linux_dsm_epyc7002/net/sched/sch_api.c

2038 lines
47 KiB
C
Raw Normal View History

/*
* net/sched/sch_api.c Packet scheduler API.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
* Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
*
* Fixes:
*
* Rani Assaf <rani@magic.metawire.com> :980802: JIFFIES and CPU clock sources are repaired.
* Eduardo J. Blanco <ejbs@netlabs.com.uy> :990222: kmod support
* Jamal Hadi Salim <hadi@nortelnetworks.com>: 990601: ingress support
*/
#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/kmod.h>
#include <linux/list.h>
#include <linux/hrtimer.h>
#include <linux/lockdep.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 15:04:11 +07:00
#include <linux/slab.h>
#include <linux/hashtable.h>
#include <net/net_namespace.h>
#include <net/sock.h>
#include <net/netlink.h>
#include <net/pkt_sched.h>
#include <net/pkt_cls.h>
/*
Short review.
-------------
This file consists of two interrelated parts:
1. queueing disciplines manager frontend.
2. traffic classes manager frontend.
Generally, queueing discipline ("qdisc") is a black box,
which is able to enqueue packets and to dequeue them (when
device is ready to send something) in order and at times
determined by algorithm hidden in it.
qdisc's are divided to two categories:
- "queues", which have no internal structure visible from outside.
- "schedulers", which split all the packets to "traffic classes",
using "packet classifiers" (look at cls_api.c)
In turn, classes may have child qdiscs (as rule, queues)
attached to them etc. etc. etc.
The goal of the routines in this file is to translate
information supplied by user in the form of handles
to more intelligible for kernel form, to make some sanity
checks and part of work, which is common to all qdiscs
and to provide rtnetlink notifications.
All real intelligent work is done inside qdisc modules.
Every discipline has two major routines: enqueue and dequeue.
---dequeue
dequeue usually returns a skb to send. It is allowed to return NULL,
but it does not mean that queue is empty, it just means that
discipline does not want to send anything this time.
Queue is really empty if q->q.qlen == 0.
For complicated disciplines with multiple queues q->q is not
real packet queue, but however q->q.qlen must be valid.
---enqueue
enqueue returns 0, if packet was enqueued successfully.
If packet (this one or another one) was dropped, it returns
not zero error code.
NET_XMIT_DROP - this packet dropped
Expected action: do not backoff, but wait until queue will clear.
NET_XMIT_CN - probably this packet enqueued, but another one dropped.
Expected action: backoff or ignore
Auxiliary routines:
---peek
like dequeue but without removing a packet from the queue
---reset
returns qdisc to initial state: purge all buffers, clear all
timers, counters (except for statistics) etc.
---init
initializes newly created qdisc.
---destroy
destroys resources allocated by init and during lifetime of qdisc.
---change
changes qdisc parameters.
*/
/* Protects list of registered TC modules. It is pure SMP lock. */
static DEFINE_RWLOCK(qdisc_mod_lock);
/************************************************
* Queueing disciplines manipulation. *
************************************************/
/* The list of all installed queueing disciplines. */
static struct Qdisc_ops *qdisc_base;
/* Register/unregister queueing discipline */
int register_qdisc(struct Qdisc_ops *qops)
{
struct Qdisc_ops *q, **qp;
int rc = -EEXIST;
write_lock(&qdisc_mod_lock);
for (qp = &qdisc_base; (q = *qp) != NULL; qp = &q->next)
if (!strcmp(qops->id, q->id))
goto out;
if (qops->enqueue == NULL)
qops->enqueue = noop_qdisc_ops.enqueue;
if (qops->peek == NULL) {
if (qops->dequeue == NULL)
qops->peek = noop_qdisc_ops.peek;
else
goto out_einval;
}
if (qops->dequeue == NULL)
qops->dequeue = noop_qdisc_ops.dequeue;
if (qops->cl_ops) {
const struct Qdisc_class_ops *cops = qops->cl_ops;
if (!(cops->find && cops->walk && cops->leaf))
goto out_einval;
if (cops->tcf_block && !(cops->bind_tcf && cops->unbind_tcf))
goto out_einval;
}
qops->next = NULL;
*qp = qops;
rc = 0;
out:
write_unlock(&qdisc_mod_lock);
return rc;
out_einval:
rc = -EINVAL;
goto out;
}
EXPORT_SYMBOL(register_qdisc);
int unregister_qdisc(struct Qdisc_ops *qops)
{
struct Qdisc_ops *q, **qp;
int err = -ENOENT;
write_lock(&qdisc_mod_lock);
for (qp = &qdisc_base; (q = *qp) != NULL; qp = &q->next)
if (q == qops)
break;
if (q) {
*qp = q->next;
q->next = NULL;
err = 0;
}
write_unlock(&qdisc_mod_lock);
return err;
}
EXPORT_SYMBOL(unregister_qdisc);
/* Get default qdisc if not otherwise specified */
void qdisc_get_default(char *name, size_t len)
{
read_lock(&qdisc_mod_lock);
strlcpy(name, default_qdisc_ops->id, len);
read_unlock(&qdisc_mod_lock);
}
static struct Qdisc_ops *qdisc_lookup_default(const char *name)
{
struct Qdisc_ops *q = NULL;
for (q = qdisc_base; q; q = q->next) {
if (!strcmp(name, q->id)) {
if (!try_module_get(q->owner))
q = NULL;
break;
}
}
return q;
}
/* Set new default qdisc to use */
int qdisc_set_default(const char *name)
{
const struct Qdisc_ops *ops;
if (!capable(CAP_NET_ADMIN))
return -EPERM;
write_lock(&qdisc_mod_lock);
ops = qdisc_lookup_default(name);
if (!ops) {
/* Not found, drop lock and try to load module */
write_unlock(&qdisc_mod_lock);
request_module("sch_%s", name);
write_lock(&qdisc_mod_lock);
ops = qdisc_lookup_default(name);
}
if (ops) {
/* Set new default */
module_put(default_qdisc_ops->owner);
default_qdisc_ops = ops;
}
write_unlock(&qdisc_mod_lock);
return ops ? 0 : -ENOENT;
}
#ifdef CONFIG_NET_SCH_DEFAULT
/* Set default value from kernel config */
static int __init sch_default_qdisc(void)
{
return qdisc_set_default(CONFIG_DEFAULT_NET_SCH);
}
late_initcall(sch_default_qdisc);
#endif
/* We know handle. Find qdisc among all qdisc's attached to device
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
* (root qdisc, all its children, children of children etc.)
* Note: caller either uses rtnl or rcu_read_lock()
*/
static struct Qdisc *qdisc_match_from_root(struct Qdisc *root, u32 handle)
{
struct Qdisc *q;
net: sched: fix handling of singleton qdiscs with qdisc_hash qdisc_match_from_root() is now iterating over per-netdevice qdisc hashtable instead of going through a linked-list of qdiscs (independently on the actual underlying netdev), which was the case before the switch to hashtable for qdiscs. For singleton qdiscs, there is no underlying netdev associated though, and therefore dumping a singleton qdisc will panic, as qdisc_dev(root) will always be NULL. BUG: unable to handle kernel NULL pointer dereference at 0000000000000410 IP: [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 PGD 1aceba067 PUD 1aceb7067 PMD 0 Oops: 0000 [#1] PREEMPT SMP [ ... ] task: ffff8801ec996e00 task.stack: ffff8801ec934000 RIP: 0010:[<ffffffff8167efac>] [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 RSP: 0018:ffff8801ec937ab0 EFLAGS: 00010203 RAX: 0000000000000408 RBX: ffff88025e612000 RCX: ffffffffffffffd8 RDX: 0000000000000000 RSI: 00000000ffff0000 RDI: ffffffff81cf8100 RBP: ffff8801ec937ab0 R08: 000000000001c160 R09: ffff8802668032c0 R10: ffffffff81cf8100 R11: 0000000000000030 R12: 00000000ffff0000 R13: ffff88025e612000 R14: ffffffff81cf3140 R15: 0000000000000000 FS: 00007f24b9af6740(0000) GS:ffff88026f280000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000410 CR3: 00000001aceec000 CR4: 00000000001406e0 Stack: ffff8801ec937ad0 ffffffff81681210 ffff88025dd51a00 00000000fffffff1 ffff8801ec937b88 ffffffff81681e4e ffffffff81c42bc0 ffff880262431500 ffffffff81cf3140 ffff88025dd51a10 ffff88025dd51a24 00000000ec937b38 Call Trace: [<ffffffff81681210>] qdisc_lookup+0x40/0x50 [<ffffffff81681e4e>] tc_modify_qdisc+0x21e/0x550 [<ffffffff8166ae25>] rtnetlink_rcv_msg+0x95/0x220 [<ffffffff81209602>] ? __kmalloc_track_caller+0x172/0x230 [<ffffffff8166ad90>] ? rtnl_newlink+0x870/0x870 [<ffffffff816897b7>] netlink_rcv_skb+0xa7/0xc0 [<ffffffff816657c8>] rtnetlink_rcv+0x28/0x30 [<ffffffff8168919b>] netlink_unicast+0x15b/0x210 [<ffffffff81689569>] netlink_sendmsg+0x319/0x390 [<ffffffff816379f8>] sock_sendmsg+0x38/0x50 [<ffffffff81638296>] ___sys_sendmsg+0x256/0x260 [<ffffffff811b1275>] ? __pagevec_lru_add_fn+0x135/0x280 [<ffffffff811b1a90>] ? pagevec_lru_move_fn+0xd0/0xf0 [<ffffffff811b1140>] ? trace_event_raw_event_mm_lru_insertion+0x180/0x180 [<ffffffff811b1b85>] ? __lru_cache_add+0x75/0xb0 [<ffffffff817708a6>] ? _raw_spin_unlock+0x16/0x40 [<ffffffff811d8dff>] ? handle_mm_fault+0x39f/0x1160 [<ffffffff81638b15>] __sys_sendmsg+0x45/0x80 [<ffffffff81638b62>] SyS_sendmsg+0x12/0x20 [<ffffffff810038e7>] do_syscall_64+0x57/0xb0 Fix this by special-casing singleton qdiscs (those that don't have underlying netdevice) and introduce immediate handling of those rather than trying to go over an underlying netdevice. We're in the same situation in tc_dump_qdisc_root() and tc_dump_tclass_root(). Ultimately, this will have to be slightly reworked so that we are actually able to show singleton qdiscs (noop) in the dump properly; but we're not currently doing that anyway, so no regression there, and better do this in a gradual manner. Fixes: 59cc1f61f ("net: sched: convert qdisc linked list to hashtable") Reported-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Reported-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-08-17 04:52:58 +07:00
if (!qdisc_dev(root))
return (root->handle == handle ? root : NULL);
if (!(root->flags & TCQ_F_BUILTIN) &&
root->handle == handle)
return root;
hash_for_each_possible_rcu(qdisc_dev(root)->qdisc_hash, q, hash, handle) {
if (q->handle == handle)
return q;
}
return NULL;
}
void qdisc_hash_add(struct Qdisc *q, bool invisible)
{
if ((q->parent != TC_H_ROOT) && !(q->flags & TCQ_F_INGRESS)) {
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
ASSERT_RTNL();
hash_add_rcu(qdisc_dev(q)->qdisc_hash, &q->hash, q->handle);
if (invisible)
q->flags |= TCQ_F_INVISIBLE;
}
}
EXPORT_SYMBOL(qdisc_hash_add);
void qdisc_hash_del(struct Qdisc *q)
{
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
if ((q->parent != TC_H_ROOT) && !(q->flags & TCQ_F_INGRESS)) {
ASSERT_RTNL();
hash_del_rcu(&q->hash);
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
}
}
EXPORT_SYMBOL(qdisc_hash_del);
struct Qdisc *qdisc_lookup(struct net_device *dev, u32 handle)
{
struct Qdisc *q;
if (!handle)
return NULL;
q = qdisc_match_from_root(dev->qdisc, handle);
if (q)
goto out;
if (dev_ingress_queue(dev))
q = qdisc_match_from_root(
dev_ingress_queue(dev)->qdisc_sleeping,
handle);
out:
return q;
}
static struct Qdisc *qdisc_leaf(struct Qdisc *p, u32 classid)
{
unsigned long cl;
struct Qdisc *leaf;
const struct Qdisc_class_ops *cops = p->ops->cl_ops;
if (cops == NULL)
return NULL;
cl = cops->find(p, classid);
if (cl == 0)
return NULL;
leaf = cops->leaf(p, cl);
return leaf;
}
/* Find queueing discipline by name */
static struct Qdisc_ops *qdisc_lookup_ops(struct nlattr *kind)
{
struct Qdisc_ops *q = NULL;
if (kind) {
read_lock(&qdisc_mod_lock);
for (q = qdisc_base; q; q = q->next) {
if (nla_strcmp(kind, q->id) == 0) {
if (!try_module_get(q->owner))
q = NULL;
break;
}
}
read_unlock(&qdisc_mod_lock);
}
return q;
}
/* The linklayer setting were not transferred from iproute2, in older
* versions, and the rate tables lookup systems have been dropped in
* the kernel. To keep backward compatible with older iproute2 tc
* utils, we detect the linklayer setting by detecting if the rate
* table were modified.
*
* For linklayer ATM table entries, the rate table will be aligned to
* 48 bytes, thus some table entries will contain the same value. The
* mpu (min packet unit) is also encoded into the old rate table, thus
* starting from the mpu, we find low and high table entries for
* mapping this cell. If these entries contain the same value, when
* the rate tables have been modified for linklayer ATM.
*
* This is done by rounding mpu to the nearest 48 bytes cell/entry,
* and then roundup to the next cell, calc the table entry one below,
* and compare.
