This is AF_PACKET mmap compat support.
I confirmed that 32-bit and 64-bit tcpdump binaries using AF_PACKET
mmap'ed pcap can dump packets properly with a 64-bit kernel.
Thanks to David for all your help.
Signed-off-by: FUJITA Tomonori <[EMAIL PROTECTED]>
---
net/packet/af_packet.c | 156 +++++++++++++++++++++++++++++++++++++++---------
1 files changed, 128 insertions(+), 28 deletions(-)
a4fd7acdecf12871d91976610cd3d1442807b206
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 9db7dbd..71ab97a 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -78,6 +78,7 @@
#include <linux/poll.h>
#include <linux/module.h>
#include <linux/init.h>
+#include <linux/compat.h>
#ifdef CONFIG_INET
#include <net/inet_common.h>
@@ -182,6 +183,31 @@ struct packet_mreq_max
#endif
#ifdef CONFIG_PACKET_MMAP
static int packet_set_ring(struct sock *sk, struct tpacket_req *req, int
closing);
+
+#ifdef CONFIG_COMPAT
+struct tpacket_hdr_compat {
+ compat_ulong_t tp_status;
+ compat_uint_t tp_len;
+ compat_uint_t tp_snaplen;
+ unsigned short tp_mac;
+ unsigned short tp_net;
+ compat_uint_t tp_sec;
+ compat_uint_t tp_usec;
+};
+
+#ifdef CONFIG_X86_64
+# define COMPAT_TEST is_compat_task()
+#elif defined(CONFIG_IA64)
+# define COMPAT_TEST IS_IA32_PROCESS(task_pt_regs(current))
+#elif defined(CONFIG_S390)
+# define COMPAT_TEST test_thread_flag(TIF_31BIT)
+#elif defined(CONFIG_MIPS)
+# define COMPAT_TEST (current->thread.mflags & MF_32BIT_ADDR)
+#else
+# define COMPAT_TEST test_thread_flag(TIF_32BIT)
+#endif
+#endif
+
#endif
static void packet_flush_mclist(struct sock *sk);
@@ -211,6 +237,9 @@ struct packet_sock {
unsigned int pg_vec_order;
unsigned int pg_vec_pages;
unsigned int pg_vec_len;
+#ifdef CONFIG_COMPAT
+ int compat;
+#endif
#endif
};
@@ -562,15 +591,93 @@ drop:
}
#ifdef CONFIG_PACKET_MMAP
+#define __tpacket_hdr_update(h, skb, snaplen, mac, net, status)
\
+do { \
+ h->tp_len = (skb)->len; \
+ h->tp_snaplen = snaplen; \
+ h->tp_mac = mac; \
+ h->tp_net = net; \
+ h->tp_sec = (skb)->tstamp.off_sec; \
+ h->tp_usec = (skb)->tstamp.off_usec; \
+ h->tp_status = status; \
+} while (0)
+
+#ifdef CONFIG_COMPAT
+static void tpacket_hdr_update(struct packet_sock *po, unsigned int idx,
+ struct sk_buff *skb,
+ unsigned int snaplen, unsigned short mac,
+ unsigned short net, unsigned long status)
+{
+ if (po->compat) {
+ struct tpacket_hdr_compat *h;
+ h = (struct tpacket_hdr_compat *)
+ packet_lookup_frame(po, idx);
+ __tpacket_hdr_update(h, skb, snaplen, mac, net, status);
+ } else {
+ struct tpacket_hdr *h;
+ h = (struct tpacket_hdr *) packet_lookup_frame(po, idx);
+ __tpacket_hdr_update(h, skb, snaplen, mac, net, status);
+ }
+ mb();
+}
+
+static unsigned int tpacket_hdr_status(struct packet_sock *po, unsigned int
idx)
+{
+ if (po->compat) {
+ struct tpacket_hdr_compat *h;
+ h = (struct tpacket_hdr_compat *)
+ packet_lookup_frame(po, idx);
+ return h->tp_status;
+ } else {
+ struct tpacket_hdr *h;
+ h = (struct tpacket_hdr *) packet_lookup_frame(po, idx);
+ return h->tp_status;
+ }
+}
+
+static size_t tpacket_hdr_size(struct packet_sock *po)
+{
+ return po->compat ? sizeof(struct tpacket_hdr_compat) :
+ sizeof(struct tpacket_hdr);
+}
+#else
+
+static void tpacket_hdr_update(struct packet_sock *po, unsigned int idx,
+ struct sk_buff *skb,
+ unsigned int snaplen, unsigned short mac,
+ unsigned short net, unsigned long status)
+{
+ struct tpacket_hdr *h;
+ h = (struct tpacket_hdr *) packet_lookup_frame(po, idx);
+ __tpacket_hdr_update(h, skb, snaplen, mac, net, status);
+}
+
+static unsigned long tpacket_hdr_status(struct packet_sock *po, unsigned int
idx)
+{
+ struct tpacket_hdr *h;
+ h = (struct tpacket_hdr *) packet_lookup_frame(po, idx);
+ return h->tp_status;
+}
+
+static size_t tpacket_hdr_size(struct packet_sock *po)
+{
+ return sizeof(struct tpacket_hdr);
+}
