linux/net/atm/clip.c

930 lines
22 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/* net/atm/clip.c - RFC1577 Classical IP over ATM */
/* Written 1995-2000 by Werner Almesberger, EPFL LRC/ICA */
#define pr_fmt(fmt) KBUILD_MODNAME ":%s: " fmt, __func__
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/kernel.h> /* for UINT_MAX */
#include <linux/module.h>
#include <linux/init.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/wait.h>
#include <linux/timer.h>
#include <linux/if_arp.h> /* for some manifest constants */
#include <linux/notifier.h>
#include <linux/atm.h>
#include <linux/atmdev.h>
#include <linux/atmclip.h>
#include <linux/atmarp.h>
#include <linux/capability.h>
#include <linux/ip.h> /* for net/route.h */
#include <linux/in.h> /* for struct sockaddr_in */
#include <linux/if.h> /* for IFF_UP */
#include <linux/inetdevice.h>
#include <linux/bitops.h>
#include <linux/poison.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/rcupdate.h>
#include <linux/jhash.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <net/route.h> /* for struct rtable and routing */
#include <net/icmp.h> /* icmp_send */
#include <net/arp.h>
#include <linux/param.h> /* for HZ */
#include <linux/uaccess.h>
#include <asm/byteorder.h> /* for htons etc. */
#include <linux/atomic.h>
#include "common.h"
#include "resources.h"
#include <net/atmclip.h>
static struct net_device *clip_devs;
static struct atm_vcc *atmarpd;
static struct timer_list idle_timer;
static const struct neigh_ops clip_neigh_ops;
static int to_atmarpd(enum atmarp_ctrl_type type, int itf, __be32 ip)
{
struct sock *sk;
struct atmarp_ctrl *ctrl;
struct sk_buff *skb;
pr_debug("(%d)\n", type);
if (!atmarpd)
return -EUNATCH;
skb = alloc_skb(sizeof(struct atmarp_ctrl), GFP_ATOMIC);
if (!skb)
return -ENOMEM;
ctrl = skb_put(skb, sizeof(struct atmarp_ctrl));
ctrl->type = type;
ctrl->itf_num = itf;
ctrl->ip = ip;
atm_force_charge(atmarpd, skb->truesize);
sk = sk_atm(atmarpd);
skb_queue_tail(&sk->sk_receive_queue, skb);
sk->sk_data_ready(sk);
return 0;
}
static void link_vcc(struct clip_vcc *clip_vcc, struct atmarp_entry *entry)
{
pr_debug("%p to entry %p (neigh %p)\n", clip_vcc, entry, entry->neigh);
clip_vcc->entry = entry;
clip_vcc->xoff = 0; /* @@@ may overrun buffer by one packet */
clip_vcc->next = entry->vccs;
entry->vccs = clip_vcc;
entry->neigh->used = jiffies;
}
static void unlink_clip_vcc(struct clip_vcc *clip_vcc)
{
struct atmarp_entry *entry = clip_vcc->entry;
struct clip_vcc **walk;
if (!entry) {
pr_err("!clip_vcc->entry (clip_vcc %p)\n", clip_vcc);
return;
}
[NET]: Add netif_tx_lock Various drivers use xmit_lock internally to synchronise with their transmission routines. They do so without setting xmit_lock_owner. This is fine as long as netpoll is not in use. With netpoll it is possible for deadlocks to occur if xmit_lock_owner isn't set. This is because if a printk occurs while xmit_lock is held and xmit_lock_owner is not set can cause netpoll to attempt to take xmit_lock recursively. While it is possible to resolve this by getting netpoll to use trylock, it is suboptimal because netpoll's sole objective is to maximise the chance of getting the printk out on the wire. So delaying or dropping the message is to be avoided as much as possible. So the only alternative is to always set xmit_lock_owner. The following patch does this by introducing the netif_tx_lock family of functions that take care of setting/unsetting xmit_lock_owner. I renamed xmit_lock to _xmit_lock to indicate that it should not be used directly. I didn't provide irq versions of the netif_tx_lock functions since xmit_lock is meant to be a BH-disabling lock. This is pretty much a straight text substitution except for a small bug fix in winbond. It currently uses netif_stop_queue/spin_unlock_wait to stop transmission. This is unsafe as an IRQ can potentially wake up the queue. So it is safer to use netif_tx_disable. The hamradio bits used spin_lock_irq but it is unnecessary as xmit_lock must never be taken in an IRQ handler. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-09 19:20:56 +00:00
netif_tx_lock_bh(entry->neigh->dev); /* block clip_start_xmit() */
entry->neigh->used = jiffies;
for (walk = &entry->vccs; *walk; walk = &(*walk)->next)
if (*walk == clip_vcc) {
int error;
*walk = clip_vcc->next; /* atomic */
clip_vcc->entry = NULL;
if (clip_vcc->xoff)
netif_wake_queue(entry->neigh->dev);
if (entry->vccs)
goto out;
entry->expires = jiffies - 1;
/* force resolution or expiration */
error = neigh_update(entry->neigh, NULL, NUD_NONE,
NEIGH_UPDATE_F_ADMIN, 0);