*/
static __u8 __detect_linklayer(struct tc_ratespec *r, __u32 *rtab)
{
int low = roundup(r->mpu, 48);
int high = roundup(low+1, 48);
int cell_low = low >> r->cell_log;
int cell_high = (high >> r->cell_log) - 1;
/* rtab is too inaccurate at rates > 100Mbit/s */
if ((r->rate > (100000000/8)) || (rtab[0] == 0)) {
pr_debug("TC linklayer: Giving up ATM detection\n");
return TC_LINKLAYER_ETHERNET;
}
if ((cell_high > cell_low) && (cell_high < 256)
&& (rtab[cell_low] == rtab[cell_high])) {
pr_debug("TC linklayer: Detected ATM, low(%d)=high(%d)=%u\n",
cell_low, cell_high, rtab[cell_high]);
return TC_LINKLAYER_ATM;
}
return TC_LINKLAYER_ETHERNET;
}
static struct qdisc_rate_table *qdisc_rtab_list;
struct qdisc_rate_table *qdisc_get_rtab(struct tc_ratespec *r,
struct nlattr *tab)
{
struct qdisc_rate_table *rtab;
if (tab == NULL || r->rate == 0 || r->cell_log == 0 ||
nla_len(tab) != TC_RTAB_SIZE)
return NULL;
for (rtab = qdisc_rtab_list; rtab; rtab = rtab->next) {
if (!memcmp(&rtab->rate, r, sizeof(struct tc_ratespec)) &&
!memcmp(&rtab->data, nla_data(tab), 1024)) {
rtab->refcnt++;
return rtab;
}
}
rtab = kmalloc(sizeof(*rtab), GFP_KERNEL);
if (rtab) {
rtab->rate = *r;
rtab->refcnt = 1;
memcpy(rtab->data, nla_data(tab), 1024);
if (r->linklayer == TC_LINKLAYER_UNAWARE)
r->linklayer = __detect_linklayer(r, rtab->data);
rtab->next = qdisc_rtab_list;
qdisc_rtab_list = rtab;
}
return rtab;
}
EXPORT_SYMBOL(qdisc_get_rtab);
void qdisc_put_rtab(struct qdisc_rate_table *tab)
{
struct qdisc_rate_table *rtab, **rtabp;
if (!tab || --tab->refcnt)
return;
for (rtabp = &qdisc_rtab_list;
(rtab = *rtabp) != NULL;
rtabp = &rtab->next) {
if (rtab == tab) {
*rtabp = rtab->next;
kfree(rtab);
return;
}
}
}
EXPORT_SYMBOL(qdisc_put_rtab);
static LIST_HEAD(qdisc_stab_list);
static const struct nla_policy stab_policy[TCA_STAB_MAX + 1] = {
[TCA_STAB_BASE] = { .len = sizeof(struct tc_sizespec) },
[TCA_STAB_DATA] = { .type = NLA_BINARY },
};
static struct qdisc_size_table *qdisc_get_stab(struct nlattr *opt)
{
struct nlattr *tb[TCA_STAB_MAX + 1];
struct qdisc_size_table *stab;
struct tc_sizespec *s;
unsigned int tsize = 0;
u16 *tab = NULL;
int err;
err = nla_parse_nested(tb, TCA_STAB_MAX, opt, stab_policy, NULL);
if (err < 0)
return ERR_PTR(err);
if (!tb[TCA_STAB_BASE])
return ERR_PTR(-EINVAL);
s = nla_data(tb[TCA_STAB_BASE]);
if (s->tsize > 0) {
if (!tb[TCA_STAB_DATA])
return ERR_PTR(-EINVAL);
tab = nla_data(tb[TCA_STAB_DATA]);
tsize = nla_len(tb[TCA_STAB_DATA]) / sizeof(u16);
}
if (tsize != s->tsize || (!tab && tsize > 0))
return ERR_PTR(-EINVAL);
list_for_each_entry(stab, &qdisc_stab_list, list) {
if (memcmp(&stab->szopts, s, sizeof(*s)))
continue;
if (tsize > 0 && memcmp(stab->data, tab, tsize * sizeof(u16)))
continue;
stab->refcnt++;
return stab;
}
stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
if (!stab)
return ERR_PTR(-ENOMEM);
stab->refcnt = 1;
stab->szopts = *s;
if (tsize > 0)
memcpy(stab->data, tab, tsize * sizeof(u16));
list_add_tail(&stab->list, &qdisc_stab_list);
return stab;
}
static void stab_kfree_rcu(struct rcu_head *head)
{
kfree(container_of(head, struct qdisc_size_table, rcu));
}
void qdisc_put_stab(struct qdisc_size_table *tab)
{
if (!tab)
return;
if (--tab->refcnt == 0) {
list_del(&tab->list);
call_rcu_bh(&tab->rcu, stab_kfree_rcu);
}
}
EXPORT_SYMBOL(qdisc_put_stab);
static int qdisc_dump_stab(struct sk_buff *skb, struct qdisc_size_table *stab)
{
struct nlattr *nest;
nest = nla_nest_start(skb, TCA_STAB);
if (nest == NULL)
goto nla_put_failure;
if (nla_put(skb, TCA_STAB_BASE, sizeof(stab->szopts), &stab->szopts))
goto nla_put_failure;
nla_nest_end(skb, nest);
return skb->len;
nla_put_failure:
return -1;
}
void __qdisc_calculate_pkt_len(struct sk_buff *skb,
const struct qdisc_size_table *stab)
{
int pkt_len, slot;
pkt_len = skb->len + stab->szopts.overhead;
if (unlikely(!stab->szopts.tsize))
goto out;
slot = pkt_len + stab->szopts.cell_align;
if (unlikely(slot < 0))
slot = 0;
slot >>= stab->szopts.cell_log;
if (likely(slot < stab->szopts.tsize))
pkt_len = stab->data[slot];
else
pkt_len = stab->data[stab->szopts.tsize - 1] *
(slot / stab->szopts.tsize) +
stab->data[slot % stab->szopts.tsize];
pkt_len <<= stab->szopts.size_log;
out:
if (unlikely(pkt_len < 1))
pkt_len = 1;
qdisc_skb_cb(skb)->pkt_len = pkt_len;
}
EXPORT_SYMBOL(__qdisc_calculate_pkt_len);
void qdisc_warn_nonwc(const char *txt, struct Qdisc *qdisc)
{
if (!(qdisc->flags & TCQ_F_WARN_NONWC)) {
pr_warn("%s: %s qdisc %X: is non-work-conserving?\n",
txt, qdisc->ops->id, qdisc->handle >> 16);
qdisc->flags |= TCQ_F_WARN_NONWC;
}
}
EXPORT_SYMBOL(qdisc_warn_nonwc);
static enum hrtimer_restart qdisc_watchdog(struct hrtimer *timer)
{
struct qdisc_watchdog *wd = container_of(timer, struct qdisc_watchdog,
timer);
net: sched: suspicious RCU usage in qdisc_watchdog Suspicious RCU usage in qdisc_watchdog call needs to be done inside rcu_read_lock/rcu_read_unlock. And then Qdisc destroy operations need to ensure timer is cancelled before removing qdisc structure. [ 3992.191339] =============================== [ 3992.191340] [ INFO: suspicious RCU usage. ] [ 3992.191343] 3.17.0-rc6net-next+ #72 Not tainted [ 3992.191345] ------------------------------- [ 3992.191347] include/net/sch_generic.h:272 suspicious rcu_dereference_check() usage! [ 3992.191348] [ 3992.191348] other info that might help us debug this: [ 3992.191348] [ 3992.191351] [ 3992.191351] rcu_scheduler_active = 1, debug_locks = 1 [ 3992.191353] no locks held by swapper/1/0. [ 3992.191355] [ 3992.191355] stack backtrace: [ 3992.191358] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 3.17.0-rc6net-next+ #72 [ 3992.191360] Hardware name: /DZ77RE-75K, BIOS GAZ7711H.86A.0060.2012.1115.1750 11/15/2012 [ 3992.191362] 0000000000000001 ffff880235803e48 ffffffff8178f92c 0000000000000000 [ 3992.191366] ffff8802322224a0 ffff880235803e78 ffffffff810c9966 ffff8800a5fe3000 [ 3992.191370] ffff880235803f30 ffff8802359cd768 ffff8802359cd6e0 ffff880235803e98 [ 3992.191374] Call Trace: [ 3992.191376] <IRQ> [<ffffffff8178f92c>] dump_stack+0x4e/0x68 [ 3992.191387] [<ffffffff810c9966>] lockdep_rcu_suspicious+0xe6/0x130 [ 3992.191392] [<ffffffff8167213a>] qdisc_watchdog+0x8a/0xb0 [ 3992.191396] [<ffffffff810f93f2>] __run_hrtimer+0x72/0x420 [ 3992.191399] [<ffffffff810f9bcd>] ? hrtimer_interrupt+0x7d/0x240 [ 3992.191403] [<ffffffff816720b0>] ? tc_classify+0xc0/0xc0 [ 3992.191406] [<ffffffff810f9c4f>] hrtimer_interrupt+0xff/0x240 [ 3992.191410] [<ffffffff8109e4a5>] ? __atomic_notifier_call_chain+0x5/0x140 [ 3992.191415] [<ffffffff8103577b>] local_apic_timer_interrupt+0x3b/0x60 [ 3992.191419] [<ffffffff8179c2b5>] smp_apic_timer_interrupt+0x45/0x60 [ 3992.191422] [<ffffffff8179a6bf>] apic_timer_interrupt+0x6f/0x80 [ 3992.191424] <EOI> [<ffffffff815ed233>] ? cpuidle_enter_state+0x73/0x2e0 [ 3992.191432] [<ffffffff815ed22e>] ? cpuidle_enter_state+0x6e/0x2e0 [ 3992.191437] [<ffffffff815ed567>] cpuidle_enter+0x17/0x20 [ 3992.191441] [<ffffffff810c0741>] cpu_startup_entry+0x3d1/0x4a0 [ 3992.191445] [<ffffffff81106fc6>] ? clockevents_config_and_register+0x26/0x30 [ 3992.191448] [<ffffffff81033c16>] start_secondary+0x1b6/0x260 Fixes: b26b0d1e8b1 ("net: qdisc: use rcu prefix and silence sparse warnings") Signed-off-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <cwang@twopensource.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-10-03 12:43:09 +07:00
rcu_read_lock();
__netif_schedule(qdisc_root(wd->qdisc));
net: sched: suspicious RCU usage in qdisc_watchdog Suspicious RCU usage in qdisc_watchdog call needs to be done inside rcu_read_lock/rcu_read_unlock. And then Qdisc destroy operations need to ensure timer is cancelled before removing qdisc structure. [ 3992.191339] =============================== [ 3992.191340] [ INFO: suspicious RCU usage. ] [ 3992.191343] 3.17.0-rc6net-next+ #72 Not tainted [ 3992.191345] ------------------------------- [ 3992.191347] include/net/sch_generic.h:272 suspicious rcu_dereference_check() usage! [ 3992.191348] [ 3992.191348] other info that might help us debug this: [ 3992.191348] [ 3992.191351] [ 3992.191351] rcu_scheduler_active = 1, debug_locks = 1 [ 3992.191353] no locks held by swapper/1/0. [ 3992.191355] [ 3992.191355] stack backtrace: [ 3992.191358] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 3.17.0-rc6net-next+ #72 [ 3992.191360] Hardware name: /DZ77RE-75K, BIOS GAZ7711H.86A.0060.2012.1115.1750 11/15/2012 [ 3992.191362] 0000000000000001 ffff880235803e48 ffffffff8178f92c 0000000000000000 [ 3992.191366] ffff8802322224a0 ffff880235803e78 ffffffff810c9966 ffff8800a5fe3000 [ 3992.191370] ffff880235803f30 ffff8802359cd768 ffff8802359cd6e0 ffff880235803e98 [ 3992.191374] Call Trace: [ 3992.191376] <IRQ> [<ffffffff8178f92c>] dump_stack+0x4e/0x68 [ 3992.191387] [<ffffffff810c9966>] lockdep_rcu_suspicious+0xe6/0x130 [ 3992.191392] [<ffffffff8167213a>] qdisc_watchdog+0x8a/0xb0 [ 3992.191396] [<ffffffff810f93f2>] __run_hrtimer+0x72/0x420 [ 3992.191399] [<ffffffff810f9bcd>] ? hrtimer_interrupt+0x7d/0x240 [ 3992.191403] [<ffffffff816720b0>] ? tc_classify+0xc0/0xc0 [ 3992.191406] [<ffffffff810f9c4f>] hrtimer_interrupt+0xff/0x240 [ 3992.191410] [<ffffffff8109e4a5>] ? __atomic_notifier_call_chain+0x5/0x140 [ 3992.191415] [<ffffffff8103577b>] local_apic_timer_interrupt+0x3b/0x60 [ 3992.191419] [<ffffffff8179c2b5>] smp_apic_timer_interrupt+0x45/0x60 [ 3992.191422] [<ffffffff8179a6bf>] apic_timer_interrupt+0x6f/0x80 [ 3992.191424] <EOI> [<ffffffff815ed233>] ? cpuidle_enter_state+0x73/0x2e0 [ 3992.191432] [<ffffffff815ed22e>] ? cpuidle_enter_state+0x6e/0x2e0 [ 3992.191437] [<ffffffff815ed567>] cpuidle_enter+0x17/0x20 [ 3992.191441] [<ffffffff810c0741>] cpu_startup_entry+0x3d1/0x4a0 [ 3992.191445] [<ffffffff81106fc6>] ? clockevents_config_and_register+0x26/0x30 [ 3992.191448] [<ffffffff81033c16>] start_secondary+0x1b6/0x260 Fixes: b26b0d1e8b1 ("net: qdisc: use rcu prefix and silence sparse warnings") Signed-off-by: John Fastabend <john.r.fastabend@intel.com> Acked-by: Cong Wang <cwang@twopensource.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-10-03 12:43:09 +07:00
rcu_read_unlock();
return HRTIMER_NORESTART;
}
void qdisc_watchdog_init(struct qdisc_watchdog *wd, struct Qdisc *qdisc)
{
net: sched: use pinned timers While using a MQ + NETEM setup, I had confirmation that the default timer migration ( /proc/sys/kernel/timer_migration ) is killing us. Installing this on a receiver side of a TCP_STREAM test, (NIC has 8 TX queues) : EST="est 1sec 4sec" for ETH in eth1 do tc qd del dev $ETH root 2>/dev/null tc qd add dev $ETH root handle 1: mq tc qd add dev $ETH parent 1:1 $EST netem limit 70000 delay 6ms tc qd add dev $ETH parent 1:2 $EST netem limit 70000 delay 8ms tc qd add dev $ETH parent 1:3 $EST netem limit 70000 delay 10ms tc qd add dev $ETH parent 1:4 $EST netem limit 70000 delay 12ms tc qd add dev $ETH parent 1:5 $EST netem limit 70000 delay 14ms tc qd add dev $ETH parent 1:6 $EST netem limit 70000 delay 16ms tc qd add dev $ETH parent 1:7 $EST netem limit 80000 delay 18ms tc qd add dev $ETH parent 1:8 $EST netem limit 90000 delay 20ms done We can see that timers get migrated into a single cpu, presumably idle at the time timers are set up. Then all qdisc dequeues run from this cpu and huge lock contention happens. This single cpu is stuck in softirq mode and cannot dequeue fast enough. 39.24% [kernel] [k] _raw_spin_lock 2.65% [kernel] [k] netem_enqueue 1.80% [kernel] [k] netem_dequeue 1.63% [kernel] [k] copy_user_enhanced_fast_string 1.45% [kernel] [k] _raw_spin_lock_bh By pinning qdisc timers on the cpu running the qdisc, we respect proper XPS setting and remove this lock contention. 5.84% [kernel] [k] netem_enqueue 4.83% [kernel] [k] _raw_spin_lock 2.92% [kernel] [k] copy_user_enhanced_fast_string Current Qdiscs that benefit from this change are : netem, cbq, fq, hfsc, tbf, htb. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-09-21 08:01:30 +07:00
hrtimer_init(&wd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED);
wd->timer.function = qdisc_watchdog;
wd->qdisc = qdisc;
}
EXPORT_SYMBOL(qdisc_watchdog_init);
void qdisc_watchdog_schedule_ns(struct qdisc_watchdog *wd, u64 expires)
{
if (test_bit(__QDISC_STATE_DEACTIVATED,
&qdisc_root_sleeping(wd->qdisc)->state))
return;
if (wd->last_expires == expires)
return;
wd->last_expires = expires;
hrtimer_start(&wd->timer,
ns_to_ktime(expires),
net: sched: use pinned timers While using a MQ + NETEM setup, I had confirmation that the default timer migration ( /proc/sys/kernel/timer_migration ) is killing us. Installing this on a receiver side of a TCP_STREAM test, (NIC has 8 TX queues) : EST="est 1sec 4sec" for ETH in eth1 do tc qd del dev $ETH root 2>/dev/null tc qd add dev $ETH root handle 1: mq tc qd add dev $ETH parent 1:1 $EST netem limit 70000 delay 6ms tc qd add dev $ETH parent 1:2 $EST netem limit 70000 delay 8ms tc qd add dev $ETH parent 1:3 $EST netem limit 70000 delay 10ms tc qd add dev $ETH parent 1:4 $EST netem limit 70000 delay 12ms tc qd add dev $ETH parent 1:5 $EST netem limit 70000 delay 14ms tc qd add dev $ETH parent 1:6 $EST netem limit 70000 delay 16ms tc qd add dev $ETH parent 1:7 $EST netem limit 80000 delay 18ms tc qd add dev $ETH parent 1:8 $EST netem limit 90000 delay 20ms done We can see that timers get migrated into a single cpu, presumably idle at the time timers are set up. Then all qdisc dequeues run from this cpu and huge lock contention happens. This single cpu is stuck in softirq mode and cannot dequeue fast enough. 39.24% [kernel] [k] _raw_spin_lock 2.65% [kernel] [k] netem_enqueue 1.80% [kernel] [k] netem_dequeue 1.63% [kernel] [k] copy_user_enhanced_fast_string 1.45% [kernel] [k] _raw_spin_lock_bh By pinning qdisc timers on the cpu running the qdisc, we respect proper XPS setting and remove this lock contention. 5.84% [kernel] [k] netem_enqueue 4.83% [kernel] [k] _raw_spin_lock 2.92% [kernel] [k] copy_user_enhanced_fast_string Current Qdiscs that benefit from this change are : netem, cbq, fq, hfsc, tbf, htb. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-09-21 08:01:30 +07:00
