2023-09-26 21:09:49

by Wen Yang

[permalink] [raw]
Subject: [PATCH] eventfd: move 'eventfd-count' printing out of spinlock

From: Wen Yang <[email protected]>

It is better to print debug messages outside of the wqh.lock
spinlock where possible.

Signed-off-by: Wen Yang <[email protected]>
Cc: Alexander Viro <[email protected]>
Cc: Jens Axboe <[email protected]>
Cc: Christian Brauner <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Dylan Yudaken <[email protected]>
Cc: David Woodhouse <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Eric Biggers <[email protected]>
Cc: [email protected]
Cc: [email protected]
---
fs/eventfd.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/eventfd.c b/fs/eventfd.c
index 33a918f9566c..6c5fe0f40aa5 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -295,11 +295,13 @@ static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t c
static void eventfd_show_fdinfo(struct seq_file *m, struct file *f)
{
struct eventfd_ctx *ctx = f->private_data;
+ unsigned long long count;

spin_lock_irq(&ctx->wqh.lock);
- seq_printf(m, "eventfd-count: %16llx\n",
- (unsigned long long)ctx->count);
+ count = ctx->count;
spin_unlock_irq(&ctx->wqh.lock);
+
+ seq_printf(m, "eventfd-count: %16llx\n", count);
seq_printf(m, "eventfd-id: %d\n", ctx->id);
seq_printf(m, "eventfd-semaphore: %d\n",
!!(ctx->flags & EFD_SEMAPHORE));
--
2.25.1


2023-09-27 14:30:03

by Jens Axboe

[permalink] [raw]
Subject: Re: [PATCH] eventfd: move 'eventfd-count' printing out of spinlock

On 9/26/23 8:05 AM, [email protected] wrote:
> From: Wen Yang <[email protected]>
>
> It is better to print debug messages outside of the wqh.lock
> spinlock where possible.

Does it really matter for fdinfo? Your commit message is a bit
light, so I'm having to guess whether this is fixing a real issue
for you, or if it's just a drive-by observation.

--
Jens Axboe


2023-10-02 19:16:57

by Wen Yang

[permalink] [raw]
Subject: Re: [PATCH] eventfd: move 'eventfd-count' printing out of spinlock


On 2023/9/27 15:32, Jens Axboe wrote:
> On 9/26/23 8:05 AM,[email protected] wrote:
>> From: Wen Yang<[email protected]>
>>
>> It is better to print debug messages outside of the wqh.lock
>> spinlock where possible.
> Does it really matter for fdinfo? Your commit message is a bit
> light, so I'm having to guess whether this is fixing a real issue
> for you, or if it's just a drive-by observation.
>
Thanks for your comments.
There is a business running on our machine that frequently calls
eventfd_write().
We will resend the V2 and provide a detailed explanation.

--

Best wishes,
Wen