* Re: [PATCH] IPoIB: Convert to netdevice internal stats
From: David Miller @ 2007-09-28 22:34 UTC (permalink / raw)
To: rdreier; +Cc: davem, jeff, general, netdev
In-Reply-To: <ada641umo2u.fsf@cisco.com>
From: Roland Dreier <rdreier@cisco.com>
Date: Fri, 28 Sep 2007 15:18:01 -0700
> Use the stats member of struct netdevice in IPoIB, so we can save
> memory by deleting the stats member of struct ipoib_dev_priv, and save
> code by deleting ipoib_get_stats().
>
> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Applied to net-2.6.24, thanks.
How is that ibm_emac NAPI conversion coming along? :-)
^ permalink raw reply
* Re: [PATCH] IPoIB: Convert to netdevice internal stats
From: Roland Dreier @ 2007-09-28 22:36 UTC (permalink / raw)
To: David Miller; +Cc: jeff, general, netdev
In-Reply-To: <20070928.153425.43869012.davem@davemloft.net>
> How is that ibm_emac NAPI conversion coming along? :-)
Sorry, trying to reduce my backlog first, but it is still on my list
of things to work on :)
- R.
^ permalink raw reply
* Re: [RFC] Make TCP prequeue configurable
From: David Miller @ 2007-09-28 22:40 UTC (permalink / raw)
To: jheffner; +Cc: shemminger, dada1, netdev
In-Reply-To: <46FC663A.6030601@psc.edu>
From: John Heffner <jheffner@psc.edu>
Date: Thu, 27 Sep 2007 22:26:02 -0400
> I think it really does help in case (4) with old NICs that don't do rx
> checksumming. I'm not sure how many people really care about this
> anymore, but probably some...?
>
> OTOH, it would be nice to get rid of sysctl_tcp_low_latency.
I know most high end apps use poll() so won't sleep in recvmsg()
directly, but occasisionally they will, and even those that have a
poll() triggered recvmsg() will run the backlog and do prequeue if
packets arrive while they are processing the existing receive packets
which is quite common.
So for any app that ends up doing a prequeue it's a win because there
is the issue of scheduling and cpu usage charging.
If the ACK's are coming out of the stack at the rate that the
application can pull data out of the receive queue, and no faster,
this will pace the sender to send precisely how fast the receiver can
get onto the cpu depending upon load.
Furthermore, prequeue puts the stack input processing work into user
context, which means that the users will be charged more fairly for
the work that is done for them.
When packets get fully processed in softirq context, that's bad
because this is cpu usage which doesn't get charged to the user, and
for TCP input processing this cpu usage is non-trivial and is
multiplied by packet count.
^ permalink raw reply
* SFQ: backport some features from ESFQ (try 4)
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev
Patchset try 2 addresses the review by Michael Buesch.
Patchset try 3 addresses the review by Patrick McHardy.
Patchset try 4 has a few cosmetic improvements.
Nobody reviewed my last set of patches, and I wasn't pushy about asking.
Since it's been a while, I ported the kernel and userspace patchsets to
current git of net-2.6 and iproute2, respectively.
The first 7 patches in this series resemble the corresponding 7 patches
in patchset try 2. There aren't any major changes--just modifications
to address errors noticed in review and slight reorganizations to make
the next patches easier.
Patches 8-10 implement parameter passing via nested compat attributes.
This is necessary for using 'tc qdisc change' to disable perturbation.
The rest of the parameters were added for consistency.
Iproute2 patches will follow shortly.
The following is the original patch text.
This set of patches adds some of ESFQ's modifications to the original
SFQ. Thus far, I have received support for this approach rather than for
trying to get ESFQ included as a separate qdisc.
http://mailman.ds9a.nl/pipermail/lartc/2007q2/021056.html
My patches here implement "tc qdisc change", user-configurable depth
(number of flows), and user-configurable divisor (for setting hash table
size). I've left out the remaining ESFQ features (usage of jhash and
different hashing methods) because Patrick McHardy intends to submit a
patch that will supersede that functionality; see the URL above.
Default values remain the same, and SFQ's default behavior remains the
same, so there should be no user disruption.
Thanks for your consideration,
Corey
include/linux/pkt_sched.h | 23 ++--
net/sched/sch_sfq.c | 353 ++++++++++++++++++++++++++++++---------------
2 files changed, 254 insertions(+), 122 deletions(-)
[PATCH 01/10] Preparatory refactoring part 1.
[PATCH 02/10] Preparatory refactoring part 2.
[PATCH 03/10] Move two functions.
[PATCH 04/10] Make "depth" (number of queues) user-configurable:
[PATCH 05/10] Add divisor.
[PATCH 06/10] Make qdisc changeable.
[PATCH 07/10] Remove comments about hardcoded values.