HRTIMER_MODE_ABS_PINNED);
}
EXPORT_SYMBOL(qdisc_watchdog_schedule_ns);
void qdisc_watchdog_cancel(struct qdisc_watchdog *wd)
{
hrtimer_cancel(&wd->timer);
}
EXPORT_SYMBOL(qdisc_watchdog_cancel);
static struct hlist_head *qdisc_class_hash_alloc(unsigned int n)
{
struct hlist_head *h;
unsigned int i;
h = kvmalloc_array(n, sizeof(struct hlist_head), GFP_KERNEL);
if (h != NULL) {
for (i = 0; i < n; i++)
INIT_HLIST_HEAD(&h[i]);
}
return h;
}
void qdisc_class_hash_grow(struct Qdisc *sch, struct Qdisc_class_hash *clhash)
{
struct Qdisc_class_common *cl;
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 08:06:00 +07:00
struct hlist_node *next;
struct hlist_head *nhash, *ohash;
unsigned int nsize, nmask, osize;
unsigned int i, h;
/* Rehash when load factor exceeds 0.75 */
if (clhash->hashelems * 4 <= clhash->hashsize * 3)
return;
nsize = clhash->hashsize * 2;
nmask = nsize - 1;
nhash = qdisc_class_hash_alloc(nsize);
if (nhash == NULL)
return;
ohash = clhash->hash;
osize = clhash->hashsize;
sch_tree_lock(sch);
for (i = 0; i < osize; i++) {
hlist: drop the node parameter from iterators I'm not sure why, but the hlist for each entry iterators were conceived list_for_each_entry(pos, head, member) The hlist ones were greedy and wanted an extra parameter: hlist_for_each_entry(tpos, pos, head, member) Why did they need an extra pos parameter? I'm not quite sure. Not only they don't really need it, it also prevents the iterator from looking exactly like the list iterator, which is unfortunate. Besides the semantic patch, there was some manual work required: - Fix up the actual hlist iterators in linux/list.h - Fix up the declaration of other iterators based on the hlist ones. - A very small amount of places were using the 'node' parameter, this was modified to use 'obj->member' instead. - Coccinelle didn't handle the hlist_for_each_entry_safe iterator properly, so those had to be fixed up manually. The semantic patch which is mostly the work of Peter Senna Tschudin is here: @@ iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host; type T; expression a,c,d,e; identifier b; statement S; @@ -T b; <+... when != b ( hlist_for_each_entry(a, - b, c, d) S | hlist_for_each_entry_continue(a, - b, c) S | hlist_for_each_entry_from(a, - b, c) S | hlist_for_each_entry_rcu(a, - b, c, d) S | hlist_for_each_entry_rcu_bh(a, - b, c, d) S | hlist_for_each_entry_continue_rcu_bh(a, - b, c) S | for_each_busy_worker(a, c, - b, d) S | ax25_uid_for_each(a, - b, c) S | ax25_for_each(a, - b, c) S | inet_bind_bucket_for_each(a, - b, c) S | sctp_for_each_hentry(a, - b, c) S | sk_for_each(a, - b, c) S | sk_for_each_rcu(a, - b, c) S | sk_for_each_from -(a, b) +(a) S + sk_for_each_from(a) S | sk_for_each_safe(a, - b, c, d) S | sk_for_each_bound(a, - b, c) S | hlist_for_each_entry_safe(a, - b, c, d, e) S | hlist_for_each_entry_continue_rcu(a, - b, c) S | nr_neigh_for_each(a, - b, c) S | nr_neigh_for_each_safe(a, - b, c, d) S | nr_node_for_each(a, - b, c) S | nr_node_for_each_safe(a, - b, c, d) S | - for_each_gfn_sp(a, c, d, b) S + for_each_gfn_sp(a, c, d) S | - for_each_gfn_indirect_valid_sp(a, c, d, b) S + for_each_gfn_indirect_valid_sp(a, c, d) S | for_each_host(a, - b, c) S | for_each_host_safe(a, - b, c, d) S | for_each_mesh_entry(a, - b, c, d) S ) ...+> [akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c] [akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c] [akpm@linux-foundation.org: checkpatch fixes] [akpm@linux-foundation.org: fix warnings] [akpm@linux-foudnation.org: redo intrusive kvm changes] Tested-by: Peter Senna Tschudin <peter.senna@gmail.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Gleb Natapov <gleb@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 08:06:00 +07:00
hlist_for_each_entry_safe(cl, next, &ohash[i], hnode) {
h = qdisc_class_hash(cl->classid, nmask);
hlist_add_head(&cl->hnode, &nhash[h]);
}
}
clhash->hash = nhash;
clhash->hashsize = nsize;
clhash->hashmask = nmask;
sch_tree_unlock(sch);
kvfree(ohash);
}
EXPORT_SYMBOL(qdisc_class_hash_grow);
int qdisc_class_hash_init(struct Qdisc_class_hash *clhash)
{
unsigned int size = 4;
clhash->hash = qdisc_class_hash_alloc(size);
if (clhash->hash == NULL)
return -ENOMEM;
clhash->hashsize = size;
clhash->hashmask = size - 1;
clhash->hashelems = 0;
return 0;
}
EXPORT_SYMBOL(qdisc_class_hash_init);
void qdisc_class_hash_destroy(struct Qdisc_class_hash *clhash)
{
kvfree(clhash->hash);
}
EXPORT_SYMBOL(qdisc_class_hash_destroy);
void qdisc_class_hash_insert(struct Qdisc_class_hash *clhash,
struct Qdisc_class_common *cl)
{
unsigned int h;
INIT_HLIST_NODE(&cl->hnode);
h = qdisc_class_hash(cl->classid, clhash->hashmask);
hlist_add_head(&cl->hnode, &clhash->hash[h]);
clhash->hashelems++;
}
EXPORT_SYMBOL(qdisc_class_hash_insert);
void qdisc_class_hash_remove(struct Qdisc_class_hash *clhash,
struct Qdisc_class_common *cl)
{
hlist_del(&cl->hnode);
clhash->hashelems--;
}
EXPORT_SYMBOL(qdisc_class_hash_remove);
/* Allocate an unique handle from space managed by kernel
* Possible range is [8000-FFFF]:0000 (0x8000 values)
*/
static u32 qdisc_alloc_handle(struct net_device *dev)
{
int i = 0x8000;
static u32 autohandle = TC_H_MAKE(0x80000000U, 0);
do {
autohandle += TC_H_MAKE(0x10000U, 0);
if (autohandle == TC_H_MAKE(TC_H_ROOT, 0))
autohandle = TC_H_MAKE(0x80000000U, 0);
if (!qdisc_lookup(dev, autohandle))
return autohandle;
cond_resched();
} while (--i > 0);
return 0;
}
void qdisc_tree_reduce_backlog(struct Qdisc *sch, unsigned int n,
unsigned int len)
{
const struct Qdisc_class_ops *cops;
unsigned long cl;
u32 parentid;
bool notify;
int drops;
if (n == 0 && len == 0)
return;
drops = max_t(int, n, 0);
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
rcu_read_lock();
while ((parentid = sch->parent)) {
if (TC_H_MAJ(parentid) == TC_H_MAJ(TC_H_INGRESS))
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
break;
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
if (sch->flags & TCQ_F_NOPARENT)
break;
/* Notify parent qdisc only if child qdisc becomes empty.
*
* If child was empty even before update then backlog
* counter is screwed and we skip notification because
* parent class is already passive.
*/
notify = !sch->q.qlen && !WARN_ON_ONCE(!n);
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
/* TODO: perform the search on a per txq basis */
sch = qdisc_lookup(qdisc_dev(sch), TC_H_MAJ(parentid));
if (sch == NULL) {
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
WARN_ON_ONCE(parentid != TC_H_ROOT);
break;
}
cops = sch->ops->cl_ops;
if (notify && cops->qlen_notify) {
cl = cops->find(sch, parentid);
cops->qlen_notify(sch, cl);
}
sch->q.qlen -= n;
sch->qstats.backlog -= len;
__qdisc_qstats_drop(sch, drops);
}
net_sched: fix qdisc_tree_decrease_qlen() races qdisc_tree_decrease_qlen() suffers from two problems on multiqueue devices. One problem is that it updates sch->q.qlen and sch->qstats.drops on the mq/mqprio root qdisc, while it should not : Daniele reported underflows errors : [ 681.774821] PAX: sch->q.qlen: 0 n: 1 [ 681.774825] PAX: size overflow detected in function qdisc_tree_decrease_qlen net/sched/sch_api.c:769 cicus.693_49 min, count: 72, decl: qlen; num: 0; context: sk_buff_head; [ 681.774954] CPU: 2 PID: 19 Comm: ksoftirqd/2 Tainted: G O 4.2.6.201511282239-1-grsec #1 [ 681.774955] Hardware name: ASUSTeK COMPUTER INC. X302LJ/X302LJ, BIOS X302LJ.202 03/05/2015 [ 681.774956] ffffffffa9a04863 0000000000000000 0000000000000000 ffffffffa990ff7c [ 681.774959] ffffc90000d3bc38 ffffffffa95d2810 0000000000000007 ffffffffa991002b [ 681.774960] ffffc90000d3bc68 ffffffffa91a44f4 0000000000000001 0000000000000001 [ 681.774962] Call Trace: [ 681.774967] [<ffffffffa95d2810>] dump_stack+0x4c/0x7f [ 681.774970] [<ffffffffa91a44f4>] report_size_overflow+0x34/0x50 [ 681.774972] [<ffffffffa94d17e2>] qdisc_tree_decrease_qlen+0x152/0x160 [ 681.774976] [<ffffffffc02694b1>] fq_codel_dequeue+0x7b1/0x820 [sch_fq_codel] [ 681.774978] [<ffffffffc02680a0>] ? qdisc_peek_dequeued+0xa0/0xa0 [sch_fq_codel] [ 681.774980] [<ffffffffa94cd92d>] __qdisc_run+0x4d/0x1d0 [ 681.774983] [<ffffffffa949b2b2>] net_tx_action+0xc2/0x160 [ 681.774985] [<ffffffffa90664c1>] __do_softirq+0xf1/0x200 [ 681.774987] [<ffffffffa90665ee>] run_ksoftirqd+0x1e/0x30 [ 681.774989] [<ffffffffa90896b0>] smpboot_thread_fn+0x150/0x260 [ 681.774991] [<ffffffffa9089560>] ? sort_range+0x40/0x40 [ 681.774992] [<ffffffffa9085fe4>] kthread+0xe4/0x100 [ 681.774994] [<ffffffffa9085f00>] ? kthread_worker_fn+0x170/0x170 [ 681.774995] [<ffffffffa95d8d1e>] ret_from_fork+0x3e/0x70 mq/mqprio have their own ways to report qlen/drops by folding stats on all their queues, with appropriate locking. A second problem is that qdisc_tree_decrease_qlen() calls qdisc_lookup() without proper locking : concurrent qdisc updates could corrupt the list that qdisc_match_from_root() parses to find a qdisc given its handle. Fix first problem adding a TCQ_F_NOPARENT qdisc flag that qdisc_tree_decrease_qlen() can use to abort its tree traversal, as soon as it meets a mq/mqprio qdisc children. Second problem can be fixed by RCU protection. Qdisc are already freed after RCU grace period, so qdisc_list_add() and qdisc_list_del() simply have to use appropriate rcu list variants. A future patch will add a per struct netdev_queue list anchor, so that qdisc_tree_decrease_qlen() can have more efficient lookups. Reported-by: Daniele Fucini <dfucini@gmail.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Cong Wang <cwang@twopensource.com> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-12-02 11:08:51 +07:00
rcu_read_unlock();
}
EXPORT_SYMBOL(qdisc_tree_reduce_backlog);
static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
u32 portid, u32 seq, u16 flags, int event)
{
struct gnet_stats_basic_cpu __percpu *cpu_bstats = NULL;
struct gnet_stats_queue __percpu *cpu_qstats = NULL;
struct tcmsg *tcm;
struct nlmsghdr *nlh;
unsigned char *b = skb_tail_pointer(skb);
struct gnet_dump d;
struct qdisc_size_table *stab;
__u32 qlen;
cond_resched();
nlh = nlmsg_put(skb, portid, seq, event, sizeof(*tcm), flags);
if (!nlh)
goto out_nlmsg_trim;
tcm = nlmsg_data(nlh);
tcm->tcm_family = AF_UNSPEC;
tcm->tcm__pad1 = 0;
tcm->tcm__pad2 = 0;
tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
tcm->tcm_parent = clid;
tcm->tcm_handle = q->handle;
tcm->tcm_info = refcount_read(&q->refcnt);
if (nla_put_string(skb, TCA_KIND, q->ops->id))
goto nla_put_failure;
if (q->ops->dump && q->ops->dump(q, skb) < 0)
goto nla_put_failure;
qlen = q->q.qlen;
stab = rtnl_dereference(q->stab);
if (stab && qdisc_dump_stab(skb, stab) < 0)
goto nla_put_failure;
if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS, TCA_XSTATS,
NULL, &d, TCA_PAD) < 0)
goto nla_put_failure;
if (q->ops->dump_stats && q->ops->dump_stats(q, &d) < 0)
goto nla_put_failure;
if (qdisc_is_percpu_stats(q)) {
cpu_bstats = q->cpu_bstats;
cpu_qstats = q->cpu_qstats;
}
if (gnet_stats_copy_basic(qdisc_root_sleeping_running(q),
&d, cpu_bstats, &q->bstats) < 0 ||
gnet_stats_copy_rate_est(&d, &q->rate_est) < 0 ||
gnet_stats_copy_queue(&d, cpu_qstats, &q->qstats, qlen) < 0)
goto nla_put_failure;
if (gnet_stats_finish_copy(&d) < 0)
goto nla_put_failure;
nlh->nlmsg_len = skb_tail_pointer(skb) - b;
return skb->len;
out_nlmsg_trim:
nla_put_failure:
nlmsg_trim(skb, b);
return -1;
}
static bool tc_qdisc_dump_ignore(struct Qdisc *q, bool dump_invisible)
{
if (q->flags & TCQ_F_BUILTIN)
return true;
if ((q->flags & TCQ_F_INVISIBLE) && !dump_invisible)
return true;
return false;
}
static int qdisc_notify(struct net *net, struct sk_buff *oskb,
struct nlmsghdr *n, u32 clid,
struct Qdisc *old, struct Qdisc *new)
{
struct sk_buff *skb;
u32 portid = oskb ? NETLINK_CB(oskb).portid : 0;
skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
if (!skb)
return -ENOBUFS;
if (old && !tc_qdisc_dump_ignore(old, false)) {
if (tc_fill_qdisc(skb, old, clid, portid, n->nlmsg_seq,
0, RTM_DELQDISC) < 0)
goto err_out;
}
if (new && !tc_qdisc_dump_ignore(new, false)) {
if (tc_fill_qdisc(skb, new, clid, portid, n->nlmsg_seq,
old ? NLM_F_REPLACE : 0, RTM_NEWQDISC) < 0)
goto err_out;
}
if (skb->len)
return rtnetlink_send(skb, net, portid, RTNLGRP_TC,
n->nlmsg_flags & NLM_F_ECHO);
err_out:
kfree_skb(skb);
return -EINVAL;
}
static void notify_and_destroy(struct net *net, struct sk_buff *skb,
struct nlmsghdr *n, u32 clid,
struct Qdisc *old, struct Qdisc *new)
{
if (new || old)
qdisc_notify(net, skb, n, clid, old, new);
if (old)
qdisc_destroy(old);
}
/* Graft qdisc "new" to class "classid" of qdisc "parent" or
* to device "dev".
