, __delayed_free_task); >> +#else /* CONFIG_MEMCG */ >> + free_task(tsk); >> +#endif /* CONFIG_MEMCG */ >> +} >> + >> /* >> * This creates a new process as a copy of the old one, >> * but does not actually start it yet. >> @@ -2123,8 +2151,10 @@ static __latent_entropy struct task_struct *copy_process( >> bad_fork_cleanup_namespaces: >> exit_task_namespaces(p); >> bad_fork_cleanup_mm: >> - if (p->mm) >> + if (p->mm) { >> + mm_clear_owner(p->mm, p); >> mmput(p->mm); >> + } >> bad_fork_cleanup_signal: >> if (!(clone_flags & CLONE_THREAD)) >> free_signal_struct(p->signal); >> @@ -2155,7 +2185,7 @@ static __latent_entropy struct task_struct *copy_process( >> bad_fork_free: >> p->state = TASK_DEAD; >> put_task_stack(p); >> - free_task(p); >> + delayed_free_task(p); > > Can't call_rcu(&p->rcu, delayed_put_task_struct) be used instead this? I mean: refcount_set(&tsk->usage, 2); call_rcu(&p->rcu, delayed_put_task_struct); And: diff --git a/kernel/fork.c b/kernel/fork.c index 3c516c6f7ce4..27cdf61b51a1 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -715,7 +715,9 @@ static inline void put_signal_struct(struct signal_struct *sig) void __put_task_struct(struct task_struct *tsk) { - WARN_ON(!tsk->exit_state); + if (!tsk->exit_state) + /* Cleanup of copy_process() */ + goto free; WARN_ON(refcount_read(&tsk->usage)); WARN_ON(tsk == current); @@ -727,6 +729,7 @@ void __put_task_struct(struct task_struct *tsk) put_signal_struct(tsk->signal); if (!profile_handoff_task(tsk)) +free: free_task(tsk); } EXPORT_SYMBOL_GPL(__put_task_struct);
On 26.03.2019 11:18, Kirill Tkhai wrote: > On 26.03.2019 11:07, Kirill Tkhai wrote: >> On 26.03.2019 01:56, Andrea Arcangeli wrote: >>> MEMCG depends on the task structure not to be freed under >>> rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences >>> mm->owner. >>> >>> An alternate possible fix would be to defer the delivery of the >>> userfaultfd contexts to the monitor until after fork() is guaranteed >>> to succeed. Such a change would require more changes because it would >>> create a strict ordering dependency where the uffd methods would need >>> to be called beyond the last potentially failing branch in order to be >>> safe. This solution as opposed only adds the dependency to common code >>> to set mm->owner to NULL and to free the task struct that was pointed >>> by mm->owner with RCU, if fork ends up failing. The userfaultfd >>> methods can still be called anywhere during the fork runtime and the >>> monitor will keep discarding orphaned "mm" coming from failed forks in >>> userland. >>> >>> This race condition couldn't trigger if CONFIG_MEMCG was set =n at >>> build time. >>> >>> Fixes: 893e26e61d04 ("userfaultfd: non-cooperative: Add fork() event") >>> Cc: stable@kernel.org >>> Tested-by: zhong jiang <zhongjiang@huawei.com> >>> Reported-by: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com >>> Signed-off-by: Andrea Arcangeli <aarcange@redhat.com> >>> --- >>> kernel/fork.c | 34 ++++++++++++++++++++++++++++++++-- >>> 1 file changed, 32 insertions(+), 2 deletions(-) >>> >>> diff --git a/kernel/fork.c b/kernel/fork.c >>> index 9dcd18aa210b..a19790e27afd 100644 >>> --- a/kernel/fork.c >>> +++ b/kernel/fork.c >>> @@ -952,6 +952,15 @@ static void mm_init_aio(struct mm_struct *mm) >>> #endif >>> } >>> >>> +static __always_inline void mm_clear_owner(struct mm_struct *mm, >>> + struct task_struct *p) >>> +{ >>> +#ifdef CONFIG_MEMCG >>> + if (mm->owner == p) >>> + WRITE_ONCE(mm->owner, NULL); >>> +#endif >>> +} >>> + >>> static void mm_init_owner(struct mm_struct *mm, struct task_struct *p) >>> { >>> #ifdef CONFIG_MEMCG >>> @@ -1331,6 +1340,7 @@ static struct mm_struct *dup_mm(struct task_struct *tsk) >>> free_pt: >>> /* don't put binfmt in mmput, we haven't got module yet */ >>> mm->binfmt = NULL; >>> + mm_init_owner(mm, NULL); >>> mmput(mm); >>> >>> fail_nomem: >>> @@ -1662,6 +1672,24 @@ static inline void rcu_copy_process(struct