[PATCH 08/10] Multiply perturb_period by HZ when used rather than when assigned.
[PATCH 09/10] Change perturb_period to unsigned.
[PATCH 10/10] Use nested compat attributes to pass parameters.
^ permalink raw reply
* [PATCH 03/10] Move two functions.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Move sfq_q_destroy() to above sfq_q_init() so that it can be used
by an error case in a later patch.
Move sfq_destroy() as well, for clarity.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 22 +++++++++++-----------
1 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 1ba3d1a..ca22cb7 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -382,6 +382,17 @@ static void sfq_perturbation(unsigned long arg)
}
}
+static void sfq_q_destroy(struct sfq_sched_data *q)
+{
+ del_timer(&q->perturb_timer);
+}
+
+static void sfq_destroy(struct Qdisc *sch)
+{
+ struct sfq_sched_data *q = qdisc_priv(sch);
+ sfq_q_destroy(q);
+}
+
static void
sfq_default_parameters(struct Qdisc *sch)
{
@@ -451,17 +462,6 @@ static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
return 0;
}
-static void sfq_q_destroy(struct sfq_sched_data *q)
-{
- del_timer(&q->perturb_timer);
-}
-
-static void sfq_destroy(struct Qdisc *sch)
-{
- struct sfq_sched_data *q = qdisc_priv(sch);
- sfq_q_destroy(q);
-}
-
static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct sfq_sched_data *q = qdisc_priv(sch);
--
1.5.3
^ permalink raw reply related
* [PATCH 02/10] Preparatory refactoring part 2.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Factor code out of sfq_init() and sfq_destroy(), again so that the
new functions can be used by sfq_change() later.
Actually, as the diff itself shows, most of the sfq_q_init() code
comes from the original sfq_change(), but sfq_change() is only
called by sfq_init() right now. Thus, it is safe to remove
sfq_change(); "tc qdisc change" doesn't yet work for sfq anyway.
Setting default parameters is moved into a separate function for
clarity.
The sfq_destroy() --> sfq_q_destroy() change looks pointless here,
but it's cleaner to split now and add code to sfq_q_destroy() in a
later patch.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 95 +++++++++++++++++++++++++++++----------------------
1 files changed, 54 insertions(+), 41 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 57485ef..1ba3d1a 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -382,43 +382,41 @@ static void sfq_perturbation(unsigned long arg)
}
}
-static int sfq_change(struct Qdisc *sch, struct rtattr *opt)
+static void
+sfq_default_parameters(struct Qdisc *sch)
{
struct sfq_sched_data *q = qdisc_priv(sch);
- struct tc_sfq_qopt *ctl = RTA_DATA(opt);
- unsigned int qlen;
-
- if (opt->rta_len < RTA_LENGTH(sizeof(*ctl)))
- return -EINVAL;
-
- sch_tree_lock(sch);
- q->quantum = ctl->quantum ? : psched_mtu(sch->dev);
- q->perturb_period = ctl->perturb_period*HZ;
- if (ctl->limit)
- q->limit = min_t(u32, ctl->limit, SFQ_DEPTH - 2);
- qlen = sch->q.qlen;
- while (sch->q.qlen > q->limit)
- sfq_drop(sch);
- qdisc_tree_decrease_qlen(sch, qlen - sch->q.qlen);
-
- del_timer(&q->perturb_timer);
- if (q->perturb_period) {
- q->perturb_timer.expires = jiffies + q->perturb_period;
- add_timer(&q->perturb_timer);
- }
- sch_tree_unlock(sch);
- return 0;
+ q->quantum = psched_mtu(sch->dev);
+ q->perturbation = 0;
+ q->perturb_period = 0;
+ q->limit = SFQ_DEPTH - 2;
}
-static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
+static int
+sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
{
- struct sfq_sched_data *q = qdisc_priv(sch);
int i;
- init_timer(&q->perturb_timer);
- q->perturb_timer.data = (unsigned long)sch;
- q->perturb_timer.function = sfq_perturbation;