*
* When appropriate send a netlink notification using 'skb'
* and "n".
*
* On success, destroy old qdisc.
*/
static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
struct sk_buff *skb, struct nlmsghdr *n, u32 classid,
struct Qdisc *new, struct Qdisc *old)
{
struct Qdisc *q = old;
struct net *net = dev_net(dev);
int err = 0;
if (parent == NULL) {
unsigned int i, num_q, ingress;
ingress = 0;
num_q = dev->num_tx_queues;
if ((q && q->flags & TCQ_F_INGRESS) ||
(new && new->flags & TCQ_F_INGRESS)) {
num_q = 1;
ingress = 1;
if (!dev_ingress_queue(dev))
return -ENOENT;
}
if (dev->flags & IFF_UP)
dev_deactivate(dev);
net_sched: invoke ->attach() after setting dev->qdisc For mq qdisc, we add per tx queue qdisc to root qdisc for display purpose, however, that happens too early, before the new dev->qdisc is finally set, this causes q->list points to an old root qdisc which is going to be freed right before assigning with a new one. Fix this by moving ->attach() after setting dev->qdisc. For the record, this fixes the following crash: ------------[ cut here ]------------ WARNING: CPU: 1 PID: 975 at lib/list_debug.c:59 __list_del_entry+0x5a/0x98() list_del corruption. prev->next should be ffff8800d1998ae8, but was 6b6b6b6b6b6b6b6b CPU: 1 PID: 975 Comm: tc Not tainted 4.1.0-rc4+ #1019 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 0000000000000009 ffff8800d73fb928 ffffffff81a44e7f 0000000047574756 ffff8800d73fb978 ffff8800d73fb968 ffffffff810790da ffff8800cfc4cd20 ffffffff814e725b ffff8800d1998ae8 ffffffff82381250 0000000000000000 Call Trace: [<ffffffff81a44e7f>] dump_stack+0x4c/0x65 [<ffffffff810790da>] warn_slowpath_common+0x9c/0xb6 [<ffffffff814e725b>] ? __list_del_entry+0x5a/0x98 [<ffffffff81079162>] warn_slowpath_fmt+0x46/0x48 [<ffffffff81820eb0>] ? dev_graft_qdisc+0x5e/0x6a [<ffffffff814e725b>] __list_del_entry+0x5a/0x98 [<ffffffff814e72a7>] list_del+0xe/0x2d [<ffffffff81822f05>] qdisc_list_del+0x1e/0x20 [<ffffffff81820cd1>] qdisc_destroy+0x30/0xd6 [<ffffffff81822676>] qdisc_graft+0x11d/0x243 [<ffffffff818233c1>] tc_get_qdisc+0x1a6/0x1d4 [<ffffffff810b5eaf>] ? mark_lock+0x2e/0x226 [<ffffffff817ff8f5>] rtnetlink_rcv_msg+0x181/0x194 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff774>] ? __rtnl_unlock+0x17/0x17 [<ffffffff81855dc6>] netlink_rcv_skb+0x4d/0x93 [<ffffffff817ff756>] rtnetlink_rcv+0x26/0x2d [<ffffffff818544b2>] netlink_unicast+0xcb/0x150 [<ffffffff81161db9>] ? might_fault+0x59/0xa9 [<ffffffff81854f78>] netlink_sendmsg+0x4fa/0x51c [<ffffffff817d6e09>] sock_sendmsg_nosec+0x12/0x1d [<ffffffff817d8967>] sock_sendmsg+0x29/0x2e [<ffffffff817d8cf3>] ___sys_sendmsg+0x1b4/0x23a [<ffffffff8100a1b8>] ? native_sched_clock+0x35/0x37 [<ffffffff810a1d83>] ? sched_clock_local+0x12/0x72 [<ffffffff810a1fd4>] ? sched_clock_cpu+0x9e/0xb7 [<ffffffff810def2a>] ? current_kernel_time+0xe/0x32 [<ffffffff810b4bc5>] ? lock_release_holdtime.part.29+0x71/0x7f [<ffffffff810ddebf>] ? read_seqcount_begin.constprop.27+0x5f/0x76 [<ffffffff810b6292>] ? trace_hardirqs_on_caller+0x17d/0x199 [<ffffffff811b14d5>] ? __fget_light+0x50/0x78 [<ffffffff817d9808>] __sys_sendmsg+0x42/0x60 [<ffffffff817d9838>] SyS_sendmsg+0x12/0x1c [<ffffffff81a50e97>] system_call_fastpath+0x12/0x6f ---[ end trace ef29d3fb28e97ae7 ]--- For long term, we probably need to clean up the qdisc_graft() code in case it hides other bugs like this. Fixes: 95dc19299f74 ("pkt_sched: give visibility to mq slave qdiscs") Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-27 06:08:48 +07:00
if (new && new->ops->attach)
goto skip;
for (i = 0; i < num_q; i++) {
struct netdev_queue *dev_queue = dev_ingress_queue(dev);
if (!ingress)
dev_queue = netdev_get_tx_queue(dev, i);
old = dev_graft_qdisc(dev_queue, new);
if (new && i > 0)
net_sched: fix a refcount_t issue with noop_qdisc syzkaller reported a refcount_t warning [1] Issue here is that noop_qdisc refcnt was never really considered as a true refcount, since qdisc_destroy() found TCQ_F_BUILTIN set : if (qdisc->flags & TCQ_F_BUILTIN || !refcount_dec_and_test(&qdisc->refcnt))) return; Meaning that all atomic_inc() we did on noop_qdisc.refcnt were not really needed, but harmless until refcount_t came. To fix this problem, we simply need to not increment noop_qdisc.refcnt, since we never decrement it. [1] refcount_t: increment on 0; use-after-free. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 21754 at lib/refcount.c:152 refcount_inc+0x47/0x50 lib/refcount.c:152 Kernel panic - not syncing: panic_on_warn set ... CPU: 0 PID: 21754 Comm: syz-executor7 Not tainted 4.13.0-rc6+ #20 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Call Trace: __dump_stack lib/dump_stack.c:16 [inline] dump_stack+0x194/0x257 lib/dump_stack.c:52 panic+0x1e4/0x417 kernel/panic.c:180 __warn+0x1c4/0x1d9 kernel/panic.c:541 report_bug+0x211/0x2d0 lib/bug.c:183 fixup_bug+0x40/0x90 arch/x86/kernel/traps.c:190 do_trap_no_signal arch/x86/kernel/traps.c:224 [inline] do_trap+0x260/0x390 arch/x86/kernel/traps.c:273 do_error_trap+0x120/0x390 arch/x86/kernel/traps.c:310 do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:323 invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:846 RIP: 0010:refcount_inc+0x47/0x50 lib/refcount.c:152 RSP: 0018:ffff8801c43477a0 EFLAGS: 00010282 RAX: 000000000000002b RBX: ffffffff86093c14 RCX: 0000000000000000 RDX: 000000000000002b RSI: ffffffff8159314e RDI: ffffed0038868ee8 RBP: ffff8801c43477a8 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff86093ac0 R13: 0000000000000001 R14: ffff8801d0f3bac0 R15: dffffc0000000000 attach_default_qdiscs net/sched/sch_generic.c:792 [inline] dev_activate+0x7d3/0xaa0 net/sched/sch_generic.c:833 __dev_open+0x227/0x330 net/core/dev.c:1380 __dev_change_flags+0x695/0x990 net/core/dev.c:6726 dev_change_flags+0x88/0x140 net/core/dev.c:6792 dev_ifsioc+0x5a6/0x930 net/core/dev_ioctl.c:256 dev_ioctl+0x2bc/0xf90 net/core/dev_ioctl.c:554 sock_do_ioctl+0x94/0xb0 net/socket.c:968 sock_ioctl+0x2c2/0x440 net/socket.c:1058 vfs_ioctl fs/ioctl.c:45 [inline] do_vfs_ioctl+0x1b1/0x1520 fs/ioctl.c:685 SYSC_ioctl fs/ioctl.c:700 [inline] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691 Fixes: 7b9364050246 ("net, sched: convert Qdisc.refcnt from atomic_t to refcount_t") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Cc: Reshetova, Elena <elena.reshetova@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-08-25 11:12:28 +07:00
qdisc_refcount_inc(new);
if (!ingress)
qdisc_destroy(old);
}
net_sched: invoke ->attach() after setting dev->qdisc For mq qdisc, we add per tx queue qdisc to root qdisc for display purpose, however, that happens too early, before the new dev->qdisc is finally set, this causes q->list points to an old root qdisc which is going to be freed right before assigning with a new one. Fix this by moving ->attach() after setting dev->qdisc. For the record, this fixes the following crash: ------------[ cut here ]------------ WARNING: CPU: 1 PID: 975 at lib/list_debug.c:59 __list_del_entry+0x5a/0x98() list_del corruption. prev->next should be ffff8800d1998ae8, but was 6b6b6b6b6b6b6b6b CPU: 1 PID: 975 Comm: tc Not tainted 4.1.0-rc4+ #1019 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 0000000000000009 ffff8800d73fb928 ffffffff81a44e7f 0000000047574756 ffff8800d73fb978 ffff8800d73fb968 ffffffff810790da ffff8800cfc4cd20 ffffffff814e725b ffff8800d1998ae8 ffffffff82381250 0000000000000000 Call Trace: [<ffffffff81a44e7f>] dump_stack+0x4c/0x65 [<ffffffff810790da>] warn_slowpath_common+0x9c/0xb6 [<ffffffff814e725b>] ? __list_del_entry+0x5a/0x98 [<ffffffff81079162>] warn_slowpath_fmt+0x46/0x48 [<ffffffff81820eb0>] ? dev_graft_qdisc+0x5e/0x6a [<ffffffff814e725b>] __list_del_entry+0x5a/0x98 [<ffffffff814e72a7>] list_del+0xe/0x2d [<ffffffff81822f05>] qdisc_list_del+0x1e/0x20 [<ffffffff81820cd1>] qdisc_destroy+0x30/0xd6 [<ffffffff81822676>] qdisc_graft+0x11d/0x243 [<ffffffff818233c1>] tc_get_qdisc+0x1a6/0x1d4 [<ffffffff810b5eaf>] ? mark_lock+0x2e/0x226 [<ffffffff817ff8f5>] rtnetlink_rcv_msg+0x181/0x194 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff774>] ? __rtnl_unlock+0x17/0x17 [<ffffffff81855dc6>] netlink_rcv_skb+0x4d/0x93 [<ffffffff817ff756>] rtnetlink_rcv+0x26/0x2d [<ffffffff818544b2>] netlink_unicast+0xcb/0x150 [<ffffffff81161db9>] ? might_fault+0x59/0xa9 [<ffffffff81854f78>] netlink_sendmsg+0x4fa/0x51c [<ffffffff817d6e09>] sock_sendmsg_nosec+0x12/0x1d [<ffffffff817d8967>] sock_sendmsg+0x29/0x2e [<ffffffff817d8cf3>] ___sys_sendmsg+0x1b4/0x23a [<ffffffff8100a1b8>] ? native_sched_clock+0x35/0x37 [<ffffffff810a1d83>] ? sched_clock_local+0x12/0x72 [<ffffffff810a1fd4>] ? sched_clock_cpu+0x9e/0xb7 [<ffffffff810def2a>] ? current_kernel_time+0xe/0x32 [<ffffffff810b4bc5>] ? lock_release_holdtime.part.29+0x71/0x7f [<ffffffff810ddebf>] ? read_seqcount_begin.constprop.27+0x5f/0x76 [<ffffffff810b6292>] ? trace_hardirqs_on_caller+0x17d/0x199 [<ffffffff811b14d5>] ? __fget_light+0x50/0x78 [<ffffffff817d9808>] __sys_sendmsg+0x42/0x60 [<ffffffff817d9838>] SyS_sendmsg+0x12/0x1c [<ffffffff81a50e97>] system_call_fastpath+0x12/0x6f ---[ end trace ef29d3fb28e97ae7 ]--- For long term, we probably need to clean up the qdisc_graft() code in case it hides other bugs like this. Fixes: 95dc19299f74 ("pkt_sched: give visibility to mq slave qdiscs") Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-27 06:08:48 +07:00
skip:
if (!ingress) {
notify_and_destroy(net, skb, n, classid,
dev->qdisc, new);
if (new && !new->ops->attach)
net_sched: fix a refcount_t issue with noop_qdisc syzkaller reported a refcount_t warning [1] Issue here is that noop_qdisc refcnt was never really considered as a true refcount, since qdisc_destroy() found TCQ_F_BUILTIN set : if (qdisc->flags & TCQ_F_BUILTIN || !refcount_dec_and_test(&qdisc->refcnt))) return; Meaning that all atomic_inc() we did on noop_qdisc.refcnt were not really needed, but harmless until refcount_t came. To fix this problem, we simply need to not increment noop_qdisc.refcnt, since we never decrement it. [1] refcount_t: increment on 0; use-after-free. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 21754 at lib/refcount.c:152 refcount_inc+0x47/0x50 lib/refcount.c:152 Kernel panic - not syncing: panic_on_warn set ... CPU: 0 PID: 21754 Comm: syz-executor7 Not tainted 4.13.0-rc6+ #20 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Call Trace: __dump_stack lib/dump_stack.c:16 [inline] dump_stack+0x194/0x257 lib/dump_stack.c:52 panic+0x1e4/0x417 kernel/panic.c:180 __warn+0x1c4/0x1d9 kernel/panic.c:541 report_bug+0x211/0x2d0 lib/bug.c:183 fixup_bug+0x40/0x90 arch/x86/kernel/traps.c:190 do_trap_no_signal arch/x86/kernel/traps.c:224 [inline] do_trap+0x260/0x390 arch/x86/kernel/traps.c:273 do_error_trap+0x120/0x390 arch/x86/kernel/traps.c:310 do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:323 invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:846 RIP: 0010:refcount_inc+0x47/0x50 lib/refcount.c:152 RSP: 0018:ffff8801c43477a0 EFLAGS: 00010282 RAX: 000000000000002b RBX: ffffffff86093c14 RCX: 0000000000000000 RDX: 000000000000002b RSI: ffffffff8159314e RDI: ffffed0038868ee8 RBP: ffff8801c43477a8 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff86093ac0 R13: 0000000000000001 R14: ffff8801d0f3bac0 R15: dffffc0000000000 attach_default_qdiscs net/sched/sch_generic.c:792 [inline] dev_activate+0x7d3/0xaa0 net/sched/sch_generic.c:833 __dev_open+0x227/0x330 net/core/dev.c:1380 __dev_change_flags+0x695/0x990 net/core/dev.c:6726 dev_change_flags+0x88/0x140 net/core/dev.c:6792 dev_ifsioc+0x5a6/0x930 net/core/dev_ioctl.c:256 dev_ioctl+0x2bc/0xf90 net/core/dev_ioctl.c:554 sock_do_ioctl+0x94/0xb0 net/socket.c:968 sock_ioctl+0x2c2/0x440 net/socket.c:1058 vfs_ioctl fs/ioctl.c:45 [inline] do_vfs_ioctl+0x1b1/0x1520 fs/ioctl.c:685 SYSC_ioctl fs/ioctl.c:700 [inline] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691 Fixes: 7b9364050246 ("net, sched: convert Qdisc.refcnt from atomic_t to refcount_t") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Cc: Reshetova, Elena <elena.reshetova@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-08-25 11:12:28 +07:00
