| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <cd0f6613-9aa9-4698-bebe-0f61286d7552@quicinc.com>
Date: Tue, 2 Jan 2024 10:19:47 +0800
From: "Aiqun Yu (Maria)" <quic_aiquny@...cinc.com>
To: Matthew Wilcox <willy@...radead.org>,
"Eric W. Biederman"
<ebiederm@...ssion.com>,
Hillf Danton <hdanton@...a.com>
CC: <kernel@...cinc.com>, <quic_pkondeti@...cinc.com>, <keescook@...omium.or>,
<viro@...iv.linux.org.uk>, <brauner@...nel.org>, <oleg@...hat.com>,
<dhowells@...hat.com>, <jarkko@...nel.org>, <paul@...l-moore.com>,
<jmorris@...ei.org>, <serge@...lyn.com>, <linux-mm@...ck.org>,
<linux-fsdevel@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
<keyrings@...r.kernel.org>, <linux-security-module@...r.kernel.org>,
<linux-arm-msm@...r.kernel.org>
Subject: Re: [PATCH] kernel: Introduce a write lock/unlock wrapper for
tasklist_lock
On 12/29/2023 6:20 AM, Matthew Wilcox wrote:
> On Wed, Dec 13, 2023 at 12:27:05PM -0600, Eric W. Biederman wrote:
>> Matthew Wilcox <willy@...radead.org> writes:
>>> I think the right way to fix this is to pass a boolean flag to
>>> queued_write_lock_slowpath() to let it know whether it can re-enable
>>> interrupts while checking whether _QW_WAITING is set.
>>
>> Yes. It seems to make sense to distinguish between write_lock_irq and
>> write_lock_irqsave and fix this for all of write_lock_irq.
>
> I wasn't planning on doing anything here, but Hillf kind of pushed me into
> it. I think it needs to be something like this. Compile tested only.
> If it ends up getting used,
Happy new year!
Thx Metthew for chiming into this. I think more thoughts will gain more
perfect designs.
>
> Signed-off-by: Matthew Wilcox (Oracle) <willy@...radead.org>
>
> diff --git a/include/asm-generic/qrwlock.h b/include/asm-generic/qrwlock.h
> index 75b8f4601b28..1152e080c719 100644
> --- a/include/asm-generic/qrwlock.h
> +++ b/include/asm-generic/qrwlock.h
> @@ -33,8 +33,8 @@
> /*
> * External function declarations
> */
> -extern void queued_read_lock_slowpath(struct qrwlock *lock);
> -extern void queued_write_lock_slowpath(struct qrwlock *lock);
> +void queued_read_lock_slowpath(struct qrwlock *lock);
> +void queued_write_lock_slowpath(struct qrwlock *lock, bool irq);
>
> /**
> * queued_read_trylock - try to acquire read lock of a queued rwlock
> @@ -98,7 +98,21 @@ static inline void queued_write_lock(struct qrwlock *lock)
> if (likely(atomic_try_cmpxchg_acquire(&lock->cnts, &cnts, _QW_LOCKED)))
> return;
>
> - queued_write_lock_slowpath(lock);
> + queued_write_lock_slowpath(lock, false);
> +}
> +
> +/**
> + * queued_write_lock_irq - acquire write lock of a queued rwlock
> + * @lock : Pointer to queued rwlock structure
> + */
> +static inline void queued_write_lock_irq(struct qrwlock *lock)
> +{
> + int cnts = 0;
> + /* Optimize for the unfair lock case where the fair flag is 0. */
> + if (likely(atomic_try_cmpxchg_acquire(&lock->cnts, &cnts, _QW_LOCKED)))
> + return;
> +
> + queued_write_lock_slowpath(lock, true);
> }
>
> /**
> @@ -138,6 +152,7 @@ static inline int queued_rwlock_is_contended(struct qrwlock *lock)
> */
> #define arch_read_lock(l) queued_read_lock(l)
> #define arch_write_lock(l) queued_write_lock(l)
> +#define arch_write_lock_irq(l) queued_write_lock_irq(l)
> #define arch_read_trylock(l) queued_read_trylock(l)
> #define arch_write_trylock(l) queued_write_trylock(l)
> #define arch_read_unlock(l) queued_read_unlock(l)
> diff --git a/include/linux/rwlock.h b/include/linux/rwlock.h
> index c0ef596f340b..897010b6ba0a 100644
> --- a/include/linux/rwlock.h
> +++ b/include/linux/rwlock.h
> @@ -33,6 +33,7 @@ do { \
> extern int do_raw_read_trylock(rwlock_t *lock);
> extern void do_raw_read_unlock(rwlock_t *lock) __releases(lock);
> extern void do_raw_write_lock(rwlock_t *lock) __acquires(lock);
