On 16/08/2021 06:05, bongsu.jeon2@gmail.com wrote:
From: Bongsu Jeon bongsu.jeon@samsung.com
In previous version, the user level virtual device application that used this driver should have the polling scheme to read a NCI frame. To remove this polling scheme, changed the driver code to use Wait Queue.
Subject - please prefix it with: "nfc: virtual_ncidev: "
Also make it simpler (skipping unnecessary words like "change", "device driver"), so: "nfc: virtual_ncidev: use wait queue instead of polling"
Signed-off-by: Bongsu Jeon bongsu.jeon@samsung.com
drivers/nfc/virtual_ncidev.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/drivers/nfc/virtual_ncidev.c b/drivers/nfc/virtual_ncidev.c index 2ee0ec4bb739..1953904176a2 100644 --- a/drivers/nfc/virtual_ncidev.c +++ b/drivers/nfc/virtual_ncidev.c @@ -10,6 +10,7 @@ #include <linux/module.h> #include <linux/miscdevice.h> #include <linux/mutex.h> +#include <linux/wait.h> #include <net/nfc/nci_core.h> enum virtual_ncidev_mode { @@ -27,6 +28,7 @@ enum virtual_ncidev_mode { NFC_PROTO_ISO15693_MASK) static enum virtual_ncidev_mode state; +static DECLARE_WAIT_QUEUE_HEAD(wq); static struct miscdevice miscdev; static struct sk_buff *send_buff; static struct nci_dev *ndev; @@ -61,6 +63,7 @@ static int virtual_nci_send(struct nci_dev *ndev, struct sk_buff *skb) } send_buff = skb_copy(skb, GFP_KERNEL); mutex_unlock(&nci_mutex);
- wake_up_interruptible(&wq);
return 0; } @@ -76,12 +79,11 @@ static ssize_t virtual_ncidev_read(struct file *file, char __user *buf, { size_t actual_len;
- mutex_lock(&nci_mutex);
- if (!send_buff) {
mutex_unlock(&nci_mutex);
- wait_event_interruptible(wq, send_buff);
- if (!send_buff)
I think access to send_buff should still be protected by mutex. What happens if you have to readers?
return 0;
- }
- mutex_lock(&nci_mutex); actual_len = min_t(size_t, count, send_buff->len);
if (copy_to_user(buf, send_buff->data, actual_len)) {
Best regards, Krzysztof