qdisc_refcount_inc(new);
dev->qdisc = new ? : &noop_qdisc;
net_sched: invoke ->attach() after setting dev->qdisc For mq qdisc, we add per tx queue qdisc to root qdisc for display purpose, however, that happens too early, before the new dev->qdisc is finally set, this causes q->list points to an old root qdisc which is going to be freed right before assigning with a new one. Fix this by moving ->attach() after setting dev->qdisc. For the record, this fixes the following crash: ------------[ cut here ]------------ WARNING: CPU: 1 PID: 975 at lib/list_debug.c:59 __list_del_entry+0x5a/0x98() list_del corruption. prev->next should be ffff8800d1998ae8, but was 6b6b6b6b6b6b6b6b CPU: 1 PID: 975 Comm: tc Not tainted 4.1.0-rc4+ #1019 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 0000000000000009 ffff8800d73fb928 ffffffff81a44e7f 0000000047574756 ffff8800d73fb978 ffff8800d73fb968 ffffffff810790da ffff8800cfc4cd20 ffffffff814e725b ffff8800d1998ae8 ffffffff82381250 0000000000000000 Call Trace: [<ffffffff81a44e7f>] dump_stack+0x4c/0x65 [<ffffffff810790da>] warn_slowpath_common+0x9c/0xb6 [<ffffffff814e725b>] ? __list_del_entry+0x5a/0x98 [<ffffffff81079162>] warn_slowpath_fmt+0x46/0x48 [<ffffffff81820eb0>] ? dev_graft_qdisc+0x5e/0x6a [<ffffffff814e725b>] __list_del_entry+0x5a/0x98 [<ffffffff814e72a7>] list_del+0xe/0x2d [<ffffffff81822f05>] qdisc_list_del+0x1e/0x20 [<ffffffff81820cd1>] qdisc_destroy+0x30/0xd6 [<ffffffff81822676>] qdisc_graft+0x11d/0x243 [<ffffffff818233c1>] tc_get_qdisc+0x1a6/0x1d4 [<ffffffff810b5eaf>] ? mark_lock+0x2e/0x226 [<ffffffff817ff8f5>] rtnetlink_rcv_msg+0x181/0x194 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff72e>] ? rtnl_lock+0x17/0x19 [<ffffffff817ff774>] ? __rtnl_unlock+0x17/0x17 [<ffffffff81855dc6>] netlink_rcv_skb+0x4d/0x93 [<ffffffff817ff756>] rtnetlink_rcv+0x26/0x2d [<ffffffff818544b2>] netlink_unicast+0xcb/0x150 [<ffffffff81161db9>] ? might_fault+0x59/0xa9 [<ffffffff81854f78>] netlink_sendmsg+0x4fa/0x51c [<ffffffff817d6e09>] sock_sendmsg_nosec+0x12/0x1d [<ffffffff817d8967>] sock_sendmsg+0x29/0x2e [<ffffffff817d8cf3>] ___sys_sendmsg+0x1b4/0x23a [<ffffffff8100a1b8>] ? native_sched_clock+0x35/0x37 [<ffffffff810a1d83>] ? sched_clock_local+0x12/0x72 [<ffffffff810a1fd4>] ? sched_clock_cpu+0x9e/0xb7 [<ffffffff810def2a>] ? current_kernel_time+0xe/0x32 [<ffffffff810b4bc5>] ? lock_release_holdtime.part.29+0x71/0x7f [<ffffffff810ddebf>] ? read_seqcount_begin.constprop.27+0x5f/0x76 [<ffffffff810b6292>] ? trace_hardirqs_on_caller+0x17d/0x199 [<ffffffff811b14d5>] ? __fget_light+0x50/0x78 [<ffffffff817d9808>] __sys_sendmsg+0x42/0x60 [<ffffffff817d9838>] SyS_sendmsg+0x12/0x1c [<ffffffff81a50e97>] system_call_fastpath+0x12/0x6f ---[ end trace ef29d3fb28e97ae7 ]--- For long term, we probably need to clean up the qdisc_graft() code in case it hides other bugs like this. Fixes: 95dc19299f74 ("pkt_sched: give visibility to mq slave qdiscs") Cc: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Acked-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-27 06:08:48 +07:00
if (new && new->ops->attach)
new->ops->attach(new);
} else {
notify_and_destroy(net, skb, n, classid, old, new);
}
if (dev->flags & IFF_UP)
dev_activate(dev);
} else {
const struct Qdisc_class_ops *cops = parent->ops->cl_ops;
err = -EOPNOTSUPP;
if (cops && cops->graft) {
unsigned long cl = cops->find(parent, classid);
if (cl)
err = cops->graft(parent, cl, new, &old);
else
err = -ENOENT;
}
if (!err)
notify_and_destroy(net, skb, n, classid, old, new);
}
return err;
}
/* lockdep annotation is needed for ingress; egress gets it only for name */
static struct lock_class_key qdisc_tx_lock;
static struct lock_class_key qdisc_rx_lock;
/*
Allocate and initialize new qdisc.
Parameters are passed via opt.
*/
static struct Qdisc *qdisc_create(struct net_device *dev,
struct netdev_queue *dev_queue,
struct Qdisc *p, u32 parent, u32 handle,
struct nlattr **tca, int *errp)
{
int err;
struct nlattr *kind = tca[TCA_KIND];
struct Qdisc *sch;
struct Qdisc_ops *ops;
struct qdisc_size_table *stab;
ops = qdisc_lookup_ops(kind);
#ifdef CONFIG_MODULES
if (ops == NULL && kind != NULL) {
char name[IFNAMSIZ];
if (nla_strlcpy(name, kind, IFNAMSIZ) < IFNAMSIZ) {
/* We dropped the RTNL semaphore in order to
* perform the module load. So, even if we
* succeeded in loading the module we have to
* tell the caller to replay the request. We
* indicate this using -EAGAIN.
* We replay the request because the device may
* go away in the mean time.
*/
rtnl_unlock();
request_module("sch_%s", name);
rtnl_lock();
ops = qdisc_lookup_ops(kind);
if (ops != NULL) {
/* We will try again qdisc_lookup_ops,
* so don't keep a reference.
*/
module_put(ops->owner);
err = -EAGAIN;
goto err_out;
}
}
}
#endif
err = -ENOENT;
if (ops == NULL)
goto err_out;
sch = qdisc_alloc(dev_queue, ops);
if (IS_ERR(sch)) {
err = PTR_ERR(sch);
goto err_out2;
}
sch->parent = parent;
if (handle == TC_H_INGRESS) {
sch->flags |= TCQ_F_INGRESS;
handle = TC_H_MAKE(TC_H_INGRESS, 0);
lockdep_set_class(qdisc_lock(sch), &qdisc_rx_lock);
} else {
if (handle == 0) {
handle = qdisc_alloc_handle(dev);
err = -ENOMEM;
if (handle == 0)
goto err_out3;
}
lockdep_set_class(qdisc_lock(sch), &qdisc_tx_lock);
if (!netif_is_multiqueue(dev))
sch->flags |= TCQ_F_ONETXQUEUE;
}
sch->handle = handle;
/* This exist to keep backward compatible with a userspace
* loophole, what allowed userspace to get IFF_NO_QUEUE
* facility on older kernels by setting tx_queue_len=0 (prior
* to qdisc init), and then forgot to reinit tx_queue_len
* before again attaching a qdisc.
*/
if ((dev->priv_flags & IFF_NO_QUEUE) && (dev->tx_queue_len == 0)) {
dev->tx_queue_len = DEFAULT_TX_QUEUE_LEN;
netdev_info(dev, "Caught tx_queue_len zero misconfig\n");
}
if (!ops->init || (err = ops->init(sch, tca[TCA_OPTIONS])) == 0) {
if (qdisc_is_percpu_stats(sch)) {
sch->cpu_bstats =
netdev_alloc_pcpu_stats(struct gnet_stats_basic_cpu);
if (!sch->cpu_bstats)
goto err_out4;
sch->cpu_qstats = alloc_percpu(struct gnet_stats_queue);
if (!sch->cpu_qstats)
goto err_out4;
}
if (tca[TCA_STAB]) {
stab = qdisc_get_stab(tca[TCA_STAB]);
if (IS_ERR(stab)) {
err = PTR_ERR(stab);
goto err_out4;
}
rcu_assign_pointer(sch->stab, stab);
}
if (tca[TCA_RATE]) {
seqcount_t *running;
err = -EOPNOTSUPP;
if (sch->flags & TCQ_F_MQROOT)
goto err_out4;
if ((sch->parent != TC_H_ROOT) &&
!(sch->flags & TCQ_F_INGRESS) &&
(!p || !(p->flags & TCQ_F_MQROOT)))
running = qdisc_root_sleeping_running(sch);
else
running = &sch->running;
err = gen_new_estimator(&sch->bstats,
sch->cpu_bstats,
&sch->rate_est,
NULL,
running,
tca[TCA_RATE]);
if (err)
goto err_out4;
}
qdisc_hash_add(sch, false);
return sch;
}
net_sched: fix error recovery at qdisc creation Dmitry reported uses after free in qdisc code [1] The problem here is that ops->init() can return an error. qdisc_create_dflt() then call ops->destroy(), while qdisc_create() does _not_ call it. Four qdisc chose to call their own ops->destroy(), assuming their caller would not. This patch makes sure qdisc_create() calls ops->destroy() and fixes the four qdisc to avoid double free. [1] BUG: KASAN: use-after-free in mq_destroy+0x242/0x290 net/sched/sch_mq.c:33 at addr ffff8801d415d440 Read of size 8 by task syz-executor2/5030 CPU: 0 PID: 5030 Comm: syz-executor2 Not tainted 4.3.5-smp-DEV #119 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 0000000000000046 ffff8801b435b870 ffffffff81bbbed4 ffff8801db000400 ffff8801d415d440 ffff8801d415dc40 ffff8801c4988510 ffff8801b435b898 ffffffff816682b1 ffff8801b435b928 ffff8801d415d440 ffff8801c49880c0 Call Trace: [<ffffffff81bbbed4>] __dump_stack lib/dump_stack.c:15 [inline] [<ffffffff81bbbed4>] dump_stack+0x6c/0x98 lib/dump_stack.c:51 [<ffffffff816682b1>] kasan_object_err+0x21/0x70 mm/kasan/report.c:158 [<ffffffff81668524>] print_address_description mm/kasan/report.c:196 [inline] [<ffffffff81668524>] kasan_report_error+0x1b4/0x4b0 mm/kasan/report.c:285 [<ffffffff81668953>] kasan_report mm/kasan/report.c:305 [inline] [<ffffffff81668953>] __asan_report_load8_noabort+0x43/0x50 mm/kasan/report.c:326 [<ffffffff82527b02>] mq_destroy+0x242/0x290 net/sched/sch_mq.c:33 [<ffffffff82524bdd>] qdisc_destroy+0x12d/0x290 net/sched/sch_generic.c:953 [<ffffffff82524e30>] qdisc_create_dflt+0xf0/0x120 net/sched/sch_generic.c:848 [<ffffffff8252550d>] attach_default_qdiscs net/sched/sch_generic.c:1029 [inline] [<ffffffff8252550d>] dev_activate+0x6ad/0x880 net/sched/sch_generic.c:1064 [<ffffffff824b1db1>] __dev_open+0x221/0x320 net/core/dev.c:1403 [<ffffffff824b24ce>] __dev_change_flags+0x15e/0x3e0 net/core/dev.c:6858 [<ffffffff824b27de>] dev_change_flags+0x8e/0x140 net/core/dev.c:6926 [<ffffffff824f5bf6>] dev_ifsioc+0x446/0x890 net/core/dev_ioctl.c:260 [<ffffffff824f61fa>] dev_ioctl+0x1ba/0xb80 net/core/dev_ioctl.c:546 [<ffffffff82430509>] sock_do_ioctl+0x99/0xb0 net/socket.c:879 [<ffffffff82430d30>] sock_ioctl+0x2a0/0x390 net/socket.c:958 [<ffffffff816f3b68>] vfs_ioctl fs/ioctl.c:44 [inline] [<ffffffff816f3b68>] do_vfs_ioctl+0x8a8/0xe50 fs/ioctl.c:611 [<ffffffff816f41a4>] SYSC_ioctl fs/ioctl.c:626 [inline] [<ffffffff816f41a4>] SyS_ioctl+0x94/0xc0 fs/ioctl.c:617 [<ffffffff8123e357>] entry_SYSCALL_64_fastpath+0x12/0x17 Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-02-11 01:31:49 +07:00
/* ops->init() failed, we call ->destroy() like qdisc_create_dflt() */
if (ops->destroy)
ops->destroy(sch);
err_out3:
dev_put(dev);
kfree((char *) sch - sch->padded);
err_out2:
module_put(ops->owner);
err_out:
*errp = err;
return NULL;
err_out4:
free_percpu(sch->cpu_bstats);
free_percpu(sch->cpu_qstats);
/*
* Any broken qdiscs that would require a ops->reset() here?