task_struct *p) >>> #endif /* #ifdef CONFIG_TASKS_RCU */ >>> } >>> >>> +#ifdef CONFIG_MEMCG >>> +static void __delayed_free_task(struct rcu_head *rhp) >>> +{ >>> + struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); >>> + >>> + free_task(tsk); >>> +} >>> +#endif /* CONFIG_MEMCG */ >>> + >>> +static __always_inline void delayed_free_task(struct task_struct *tsk) >>> +{ >>> +#ifdef CONFIG_MEMCG >>> + call_rcu(&tsk->rcu, __delayed_free_task); >>> +#else /* CONFIG_MEMCG */ >>> + free_task(tsk); >>> +#endif /* CONFIG_MEMCG */ >>> +} >>> + >>> /* >>> * This creates a new process as a copy of the old one, >>> * but does not actually start it yet. >>> @@ -2123,8 +2151,10 @@ static __latent_entropy struct task_struct *copy_process( >>> bad_fork_cleanup_namespaces: >>> exit_task_namespaces(p); >>> bad_fork_cleanup_mm: >>> - if (p->mm) >>> + if (p->mm) { >>> + mm_clear_owner(p->mm, p); >>> mmput(p->mm); >>> + } >>> bad_fork_cleanup_signal: >>> if (!(clone_flags & CLONE_THREAD)) >>> free_signal_struct(p->signal); >>> @@ -2155,7 +2185,7 @@ static __latent_entropy struct task_struct *copy_process( >>> bad_fork_free: >>> p->state = TASK_DEAD; >>> put_task_stack(p); >>> - free_task(p); >>> + delayed_free_task(p); >> >> Can't call_rcu(&p->rcu, delayed_put_task_struct) be used instead this? > > I mean: > > refcount_set(&tsk->usage, 2); I.e., refcount_set(&tsk->usage, 1); > call_rcu(&p->rcu, delayed_put_task_struct); > > And: > > diff --git a/kernel/fork.c b/kernel/fork.c > index 3c516c6f7ce4..27cdf61b51a1 100644 > --- a/kernel/fork.c > +++ b/kernel/fork.c > @@ -715,7 +715,9 @@ static inline void put_signal_struct(struct signal_struct *sig) > > void __put_task_struct(struct task_struct *tsk) > { > - WARN_ON(!tsk->exit_state); > + if (!tsk->exit_state) > + /* Cleanup of copy_process() */ > + goto free; > WARN_ON(refcount_read(&tsk->usage)); > WARN_ON(tsk == current); > > @@ -727,6 +729,7 @@ void __put_task_struct(struct task_struct *tsk) > put_signal_struct(tsk->signal); > > if (!profile_handoff_task(tsk)) > +free: > free_task(tsk); > } > EXPORT_SYMBOL_GPL(__put_task_struct); >
On Mon 25-03-19 18:56:35, Andrea Arcangeli wrote: > MEMCG depends on the task structure not to be freed under > rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences > mm->owner. Please state the actual problem. Your cover letter mentiones a race condition. Please make it explicit in the changelog. > An alternate possible fix would be to defer the delivery of the > userfaultfd contexts to the monitor until after fork() is guaranteed > to succeed. Such a change would require more changes because it would > create a strict ordering dependency where the uffd methods would need > to be called beyond the last potentially failing branch in order to be > safe. How much more changes are we talking about? Because ... > This solution as opposed only adds the dependency to common code > to set mm->owner to NULL and to free the task struct that was pointed > by mm->owner with RCU, if fork ends up failing. The userfaultfd > methods can still be called anywhere during the fork runtime and the > monitor will keep discarding orphaned "mm" coming from failed forks in > userland. ... this is adding a subtle hack that might break in the future because copy_process error paths are far from trivial and quite error prone IMHO. I am not opposed to the patch in principle but I would really like to see what kind of solutions we are comparing here. > This race condition couldn't trigger if CONFIG_MEMCG was set =n at > build time. All the CONFIG_MEMCG is just ugly as hell. Can we reduce that please? E.g. use if (IS_ENABLED(CONFIG_MEMCG)) where appropriate? [...] > +static __always_inline void mm_clear_owner(struct mm_struct *mm, > + struct task_struct *p) > +{ > +#ifdef CONFIG_MEMCG > + if (mm->owner == p) > + WRITE_ONCE(mm->owner, NULL); > +#endif How can we ever hit this warning and what does that mean?