+ /* At this point, parameters are set to either defaults (sfq_init) or
+ * the previous values (sfq_change). So, overwrite the parameters as
+ * specified. */
+ if (opt) {
+ struct tc_sfq_qopt *ctl = RTA_DATA(opt);
+
+ if (opt->rta_len < RTA_LENGTH(sizeof(*ctl)))
+ return -EINVAL;
+
+ if (ctl->quantum)
+ q->quantum = ctl->quantum;
+ if (ctl->perturb_period)
+ q->perturb_period = ctl->perturb_period * HZ;
+ if (ctl->limit)
+ q->limit = ctl->limit;
+ }
+ q->limit = min_t(u32, q->limit, SFQ_DEPTH - 2);
+ q->tail = SFQ_DEPTH;
+ q->max_depth = 0;
for (i=0; i<SFQ_HASH_DIVISOR; i++)
q->ht[i] = SFQ_DEPTH;
@@ -427,28 +425,43 @@ static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
q->dep[i+SFQ_DEPTH].next = i+SFQ_DEPTH;
q->dep[i+SFQ_DEPTH].prev = i+SFQ_DEPTH;
}
- q->limit = SFQ_DEPTH - 2;
- q->max_depth = 0;
- q->tail = SFQ_DEPTH;
- if (opt == NULL) {
- q->quantum = psched_mtu(sch->dev);
- q->perturb_period = 0;
- } else {
- int err = sfq_change(sch, opt);
- if (err)
- return err;
- }
+
for (i=0; i<SFQ_DEPTH; i++)
sfq_link(q, i);
return 0;
}
-static void sfq_destroy(struct Qdisc *sch)
+static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
{
struct sfq_sched_data *q = qdisc_priv(sch);
+ int err;
+
+ sfq_default_parameters(sch);
+ if ((err = sfq_q_init(q, opt)))
+ return err;
+
+ init_timer(&q->perturb_timer);
+ q->perturb_timer.data = (unsigned long)sch;
+ q->perturb_timer.function = sfq_perturbation;
+ if (q->perturb_period) {
+ q->perturb_timer.expires = jiffies + q->perturb_period;
+ add_timer(&q->perturb_timer);
+ }
+
+ return 0;
+}
+
+static void sfq_q_destroy(struct sfq_sched_data *q)
+{
del_timer(&q->perturb_timer);
}
+static void sfq_destroy(struct Qdisc *sch)
+{
+ struct sfq_sched_data *q = qdisc_priv(sch);
+ sfq_q_destroy(q);
+}
+
static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct sfq_sched_data *q = qdisc_priv(sch);
--
1.5.3
^ permalink raw reply related
* [PATCH 01/10] Preparatory refactoring part 1.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Make a new function sfq_q_enqueue() that operates directly on the
queue data. This will be useful for implementing sfq_change() in
a later patch. A pleasant side-effect is reducing most of the
duplicate code in sfq_enqueue() and sfq_requeue().
Similarly, make a new function sfq_q_dequeue().
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 72 +++++++++++++++++++++++++++------------------------
1 files changed, 38 insertions(+), 34 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 3a23e30..57485ef 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -77,6 +77,9 @@
#define SFQ_DEPTH 128
#define SFQ_HASH_DIVISOR 1024
+#define SFQ_HEAD 0
+#define SFQ_TAIL 1
+
/* This type should contain at least SFQ_DEPTH*2 values */
typedef unsigned char sfq_index;
@@ -244,10 +247,9 @@ static unsigned int sfq_drop(struct Qdisc *sch)
return 0;
}
-static int
-sfq_enqueue(struct sk_buff *skb, struct Qdisc* sch)
+static void
+sfq_q_enqueue(struct sk_buff *skb, struct sfq_sched_data *q, unsigned int end)
{
- struct sfq_sched_data *q = qdisc_priv(sch);
unsigned hash = sfq_hash(q, skb);
sfq_index x;
@@ -256,8 +258,12 @@ sfq_enqueue(struct sk_buff *skb, struct Qdisc* sch)
q->ht[hash] = x = q->dep[SFQ_DEPTH].next;
q->hash[x] = hash;
}
- sch->qstats.backlog += skb->len;
- __skb_queue_tail(&q->qs[x], skb);
+
+ if (end == SFQ_TAIL)
+ __skb_queue_tail(&q->qs[x], skb);
+ else
+ __skb_queue_head(&q->qs[x], skb);
+
sfq_inc(q, x);
if (q->qs[x].qlen == 1) { /* The flow is new */
if (q->tail == SFQ_DEPTH) { /* It is the first flow */
@@ -270,6 +276,15 @@ sfq_enqueue(struct sk_buff *skb, struct Qdisc* sch)
q->tail = x;
}
}
+}
+
+static int