if (error)
pr_err("neigh_update failed with %d\n", error);
goto out;
}
pr_err("ATMARP: failed (entry %p, vcc 0x%p)\n", entry, clip_vcc);
out:
[NET]: Add netif_tx_lock Various drivers use xmit_lock internally to synchronise with their transmission routines. They do so without setting xmit_lock_owner. This is fine as long as netpoll is not in use. With netpoll it is possible for deadlocks to occur if xmit_lock_owner isn't set. This is because if a printk occurs while xmit_lock is held and xmit_lock_owner is not set can cause netpoll to attempt to take xmit_lock recursively. While it is possible to resolve this by getting netpoll to use trylock, it is suboptimal because netpoll's sole objective is to maximise the chance of getting the printk out on the wire. So delaying or dropping the message is to be avoided as much as possible. So the only alternative is to always set xmit_lock_owner. The following patch does this by introducing the netif_tx_lock family of functions that take care of setting/unsetting xmit_lock_owner. I renamed xmit_lock to _xmit_lock to indicate that it should not be used directly. I didn't provide irq versions of the netif_tx_lock functions since xmit_lock is meant to be a BH-disabling lock. This is pretty much a straight text substitution except for a small bug fix in winbond. It currently uses netif_stop_queue/spin_unlock_wait to stop transmission. This is unsafe as an IRQ can potentially wake up the queue. So it is safer to use netif_tx_disable. The hamradio bits used spin_lock_irq but it is unnecessary as xmit_lock must never be taken in an IRQ handler. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-09 19:20:56 +00:00
netif_tx_unlock_bh(entry->neigh->dev);
}
/* The neighbour entry n->lock is held. */
static int neigh_check_cb(struct neighbour *n)
{
struct atmarp_entry *entry = neighbour_priv(n);
struct clip_vcc *cv;
if (n->ops != &clip_neigh_ops)
return 0;
for (cv = entry->vccs; cv; cv = cv->next) {
unsigned long exp = cv->last_use + cv->idle_timeout;
if (cv->idle_timeout && time_after(jiffies, exp)) {
pr_debug("releasing vcc %p->%p of entry %p\n",
cv, cv->vcc, entry);
vcc_release_async(cv->vcc, -ETIMEDOUT);
}
}
if (entry->vccs || time_before(jiffies, entry->expires))
return 0;
if (refcount_read(&n->refcnt) > 1) {
struct sk_buff *skb;
pr_debug("destruction postponed with ref %d\n",
refcount_read(&n->refcnt));
while ((skb = skb_dequeue(&n->arp_queue)) != NULL)
dev_kfree_skb(skb);
return 0;
}
pr_debug("expired neigh %p\n", n);
return 1;
}
static void idle_timer_check(struct timer_list *unused)
{
write_lock(&arp_tbl.lock);
__neigh_for_each_release(&arp_tbl, neigh_check_cb);
mod_timer(&idle_timer, jiffies + CLIP_CHECK_INTERVAL * HZ);
write_unlock(&arp_tbl.lock);
}
static int clip_arp_rcv(struct sk_buff *skb)
{
struct atm_vcc *vcc;
pr_debug("\n");
vcc = ATM_SKB(skb)->vcc;
if (!vcc || !atm_charge(vcc, skb->truesize)) {
dev_kfree_skb_any(skb);
return 0;
}
pr_debug("pushing to %p\n", vcc);
pr_debug("using %p\n", CLIP_VCC(vcc)->old_push);
CLIP_VCC(vcc)->old_push(vcc, skb);
return 0;
}
static const unsigned char llc_oui[] = {
0xaa, /* DSAP: non-ISO */
0xaa, /* SSAP: non-ISO */
0x03, /* Ctrl: Unnumbered Information Command PDU */
0x00, /* OUI: EtherType */
0x00,
0x00
};
static void clip_push(struct atm_vcc *vcc, struct sk_buff *skb)
{
struct clip_vcc *clip_vcc = CLIP_VCC(vcc);
pr_debug("\n");
if (!clip_devs) {
atm_return(vcc, skb->truesize);
kfree_skb(skb);
return;
}
if (!skb) {
pr_debug("removing VCC %p\n", clip_vcc);
if (clip_vcc->entry)
unlink_clip_vcc(clip_vcc);
clip_vcc->old_push(vcc, NULL); /* pass on the bad news */
kfree(clip_vcc);
return;
}
atm_return(vcc, skb->truesize);
skb->dev = clip_vcc->entry ? clip_vcc->entry->neigh->dev : clip_devs;
/* clip_vcc->entry == NULL if we don't have an IP address yet */
if (!skb->dev) {
dev_kfree_skb_any(skb);
return;
}
ATM_SKB(skb)->vcc = vcc;
skb_reset_mac_header(skb);
if (!clip_vcc->encap ||
skb->len < RFC1483LLC_LEN ||
memcmp(skb->data, llc_oui, sizeof(llc_oui)))
skb->protocol = htons(ETH_P_IP);
else {
skb->protocol = ((__be16 *)skb->data)[3];
skb_pull(skb, RFC1483LLC_LEN);
if (skb->protocol == htons(ETH_P_ARP)) {
skb->dev->stats.rx_packets++;
skb->dev->stats.rx_bytes += skb->len;
clip_arp_rcv(skb);
return;
}
}
clip_vcc->last_use = jiffies;
skb->dev->stats.rx_packets++;
skb->dev->stats.rx_bytes += skb->len;
memset(ATM_SKB(skb), 0, sizeof(struct atm_skb_data));
netif_rx(skb);
}
/*
* Note: these spinlocks _must_not_ block on non-SMP. The only goal is that
* clip_pop is atomic with respect to the critical section in clip_start_xmit.