On Tue, Mar 26, 2019 at 09:56:43AM +0100, Michal Hocko wrote: > On Mon 25-03-19 18:56:35, Andrea Arcangeli wrote: > > MEMCG depends on the task structure not to be freed under > > rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences > > mm->owner. > > Please state the actual problem. Your cover letter mentiones a race > condition. Please make it explicit in the changelog. The actual problem is the task structure is freed while get_mem_cgroup_from_mm() holds rcu_read_lock() and dereferences mm->owner. I thought the breakage of RCU is pretty clear, but we could add a description of the race like I did in the original thread: https://lkml.kernel.org/r/000000000000601367057a095de4@google.com https://lkml.kernel.org/r/20190316194222.GA29767@redhat.com > > An alternate possible fix would be to defer the delivery of the > > userfaultfd contexts to the monitor until after fork() is guaranteed > > to succeed. Such a change would require more changes because it would > > create a strict ordering dependency where the uffd methods would need > > to be called beyond the last potentially failing branch in order to be > > safe. > > How much more changes are we talking about? Because ... I haven't implemented but I can theorize. It should require a new hooking point and information being accumulated in RAM and passed from the current hooking point to the new hooking point and to hold off the delivery of such information to the uffd monitor (the fd reader), until the new hooking point is invoked. The new hooking point would need to be invoked after fork cannot fail anymore. We already accumulate some information in RAM there, but the first delivery happens at a point where fork can still fail. > > This solution as opposed only adds the dependency to common code > > to set mm->owner to NULL and to free the task struct that was pointed > > by mm->owner with RCU, if fork ends up failing. The userfaultfd > > methods can still be called anywhere during the fork runtime and the > > monitor will keep discarding orphaned "mm" coming from failed forks in > > userland. > > ... this is adding a subtle hack that might break in the future because > copy_process error paths are far from trivial and quite error prone > IMHO. I am not opposed to the patch in principle but I would really like > to see what kind of solutions we are comparing here. The rule of clearing mm->owner and then freeing the mm->owner memory with call_rcu is already followed everywhere else. See for example mm_update_next_owner() that sets mm->owner to NULL and only then invokes put_task_struct which frees the memory pointed by the old value of mm->owner using RCU. The "subtle hack" already happens at every exit when MEMCG=y. All the patch does is to extend the "subtle hack" to the fork failure path too which it didn't follow the rule and it didn't clear mm->owner and it just freed the task struct without waiting for a RCU grace period. In fact like pointed out by Kirill Tkhai we could reuse delayed_put_task_struct method that is already used by exit, except it does more than freeing the task structure and it relies on refcounters to be initialized so I thought the free_task -> call_rcu( free_task) conversion was simpler and more obviously safe. Sharing the other method only looked a complication that requires syncing up the refcounts. I think the only conceptual simplification possible would be again to add a new hooking point and more buildup of information until fork cannot fail, but in implementation terms I doubt the fix will become smaller or simpler that way. > > This race condition couldn't trigger if CONFIG_MEMCG was set =n at > > build time. > > All the CONFIG_MEMCG is just ugly as hell. Can we reduce that please? > E.g. use if (IS_ENABLED(CONFIG_MEMCG)) where appropriate? There's just one place where I could use that instead of #ifdef. > > +static __always_inline void mm_clear_owner(struct mm_struct *mm, > > + struct task_struct *p) > > +{ > > +#ifdef CONFIG_MEMCG > > + if (mm->owner == p) > > + WRITE_ONCE(mm->owner, NULL); > > +#endif > > How can we ever hit this warning and what does that mean? Which warning?