> + extern void do_raw_write_lock_irq(rwlock_t *lock) __acquires(lock);
> extern int do_raw_write_trylock(rwlock_t *lock);
> extern void do_raw_write_unlock(rwlock_t *lock) __releases(lock);
> #else
> @@ -40,6 +41,7 @@ do { \
> # define do_raw_read_trylock(rwlock) arch_read_trylock(&(rwlock)->raw_lock)
> # define do_raw_read_unlock(rwlock) do {arch_read_unlock(&(rwlock)->raw_lock); __release(lock); } while (0)
> # define do_raw_write_lock(rwlock) do {__acquire(lock); arch_write_lock(&(rwlock)->raw_lock); } while (0)
> +# define do_raw_write_lock_irq(rwlock) do {__acquire(lock); arch_write_lock_irq(&(rwlock)->raw_lock); } while (0)
> # define do_raw_write_trylock(rwlock) arch_write_trylock(&(rwlock)->raw_lock)
> # define do_raw_write_unlock(rwlock) do {arch_write_unlock(&(rwlock)->raw_lock); __release(lock); } while (0)
> #endif
> diff --git a/include/linux/rwlock_api_smp.h b/include/linux/rwlock_api_smp.h
> index dceb0a59b692..6257976dfb72 100644
> --- a/include/linux/rwlock_api_smp.h
> +++ b/include/linux/rwlock_api_smp.h
> @@ -193,7 +193,7 @@ static inline void __raw_write_lock_irq(rwlock_t *lock)
> local_irq_disable();
> preempt_disable();
> rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
> - LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock);
> + LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock_irq);
> }
>
> static inline void __raw_write_lock_bh(rwlock_t *lock)
> diff --git a/kernel/locking/qrwlock.c b/kernel/locking/qrwlock.c
> index d2ef312a8611..6c644a71b01d 100644
> --- a/kernel/locking/qrwlock.c
> +++ b/kernel/locking/qrwlock.c
> @@ -61,9 +61,10 @@ EXPORT_SYMBOL(queued_read_lock_slowpath);
>
> /**
> * queued_write_lock_slowpath - acquire write lock of a queued rwlock
> - * @lock : Pointer to queued rwlock structure
> + * @lock: Pointer to queued rwlock structure
> + * @irq: True if we can enable interrupts while spinning
> */
> -void __lockfunc queued_write_lock_slowpath(struct qrwlock *lock)
> +void __lockfunc queued_write_lock_slowpath(struct qrwlock *lock, bool irq)
> {
> int cnts;
>
> @@ -82,7 +83,11 @@ void __lockfunc queued_write_lock_slowpath(struct qrwlock *lock)
>
Also a new state showed up after the current design:
1. locked flag with _QW_WAITING, while irq enabled.
2. And this state will be only in interrupt context.
3. lock->wait_lock is hold by the write waiter.
So per my understanding, a different behavior also needed to be done in
queued_write_lock_slowpath:
when (unlikely(in_interrupt())) , get the lock directly.
So needed to be done in release path. This is to address Hillf's concern
on possibility of deadlock.
Add Hillf here to merge thread. I am going to have a tested patch V2
accordingly.
Feel free to let me know your thoughts prior on that.
> /* When no more readers or writers, set the locked flag */
> do {
> + if (irq)
> + local_irq_enable();
I think write_lock_irqsave also needs to be take account. So
loal_irq_save(flags) should be take into account here.
> cnts = atomic_cond_read_relaxed(&lock->cnts, VAL == _QW_WAITING);
> + if (irq)
> + local_irq_disable();
ditto.
> } while (!atomic_try_cmpxchg_acquire(&lock->cnts, &cnts, _QW_LOCKED));
> unlock:
> arch_spin_unlock(&lock->wait_lock);
> diff --git a/kernel/locking/spinlock_debug.c b/kernel/locking/spinlock_debug.c
> index 87b03d2e41db..bf94551d7435 100644
> --- a/kernel/locking/spinlock_debug.c
> +++ b/kernel/locking/spinlock_debug.c
> @@ -212,6 +212,13 @@ void do_raw_write_lock(rwlock_t *lock)
> debug_write_lock_after(lock);
> }
>
> +void do_raw_write_lock_irq(rwlock_t *lock)
> +{
> + debug_write_lock_before(lock);
> + arch_write_lock_irq(&lock->raw_lock);
> + debug_write_lock_after(lock);
> +}
> +
> int do_raw_write_trylock(rwlock_t *lock)
> {
> int ret = arch_write_trylock(&lock->raw_lock);
--
Thx and BRs,
Aiqun(Maria) Yu
Powered by blists - more mailing lists