*/
static void clip_pop(struct atm_vcc *vcc, struct sk_buff *skb)
{
struct clip_vcc *clip_vcc = CLIP_VCC(vcc);
struct net_device *dev = skb->dev;
int old;
unsigned long flags;
pr_debug("(vcc %p)\n", vcc);
clip_vcc->old_pop(vcc, skb);
/* skb->dev == NULL in outbound ARP packets */
if (!dev)
return;
spin_lock_irqsave(&PRIV(dev)->xoff_lock, flags);
if (atm_may_send(vcc, 0)) {
old = xchg(&clip_vcc->xoff, 0);
if (old)
netif_wake_queue(dev);
}
spin_unlock_irqrestore(&PRIV(dev)->xoff_lock, flags);
}
static void clip_neigh_solicit(struct neighbour *neigh, struct sk_buff *skb)
{
__be32 *ip = (__be32 *) neigh->primary_key;
pr_debug("(neigh %p, skb %p)\n", neigh, skb);
to_atmarpd(act_need, PRIV(neigh->dev)->number, *ip);
}
static void clip_neigh_error(struct neighbour *neigh, struct sk_buff *skb)
{
#ifndef CONFIG_ATM_CLIP_NO_ICMP
icmp_send(skb, ICMP_DEST_UNREACH, ICMP_HOST_UNREACH, 0);
#endif
kfree_skb(skb);
}
static const struct neigh_ops clip_neigh_ops = {
.family = AF_INET,
.solicit = clip_neigh_solicit,
.error_report = clip_neigh_error,
.output = neigh_direct_output,
.connected_output = neigh_direct_output,
};
static int clip_constructor(struct net_device *dev, struct neighbour *neigh)
{
struct atmarp_entry *entry = neighbour_priv(neigh);
if (neigh->tbl->family != AF_INET)
return -EINVAL;
if (neigh->type != RTN_UNICAST)
return -EINVAL;
neigh->nud_state = NUD_NONE;
neigh->ops = &clip_neigh_ops;
neigh->output = neigh->ops->output;
entry->neigh = neigh;
entry->vccs = NULL;
entry->expires = jiffies - 1;
return 0;
}
/* @@@ copy bh locking from arp.c -- need to bh-enable atm code before */
/*
* We play with the resolve flag: 0 and 1 have the usual meaning, but -1 means
* to allocate the neighbour entry but not to ask atmarpd for resolution. Also,
* don't increment the usage count. This is used to create entries in
* clip_setentry.