+sfq_enqueue(struct sk_buff *skb, struct Qdisc* sch)
+{
+ struct sfq_sched_data *q = qdisc_priv(sch);
+
+ sfq_q_enqueue(skb, q, SFQ_TAIL);
+ sch->qstats.backlog += skb->len;
if (++sch->q.qlen <= q->limit) {
sch->bstats.bytes += skb->len;
sch->bstats.packets++;
@@ -284,45 +299,21 @@ static int
sfq_requeue(struct sk_buff *skb, struct Qdisc* sch)
{
struct sfq_sched_data *q = qdisc_priv(sch);
- unsigned hash = sfq_hash(q, skb);
- sfq_index x;
- x = q->ht[hash];
- if (x == SFQ_DEPTH) {
- q->ht[hash] = x = q->dep[SFQ_DEPTH].next;
- q->hash[x] = hash;
- }
+ sfq_q_enqueue(skb, q, SFQ_HEAD);
sch->qstats.backlog += skb->len;
- __skb_queue_head(&q->qs[x], skb);
- sfq_inc(q, x);
- if (q->qs[x].qlen == 1) { /* The flow is new */
- if (q->tail == SFQ_DEPTH) { /* It is the first flow */
- q->tail = x;
- q->next[x] = x;
- q->allot[x] = q->quantum;
- } else {
- q->next[x] = q->next[q->tail];
- q->next[q->tail] = x;
- q->tail = x;
- }
- }
if (++sch->q.qlen <= q->limit) {
sch->qstats.requeues++;
return 0;
}
- sch->qstats.drops++;
sfq_drop(sch);
return NET_XMIT_CN;
}
-
-
-
-static struct sk_buff *
-sfq_dequeue(struct Qdisc* sch)
+static struct
+sk_buff *sfq_q_dequeue(struct sfq_sched_data *q)
{
- struct sfq_sched_data *q = qdisc_priv(sch);
struct sk_buff *skb;
sfq_index a, old_a;
@@ -335,8 +326,6 @@ sfq_dequeue(struct Qdisc* sch)
/* Grab packet */
skb = __skb_dequeue(&q->qs[a]);
sfq_dec(q, a);
- sch->q.qlen--;
- sch->qstats.backlog -= skb->len;
/* Is the slot empty? */
if (q->qs[a].qlen == 0) {
@@ -353,6 +342,21 @@ sfq_dequeue(struct Qdisc* sch)
a = q->next[a];
q->allot[a] += q->quantum;
}
+
+ return skb;
+}
+
+static struct sk_buff
+*sfq_dequeue(struct Qdisc* sch)
+{
+ struct sfq_sched_data *q = qdisc_priv(sch);
+ struct sk_buff *skb;
+
+ skb = sfq_q_dequeue(q);
+ if (skb == NULL)
+ return NULL;
+ sch->q.qlen--;
+ sch->qstats.backlog -= skb->len;
return skb;
}
--
1.5.3
^ permalink raw reply related
* [PATCH 06/10] Make qdisc changeable.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Re-implement sfq_change() and enable Qdisc_opts.change so "tc qdisc
change" will work.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++-
1 files changed, 60 insertions(+), 1 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index d72ea7c..b8e8fa5 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -416,6 +416,17 @@ sfq_default_parameters(struct Qdisc *sch)
q->limit = SFQ_DEPTH_DEFAULT - 2;
}
+static void
+sfq_copy_parameters(struct sfq_sched_data *dst, struct sfq_sched_data *src)
+{
+ dst->quantum = src->quantum;
+ dst->perturbation = src->perturbation;
+ dst->perturb_period = src->perturb_period;
+ dst->hash_divisor = src->hash_divisor;
+ dst->limit = src->limit;
+ dst->depth = src->depth;
+}
+
static int
sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
{
@@ -503,6 +514,54 @@ static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
return 0;
}
+static int sfq_change(struct Qdisc *sch, struct rtattr *opt)
+{
+ struct sfq_sched_data *q = qdisc_priv(sch);
+ struct sfq_sched_data tmp;
+ struct sk_buff *skb;
+ unsigned int qlen;
+ int err;
+
+ /* set up tmp queue */
+ memset(&tmp, 0, sizeof(struct sfq_sched_data));
+ sfq_copy_parameters(&tmp, q);
+ if ((err = sfq_q_init(&tmp, opt)))
+ return err;
+
+ /* copy packets from the old queue to the tmp queue */
+ sch_tree_lock(sch);
+ qlen = sch->q.qlen;
+ while (sch->q.qlen >= tmp.limit - 1)
+ sfq_drop(sch);
+ qdisc_tree_decrease_qlen(sch, qlen - sch->q.qlen);
+ while ((skb = sfq_q_dequeue(q)) != NULL)
+ sfq_q_enqueue(skb, &tmp, SFQ_TAIL);
+
+ /* clean up the old queue */
+ sfq_q_destroy(q);
+
+ /* copy elements of the tmp queue into the old queue */
+ sfq_copy_parameters(q, &tmp);