On Tue 26-03-19 20:16:16, Andrea Arcangeli wrote: > On Tue, Mar 26, 2019 at 09:56:43AM +0100, Michal Hocko wrote: > > On Mon 25-03-19 18:56:35, Andrea Arcangeli wrote: > > > MEMCG depends on the task structure not to be freed under > > > rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences > > > mm->owner. > > > > Please state the actual problem. Your cover letter mentiones a race > > condition. Please make it explicit in the changelog. > > The actual problem is the task structure is freed while > get_mem_cgroup_from_mm() holds rcu_read_lock() and dereferences > mm->owner. > > I thought the breakage of RCU is pretty clear, but we could add a > description of the race like I did in the original thread: > > https://lkml.kernel.org/r/000000000000601367057a095de4@google.com > https://lkml.kernel.org/r/20190316194222.GA29767@redhat.com Yes please. That really belongs to the changelog. You do not expect people chasing long email threads or code to figure that out, right? > > > An alternate possible fix would be to defer the delivery of the > > > userfaultfd contexts to the monitor until after fork() is guaranteed > > > to succeed. Such a change would require more changes because it would > > > create a strict ordering dependency where the uffd methods would need > > > to be called beyond the last potentially failing branch in order to be > > > safe. > > > > How much more changes are we talking about? Because ... > > I haven't implemented but I can theorize. It should require a new > hooking point and information being accumulated in RAM and passed from > the current hooking point to the new hooking point and to hold off the > delivery of such information to the uffd monitor (the fd reader), > until the new hooking point is invoked. The new hooking point would > need to be invoked after fork cannot fail anymore. > > We already accumulate some information in RAM there, but the first > delivery happens at a point where fork can still fail. I am sorry but this is not really clear to me. What is the problem to postpone hooking point to later and how much more data we are talking about here? > > > This solution as opposed only adds the dependency to common code > > > to set mm->owner to NULL and to free the task struct that was pointed > > > by mm->owner with RCU, if fork ends up failing. The userfaultfd > > > methods can still be called anywhere during the fork runtime and the > > > monitor will keep discarding orphaned "mm" coming from failed forks in > > > userland. > > > > ... this is adding a subtle hack that might break in the future because > > copy_process error paths are far from trivial and quite error prone > > IMHO. I am not opposed to the patch in principle but I would really like > > to see what kind of solutions we are comparing here. > > The rule of clearing mm->owner and then freeing the mm->owner memory > with call_rcu is already followed everywhere else. See for example > mm_update_next_owner() that sets mm->owner to NULL and only then > invokes put_task_struct which frees the memory pointed by the old > value of mm->owner using RCU. > > The "subtle hack" already happens at every exit when MEMCG=y. All the > patch does is to extend the "subtle hack" to the fork failure path too > which it didn't follow the rule and it didn't clear mm->owner and it > just freed the task struct without waiting for a RCU grace period. In > fact like pointed out by Kirill Tkhai we could reuse > delayed_put_task_struct method that is already used by exit, except it > does more than freeing the task structure and it relies on refcounters > to be initialized so I thought the free_task -> call_rcu( free_task) > conversion was simpler and more obviously safe. Sharing the other > method only looked a complication that requires syncing up the > refcounts. > > I think the only conceptual simplification possible would be again to > add a new hooking point and more buildup of information until fork > cannot fail, but in implementation terms I doubt the fix will become > smaller or simpler that way. Well, in general I prefer the code to be memcg neutral as much as possible. We might have this subtle dependency with memcg now but this is not specific to memcg in general. Therefore, if there is a way to make a userfault specific fix then I would prefer it. If that is not feasible then fair enough. JFYI, getting rid of mm->owner is a long term plan. This is just too ugly to live. Easier said than done, unfortunately. > > > This race condition couldn't trigger if CONFIG_MEMCG was set =n at > > > build time. > > > > All the CONFIG_MEMCG is just ugly as hell. Can we reduce that please? > > E.g. use if (IS_ENABLED(CONFIG_MEMCG)) where appropriate? > > There's just one place where I could use that instead of #ifdef. OK, I can see it now. Is there any strong reason to make the delayed freeing conditional that would spare at least part of the ugliness. > > > +static __always_inline void mm_clear_owner(struct mm_struct *mm, > > > + struct task_struct *p) > > > +{ > > > +#ifdef CONFIG_MEMCG > > > + if (mm->owner == p) > > > + WRITE_ONCE(mm->owner, NULL); > > > +#endif > > > > How can we ever hit this warning and what does that mean? > > Which warning? A brain fart, I would have sworn that I've seen WARN_ON_ONCE. Sorry about the confusion.