*/
static int clip_encap(struct atm_vcc *vcc, int mode)
{
atm: deal with setting entry before mkip was called If we didn't call ATMARP_MKIP before ATMARP_ENCAP the VCC descriptor is non-existant and we'll end up dereferencing a NULL ptr: [1033173.491930] kasan: GPF could be caused by NULL-ptr deref or user memory accessirq event stamp: 123386 [1033173.493678] general protection fault: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC KASAN [1033173.493689] Modules linked in: [1033173.493697] CPU: 9 PID: 23815 Comm: trinity-c64 Not tainted 4.2.0-next-20150911-sasha-00043-g353d875-dirty #2545 [1033173.493706] task: ffff8800630c4000 ti: ffff880063110000 task.ti: ffff880063110000 [1033173.493823] RIP: clip_ioctl (net/atm/clip.c:320 net/atm/clip.c:689) [1033173.493826] RSP: 0018:ffff880063117a88 EFLAGS: 00010203 [1033173.493828] RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 000000000000000c [1033173.493830] RDX: 0000000000000002 RSI: ffffffffb3f10720 RDI: 0000000000000014 [1033173.493832] RBP: ffff880063117b80 R08: ffff88047574d9a4 R09: 0000000000000000 [1033173.493834] R10: 0000000000000000 R11: 0000000000000000 R12: 1ffff1000c622f53 [1033173.493836] R13: ffff8800cb905500 R14: ffff8808d6da2000 R15: 00000000fffffdfd [1033173.493840] FS: 00007fa56b92d700(0000) GS:ffff880478000000(0000) knlGS:0000000000000000 [1033173.493843] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [1033173.493845] CR2: 0000000000000000 CR3: 00000000630e8000 CR4: 00000000000006a0 [1033173.493855] Stack: [1033173.493862] ffffffffb0b60444 000000000000eaea 0000000041b58ab3 ffffffffb3c3ce32 [1033173.493867] ffffffffb0b6f3e0 ffffffffb0b60444 ffffffffb5ea2e50 1ffff1000c622f5e [1033173.493873] ffff8800630c4cd8 00000000000ee09a ffffffffb3ec4888 ffffffffb5ea2de8 [1033173.493874] Call Trace: [1033173.494108] do_vcc_ioctl (net/atm/ioctl.c:170) [1033173.494113] vcc_ioctl (net/atm/ioctl.c:189) [1033173.494116] svc_ioctl (net/atm/svc.c:605) [1033173.494200] sock_do_ioctl (net/socket.c:874) [1033173.494204] sock_ioctl (net/socket.c:958) [1033173.494244] do_vfs_ioctl (fs/ioctl.c:43 fs/ioctl.c:607) [1033173.494290] SyS_ioctl (fs/ioctl.c:622 fs/ioctl.c:613) [1033173.494295] entry_SYSCALL_64_fastpath (arch/x86/entry/entry_64.S:186) [1033173.494362] Code: fa 48 c1 ea 03 80 3c 02 00 0f 85 50 09 00 00 49 8b 9e 60 06 00 00 48 b8 00 00 00 00 00 fc ff df 48 8d 7b 14 48 89 fa 48 c1 ea 03 <0f> b6 04 02 48 89 fa 83 e2 07 38 d0 7f 08 84 c0 0f 85 14 09 00 All code ======== 0: fa cli 1: 48 c1 ea 03 shr $0x3,%rdx 5: 80 3c 02 00 cmpb $0x0,(%rdx,%rax,1) 9: 0f 85 50 09 00 00 jne 0x95f f: 49 8b 9e 60 06 00 00 mov 0x660(%r14),%rbx 16: 48 b8 00 00 00 00 00 movabs $0xdffffc0000000000,%rax 1d: fc ff df 20: 48 8d 7b 14 lea 0x14(%rbx),%rdi 24: 48 89 fa mov %rdi,%rdx 27: 48 c1 ea 03 shr $0x3,%rdx 2b:* 0f b6 04 02 movzbl (%rdx,%rax,1),%eax <-- trapping instruction 2f: 48 89 fa mov %rdi,%rdx 32: 83 e2 07 and $0x7,%edx 35: 38 d0 cmp %dl,%al 37: 7f 08 jg 0x41 39: 84 c0 test %al,%al 3b: 0f 85 14 09 00 00 jne 0x955 Code starting with the faulting instruction =========================================== 0: 0f b6 04 02 movzbl (%rdx,%rax,1),%eax 4: 48 89 fa mov %rdi,%rdx 7: 83 e2 07 and $0x7,%edx a: 38 d0 cmp %dl,%al c: 7f 08 jg 0x16 e: 84 c0 test %al,%al 10: 0f 85 14 09 00 00 jne 0x92a [1033173.494366] RIP clip_ioctl (net/atm/clip.c:320 net/atm/clip.c:689) [1033173.494368] RSP <ffff880063117a88> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-09-16 19:30:21 +00:00
if (!CLIP_VCC(vcc))
return -EBADFD;
CLIP_VCC(vcc)->encap = mode;
return 0;
}
static netdev_tx_t clip_start_xmit(struct sk_buff *skb,
struct net_device *dev)
{
struct clip_priv *clip_priv = PRIV(dev);
struct dst_entry *dst = skb_dst(skb);
struct atmarp_entry *entry;
struct neighbour *n;
struct atm_vcc *vcc;
struct rtable *rt;
__be32 *daddr;
int old;
unsigned long flags;
pr_debug("(skb %p)\n", skb);
if (!dst) {
pr_err("skb_dst(skb) == NULL\n");
dev_kfree_skb(skb);
dev->stats.tx_dropped++;