+ q->tail = tmp.tail;
+ q->max_depth = tmp.max_depth;
+ q->ht = tmp.ht;
+ q->dep = tmp.dep;
+ q->next = tmp.next;
+ q->allot = tmp.allot;
+ q->hash = tmp.hash;
+ q->qs = tmp.qs;
+
+ /* finish up */
+ if (q->perturb_period) {
+ q->perturb_timer.expires = jiffies + q->perturb_period;
+ add_timer(&q->perturb_timer);
+ } else {
+ q->perturbation = 0;
+ }
+ sch_tree_unlock(sch);
+ return 0;
+}
+
static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct sfq_sched_data *q = qdisc_priv(sch);
@@ -537,7 +596,7 @@ static struct Qdisc_ops sfq_qdisc_ops = {
.init = sfq_init,
.reset = sfq_reset,
.destroy = sfq_destroy,
- .change = NULL,
+ .change = sfq_change,
.dump = sfq_dump,
.owner = THIS_MODULE,
};
--
1.5.3
^ permalink raw reply related
* [PATCH 04/10] Make "depth" (number of queues) user-configurable:
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
* replace #define with a parameter
* use old hardcoded value as a default
* kcalloc() arrays in sfq_q_init()
* free() arrays in sfq_q_destroy()
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 85 +++++++++++++++++++++++++++++++++++---------------
1 files changed, 59 insertions(+), 26 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index ca22cb7..34a299d 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -74,14 +74,16 @@
It is easy to increase these values, but not in flight. */
-#define SFQ_DEPTH 128
+#define SFQ_DEPTH_DEFAULT 128
#define SFQ_HASH_DIVISOR 1024
#define SFQ_HEAD 0
#define SFQ_TAIL 1
-/* This type should contain at least SFQ_DEPTH*2 values */
-typedef unsigned char sfq_index;
+/* This type must contain greater than depth*2 values, so depth is constrained
+ * accordingly. */
+typedef unsigned int sfq_index;
+#define SFQ_MAX_DEPTH (UINT_MAX / 2 - 1)
struct sfq_head
{
@@ -95,6 +97,7 @@ struct sfq_sched_data
int perturb_period;
unsigned quantum; /* Allotment per round: MUST BE >= MTU */
int limit;
+ unsigned depth;
/* Variables */
struct timer_list perturb_timer;
@@ -103,11 +106,11 @@ struct sfq_sched_data
sfq_index max_depth; /* Maximal depth */
sfq_index ht[SFQ_HASH_DIVISOR]; /* Hash table */
- sfq_index next[SFQ_DEPTH]; /* Active slots link */
- short allot[SFQ_DEPTH]; /* Current allotment per slot */
- unsigned short hash[SFQ_DEPTH]; /* Hash value indexed by slots */
- struct sk_buff_head qs[SFQ_DEPTH]; /* Slot queue */
- struct sfq_head dep[SFQ_DEPTH*2]; /* Linked list of slots, indexed by depth */
+ sfq_index *next; /* Active slots link */
+ short *allot; /* Current allotment per slot */
+ unsigned short *hash; /* Hash value indexed by slots */
+ struct sk_buff_head *qs; /* Slot queue */
+ struct sfq_head *dep; /* Linked list of slots, indexed by depth */
};
static __inline__ unsigned sfq_fold_hash(struct sfq_sched_data *q, u32 h, u32 h1)
@@ -164,7 +167,7 @@ static unsigned sfq_hash(struct sfq_sched_data *q, struct sk_buff *skb)
static inline void sfq_link(struct sfq_sched_data *q, sfq_index x)
{
sfq_index p, n;
- int d = q->qs[x].qlen + SFQ_DEPTH;
+ int d = q->qs[x].qlen + q->depth;
p = d;
n = q->dep[d].next;
@@ -215,7 +218,7 @@ static unsigned int sfq_drop(struct Qdisc *sch)
drop a packet from it */
if (d > 1) {
- sfq_index x = q->dep[d+SFQ_DEPTH].next;
+ sfq_index x = q->dep[d + q->depth].next;
skb = q->qs[x].prev;
len = skb->len;
__skb_unlink(skb, &q->qs[x]);
@@ -238,7 +241,7 @@ static unsigned int sfq_drop(struct Qdisc *sch)
kfree_skb(skb);
sfq_dec(q, d);
sch->q.qlen--;
- q->ht[q->hash[d]] = SFQ_DEPTH;
+ q->ht[q->hash[d]] = q->depth;
sch->qstats.drops++;
sch->qstats.backlog -= len;
return len;
@@ -254,8 +257,8 @@ sfq_q_enqueue(struct sk_buff *skb, struct sfq_sched_data *q, unsigned int end)
sfq_index x;
x = q->ht[hash];