* The qdisc was never in action so it shouldn't be necessary.
*/
qdisc_put_stab(rtnl_dereference(sch->stab));
if (ops->destroy)
ops->destroy(sch);
goto err_out3;
}
static int qdisc_change(struct Qdisc *sch, struct nlattr **tca)
{
struct qdisc_size_table *ostab, *stab = NULL;
int err = 0;
if (tca[TCA_OPTIONS]) {
if (sch->ops->change == NULL)
return -EINVAL;
err = sch->ops->change(sch, tca[TCA_OPTIONS]);
if (err)
return err;
}
if (tca[TCA_STAB]) {
stab = qdisc_get_stab(tca[TCA_STAB]);
if (IS_ERR(stab))
return PTR_ERR(stab);
}
ostab = rtnl_dereference(sch->stab);
rcu_assign_pointer(sch->stab, stab);
qdisc_put_stab(ostab);
if (tca[TCA_RATE]) {
/* NB: ignores errors from replace_estimator
because change can't be undone. */
if (sch->flags & TCQ_F_MQROOT)
goto out;
gen_replace_estimator(&sch->bstats,
sch->cpu_bstats,
&sch->rate_est,
NULL,
qdisc_root_sleeping_running(sch),
tca[TCA_RATE]);
}
out:
return 0;
}
struct check_loop_arg {
struct qdisc_walker w;
struct Qdisc *p;
int depth;
};
static int check_loop_fn(struct Qdisc *q, unsigned long cl,
struct qdisc_walker *w);
static int check_loop(struct Qdisc *q, struct Qdisc *p, int depth)
{
struct check_loop_arg arg;
if (q->ops->cl_ops == NULL)
return 0;
arg.w.stop = arg.w.skip = arg.w.count = 0;
arg.w.fn = check_loop_fn;
arg.depth = depth;
arg.p = p;
q->ops->cl_ops->walk(q, &arg.w);
return arg.w.stop ? -ELOOP : 0;
}
static int
check_loop_fn(struct Qdisc *q, unsigned long cl, struct qdisc_walker *w)
{
struct Qdisc *leaf;
const struct Qdisc_class_ops *cops = q->ops->cl_ops;
struct check_loop_arg *arg = (struct check_loop_arg *)w;
leaf = cops->leaf(q, cl);
if (leaf) {
if (leaf == arg->p || arg->depth > 7)
return -ELOOP;
return check_loop(leaf, arg->p, arg->depth + 1);
}
return 0;
}
/*
* Delete/get qdisc.
*/
static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n,
struct netlink_ext_ack *extack)
{
struct net *net = sock_net(skb->sk);
struct tcmsg *tcm = nlmsg_data(n);
struct nlattr *tca[TCA_MAX + 1];
struct net_device *dev;
u32 clid;
struct Qdisc *q = NULL;
struct Qdisc *p = NULL;
int err;
if ((n->nlmsg_type != RTM_GETQDISC) &&
!netlink_ns_capable(skb, net->user_ns, CAP_NET_ADMIN))
return -EPERM;
err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL, extack);
if (err < 0)
return err;
dev = __dev_get_by_index(net, tcm->tcm_ifindex);
if (!dev)
return -ENODEV;
clid = tcm->tcm_parent;
if (clid) {
if (clid != TC_H_ROOT) {
if (TC_H_MAJ(clid) != TC_H_MAJ(TC_H_INGRESS)) {
p = qdisc_lookup(dev, TC_H_MAJ(clid));
if (!p)
return -ENOENT;
q = qdisc_leaf(p, clid);
} else if (dev_ingress_queue(dev)) {
q = dev_ingress_queue(dev)->qdisc_sleeping;
}
} else {
q = dev->qdisc;
}
if (!q)
return -ENOENT;
if (tcm->tcm_handle && q->handle != tcm->tcm_handle)
return -EINVAL;
} else {
q = qdisc_lookup(dev, tcm->tcm_handle);
if (!q)
return -ENOENT;
}
if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
return -EINVAL;
if (n->nlmsg_type == RTM_DELQDISC) {
if (!clid)
return -EINVAL;
if (q->handle == 0)
return -ENOENT;
err = qdisc_graft(dev, p, skb, n, clid, NULL, q);
if (err != 0)
return err;
} else {
qdisc_notify(net, skb, n, clid, NULL, q);
}
return 0;
}
/*
* Create/change qdisc.
*/
static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n,
struct netlink_ext_ack *extack)
{
struct net *net = sock_net(skb->sk);
struct tcmsg *tcm;
struct nlattr *tca[TCA_MAX + 1];
struct net_device *dev;
u32 clid;
struct Qdisc *q, *p;
int err;
if (!netlink_ns_capable(skb, net->user_ns, CAP_NET_ADMIN))
return -EPERM;
replay:
/* Reinit, just in case something touches this. */
err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL, extack);
if (err < 0)
return err;
tcm = nlmsg_data(n);
clid = tcm->tcm_parent;
q = p = NULL;
dev = __dev_get_by_index(net, tcm->tcm_ifindex);
if (!dev)
return -ENODEV;
if (clid) {
if (clid != TC_H_ROOT) {
if (clid != TC_H_INGRESS) {
p = qdisc_lookup(dev, TC_H_MAJ(clid));
if (!p)
return -ENOENT;
q = qdisc_leaf(p, clid);
} else if (dev_ingress_queue_create(dev)) {
q = dev_ingress_queue(dev)->qdisc_sleeping;
}
} else {
q = dev->qdisc;
}
/* It may be default qdisc, ignore it */
if (q && q->handle == 0)
q = NULL;
if (!q || !tcm->tcm_handle || q->handle != tcm->tcm_handle) {
if (tcm->tcm_handle) {
if (q && !(n->nlmsg_flags & NLM_F_REPLACE))
return -EEXIST;
if (TC_H_MIN(tcm->tcm_handle))
return -EINVAL;
q = qdisc_lookup(dev, tcm->tcm_handle);
if (!q)
goto create_n_graft;
if (n->nlmsg_flags & NLM_F_EXCL)
return -EEXIST;
if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
return -EINVAL;
if (q == p ||
(p && check_loop(q, p, 0)))
return -ELOOP;
net_sched: fix a refcount_t issue with noop_qdisc syzkaller reported a refcount_t warning [1] Issue here is that noop_qdisc refcnt was never really considered as a true refcount, since qdisc_destroy() found TCQ_F_BUILTIN set : if (qdisc->flags & TCQ_F_BUILTIN || !refcount_dec_and_test(&qdisc->refcnt))) return; Meaning that all atomic_inc() we did on noop_qdisc.refcnt were not really needed, but harmless until refcount_t came. To fix this problem, we simply need to not increment noop_qdisc.refcnt, since we never decrement it. [1] refcount_t: increment on 0; use-after-free. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 21754 at lib/refcount.c:152 refcount_inc+0x47/0x50 lib/refcount.c:152 Kernel panic - not syncing: panic_on_warn set ... CPU: 0 PID: 21754 Comm: syz-executor7 Not tainted 4.13.0-rc6+ #20 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Call Trace: __dump_stack lib/dump_stack.c:16 [inline] dump_stack+0x194/0x257 lib/dump_stack.c:52 panic+0x1e4/0x417 kernel/panic.c:180 __warn+0x1c4/0x1d9 kernel/panic.c:541 report_bug+0x211/0x2d0 lib/bug.c:183 fixup_bug+0x40/0x90 arch/x86/kernel/traps.c:190 do_trap_no_signal arch/x86/kernel/traps.c:224 [inline] do_trap+0x260/0x390 arch/x86/kernel/traps.c:273 do_error_trap+0x120/0x390 arch/x86/kernel/traps.c:310 do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:323 invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:846 RIP: 0010:refcount_inc+0x47/0x50 lib/refcount.c:152 RSP: 0018:ffff8801c43477a0 EFLAGS: 00010282 RAX: 000000000000002b RBX: ffffffff86093c14 RCX: 0000000000000000 RDX: 000000000000002b RSI: ffffffff8159314e RDI: ffffed0038868ee8 RBP: ffff8801c43477a8 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff86093ac0 R13: 0000000000000001 R14: ffff8801d0f3bac0 R15: dffffc0000000000 attach_default_qdiscs net/sched/sch_generic.c:792 [inline] dev_activate+0x7d3/0xaa0 net/sched/sch_generic.c:833 __dev_open+0x227/0x330 net/core/dev.c:1380 __dev_change_flags+0x695/0x990 net/core/dev.c:6726 dev_change_flags+0x88/0x140 net/core/dev.c:6792 dev_ifsioc+0x5a6/0x930 net/core/dev_ioctl.c:256 dev_ioctl+0x2bc/0xf90 net/core/dev_ioctl.c:554 sock_do_ioctl+0x94/0xb0 net/socket.c:968 sock_ioctl+0x2c2/0x440 net/socket.c:1058 vfs_ioctl fs/ioctl.c:45 [inline] do_vfs_ioctl+0x1b1/0x1520 fs/ioctl.c:685 SYSC_ioctl fs/ioctl.c:700 [inline] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691 Fixes: 7b9364050246 ("net, sched: convert Qdisc.refcnt from atomic_t to refcount_t") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Cc: Reshetova, Elena <elena.reshetova@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-08-25 11:12:28 +07:00
qdisc_refcount_inc(q);
goto graft;
} else {
if (!q)
goto create_n_graft;
/* This magic test requires explanation.
*
* We know, that some child q is already
* attached to this parent and have choice:
* either to change it or to create/graft new one.
*
* 1. We are allowed to create/graft only
* if CREATE and REPLACE flags are set.
*
* 2. If EXCL is set, requestor wanted to say,
* that qdisc tcm_handle is not expected
* to exist, so that we choose create/graft too.
*
* 3. The last case is when no flags are set.
* Alas, it is sort of hole in API, we
* cannot decide what to do unambiguously.
* For now we select create/graft, if
* user gave KIND, which does not match existing.