return NETDEV_TX_OK;
}
rt = dst_rtable(dst);
if (rt->rt_gw_family == AF_INET)
daddr = &rt->rt_gw4;
else
daddr = &ip_hdr(skb)->daddr;
n = dst_neigh_lookup(dst, daddr);
if (!n) {
pr_err("NO NEIGHBOUR !\n");
dev_kfree_skb(skb);
dev->stats.tx_dropped++;
return NETDEV_TX_OK;
}
entry = neighbour_priv(n);
if (!entry->vccs) {
if (time_after(jiffies, entry->expires)) {
/* should be resolved */
entry->expires = jiffies + ATMARP_RETRY_DELAY * HZ;
to_atmarpd(act_need, PRIV(dev)->number, *((__be32 *)n->primary_key));
}
if (entry->neigh->arp_queue.qlen < ATMARP_MAX_UNRES_PACKETS)
skb_queue_tail(&entry->neigh->arp_queue, skb);
else {
dev_kfree_skb(skb);
dev->stats.tx_dropped++;
}
goto out_release_neigh;
}
pr_debug("neigh %p, vccs %p\n", entry, entry->vccs);
ATM_SKB(skb)->vcc = vcc = entry->vccs->vcc;
pr_debug("using neighbour %p, vcc %p\n", n, vcc);
if (entry->vccs->encap) {
void *here;
here = skb_push(skb, RFC1483LLC_LEN);
memcpy(here, llc_oui, sizeof(llc_oui));
((__be16 *) here)[3] = skb->protocol;
}
atm: Preserve value of skb->truesize when accounting to vcc ATM accounts for in-flight TX packets in sk_wmem_alloc of the VCC on which they are to be sent. But it doesn't take ownership of those packets from the sock (if any) which originally owned them. They should remain owned by their actual sender until they've left the box. There's a hack in pskb_expand_head() to avoid adjusting skb->truesize for certain skbs, precisely to avoid messing up sk_wmem_alloc accounting. Ideally that hack would cover the ATM use case too, but it doesn't — skbs which aren't owned by any sock, for example PPP control frames, still get their truesize adjusted when the low-level ATM driver adds headroom. This has always been an issue, it seems. The truesize of a packet increases, and sk_wmem_alloc on the VCC goes negative. But this wasn't for normal traffic, only for control frames. So I think we just got away with it, and we probably needed to send 2GiB of LCP echo frames before the misaccounting would ever have caused a problem and caused atm_may_send() to start refusing packets. Commit 14afee4b609 ("net: convert sock.sk_wmem_alloc from atomic_t to refcount_t") did exactly what it was intended to do, and turned this mostly-theoretical problem into a real one, causing PPPoATM to fail immediately as sk_wmem_alloc underflows and atm_may_send() *immediately* starts refusing to allow new packets. The least intrusive solution to this problem is to stash the value of skb->truesize that was accounted to the VCC, in a new member of the ATM_SKB(skb) structure. Then in atm_pop_raw() subtract precisely that value instead of the then-current value of skb->truesize. Fixes: 158f323b9868 ("net: adjust skb->truesize in pskb_expand_head()") Signed-off-by: David Woodhouse <dwmw2@infradead.org> Tested-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-16 10:55:44 +00:00
atm_account_tx(vcc, skb);
entry->vccs->last_use = jiffies;
pr_debug("atm_skb(%p)->vcc(%p)->dev(%p)\n", skb, vcc, vcc->dev);
old = xchg(&entry->vccs->xoff, 1); /* assume XOFF ... */
if (old) {
pr_warn("XOFF->XOFF transition\n");
goto out_release_neigh;
}
dev->stats.tx_packets++;
dev->stats.tx_bytes += skb->len;
vcc->send(vcc, skb);
if (atm_may_send(vcc, 0)) {
entry->vccs->xoff = 0;
goto out_release_neigh;
}
spin_lock_irqsave(&clip_priv->xoff_lock, flags);
netif_stop_queue(dev); /* XOFF -> throttle immediately */
barrier();
if (!entry->vccs->xoff)
netif_start_queue(dev);
/* Oh, we just raced with clip_pop. netif_start_queue should be
good enough, because nothing should really be asleep because
of the brief netif_stop_queue. If this isn't true or if it
changes, use netif_wake_queue instead. */
spin_unlock_irqrestore(&clip_priv->xoff_lock, flags);
out_release_neigh:
neigh_release(n);
return NETDEV_TX_OK;
}
static int clip_mkip(struct atm_vcc *vcc, int timeout)
{
struct clip_vcc *clip_vcc;
if (!vcc->push)
return -EBADFD;
clip_vcc = kmalloc(sizeof(struct clip_vcc), GFP_KERNEL);
if (!clip_vcc)