- if (x == SFQ_DEPTH) {
- q->ht[hash] = x = q->dep[SFQ_DEPTH].next;
+ if (x == q->depth) {
+ q->ht[hash] = x = q->dep[q->depth].next;
q->hash[x] = hash;
}
@@ -266,7 +269,7 @@ sfq_q_enqueue(struct sk_buff *skb, struct sfq_sched_data *q, unsigned int end)
sfq_inc(q, x);
if (q->qs[x].qlen == 1) { /* The flow is new */
- if (q->tail == SFQ_DEPTH) { /* It is the first flow */
+ if (q->tail == q->depth) { /* It is the first flow */
q->tail = x;
q->next[x] = x;
q->allot[x] = q->quantum;
@@ -318,7 +321,7 @@ sk_buff *sfq_q_dequeue(struct sfq_sched_data *q)
sfq_index a, old_a;
/* No active slots */
- if (q->tail == SFQ_DEPTH)
+ if (q->tail == q->depth)
return NULL;
a = old_a = q->next[q->tail];
@@ -329,10 +332,10 @@ sk_buff *sfq_q_dequeue(struct sfq_sched_data *q)
/* Is the slot empty? */
if (q->qs[a].qlen == 0) {
- q->ht[q->hash[a]] = SFQ_DEPTH;
+ q->ht[q->hash[a]] = q->depth;
a = q->next[a];
if (a == old_a) {
- q->tail = SFQ_DEPTH;
+ q->tail = q->depth;
return skb;
}
q->next[q->tail] = a;
@@ -385,6 +388,11 @@ static void sfq_perturbation(unsigned long arg)
static void sfq_q_destroy(struct sfq_sched_data *q)
{
del_timer(&q->perturb_timer);
+ kfree(q->dep);
+ kfree(q->next);
+ kfree(q->allot);
+ kfree(q->hash);
+ kfree(q->qs);
}
static void sfq_destroy(struct Qdisc *sch)
@@ -401,7 +409,8 @@ sfq_default_parameters(struct Qdisc *sch)
q->quantum = psched_mtu(sch->dev);
q->perturbation = 0;
q->perturb_period = 0;
- q->limit = SFQ_DEPTH - 2;
+ q->depth = SFQ_DEPTH_DEFAULT;
+ q->limit = SFQ_DEPTH_DEFAULT - 2;
}
static int
@@ -422,24 +431,48 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
q->quantum = ctl->quantum;
if (ctl->perturb_period)
q->perturb_period = ctl->perturb_period * HZ;
+ if (ctl->flows)
+ q->depth = ctl->flows;
if (ctl->limit)
q->limit = ctl->limit;
+
+ if (q->depth > SFQ_MAX_DEPTH)
+ return -EINVAL;
}
- q->limit = min_t(u32, q->limit, SFQ_DEPTH - 2);
- q->tail = SFQ_DEPTH;
+ q->limit = min_t(u32, q->limit, q->depth - 2);
+ q->tail = q->depth;
q->max_depth = 0;
+ q->dep = kcalloc(1 + q->depth*2, sizeof(struct sfq_head), GFP_KERNEL);
+ if (!q->dep)
+ goto err_case;
+ q->next = kcalloc(q->depth, sizeof(sfq_index), GFP_KERNEL);
+ if (!q->next)
+ goto err_case;
+ q->allot = kcalloc(q->depth, sizeof(short), GFP_KERNEL);
+ if (!q->allot)
+ goto err_case;
+ q->hash = kcalloc(q->depth, sizeof(unsigned short), GFP_KERNEL);
+ if (!q->hash)
+ goto err_case;
+ q->qs = kcalloc(q->depth, sizeof(struct sk_buff_head), GFP_KERNEL);
+ if (!q->qs)
+ goto err_case;
+
for (i=0; i<SFQ_HASH_DIVISOR; i++)
- q->ht[i] = SFQ_DEPTH;
- for (i=0; i<SFQ_DEPTH; i++) {
+ q->ht[i] = q->depth;
+ for (i=0; i < q->depth; i++) {
skb_queue_head_init(&q->qs[i]);
- q->dep[i+SFQ_DEPTH].next = i+SFQ_DEPTH;
- q->dep[i+SFQ_DEPTH].prev = i+SFQ_DEPTH;
+ q->dep[i + q->depth].next = i + q->depth;
+ q->dep[i + q->depth].prev = i + q->depth;
}
- for (i=0; i<SFQ_DEPTH; i++)
+ for (i=0; i < q->depth; i++)
sfq_link(q, i);
return 0;
+err_case:
+ sfq_q_destroy(q);
+ return -ENOBUFS;
}
static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
@@ -473,7 +506,7 @@ static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
opt.limit = q->limit;
opt.divisor = SFQ_HASH_DIVISOR;
- opt.flows = q->limit;
+ opt.flows = q->depth;
RTA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
--
1.5.3
^ permalink raw reply related
* [PATCH 08/10] Multiply perturb_period by HZ when used rather than when assigned.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
perturb_period is the only parameter that doesn't match 1:1 with the
value from userspace. This change makes it easy and clean to use a
small macro for setting parameters (in a subsequent patch).