*/
if ((n->nlmsg_flags & NLM_F_CREATE) &&
(n->nlmsg_flags & NLM_F_REPLACE) &&
((n->nlmsg_flags & NLM_F_EXCL) ||
(tca[TCA_KIND] &&
nla_strcmp(tca[TCA_KIND], q->ops->id))))
goto create_n_graft;
}
}
} else {
if (!tcm->tcm_handle)
return -EINVAL;
q = qdisc_lookup(dev, tcm->tcm_handle);
}
/* Change qdisc parameters */
if (q == NULL)
return -ENOENT;
if (n->nlmsg_flags & NLM_F_EXCL)
return -EEXIST;
if (tca[TCA_KIND] && nla_strcmp(tca[TCA_KIND], q->ops->id))
return -EINVAL;
err = qdisc_change(q, tca);
if (err == 0)
qdisc_notify(net, skb, n, clid, NULL, q);
return err;
create_n_graft:
if (!(n->nlmsg_flags & NLM_F_CREATE))
return -ENOENT;
if (clid == TC_H_INGRESS) {
if (dev_ingress_queue(dev))
q = qdisc_create(dev, dev_ingress_queue(dev), p,
tcm->tcm_parent, tcm->tcm_parent,
tca, &err);
else
err = -ENOENT;
} else {
struct netdev_queue *dev_queue;
if (p && p->ops->cl_ops && p->ops->cl_ops->select_queue)
dev_queue = p->ops->cl_ops->select_queue(p, tcm);
else if (p)
dev_queue = p->dev_queue;
else
dev_queue = netdev_get_tx_queue(dev, 0);
q = qdisc_create(dev, dev_queue, p,
tcm->tcm_parent, tcm->tcm_handle,
tca, &err);
}
if (q == NULL) {
if (err == -EAGAIN)
goto replay;
return err;
}
graft:
err = qdisc_graft(dev, p, skb, n, clid, q, NULL);
if (err) {
if (q)
qdisc_destroy(q);
return err;
}
return 0;
}
static int tc_dump_qdisc_root(struct Qdisc *root, struct sk_buff *skb,
struct netlink_callback *cb,
int *q_idx_p, int s_q_idx, bool recur,
bool dump_invisible)
{
int ret = 0, q_idx = *q_idx_p;
struct Qdisc *q;
int b;
if (!root)
return 0;
q = root;
if (q_idx < s_q_idx) {
q_idx++;
} else {
if (!tc_qdisc_dump_ignore(q, dump_invisible) &&
tc_fill_qdisc(skb, q, q->parent, NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq, NLM_F_MULTI,
RTM_NEWQDISC) <= 0)
goto done;
q_idx++;
}
net: sched: fix handling of singleton qdiscs with qdisc_hash qdisc_match_from_root() is now iterating over per-netdevice qdisc hashtable instead of going through a linked-list of qdiscs (independently on the actual underlying netdev), which was the case before the switch to hashtable for qdiscs. For singleton qdiscs, there is no underlying netdev associated though, and therefore dumping a singleton qdisc will panic, as qdisc_dev(root) will always be NULL. BUG: unable to handle kernel NULL pointer dereference at 0000000000000410 IP: [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 PGD 1aceba067 PUD 1aceb7067 PMD 0 Oops: 0000 [#1] PREEMPT SMP [ ... ] task: ffff8801ec996e00 task.stack: ffff8801ec934000 RIP: 0010:[<ffffffff8167efac>] [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 RSP: 0018:ffff8801ec937ab0 EFLAGS: 00010203 RAX: 0000000000000408 RBX: ffff88025e612000 RCX: ffffffffffffffd8 RDX: 0000000000000000 RSI: 00000000ffff0000 RDI: ffffffff81cf8100 RBP: ffff8801ec937ab0 R08: 000000000001c160 R09: ffff8802668032c0 R10: ffffffff81cf8100 R11: 0000000000000030 R12: 00000000ffff0000 R13: ffff88025e612000 R14: ffffffff81cf3140 R15: 0000000000000000 FS: 00007f24b9af6740(0000) GS:ffff88026f280000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000410 CR3: 00000001aceec000 CR4: 00000000001406e0 Stack: ffff8801ec937ad0 ffffffff81681210 ffff88025dd51a00 00000000fffffff1 ffff8801ec937b88 ffffffff81681e4e ffffffff81c42bc0 ffff880262431500 ffffffff81cf3140 ffff88025dd51a10 ffff88025dd51a24 00000000ec937b38 Call Trace: [<ffffffff81681210>] qdisc_lookup+0x40/0x50 [<ffffffff81681e4e>] tc_modify_qdisc+0x21e/0x550 [<ffffffff8166ae25>] rtnetlink_rcv_msg+0x95/0x220 [<ffffffff81209602>] ? __kmalloc_track_caller+0x172/0x230 [<ffffffff8166ad90>] ? rtnl_newlink+0x870/0x870 [<ffffffff816897b7>] netlink_rcv_skb+0xa7/0xc0 [<ffffffff816657c8>] rtnetlink_rcv+0x28/0x30 [<ffffffff8168919b>] netlink_unicast+0x15b/0x210 [<ffffffff81689569>] netlink_sendmsg+0x319/0x390 [<ffffffff816379f8>] sock_sendmsg+0x38/0x50 [<ffffffff81638296>] ___sys_sendmsg+0x256/0x260 [<ffffffff811b1275>] ? __pagevec_lru_add_fn+0x135/0x280 [<ffffffff811b1a90>] ? pagevec_lru_move_fn+0xd0/0xf0 [<ffffffff811b1140>] ? trace_event_raw_event_mm_lru_insertion+0x180/0x180 [<ffffffff811b1b85>] ? __lru_cache_add+0x75/0xb0 [<ffffffff817708a6>] ? _raw_spin_unlock+0x16/0x40 [<ffffffff811d8dff>] ? handle_mm_fault+0x39f/0x1160 [<ffffffff81638b15>] __sys_sendmsg+0x45/0x80 [<ffffffff81638b62>] SyS_sendmsg+0x12/0x20 [<ffffffff810038e7>] do_syscall_64+0x57/0xb0 Fix this by special-casing singleton qdiscs (those that don't have underlying netdevice) and introduce immediate handling of those rather than trying to go over an underlying netdevice. We're in the same situation in tc_dump_qdisc_root() and tc_dump_tclass_root(). Ultimately, this will have to be slightly reworked so that we are actually able to show singleton qdiscs (noop) in the dump properly; but we're not currently doing that anyway, so no regression there, and better do this in a gradual manner. Fixes: 59cc1f61f ("net: sched: convert qdisc linked list to hashtable") Reported-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Reported-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-08-17 04:52:58 +07:00
net: sched: avoid duplicates in qdisc dump tc_dump_qdisc() performs dumping of the per-device qdiscs in two phases; first, the "standard" dev->qdisc is being dumped. Second, if there is/are ingress queue(s), they are being dumped as well. After conversion of netdevice's qdisc linked-list into hashtable, these two sets are not in two disjunctive sets/lists any more, but are both "reachable" directly from netdevice's hashtable. As a consequence, the "full-depth" dump of the ingress qdiscs results in immediately hitting the netdevice hashtable again, and duplicating the dump that has already been performed for dev->qdisc. What in fact needs to be dumped in case of ingress queue is "just" the top-level ingress qdisc, as everything else has been dumped already. Fix this by extending tc_dump_qdisc_root() in a way that it can be instructed whether it should (while performing the "full" per-netdev qdisc dump) perform the whole recursion, or just dump "additional" top-level (ingress) qdiscs without performing any kind of recursion. This fixes duplicate dumps such as qdisc mq 0: root qdisc pfifo_fast 0: parent :4 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :3 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :1 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc clsact ffff: parent ffff:fff1 qdisc pfifo_fast 0: parent :4 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :3 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: parent :1 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 Fixes: 59cc1f61f ("net: sched: convert qdisc linked list to hashtable") Reported-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-08-17 04:53:46 +07:00
/* If dumping singletons, there is no qdisc_dev(root) and the singleton
* itself has already been dumped.
*
* If we've already dumped the top-level (ingress) qdisc above and the global
* qdisc hashtable, we don't want to hit it again
*/
if (!qdisc_dev(root) || !recur)
net: sched: fix handling of singleton qdiscs with qdisc_hash qdisc_match_from_root() is now iterating over per-netdevice qdisc hashtable instead of going through a linked-list of qdiscs (independently on the actual underlying netdev), which was the case before the switch to hashtable for qdiscs. For singleton qdiscs, there is no underlying netdev associated though, and therefore dumping a singleton qdisc will panic, as qdisc_dev(root) will always be NULL. BUG: unable to handle kernel NULL pointer dereference at 0000000000000410 IP: [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 PGD 1aceba067 PUD 1aceb7067 PMD 0 Oops: 0000 [#1] PREEMPT SMP [ ... ] task: ffff8801ec996e00 task.stack: ffff8801ec934000 RIP: 0010:[<ffffffff8167efac>] [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 RSP: 0018:ffff8801ec937ab0 EFLAGS: 00010203 RAX: 0000000000000408 RBX: ffff88025e612000 RCX: ffffffffffffffd8 RDX: 0000000000000000 RSI: 00000000ffff0000 RDI: ffffffff81cf8100 RBP: ffff8801ec937ab0 R08: 000000000001c160 R09: ffff8802668032c0 R10: ffffffff81cf8100 R11: 0000000000000030 R12: 00000000ffff0000 R13: ffff88025e612000 R14: ffffffff81cf3140 R15: 0000000000000000 FS: 00007f24b9af6740(0000) GS:ffff88026f280000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000410 CR3: 00000001aceec000 CR4: 00000000001406e0 Stack: ffff8801ec937ad0 ffffffff81681210 ffff88025dd51a00 00000000fffffff1 ffff8801ec937b88 ffffffff81681e4e ffffffff81c42bc0 ffff880262431500 ffffffff81cf3140 ffff88025dd51a10 ffff88025dd51a24 00000000ec937b38 Call Trace: [<ffffffff81681210>] qdisc_lookup+0x40/0x50 [<ffffffff81681e4e>] tc_modify_qdisc+0x21e/0x550 [<ffffffff8166ae25>] rtnetlink_rcv_msg+0x95/0x220 [<ffffffff81209602>] ? __kmalloc_track_caller+0x172/0x230 [<ffffffff8166ad90>] ? rtnl_newlink+0x870/0x870 [<ffffffff816897b7>] netlink_rcv_skb+0xa7/0xc0 [<ffffffff816657c8>] rtnetlink_rcv+0x28/0x30 [<ffffffff8168919b>] netlink_unicast+0x15b/0x210 [<ffffffff81689569>] netlink_sendmsg+0x319/0x390 [<ffffffff816379f8>] sock_sendmsg+0x38/0x50 [<ffffffff81638296>] ___sys_sendmsg+0x256/0x260 [<ffffffff811b1275>] ? __pagevec_lru_add_fn+0x135/0x280 [<ffffffff811b1a90>] ? pagevec_lru_move_fn+0xd0/0xf0 [<ffffffff811b1140>] ? trace_event_raw_event_mm_lru_insertion+0x180/0x180 [<ffffffff811b1b85>] ? __lru_cache_add+0x75/0xb0 [<ffffffff817708a6>] ? _raw_spin_unlock+0x16/0x40 [<ffffffff811d8dff>] ? handle_mm_fault+0x39f/0x1160 [<ffffffff81638b15>] __sys_sendmsg+0x45/0x80 [<ffffffff81638b62>] SyS_sendmsg+0x12/0x20 [<ffffffff810038e7>] do_syscall_64+0x57/0xb0 Fix this by special-casing singleton qdiscs (those that don't have underlying netdevice) and introduce immediate handling of those rather than trying to go over an underlying netdevice. We're in the same situation in tc_dump_qdisc_root() and tc_dump_tclass_root(). Ultimately, this will have to be slightly reworked so that we are actually able to show singleton qdiscs (noop) in the dump properly; but we're not currently doing that anyway, so no regression there, and better do this in a gradual manner. Fixes: 59cc1f61f ("net: sched: convert qdisc linked list to hashtable") Reported-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Reported-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-08-17 04:52:58 +07:00
goto out;
hash_for_each(qdisc_dev(root)->qdisc_hash, b, q, hash) {
if (q_idx < s_q_idx) {
q_idx++;
continue;
}
if (!tc_qdisc_dump_ignore(q, dump_invisible) &&
tc_fill_qdisc(skb, q, q->parent, NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq, NLM_F_MULTI,
RTM_NEWQDISC) <= 0)
goto done;
q_idx++;
}
out:
*q_idx_p = q_idx;
return ret;
done:
ret = -1;
goto out;
}
static int tc_dump_qdisc(struct sk_buff *skb, struct netlink_callback *cb)
{
struct net *net = sock_net(skb->sk);
int idx, q_idx;
int s_idx, s_q_idx;
struct net_device *dev;
const struct nlmsghdr *nlh = cb->nlh;
struct nlattr *tca[TCA_MAX + 1];
int err;
s_idx = cb->args[0];
s_q_idx = q_idx = cb->args[1];
idx = 0;
ASSERT_RTNL();
err = nlmsg_parse(nlh, sizeof(struct tcmsg), tca, TCA_MAX, NULL, NULL);
if (err < 0)
return err;
for_each_netdev(net, dev) {
struct netdev_queue *dev_queue;
if (idx < s_idx)
goto cont;
if (idx > s_idx)
s_q_idx = 0;
q_idx = 0;
if (tc_dump_qdisc_root(dev->qdisc, skb, cb, &q_idx, s_q_idx,
true, tca[TCA_DUMP_INVISIBLE]) < 0)
goto done;
dev_queue = dev_ingress_queue(dev);
if (dev_queue &&
tc_dump_qdisc_root(dev_queue->qdisc_sleeping, skb, cb,
&q_idx, s_q_idx, false,
tca[TCA_DUMP_INVISIBLE]) < 0)
goto done;
cont:
idx++;
}
done:
cb->args[0] = idx;
cb->args[1] = q_idx;
return skb->len;
}
/************************************************
* Traffic classes manipulation. *
************************************************/
static int tc_fill_tclass(struct sk_buff *skb, struct Qdisc *q,
unsigned long cl,
u32 portid, u32 seq, u16 flags, int event)
{
struct tcmsg *tcm;
struct nlmsghdr *nlh;
unsigned char *b = skb_tail_pointer(skb);
struct gnet_dump d;
const struct Qdisc_class_ops *cl_ops = q->ops->cl_ops;
cond_resched();
nlh = nlmsg_put(skb, portid, seq, event, sizeof(*tcm), flags);
if (!nlh)
goto out_nlmsg_trim;
tcm = nlmsg_data(nlh);
tcm->tcm_family = AF_UNSPEC;
tcm->tcm__pad1 = 0;
tcm->tcm__pad2 = 0;
tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
tcm->tcm_parent = q->handle;
tcm->tcm_handle = q->handle;
tcm->tcm_info = 0;
if (nla_put_string(skb, TCA_KIND, q->ops->id))
goto nla_put_failure;
if (cl_ops->dump && cl_ops->dump(q, cl, skb, tcm) < 0)
goto nla_put_failure;
if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS, TCA_XSTATS,
NULL, &d, TCA_PAD) < 0)
goto nla_put_failure;
if (cl_ops->dump_stats && cl_ops->dump_stats(q, cl, &d) < 0)
goto nla_put_failure;
if (gnet_stats_finish_copy(&d) < 0)
goto nla_put_failure;
nlh->nlmsg_len = skb_tail_pointer(skb) - b;
return skb->len;
out_nlmsg_trim:
nla_put_failure:
nlmsg_trim(skb, b);
return -1;
}
static int tclass_notify(struct net *net, struct sk_buff *oskb,
struct nlmsghdr *n, struct Qdisc *q,
unsigned long cl, int event)
{
struct sk_buff *skb;
u32 portid = oskb ? NETLINK_CB(oskb).portid : 0;
skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
if (!skb)
return -ENOBUFS;
if (tc_fill_tclass(skb, q, cl, portid, n->nlmsg_seq, 0, event) < 0) {
kfree_skb(skb);
return -EINVAL;
}
return rtnetlink_send(skb, net, portid, RTNLGRP_TC,
n->nlmsg_flags & NLM_F_ECHO);
}
static int tclass_del_notify(struct net *net,
const struct Qdisc_class_ops *cops,
struct sk_buff *oskb, struct nlmsghdr *n,
struct Qdisc *q, unsigned long cl)
{
u32 portid = oskb ? NETLINK_CB(oskb).portid : 0;
struct sk_buff *skb;
int err = 0;
if (!cops->delete)
return -EOPNOTSUPP;
skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
if (!skb)
return -ENOBUFS;
if (tc_fill_tclass(skb, q, cl, portid, n->nlmsg_seq, 0,
RTM_DELTCLASS) < 0) {
kfree_skb(skb);
return -EINVAL;
}
err = cops->delete(q, cl);
if (err) {
kfree_skb(skb);
return err;
}
return rtnetlink_send(skb, net, portid, RTNLGRP_TC,
n->nlmsg_flags & NLM_F_ECHO);
}
#ifdef CONFIG_NET_CLS
struct tcf_bind_args {
struct tcf_walker w;
u32 classid;
unsigned long cl;
};
static int tcf_node_bind(struct tcf_proto *tp, void *n, struct tcf_walker *arg)
{
struct tcf_bind_args *a = (void *)arg;
if (tp->ops->bind_class) {
struct Qdisc *q = tcf_block_q(tp->chain->block);
sch_tree_lock(q);
tp->ops->bind_class(n, a->classid, a->cl);
sch_tree_unlock(q);
}
return 0;
}
static void tc_bind_tclass(struct Qdisc *q, u32 portid, u32 clid,
unsigned long new_cl)
{
const struct Qdisc_class_ops *cops = q->ops->cl_ops;
struct tcf_block *block;
struct tcf_chain *chain;
unsigned long cl;
cl = cops->find(q, portid);
if (!cl)
return;
block = cops->tcf_block(q, cl);
if (!block)
return;
list_for_each_entry(chain, &block->chain_list, list) {
struct tcf_proto *tp;
for (tp = rtnl_dereference(chain->filter_chain);
tp; tp = rtnl_dereference(tp->next)) {
struct tcf_bind_args arg = {};
arg.w.fn = tcf_node_bind;
arg.classid = clid;
arg.cl = new_cl;
tp->ops->walk(tp, &arg.w);
}
}
}
#else
static void tc_bind_tclass(struct Qdisc *q, u32 portid, u32 clid,
unsigned long new_cl)
{
}
#endif
static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n,
struct netlink_ext_ack *extack)
{
struct net *net = sock_net(skb->sk);
struct tcmsg *tcm = nlmsg_data(n);
struct nlattr *tca[TCA_MAX + 1];
struct net_device *dev;
struct Qdisc *q = NULL;
const struct Qdisc_class_ops *cops;
unsigned long cl = 0;
unsigned long new_cl;
u32 portid;
u32 clid;
u32 qid;
int err;
if ((n->nlmsg_type != RTM_GETTCLASS) &&
!netlink_ns_capable(skb, net->user_ns, CAP_NET_ADMIN))
return -EPERM;
err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL, extack);
if (err < 0)
return err;
dev = __dev_get_by_index(net, tcm->tcm_ifindex);
if (!dev)
return -ENODEV;
/*
parent == TC_H_UNSPEC - unspecified parent.