return -ENOMEM;
pr_debug("%p vcc %p\n", clip_vcc, vcc);
clip_vcc->vcc = vcc;
vcc->user_back = clip_vcc;
set_bit(ATM_VF_IS_CLIP, &vcc->flags);
clip_vcc->entry = NULL;
clip_vcc->xoff = 0;
clip_vcc->encap = 1;
clip_vcc->last_use = jiffies;
clip_vcc->idle_timeout = timeout * HZ;
clip_vcc->old_push = vcc->push;
clip_vcc->old_pop = vcc->pop;
vcc->push = clip_push;
vcc->pop = clip_pop;
/* re-process everything received between connection setup and MKIP */
vcc_process_recv_queue(vcc);
return 0;
}
static int clip_setentry(struct atm_vcc *vcc, __be32 ip)
{
struct neighbour *neigh;
struct atmarp_entry *entry;
int error;
struct clip_vcc *clip_vcc;
struct rtable *rt;
if (vcc->push != clip_push) {
pr_warn("non-CLIP VCC\n");
return -EBADF;
}
clip_vcc = CLIP_VCC(vcc);
if (!ip) {
if (!clip_vcc->entry) {
pr_err("hiding hidden ATMARP entry\n");
return 0;
}
pr_debug("remove\n");
unlink_clip_vcc(clip_vcc);
return 0;
}
rt = ip_route_output(&init_net, ip, 0, 0, 0, RT_SCOPE_LINK);
if (IS_ERR(rt))
return PTR_ERR(rt);
neigh = __neigh_lookup(&arp_tbl, &ip, rt->dst.dev, 1);
ip_rt_put(rt);
if (!neigh)
return -ENOMEM;
entry = neighbour_priv(neigh);
if (entry != clip_vcc->entry) {
if (!clip_vcc->entry)
pr_debug("add\n");
else {
pr_debug("update\n");
unlink_clip_vcc(clip_vcc);
}
link_vcc(clip_vcc, entry);
}
error = neigh_update(neigh, llc_oui, NUD_PERMANENT,
NEIGH_UPDATE_F_OVERRIDE | NEIGH_UPDATE_F_ADMIN, 0);
neigh_release(neigh);
return error;
}
static const struct net_device_ops clip_netdev_ops = {
.ndo_start_xmit = clip_start_xmit,
.ndo_neigh_construct = clip_constructor,
};
static void clip_setup(struct net_device *dev)
{
dev->netdev_ops = &clip_netdev_ops;
dev->type = ARPHRD_ATM;
dev->neigh_priv_len = sizeof(struct atmarp_entry);
dev->hard_header_len = RFC1483LLC_LEN;
dev->mtu = RFC1626_MTU;
dev->tx_queue_len = 100; /* "normal" queue (packets) */
/* When using a "real" qdisc, the qdisc determines the queue */
/* length. tx_queue_len is only used for the default case, */
/* without any more elaborate queuing. 100 is a reasonable */
/* compromise between decent burst-tolerance and protection */
/* against memory hogs. */
netif_keep_dst(dev);
}
static int clip_create(int number)
{
struct net_device *dev;
struct clip_priv *clip_priv;
int error;
if (number != -1) {
for (dev = clip_devs; dev; dev = PRIV(dev)->next)
if (PRIV(dev)->number == number)
return -EEXIST;
} else {
number = 0;
for (dev = clip_devs; dev; dev = PRIV(dev)->next)
if (PRIV(dev)->number >= number)
number = PRIV(dev)->number + 1;
}
dev = alloc_netdev(sizeof(struct clip_priv), "", NET_NAME_UNKNOWN,
clip_setup);
if (!dev)
return -ENOMEM;
clip_priv = PRIV(dev);
sprintf(dev->name, "atm%d", number);
spin_lock_init(&clip_priv->xoff_lock);
clip_priv->number = number;
error = register_netdev(dev);
if (error) {
free_netdev(dev);
return error;
}
clip_priv->next = clip_devs;
clip_devs = dev;
pr_debug("registered (net:%s)\n", dev->name);
return number;
}
static int clip_device_event(struct notifier_block *this, unsigned long event,
void *ptr)
{
struct net_device *dev = netdev_notifier_info_to_dev(ptr);
if (!net_eq(dev_net(dev), &init_net))
return NOTIFY_DONE;
if (event == NETDEV_UNREGISTER)
return NOTIFY_DONE;
/* ignore non-CLIP devices */
if (dev->type != ARPHRD_ATM || dev->netdev_ops != &clip_netdev_ops)
return NOTIFY_DONE;
switch (event) {
case NETDEV_UP:
pr_debug("NETDEV_UP\n");
to_atmarpd(act_up, PRIV(dev)->number, 0);
break;
case NETDEV_GOING_DOWN:
pr_debug("NETDEV_DOWN\n");
to_atmarpd(act_down, PRIV(dev)->number, 0);
break;
case NETDEV_CHANGE:
case NETDEV_CHANGEMTU:
pr_debug("NETDEV_CHANGE*\n");
to_atmarpd(act_change, PRIV(dev)->number, 0);
break;
}
return NOTIFY_DONE;
}
static int clip_inet_event(struct notifier_block *this, unsigned long event,
void *ifa)
{
struct in_device *in_dev;
struct netdev_notifier_info info;
in_dev = ((struct in_ifaddr *)ifa)->ifa_dev;
/*
* Transitions are of the down-change-up type, so it's sufficient to
* handle the change on up.