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 10 +++++-----
1 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 341a9a1..2d3cc38 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -370,7 +370,7 @@ static void sfq_perturbation(unsigned long arg)
q->perturbation = net_random()&0x1F;
if (q->perturb_period) {
- q->perturb_timer.expires = jiffies + q->perturb_period;
+ q->perturb_timer.expires = jiffies + q->perturb_period * HZ;
add_timer(&q->perturb_timer);
}
}
@@ -433,7 +433,7 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
if (ctl->quantum)
q->quantum = ctl->quantum;
if (ctl->perturb_period)
- q->perturb_period = ctl->perturb_period * HZ;
+ q->perturb_period = ctl->perturb_period;
if (ctl->divisor)
q->hash_divisor = ctl->divisor;
if (ctl->flows)
@@ -496,7 +496,7 @@ static int sfq_init(struct Qdisc *sch, struct rtattr *opt)
q->perturb_timer.data = (unsigned long)sch;
q->perturb_timer.function = sfq_perturbation;
if (q->perturb_period) {
- q->perturb_timer.expires = jiffies + q->perturb_period;
+ q->perturb_timer.expires = jiffies + q->perturb_period * HZ;
add_timer(&q->perturb_timer);
}
@@ -542,7 +542,7 @@ static int sfq_change(struct Qdisc *sch, struct rtattr *opt)
/* finish up */
if (q->perturb_period) {
- q->perturb_timer.expires = jiffies + q->perturb_period;
+ q->perturb_timer.expires = jiffies + q->perturb_period * HZ;
add_timer(&q->perturb_timer);
} else {
q->perturbation = 0;
@@ -558,7 +558,7 @@ static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
struct tc_sfq_qopt opt;
opt.quantum = q->quantum;
- opt.perturb_period = q->perturb_period/HZ;
+ opt.perturb_period = q->perturb_period;
opt.limit = q->limit;
opt.divisor = q->hash_divisor;
--
1.5.3
^ permalink raw reply related
* [PATCH 07/10] Remove comments about hardcoded values.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
None of these are true anymore (hooray!).
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 8 --------
net/sched/sch_sfq.c | 17 +++--------------
2 files changed, 3 insertions(+), 22 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 268c515..58a0ea6 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -148,14 +148,6 @@ struct tc_sfq_qopt
unsigned flows; /* Maximal number of flows */
};
-/*
- * NOTE: limit, divisor and flows are hardwired to code at the moment.
- *
- * limit=flows=128, divisor=1024;
- *
- * The only reason for this is efficiency, it is possible
- * to change these parameters in compile time.
- */
/* RED section */
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index b8e8fa5..341a9a1 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -61,18 +61,7 @@
We still need true WFQ for top level CSZ, but using WFQ
for the best effort traffic is absolutely pointless:
- SFQ is superior for this purpose.
-
- IMPLEMENTATION:
- This implementation limits maximal queue length to 128;
- maximal mtu to 2^15-1; number of hash buckets to 1024.
- The only goal of this restrictions was that all data
- fit into one 4K page :-). Struct sfq_sched_data is
- organized in anti-cache manner: all the data for a bucket
- are scattered over different locations. This is not good,
- but it allowed me to put it into 4K.
-
- It is easy to increase these values, but not in flight. */
+ SFQ is superior for this purpose. */
#define SFQ_DEPTH_DEFAULT 128
#define SFQ_DIVISOR_DEFAULT 1024
@@ -521,7 +510,7 @@ static int sfq_change(struct Qdisc *sch, struct rtattr *opt)
struct sk_buff *skb;
unsigned int qlen;
int err;
-
+
/* set up tmp queue */
memset(&tmp, 0, sizeof(struct sfq_sched_data));
sfq_copy_parameters(&tmp, q);
@@ -536,7 +525,7 @@ static int sfq_change(struct Qdisc *sch, struct rtattr *opt)
qdisc_tree_decrease_qlen(sch, qlen - sch->q.qlen);
while ((skb = sfq_q_dequeue(q)) != NULL)
sfq_q_enqueue(skb, &tmp, SFQ_TAIL);
-
+
/* clean up the old queue */
sfq_q_destroy(q);
--
1.5.3
^ permalink raw reply related
* [PATCH 10/10] Use nested compat attributes to pass parameters.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
This fixes the ambiguity between, for example:
tc qdisc change ... perturb 0
tc qdisc change ...
