From: Wen Yang <[email protected]>
wait_event_interruptible_locked_irq was introduced by commit 22c43c81a51e
("wait_event_interruptible_locked() interface"), but older code such as
eventfd_{write,read} still uses the open code implementation.
Inspired by commit 8120a8aadb20
("fs/timerfd.c: make use of wait_event_interruptible_locked_irq()"), this
patch replaces the open code implementation with a single macro call.
No functional change intended.
Signed-off-by: Wen Yang <[email protected]>
Reviewed-by: Eric Biggers <[email protected]>
Cc: Alexander Viro <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Dylan Yudaken <[email protected]>
Cc: Jens Axboe <[email protected]>
Cc: David Woodhouse <[email protected]>
Cc: Fu Wei <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Michal Nazarewicz <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: [email protected]
Cc: [email protected]
---
fs/eventfd.c | 41 +++++++----------------------------------
1 file changed, 7 insertions(+), 34 deletions(-)
diff --git a/fs/eventfd.c b/fs/eventfd.c
index 249ca6c0b784..95850a13ce8d 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -228,7 +228,6 @@ static ssize_t eventfd_read(struct kiocb *iocb, struct iov_iter *to)
struct file *file = iocb->ki_filp;
struct eventfd_ctx *ctx = file->private_data;
__u64 ucnt = 0;
- DECLARE_WAITQUEUE(wait, current);
if (iov_iter_count(to) < sizeof(ucnt))
return -EINVAL;
@@ -239,23 +238,11 @@ static ssize_t eventfd_read(struct kiocb *iocb, struct iov_iter *to)
spin_unlock_irq(&ctx->wqh.lock);
return -EAGAIN;
}
- __add_wait_queue(&ctx->wqh, &wait);
- for (;;) {
- set_current_state(TASK_INTERRUPTIBLE);
- if (ctx->count)
- break;
- if (signal_pending(current)) {
- __remove_wait_queue(&ctx->wqh, &wait);
- __set_current_state(TASK_RUNNING);
- spin_unlock_irq(&ctx->wqh.lock);
- return -ERESTARTSYS;
- }
+
+ if (wait_event_interruptible_locked_irq(ctx->wqh, ctx->count)) {
spin_unlock_irq(&ctx->wqh.lock);
- schedule();
- spin_lock_irq(&ctx->wqh.lock);
+ return -ERESTARTSYS;
}
- __remove_wait_queue(&ctx->wqh, &wait);
- __set_current_state(TASK_RUNNING);
}
eventfd_ctx_do_read(ctx, &ucnt);
current->in_eventfd = 1;
@@ -275,7 +262,6 @@ static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t c
struct eventfd_ctx *ctx = file->private_data;
ssize_t res;
__u64 ucnt;
- DECLARE_WAITQUEUE(wait, current);
if (count < sizeof(ucnt))
return -EINVAL;
@@ -288,23 +274,10 @@ static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t c
if (ULLONG_MAX - ctx->count > ucnt)
res = sizeof(ucnt);
else if (!(file->f_flags & O_NONBLOCK)) {
- __add_wait_queue(&ctx->wqh, &wait);
- for (res = 0;;) {
- set_current_state(TASK_INTERRUPTIBLE);
- if (ULLONG_MAX - ctx->count > ucnt) {
- res = sizeof(ucnt);
- break;
- }
- if (signal_pending(current)) {
- res = -ERESTARTSYS;
- break;
- }
- spin_unlock_irq(&ctx->wqh.lock);
- schedule();
- spin_lock_irq(&ctx->wqh.lock);
- }
- __remove_wait_queue(&ctx->wqh, &wait);
- __set_current_state(TASK_RUNNING);
+ res = wait_event_interruptible_locked_irq(ctx->wqh,
+ ULLONG_MAX - ctx->count > ucnt);
+ if (!res)
+ res = sizeof(ucnt);
}
if (likely(res > 0)) {
ctx->count += ucnt;
--
2.37.2
On 4/5/23 1:20 PM, [email protected] wrote:
> From: Wen Yang <[email protected]>
>
> wait_event_interruptible_locked_irq was introduced by commit 22c43c81a51e
> ("wait_event_interruptible_locked() interface"), but older code such as
> eventfd_{write,read} still uses the open code implementation.
> Inspired by commit 8120a8aadb20
> ("fs/timerfd.c: make use of wait_event_interruptible_locked_irq()"), this
> patch replaces the open code implementation with a single macro call.
>
> No functional change intended.
Looks pretty reasonable to me. How did you test it?
--
Jens Axboe
在 2023/4/6 03:26, Jens Axboe 写道:
> On 4/5/23 1:20 PM, [email protected] wrote:
>> From: Wen Yang <[email protected]>
>>
>> wait_event_interruptible_locked_irq was introduced by commit 22c43c81a51e
>> ("wait_event_interruptible_locked() interface"), but older code such as
>> eventfd_{write,read} still uses the open code implementation.
>> Inspired by commit 8120a8aadb20
>> ("fs/timerfd.c: make use of wait_event_interruptible_locked_irq()"), this
>> patch replaces the open code implementation with a single macro call.
>>
>> No functional change intended.
> Looks pretty reasonable to me. How did you test it?
>
Thanks.
We have verified it in some local testing environments, and the
intel-lab-lkp has also tested it more than a month, as follows:
https://github.com/intel-lab-lkp/linux/tree/wenyang-linux-foxmail-com/eventfd-use-wait_event_interruptible_locked_irq-helper/20230217-023039
--
Best wishes,
Wen
On 4/5/23 1:40 PM, Wen Yang wrote:
>
> 在 2023/4/6 03:26, Jens Axboe 写道:
>> On 4/5/23 1:20 PM, [email protected] wrote:
>>> From: Wen Yang <[email protected]>
>>>
>>> wait_event_interruptible_locked_irq was introduced by commit 22c43c81a51e
>>> ("wait_event_interruptible_locked() interface"), but older code such as
>>> eventfd_{write,read} still uses the open code implementation.
>>> Inspired by commit 8120a8aadb20
>>> ("fs/timerfd.c: make use of wait_event_interruptible_locked_irq()"), this
>>> patch replaces the open code implementation with a single macro call.
>>>
>>> No functional change intended.
>> Looks pretty reasonable to me. How did you test it?
>>
> Thanks.
>
> We have verified it in some local testing environments, and the intel-lab-lkp has also tested it more than a month, as follows:
>
> https://github.com/intel-lab-lkp/linux/tree/wenyang-linux-foxmail-com/eventfd-use-wait_event_interruptible_locked_irq-helper/20230217-023039
Nice, you can add my:
Reviewed-by: Jens Axboe <[email protected]>
to it as well. Adding Christian as well, as I guess this should go
through his/al/vfs tree.
--
Jens Axboe
On Thu, 06 Apr 2023 03:20:02 +0800, [email protected] wrote:
> wait_event_interruptible_locked_irq was introduced by commit 22c43c81a51e
> ("wait_event_interruptible_locked() interface"), but older code such as
> eventfd_{write,read} still uses the open code implementation.
> Inspired by commit 8120a8aadb20
> ("fs/timerfd.c: make use of wait_event_interruptible_locked_irq()"), this
> patch replaces the open code implementation with a single macro call.
>
> [...]
I ran LTP with ./runltp -f syscalls -s eventfd passes and aligns with what was
done for timerfd.
Applied, thanks!
tree: git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping.git
branch: fs.misc
[1/1] eventfd: use wait_event_interruptible_locked_irq() helper
commit: 113348a44b8622b497fb884f41c8659481ad0b04