*/
if (event != NETDEV_UP)
return NOTIFY_DONE;
netdev_notifier_info_init(&info, in_dev->dev);
return clip_device_event(this, NETDEV_CHANGE, &info);
}
static struct notifier_block clip_dev_notifier = {
.notifier_call = clip_device_event,
};
static struct notifier_block clip_inet_notifier = {
.notifier_call = clip_inet_event,
};
static void atmarpd_close(struct atm_vcc *vcc)
{
pr_debug("\n");
rtnl_lock();
atmarpd = NULL;
skb_queue_purge(&sk_atm(vcc)->sk_receive_queue);
rtnl_unlock();
pr_debug("(done)\n");
module_put(THIS_MODULE);
}
static const struct atmdev_ops atmarpd_dev_ops = {
.close = atmarpd_close
};
static struct atm_dev atmarpd_dev = {
.ops = &atmarpd_dev_ops,
.type = "arpd",
.number = 999,
.lock = __SPIN_LOCK_UNLOCKED(atmarpd_dev.lock)
};
static int atm_init_atmarp(struct atm_vcc *vcc)
{
rtnl_lock();
if (atmarpd) {
rtnl_unlock();
return -EADDRINUSE;
}
mod_timer(&idle_timer, jiffies + CLIP_CHECK_INTERVAL * HZ);
atmarpd = vcc;
set_bit(ATM_VF_META, &vcc->flags);
set_bit(ATM_VF_READY, &vcc->flags);
/* allow replies and avoid getting closed if signaling dies */
vcc->dev = &atmarpd_dev;
vcc_insert_socket(sk_atm(vcc));
vcc->push = NULL;
vcc->pop = NULL; /* crash */
vcc->push_oam = NULL; /* crash */
rtnl_unlock();
return 0;
}
static int clip_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
struct atm_vcc *vcc = ATM_SD(sock);
int err = 0;
switch (cmd) {
case SIOCMKCLIP:
case ATMARPD_CTRL:
case ATMARP_MKIP:
case ATMARP_SETENTRY:
case ATMARP_ENCAP:
if (!capable(CAP_NET_ADMIN))
return -EPERM;
break;
default:
return -ENOIOCTLCMD;
}
switch (cmd) {
case SIOCMKCLIP:
err = clip_create(arg);
break;
case ATMARPD_CTRL:
err = atm_init_atmarp(vcc);
if (!err) {
sock->state = SS_CONNECTED;
__module_get(THIS_MODULE);
}
break;
case ATMARP_MKIP:
err = clip_mkip(vcc, arg);
break;
case ATMARP_SETENTRY:
err = clip_setentry(vcc, (__force __be32)arg);
break;
case ATMARP_ENCAP:
err = clip_encap(vcc, arg);
break;
}
return err;
}
static struct atm_ioctl clip_ioctl_ops = {
.owner = THIS_MODULE,
.ioctl = clip_ioctl,
};
#ifdef CONFIG_PROC_FS
static void svc_addr(struct seq_file *seq, struct sockaddr_atmsvc *addr)
{
static int code[] = { 1, 2, 10, 6, 1, 0 };
static int e164[] = { 1, 8, 4, 6, 1, 0 };
if (*addr->sas_addr.pub) {
seq_printf(seq, "%s", addr->sas_addr.pub);
if (*addr->sas_addr.prv)
seq_putc(seq, '+');
} else if (!*addr->sas_addr.prv) {
seq_printf(seq, "%s", "(none)");
return;
}
if (*addr->sas_addr.prv) {
unsigned char *prv = addr->sas_addr.prv;
int *fields;
int i, j;
fields = *prv == ATM_AFI_E164 ? e164 : code;
for (i = 0; fields[i]; i++) {
for (j = fields[i]; j; j--)
seq_printf(seq, "%02X", *prv++);
if (fields[i + 1])
seq_putc(seq, '.');
}
}
}
/* This means the neighbour entry has no attached VCC objects. */
#define SEQ_NO_VCC_TOKEN ((void *) 2)
static void atmarp_info(struct seq_file *seq, struct neighbour *n,
struct atmarp_entry *entry, struct clip_vcc *clip_vcc)
{
struct net_device *dev = n->dev;
unsigned long exp;
char buf[17];
int svc, llc, off;
svc = ((clip_vcc == SEQ_NO_VCC_TOKEN) ||
(sk_atm(clip_vcc->vcc)->sk_family == AF_ATMSVC));
llc = ((clip_vcc == SEQ_NO_VCC_TOKEN) || clip_vcc->encap);
if (clip_vcc == SEQ_NO_VCC_TOKEN)
exp = entry->neigh->used;
else
exp = clip_vcc->last_use;
exp = (jiffies - exp) / HZ;
seq_printf(seq, "%-6s%-4s%-4s%5ld ",
dev->name, svc ? "SVC" : "PVC", llc ? "LLC" : "NULL", exp);
off = scnprintf(buf, sizeof(buf) - 1, "%pI4", n->primary_key);