This patch is presently stuck. AFAICT we just need a changelog update to reflect Michal's observations? From: Andrea Arcangeli <aarcange@redhat.com> Subject: userfaultfd: use RCU to free the task struct when fork fails MEMCG depends on the task structure not to be freed under rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences mm->owner. An alternate possible fix would be to defer the delivery of the userfaultfd contexts to the monitor until after fork() is guaranteed to succeed. Such a change would require more changes because it would create a strict ordering dependency where the uffd methods would need to be called beyond the last potentially failing branch in order to be safe. This solution as opposed only adds the dependency to common code to set mm->owner to NULL and to free the task struct that was pointed by mm->owner with RCU, if fork ends up failing. The userfaultfd methods can still be called anywhere during the fork runtime and the monitor will keep discarding orphaned "mm" coming from failed forks in userland. This race condition couldn't trigger if CONFIG_MEMCG was set =n at build time. Link: http://lkml.kernel.org/r/20190325225636.11635-2-aarcange@redhat.com Fixes: 893e26e61d04 ("userfaultfd: non-cooperative: Add fork() event") Signed-off-by: Andrea Arcangeli <aarcange@redhat.com> Tested-by: zhong jiang <zhongjiang@huawei.com> Reported-by: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com Cc: Oleg Nesterov <oleg@redhat.com> Cc: Jann Horn <jannh@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Peter Xu <peterx@redhat.com> Cc: Jason Gunthorpe <jgg@mellanox.com> Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com> Cc: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Cc: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com Signed-off-by: Andrew Morton <akpm@linux-foundation.org> --- kernel/fork.c | 34 ++++++++++++++++++++++++++++++++-- 1 file changed, 32 insertions(+), 2 deletions(-) --- a/kernel/fork.c~userfaultfd-use-rcu-to-free-the-task-struct-when-fork-fails +++ a/kernel/fork.c @@ -952,6 +952,15 @@ static void mm_init_aio(struct mm_struct #endif } +static __always_inline void mm_clear_owner(struct mm_struct *mm, + struct task_struct *p) +{ +#ifdef CONFIG_MEMCG + if (mm->owner == p) + WRITE_ONCE(mm->owner, NULL); +#endif +} + static void mm_init_owner(struct mm_struct *mm, struct task_struct *p) { #ifdef CONFIG_MEMCG @@ -1331,6 +1340,7 @@ static struct mm_struct *dup_mm(struct t free_pt: /* don't put binfmt in mmput, we haven't got module yet */ mm->binfmt = NULL; + mm_init_owner(mm, NULL); mmput(mm); fail_nomem: @@ -1662,6 +1672,24 @@ static inline void rcu_copy_process(stru #endif /* #ifdef CONFIG_TASKS_RCU */ } +#ifdef CONFIG_MEMCG +static void __delayed_free_task(struct rcu_head *rhp) +{ + struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); + + free_task(tsk); +} +#endif /* CONFIG_MEMCG */ + +static __always_inline void delayed_free_task(struct task_struct *tsk) +{ +#ifdef CONFIG_MEMCG + call_rcu(&tsk->rcu, __delayed_free_task); +#else /* CONFIG_MEMCG */ + free_task(tsk); +#endif /* CONFIG_MEMCG */ +} + /* * This creates a new process as a copy of the old one, * but does not actually start it yet. @@ -2123,8 +2151,10 @@ bad_fork_cleanup_io: bad_fork_cleanup_namespaces: exit_task_namespaces(p); bad_fork_cleanup_mm: - if (p->mm) + if (p->mm) { + mm_clear_owner(p->mm, p); mmput(p->mm); + } bad_fork_cleanup_signal: if (!(clone_flags & CLONE_THREAD)) free_signal_struct(p->signal); @@ -2155,7 +2185,7 @@ bad_fork_cleanup_count: bad_fork_free: p->state = TASK_DEAD; put_task_stack(p); - free_task(p); + delayed_free_task(p); fork_out: spin_lock_irq(¤t->sighand->siglock); hlist_del_init(&delayed.node);
