Otherwise we risk sleeping with the pipe locked for indeterminate
lengths of time ‒ given:
cat > kcm.c <<^D
#define _GNU_SOURCE
#include <fcntl.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <linux/kcm.h>
int main()
{
int kcm = socket(AF_KCM, SOCK_SEQPACKET, KCMPROTO_CONNECTED);
for (;;)
splice(kcm, 0, 1, 0, 128 * 1024 * 1024, 0);
}
^D
cc kcm.c -o kcm
mkfifo fifo
./kcm > fifo &
read -r _ < fifo &
sleep 0.1
echo zupa > fifo
kcm used to sleep in splice and the shell used to enter an
uninterruptible sleep in open("fifo");
now the splice returns -EAGAIN and the whole program completes.
Also: don't pass the SPLICE_F_*-style flags argument to
skb_recv_datagram(), which expects MSG_*-style flags.
This fixes SPLICE_F_NONBLOCK not having worked.
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@xxxxxxxxxxxxxxxxxx>
---
net/kcm/kcmsock.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/kcm/kcmsock.c b/net/kcm/kcmsock.c
index 65d1f6755f98..ccfc46f31891 100644
--- a/net/kcm/kcmsock.c
+++ b/net/kcm/kcmsock.c
@@ -1028,7 +1028,7 @@ static ssize_t kcm_splice_read(struct socket *sock, loff_t *ppos,
/* Only support splice for SOCKSEQPACKET */
- skb = skb_recv_datagram(sk, flags, &err);
+ skb = skb_recv_datagram(sk, MSG_DONTWAIT, &err);
if (!skb)
goto err_out;
--
2.39.2
Attachment:
signature.asc
Description: PGP signature