while (off < 16)
buf[off++] = ' ';
buf[off] = '\0';
seq_printf(seq, "%s", buf);
if (clip_vcc == SEQ_NO_VCC_TOKEN) {
if (time_before(jiffies, entry->expires))
seq_printf(seq, "(resolving)\n");
else
seq_printf(seq, "(expired, ref %d)\n",
refcount_read(&entry->neigh->refcnt));
} else if (!svc) {
seq_printf(seq, "%d.%d.%d\n",
clip_vcc->vcc->dev->number,
clip_vcc->vcc->vpi, clip_vcc->vcc->vci);
} else {
svc_addr(seq, &clip_vcc->vcc->remote);
seq_putc(seq, '\n');
}
}
struct clip_seq_state {
/* This member must be first. */
struct neigh_seq_state ns;
/* Local to clip specific iteration. */
struct clip_vcc *vcc;
};
static struct clip_vcc *clip_seq_next_vcc(struct atmarp_entry *e,
struct clip_vcc *curr)
{
if (!curr) {
curr = e->vccs;
if (!curr)
return SEQ_NO_VCC_TOKEN;
return curr;
}
if (curr == SEQ_NO_VCC_TOKEN)
return NULL;
curr = curr->next;
return curr;
}
static void *clip_seq_vcc_walk(struct clip_seq_state *state,
struct atmarp_entry *e, loff_t * pos)
{
struct clip_vcc *vcc = state->vcc;
vcc = clip_seq_next_vcc(e, vcc);
if (vcc && pos != NULL) {
while (*pos) {
vcc = clip_seq_next_vcc(e, vcc);
if (!vcc)
break;
--(*pos);
}
}
state->vcc = vcc;
return vcc;
}
static void *clip_seq_sub_iter(struct neigh_seq_state *_state,
struct neighbour *n, loff_t * pos)
{
struct clip_seq_state *state = (struct clip_seq_state *)_state;
if (n->dev->type != ARPHRD_ATM)
return NULL;
return clip_seq_vcc_walk(state, neighbour_priv(n), pos);
}
static void *clip_seq_start(struct seq_file *seq, loff_t * pos)
{
struct clip_seq_state *state = seq->private;
state->ns.neigh_sub_iter = clip_seq_sub_iter;
return neigh_seq_start(seq, pos, &arp_tbl, NEIGH_SEQ_NEIGH_ONLY);
}
static int clip_seq_show(struct seq_file *seq, void *v)
{
static char atm_arp_banner[] =
"IPitf TypeEncp Idle IP address ATM address\n";
if (v == SEQ_START_TOKEN) {
seq_puts(seq, atm_arp_banner);
} else {
struct clip_seq_state *state = seq->private;
struct clip_vcc *vcc = state->vcc;
struct neighbour *n = v;
atmarp_info(seq, n, neighbour_priv(n), vcc);
}
return 0;
}
static const struct seq_operations arp_seq_ops = {
.start = clip_seq_start,
.next = neigh_seq_next,
.stop = neigh_seq_stop,
.show = clip_seq_show,
};
#endif
static void atm_clip_exit_noproc(void);
static int __init atm_clip_init(void)
{
register_atm_ioctl(&clip_ioctl_ops);
register_netdevice_notifier(&clip_dev_notifier);
register_inetaddr_notifier(&clip_inet_notifier);
timer_setup(&idle_timer, idle_timer_check, 0);
#ifdef CONFIG_PROC_FS
{
struct proc_dir_entry *p;
p = proc_create_net("arp", 0444, atm_proc_root, &arp_seq_ops,
sizeof(struct clip_seq_state));
if (!p) {
pr_err("Unable to initialize /proc/net/atm/arp\n");
atm_clip_exit_noproc();
return -ENOMEM;
}
}
#endif
return 0;
}
static void atm_clip_exit_noproc(void)
{
struct net_device *dev, *next;
unregister_inetaddr_notifier(&clip_inet_notifier);
unregister_netdevice_notifier(&clip_dev_notifier);
deregister_atm_ioctl(&clip_ioctl_ops);
/* First, stop the idle timer, so it stops banging
* on the table.
*/
del_timer_sync(&idle_timer);
dev = clip_devs;
while (dev) {
next = PRIV(dev)->next;
unregister_netdev(dev);
free_netdev(dev);
dev = next;
}
}
static void __exit atm_clip_exit(void)
{
remove_proc_entry("arp", atm_proc_root);
atm_clip_exit_noproc();
}
module_init(atm_clip_init);
module_exit(atm_clip_exit);
MODULE_AUTHOR("Werner Almesberger");
MODULE_DESCRIPTION("Classical/IP over ATM interface");
MODULE_LICENSE("GPL");