Userspace generally expects APIs that return EMSGSIZE to allow for them
to adjust their buffer size and retry the operation. However, the
fscontext log would previously clear the message even in the EMSGSIZE
case.

Given that it is very cheap for us to check whether the buffer is too
small before we remove the message from the ring buffer, let's just do
that instead.

Fixes: 007ec26cdc9f ("vfs: Implement logging through fs_context")
Cc: David Howells <dhowells@redhat.com>
Cc: <stable@vger.kernel.org> # v5.2+
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
---
 fs/fsopen.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/fs/fsopen.c b/fs/fsopen.c
index 1aaf4cb2afb2..f5fdaa97965b 100644
--- a/fs/fsopen.c
+++ b/fs/fsopen.c
@@ -36,23 +36,25 @@ static ssize_t fscontext_read(struct file *file,
 	if (ret < 0)
 		return ret;
 
-	if (log->head == log->tail) {
-		mutex_unlock(&fc->uapi_mutex);
-		return -ENODATA;
-	}
+	ret = -ENODATA;
+	if (log->head == log->tail)
+		goto err_unlock_nomsg;
 
 	index = log->tail & (logsize - 1);
 	p = log->buffer[index];
+	n = strlen(p);
+
+	ret = -EMSGSIZE;
+	if (n > len)
+		goto err_unlock_nomsg;
+
+	/* Consume the message from the queue. */
 	need_free = log->need_free & (1 << index);
 	log->buffer[index] = NULL;
 	log->need_free &= ~(1 << index);
 	log->tail++;
 	mutex_unlock(&fc->uapi_mutex);
 
-	ret = -EMSGSIZE;
-	n = strlen(p);
-	if (n > len)
-		goto err_free;
 	ret = -EFAULT;
 	if (copy_to_user(_buf, p, n) != 0)
 		goto err_free;
@@ -62,6 +64,10 @@ static ssize_t fscontext_read(struct file *file,
 	if (need_free)
 		kfree(p);
 	return ret;
+
+err_unlock_nomsg:
+	mutex_unlock(&fc->uapi_mutex);
+	return ret;
 }
 
 static int fscontext_release(struct inode *inode, struct file *file)

-- 
2.50.1

On Wed, Aug 06, 2025 at 03:31:10PM +1000, Aleksa Sarai wrote:
> Userspace generally expects APIs that return EMSGSIZE to allow for them
> to adjust their buffer size and retry the operation. However, the
> fscontext log would previously clear the message even in the EMSGSIZE
> case.
> 
> Given that it is very cheap for us to check whether the buffer is too
> small before we remove the message from the ring buffer, let's just do
> that instead.
> 
> Fixes: 007ec26cdc9f ("vfs: Implement logging through fs_context")
> Cc: David Howells <dhowells@redhat.com>
> Cc: <stable@vger.kernel.org> # v5.2+
> Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
> ---
>  fs/fsopen.c | 22 ++++++++++++++--------
>  1 file changed, 14 insertions(+), 8 deletions(-)
> 
> diff --git a/fs/fsopen.c b/fs/fsopen.c
> index 1aaf4cb2afb2..f5fdaa97965b 100644
> --- a/fs/fsopen.c
> +++ b/fs/fsopen.c
> @@ -36,23 +36,25 @@ static ssize_t fscontext_read(struct file *file,
>  	if (ret < 0)
>  		return ret;
>  
> -	if (log->head == log->tail) {
> -		mutex_unlock(&fc->uapi_mutex);
> -		return -ENODATA;
> -	}
> +	ret = -ENODATA;
> +	if (log->head == log->tail)
> +		goto err_unlock_nomsg;
>  
>  	index = log->tail & (logsize - 1);
>  	p = log->buffer[index];
> +	n = strlen(p);
> +
> +	ret = -EMSGSIZE;
> +	if (n > len)
> +		goto err_unlock_nomsg;
> +

FWIW, I would rather turn that into a helper taking log and len and
returning p or ERR_PTR(...); something like

static inline const char *fetch_message(struct fc_log *log, size_t len,
					bool *need_free)
{
	int index = log->tail & (ARRAY_SIZE(log->buffer) - 1);
	const char *p = log->buffer[index];

	if (unlikely(log->head == log->tail))
		return ERR_PTR(-ENODATA);

	n = strlen(p);
	if (unlikely(n > len))
		return ERR_PTR(-EMSGSIZE);

	log->buffer[index] = NULL;
	*need_free = log->need_free & (1 << index);
	log->need_free &= ~(1 << index);
	log->tail++;

	return p;
}

with caller being

        ret = mutex_lock_interruptible(&fc->uapi_mutex);
	if (ret < 0)
		return ret;
	p = fetch_message(log, len, &need_free);
	mutex_unlock(&fc->uapi_mutex);
	if (IS_ERR(p))
		return PTR_ERR(p);
	n = strlen(p);
	if (copy_to_user(_buf, p, n))
		n = -EFAULT;
	if (need_free)
		kfree(p);
	return n;

and that's it.