parent == TC_H_ROOT - class is root, which has no parent.
parent == X:0 - parent is root class.
parent == X:Y - parent is a node in hierarchy.
parent == 0:Y - parent is X:Y, where X:0 is qdisc.
handle == 0:0 - generate handle from kernel pool.
handle == 0:Y - class is X:Y, where X:0 is qdisc.
handle == X:Y - clear.
handle == X:0 - root class.
*/
/* Step 1. Determine qdisc handle X:0 */
portid = tcm->tcm_parent;
clid = tcm->tcm_handle;
qid = TC_H_MAJ(clid);
if (portid != TC_H_ROOT) {
u32 qid1 = TC_H_MAJ(portid);
if (qid && qid1) {
/* If both majors are known, they must be identical. */
if (qid != qid1)
return -EINVAL;
} else if (qid1) {
qid = qid1;
} else if (qid == 0)
qid = dev->qdisc->handle;
/* Now qid is genuine qdisc handle consistent
* both with parent and child.
*
* TC_H_MAJ(portid) still may be unspecified, complete it now.
*/
if (portid)
portid = TC_H_MAKE(qid, portid);
} else {
if (qid == 0)
qid = dev->qdisc->handle;
}
/* OK. Locate qdisc */
q = qdisc_lookup(dev, qid);
if (!q)
return -ENOENT;
/* An check that it supports classes */
cops = q->ops->cl_ops;
if (cops == NULL)
return -EINVAL;
/* Now try to get class */
if (clid == 0) {
if (portid == TC_H_ROOT)
clid = qid;
} else
clid = TC_H_MAKE(qid, clid);
if (clid)
cl = cops->find(q, clid);
if (cl == 0) {
err = -ENOENT;
if (n->nlmsg_type != RTM_NEWTCLASS ||
!(n->nlmsg_flags & NLM_F_CREATE))
goto out;
} else {
switch (n->nlmsg_type) {
case RTM_NEWTCLASS:
err = -EEXIST;
if (n->nlmsg_flags & NLM_F_EXCL)
goto out;
break;
case RTM_DELTCLASS:
err = tclass_del_notify(net, cops, skb, n, q, cl);
/* Unbind the class with flilters with 0 */
tc_bind_tclass(q, portid, clid, 0);
goto out;
case RTM_GETTCLASS:
err = tclass_notify(net, skb, n, q, cl, RTM_NEWTCLASS);
goto out;
default:
err = -EINVAL;
goto out;
}
}
new_cl = cl;
err = -EOPNOTSUPP;
if (cops->change)
err = cops->change(q, clid, portid, tca, &new_cl);
if (err == 0) {
tclass_notify(net, skb, n, q, new_cl, RTM_NEWTCLASS);
/* We just create a new class, need to do reverse binding. */
if (cl != new_cl)
tc_bind_tclass(q, portid, clid, new_cl);
}
out:
return err;
}
struct qdisc_dump_args {
struct qdisc_walker w;
struct sk_buff *skb;
struct netlink_callback *cb;
};
static int qdisc_class_dump(struct Qdisc *q, unsigned long cl,
struct qdisc_walker *arg)
{
struct qdisc_dump_args *a = (struct qdisc_dump_args *)arg;
return tc_fill_tclass(a->skb, q, cl, NETLINK_CB(a->cb->skb).portid,
a->cb->nlh->nlmsg_seq, NLM_F_MULTI,
RTM_NEWTCLASS);
}
static int tc_dump_tclass_qdisc(struct Qdisc *q, struct sk_buff *skb,
struct tcmsg *tcm, struct netlink_callback *cb,
int *t_p, int s_t)
{
struct qdisc_dump_args arg;
if (tc_qdisc_dump_ignore(q, false) ||
*t_p < s_t || !q->ops->cl_ops ||
(tcm->tcm_parent &&
TC_H_MAJ(tcm->tcm_parent) != q->handle)) {
(*t_p)++;
return 0;
}
if (*t_p > s_t)
memset(&cb->args[1], 0, sizeof(cb->args)-sizeof(cb->args[0]));
arg.w.fn = qdisc_class_dump;
arg.skb = skb;
arg.cb = cb;
arg.w.stop = 0;
arg.w.skip = cb->args[1];
arg.w.count = 0;
q->ops->cl_ops->walk(q, &arg.w);
cb->args[1] = arg.w.count;
if (arg.w.stop)
return -1;
(*t_p)++;
return 0;
}
static int tc_dump_tclass_root(struct Qdisc *root, struct sk_buff *skb,
struct tcmsg *tcm, struct netlink_callback *cb,
int *t_p, int s_t)
{
struct Qdisc *q;
int b;
if (!root)
return 0;
if (tc_dump_tclass_qdisc(root, skb, tcm, cb, t_p, s_t) < 0)
return -1;
net: sched: fix handling of singleton qdiscs with qdisc_hash qdisc_match_from_root() is now iterating over per-netdevice qdisc hashtable instead of going through a linked-list of qdiscs (independently on the actual underlying netdev), which was the case before the switch to hashtable for qdiscs. For singleton qdiscs, there is no underlying netdev associated though, and therefore dumping a singleton qdisc will panic, as qdisc_dev(root) will always be NULL. BUG: unable to handle kernel NULL pointer dereference at 0000000000000410 IP: [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 PGD 1aceba067 PUD 1aceb7067 PMD 0 Oops: 0000 [#1] PREEMPT SMP [ ... ] task: ffff8801ec996e00 task.stack: ffff8801ec934000 RIP: 0010:[<ffffffff8167efac>] [<ffffffff8167efac>] qdisc_match_from_root+0x2c/0x70 RSP: 0018:ffff8801ec937ab0 EFLAGS: 00010203 RAX: 0000000000000408 RBX: ffff88025e612000 RCX: ffffffffffffffd8 RDX: 0000000000000000 RSI: 00000000ffff0000 RDI: ffffffff81cf8100 RBP: ffff8801ec937ab0 R08: 000000000001c160 R09: ffff8802668032c0 R10: ffffffff81cf8100 R11: 0000000000000030 R12: 00000000ffff0000 R13: ffff88025e612000 R14: ffffffff81cf3140 R15: 0000000000000000 FS: 00007f24b9af6740(0000) GS:ffff88026f280000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000410 CR3: 00000001aceec000 CR4: 00000000001406e0 Stack: ffff8801ec937ad0 ffffffff81681210 ffff88025dd51a00 00000000fffffff1 ffff8801ec937b88 ffffffff81681e4e ffffffff81c42bc0 ffff880262431500 ffffffff81cf3140 ffff88025dd51a10 ffff88025dd51a24 00000000ec937b38 Call Trace: [<ffffffff81681210>] qdisc_lookup+0x40/0x50 [<ffffffff81681e4e>] tc_modify_qdisc+0x21e/0x550 [<ffffffff8166ae25>] rtnetlink_rcv_msg+0x95/0x220 [<ffffffff81209602>] ? __kmalloc_track_caller+0x172/0x230 [<ffffffff8166ad90>] ? rtnl_newlink+0x870/0x870 [<ffffffff816897b7>] netlink_rcv_skb+0xa7/0xc0 [<ffffffff816657c8>] rtnetlink_rcv+0x28/0x30 [<ffffffff8168919b>] netlink_unicast+0x15b/0x210 [<ffffffff81689569>] netlink_sendmsg+0x319/0x390 [<ffffffff816379f8>] sock_sendmsg+0x38/0x50 [<ffffffff81638296>] ___sys_sendmsg+0x256/0x260 [<ffffffff811b1275>] ? __pagevec_lru_add_fn+0x135/0x280 [<ffffffff811b1a90>] ? pagevec_lru_move_fn+0xd0/0xf0 [<ffffffff811b1140>] ? trace_event_raw_event_mm_lru_insertion+0x180/0x180 [<ffffffff811b1b85>] ? __lru_cache_add+0x75/0xb0 [<ffffffff817708a6>] ? _raw_spin_unlock+0x16/0x40 [<ffffffff811d8dff>] ? handle_mm_fault+0x39f/0x1160 [<ffffffff81638b15>] __sys_sendmsg+0x45/0x80 [<ffffffff81638b62>] SyS_sendmsg+0x12/0x20 [<ffffffff810038e7>] do_syscall_64+0x57/0xb0 Fix this by special-casing singleton qdiscs (those that don't have underlying netdevice) and introduce immediate handling of those rather than trying to go over an underlying netdevice. We're in the same situation in tc_dump_qdisc_root() and tc_dump_tclass_root(). Ultimately, this will have to be slightly reworked so that we are actually able to show singleton qdiscs (noop) in the dump properly; but we're not currently doing that anyway, so no regression there, and better do this in a gradual manner. Fixes: 59cc1f61f ("net: sched: convert qdisc linked list to hashtable") Reported-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Daniel Borkmann <daniel@iogearbox.net> Reported-by: David Ahern <dsa@cumulusnetworks.com> Tested-by: David Ahern <dsa@cumulusnetworks.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: David S. Miller <davem@davemloft.net>
2016-08-17 04:52:58 +07:00
if (!qdisc_dev(root))
return 0;
if (tcm->tcm_parent) {
q = qdisc_match_from_root(root, TC_H_MAJ(tcm->tcm_parent));
if (q && tc_dump_tclass_qdisc(q, skb, tcm, cb, t_p, s_t) < 0)
return -1;
return 0;
}
hash_for_each(qdisc_dev(root)->qdisc_hash, b, q, hash) {
if (tc_dump_tclass_qdisc(q, skb, tcm, cb, t_p, s_t) < 0)
return -1;
}
return 0;
}
static int tc_dump_tclass(struct sk_buff *skb, struct netlink_callback *cb)
{
struct tcmsg *tcm = nlmsg_data(cb->nlh);
struct net *net = sock_net(skb->sk);
struct netdev_queue *dev_queue;
struct net_device *dev;
int t, s_t;
if (nlmsg_len(cb->nlh) < sizeof(*tcm))
return 0;
dev = dev_get_by_index(net, tcm->tcm_ifindex);
if (!dev)
return 0;
s_t = cb->args[0];
t = 0;
if (tc_dump_tclass_root(dev->qdisc, skb, tcm, cb, &t, s_t) < 0)
goto done;
dev_queue = dev_ingress_queue(dev);
if (dev_queue &&
tc_dump_tclass_root(dev_queue->qdisc_sleeping, skb, tcm, cb,
&t, s_t) < 0)
goto done;
done:
cb->args[0] = t;
dev_put(dev);
return skb->len;
}
#ifdef CONFIG_PROC_FS
static int psched_show(struct seq_file *seq, void *v)
{
seq_printf(seq, "%08x %08x %08x %08x\n",
(u32)NSEC_PER_USEC, (u32)PSCHED_TICKS2NS(1),
1000000,
(u32)NSEC_PER_SEC / hrtimer_resolution);
return 0;
}
static int psched_open(struct inode *inode, struct file *file)
{
return single_open(file, psched_show, NULL);
}
static const struct file_operations psched_fops = {
.owner = THIS_MODULE,
.open = psched_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static int __net_init psched_net_init(struct net *net)
{
struct proc_dir_entry *e;
e = proc_create("psched", 0, net->proc_net, &psched_fops);
if (e == NULL)
return -ENOMEM;
return 0;
}
static void __net_exit psched_net_exit(struct net *net)
{
remove_proc_entry("psched", net->proc_net);
}
#else
static int __net_init psched_net_init(struct net *net)
{
return 0;
}
static void __net_exit psched_net_exit(struct net *net)
{
}
#endif
static struct pernet_operations psched_net_ops = {
.init = psched_net_init,
.exit = psched_net_exit,
};
static int __init pktsched_init(void)
{
int err;
err = register_pernet_subsys(&psched_net_ops);
if (err) {
pr_err("pktsched_init: "
"cannot initialize per netns operations\n");
return err;
}
register_qdisc(&pfifo_fast_ops);
register_qdisc(&pfifo_qdisc_ops);
register_qdisc(&bfifo_qdisc_ops);
register_qdisc(&pfifo_head_drop_qdisc_ops);
register_qdisc(&mq_qdisc_ops);
register_qdisc(&noqueue_qdisc_ops);
rtnl_register(PF_UNSPEC, RTM_NEWQDISC, tc_modify_qdisc, NULL, 0);
rtnl_register(PF_UNSPEC, RTM_DELQDISC, tc_get_qdisc, NULL, 0);
rtnl_register(PF_UNSPEC, RTM_GETQDISC, tc_get_qdisc, tc_dump_qdisc,
0);
rtnl_register(PF_UNSPEC, RTM_NEWTCLASS, tc_ctl_tclass, NULL, 0);
rtnl_register(PF_UNSPEC, RTM_DELTCLASS, tc_ctl_tclass, NULL, 0);
rtnl_register(PF_UNSPEC, RTM_GETTCLASS, tc_ctl_tclass, tc_dump_tclass,
0);
return 0;
}
subsys_initcall(pktsched_init);