+#endif
+
+static int tpacket_hdrlen(struct packet_sock *po)
+{
+ return TPACKET_ALIGN(tpacket_hdr_size(po)) + sizeof(struct sockaddr_ll);
+}
+
static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev, struct
packet_type *pt, struct net_device *orig_dev)
{
struct sock *sk;
struct packet_sock *po;
struct sockaddr_ll *sll;
- struct tpacket_hdr *h;
- u8 * skb_head = skb->data;
+ u8 * skb_head = skb->data, *cur_frame;
int skb_len = skb->len;
- unsigned snaplen;
+ unsigned snaplen, idx;
unsigned long status = TP_STATUS_LOSING|TP_STATUS_USER;
unsigned short macoff, netoff;
struct sk_buff *copy_skb = NULL;
@@ -603,10 +710,10 @@ static int tpacket_rcv(struct sk_buff *s
}
if (sk->sk_type == SOCK_DGRAM) {
- macoff = netoff = TPACKET_ALIGN(TPACKET_HDRLEN) + 16;
+ macoff = netoff = TPACKET_ALIGN(tpacket_hdrlen(po)) + 16;
} else {
unsigned maclen = skb->nh.raw - skb->data;
- netoff = TPACKET_ALIGN(TPACKET_HDRLEN + (maclen < 16 ? 16 :
maclen));
+ netoff = TPACKET_ALIGN(tpacket_hdrlen(po) + (maclen < 16 ? 16 :
maclen));
macoff = netoff - maclen;
}
@@ -631,9 +738,9 @@ static int tpacket_rcv(struct sk_buff *s
snaplen = skb->len-skb->data_len;
spin_lock(&sk->sk_receive_queue.lock);
- h = (struct tpacket_hdr *)packet_lookup_frame(po, po->head);
-
- if (h->tp_status)
+ idx = po->head;
+ cur_frame = packet_lookup_frame(po, idx);
+ if (tpacket_hdr_status(po, idx))
goto ring_is_full;
po->head = po->head != po->frame_max ? po->head+1 : 0;
po->stats.tp_packets++;
@@ -645,20 +752,14 @@ static int tpacket_rcv(struct sk_buff *s
status &= ~TP_STATUS_LOSING;
spin_unlock(&sk->sk_receive_queue.lock);
- memcpy((u8*)h + macoff, skb->data, snaplen);
+ memcpy(cur_frame + macoff, skb->data, snaplen);
- h->tp_len = skb->len;
- h->tp_snaplen = snaplen;
- h->tp_mac = macoff;
- h->tp_net = netoff;
- if (skb->tstamp.off_sec == 0) {
+ if (skb->tstamp.off_sec == 0) {
__net_timestamp(skb);
sock_enable_timestamp(sk);
}
- h->tp_sec = skb->tstamp.off_sec;
- h->tp_usec = skb->tstamp.off_usec;
- sll = (struct sockaddr_ll*)((u8*)h + TPACKET_ALIGN(sizeof(*h)));
+ sll = (struct sockaddr_ll*)(cur_frame +
TPACKET_ALIGN(tpacket_hdr_size(po)));
sll->sll_halen = 0;
if (dev->hard_header_parse)
sll->sll_halen = dev->hard_header_parse(skb, sll->sll_addr);
@@ -668,14 +769,12 @@ static int tpacket_rcv(struct sk_buff *s
sll->sll_pkttype = skb->pkt_type;
sll->sll_ifindex = dev->ifindex;
- h->tp_status = status;
- mb();
-
+ tpacket_hdr_update(po, idx, skb, snaplen, macoff, netoff, status);
{
struct page *p_start, *p_end;
- u8 *h_end = (u8 *)h + macoff + snaplen - 1;
+ u8 *h_end = cur_frame + macoff + snaplen - 1;
- p_start = virt_to_page(h);
+ p_start = virt_to_page(cur_frame);
p_end = virt_to_page(h_end);
while (p_start <= p_end) {
flush_dcache_page(p_start);
@@ -1542,11 +1641,8 @@ static unsigned int packet_poll(struct f
spin_lock_bh(&sk->sk_receive_queue.lock);
if (po->pg_vec) {
unsigned last = po->head ? po->head-1 : po->frame_max;
- struct tpacket_hdr *h;
-
- h = (struct tpacket_hdr *)packet_lookup_frame(po, last);
- if (h->tp_status)
+ if (tpacket_hdr_status(po, last))
mask |= POLLIN | POLLRDNORM;
}
spin_unlock_bh(&sk->sk_receive_queue.lock);
@@ -1636,7 +1732,11 @@ static int packet_set_ring(struct sock *
struct packet_sock *po = pkt_sk(sk);
int was_running, num, order = 0;
int err = 0;
-
+
+#ifdef CONFIG_COMPAT
+ po->compat = COMPAT_TEST;
+#endif
+
if (req->tp_block_nr) {
int i, l;
@@ -1649,7 +1749,7 @@ static int packet_set_ring(struct sock *
return -EINVAL;
if (unlikely(req->tp_block_size & (PAGE_SIZE - 1)))
return -EINVAL;
- if (unlikely(req->tp_frame_size < TPACKET_HDRLEN))
+ if (unlikely(req->tp_frame_size < tpacket_hdrlen(po)))
return -EINVAL;
if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
return -EINVAL;
--
1.1.3
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