diff --git a/kernel/fork.c b/kernel/fork.c index 9dcd18aa210b..a19790e27afd 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -952,6 +952,15 @@ static void mm_init_aio(struct mm_struct *mm) #endif } +static __always_inline void mm_clear_owner(struct mm_struct *mm, + struct task_struct *p) +{ +#ifdef CONFIG_MEMCG + if (mm->owner == p) + WRITE_ONCE(mm->owner, NULL); +#endif +} + static void mm_init_owner(struct mm_struct *mm, struct task_struct *p) { #ifdef CONFIG_MEMCG @@ -1331,6 +1340,7 @@ static struct mm_struct *dup_mm(struct task_struct *tsk) free_pt: /* don't put binfmt in mmput, we haven't got module yet */ mm->binfmt = NULL; + mm_init_owner(mm, NULL); mmput(mm); fail_nomem: @@ -1662,6 +1672,24 @@ static inline void rcu_copy_process(struct task_struct *p) #endif /* #ifdef CONFIG_TASKS_RCU */ } +#ifdef CONFIG_MEMCG +static void __delayed_free_task(struct rcu_head *rhp) +{ + struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); + + free_task(tsk); +} +#endif /* CONFIG_MEMCG */ + +static __always_inline void delayed_free_task(struct task_struct *tsk) +{ +#ifdef CONFIG_MEMCG + call_rcu(&tsk->rcu, __delayed_free_task); +#else /* CONFIG_MEMCG */ + free_task(tsk); +#endif /* CONFIG_MEMCG */ +} + /* * This creates a new process as a copy of the old one, * but does not actually start it yet. @@ -2123,8 +2151,10 @@ static __latent_entropy struct task_struct *copy_process( bad_fork_cleanup_namespaces: exit_task_namespaces(p); bad_fork_cleanup_mm: - if (p->mm) + if (p->mm) { + mm_clear_owner(p->mm, p); mmput(p->mm); + } bad_fork_cleanup_signal: if (!(clone_flags & CLONE_THREAD)) free_signal_struct(p->signal); @@ -2155,7 +2185,7 @@ static __latent_entropy struct task_struct *copy_process( bad_fork_free: p->state = TASK_DEAD; put_task_stack(p); - free_task(p); + delayed_free_task(p); fork_out: spin_lock_irq(¤t->sighand->siglock); hlist_del_init(&delayed.node);
MEMCG depends on the task structure not to be freed under rcu_read_lock() in get_mem_cgroup_from_mm() after it dereferences mm->owner. An alternate possible fix would be to defer the delivery of the userfaultfd contexts to the monitor until after fork() is guaranteed to succeed. Such a change would require more changes because it would create a strict ordering dependency where the uffd methods would need to be called beyond the last potentially failing branch in order to be safe. This solution as opposed only adds the dependency to common code to set mm->owner to NULL and to free the task struct that was pointed by mm->owner with RCU, if fork ends up failing. The userfaultfd methods can still be called anywhere during the fork runtime and the monitor will keep discarding orphaned "mm" coming from failed forks in userland. This race condition couldn't trigger if CONFIG_MEMCG was set =n at build time. Fixes: 893e26e61d04 ("userfaultfd: non-cooperative: Add fork() event") Cc: stable@kernel.org Tested-by: zhong jiang <zhongjiang@huawei.com> Reported-by: syzbot+cbb52e396df3e565ab02@syzkaller.appspotmail.com Signed-off-by: Andrea Arcangeli <aarcange@redhat.com> --- kernel/fork.c | 34 ++++++++++++++++++++++++++++++++-- 1 file changed, 32 insertions(+), 2 deletions(-)