Without this patch, there is no way for SFQ to differentiate between
a parameter specified to be 0 and a parameter that was omitted.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 13 +++++++++++
net/sched/sch_sfq.c | 53 +++++++++++++++++++++++++++++---------------
2 files changed, 48 insertions(+), 18 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 8559974..aad04eb 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -148,6 +148,19 @@ struct tc_sfq_qopt
unsigned flows; /* Maximal number of flows */
};
+enum
+{
+ TCA_SFQ_UNSPEC,
+ TCA_SFQ_COMPAT,
+ TCA_SFQ_QUANTUM,
+ TCA_SFQ_PERTURB,
+ TCA_SFQ_LIMIT,
+ TCA_SFQ_DIVISOR,
+ TCA_SFQ_FLOWS,
+ __TCA_SFQ_MAX,
+};
+
+#define TCA_SFQ_MAX (__TCA_SFQ_MAX - 1)
/* RED section */
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 170fd37..36197f6 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -428,25 +428,31 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
* the previous values (sfq_change). So, overwrite the parameters as
* specified. */
if (opt) {
- struct tc_sfq_qopt *ctl = RTA_DATA(opt);
-
- if (opt->rta_len < RTA_LENGTH(sizeof(*ctl)))
- return -EINVAL;
-
- if (ctl->quantum)
- q->quantum = ctl->quantum;
- if (ctl->perturb_period)
- q->perturb_period = ctl->perturb_period;
- if (ctl->divisor)
- q->hash_divisor = ctl->divisor;
- if (ctl->flows)
- q->depth = ctl->flows;
- if (ctl->limit)
- q->limit = ctl->limit;
-
+ struct tc_sfq_qopt *ctl;
+ struct rtattr *tb[TCA_SFQ_MAX];
+
+ if (rtattr_parse_nested_compat(tb, TCA_SFQ_MAX, opt, ctl,
+ sizeof(*ctl)))
+ goto rtattr_failure;
+
+#define GET_PARAM(dst, nest, compat) do { \
+ struct rtattr *rta = tb[(nest) - 1]; \
+ if (rta) \
+ (dst) = RTA_GET_U32(rta); \
+ else if ((compat)) \
+ (dst) = (compat); \
+} while (0)
+
+ GET_PARAM(q->quantum, TCA_SFQ_QUANTUM, ctl->quantum);
+ GET_PARAM(q->perturb_period, TCA_SFQ_PERTURB,
+ ctl->perturb_period);
+ GET_PARAM(q->hash_divisor, TCA_SFQ_DIVISOR, ctl->divisor);
+ GET_PARAM(q->depth, TCA_SFQ_FLOWS, ctl->flows);
+ GET_PARAM(q->limit, TCA_SFQ_LIMIT, ctl->limit);
+
if (q->perturb_period > SFQ_MAX_PERTURB ||
q->depth > SFQ_MAX_DEPTH)
- return -EINVAL;
+ goto rtattr_failure;
}
q->limit = min_t(u32, q->limit, q->depth - 2);
q->tail = q->depth;
@@ -482,6 +488,8 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
for (i=0; i < q->depth; i++)
sfq_link(q, i);
return 0;
+rtattr_failure:
+ return -EINVAL;
err_case:
sfq_q_destroy(q);
return -ENOBUFS;
@@ -559,17 +567,26 @@ static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct sfq_sched_data *q = qdisc_priv(sch);
unsigned char *b = skb_tail_pointer(skb);
+ struct rtattr *nest;
struct tc_sfq_qopt opt;
opt.quantum = q->quantum;
opt.perturb_period = q->perturb_period;
-
opt.limit = q->limit;
opt.divisor = q->hash_divisor;
opt.flows = q->depth;
+ nest = RTA_NEST_COMPAT(skb, TCA_OPTIONS, sizeof(opt), &opt);
+
+ RTA_PUT_U32(skb, TCA_SFQ_QUANTUM, q->quantum);
+ RTA_PUT_U32(skb, TCA_SFQ_PERTURB, q->perturb_period);
+ RTA_PUT_U32(skb, TCA_SFQ_LIMIT, q->limit);
+ RTA_PUT_U32(skb, TCA_SFQ_DIVISOR, q->hash_divisor);
+ RTA_PUT_U32(skb, TCA_SFQ_FLOWS, q->depth);
RTA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
+ RTA_NEST_COMPAT_END(skb, nest);
+
return skb->len;
rtattr_failure:
--
1.5.3
^ permalink raw reply related
* [PATCH 05/10] Add divisor.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Make hash divisor user-configurable.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
net/sched/sch_sfq.c | 18 +++++++++++++-----
1 files changed, 13 insertions(+), 5 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 34a299d..d72ea7c 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -75,7 +75,7 @@
It is easy to increase these values, but not in flight. */
#define SFQ_DEPTH_DEFAULT 128
-#define SFQ_HASH_DIVISOR 1024
+#define SFQ_DIVISOR_DEFAULT 1024
#define SFQ_HEAD 0
#define SFQ_TAIL 1
@@ -98,6 +98,7 @@ struct sfq_sched_data
unsigned quantum; /* Allotment per round: MUST BE >= MTU */
int limit;
unsigned depth;
+ unsigned hash_divisor;
/* Variables */
struct timer_list perturb_timer;
@@ -105,7 +106,7 @@ struct sfq_sched_data
sfq_index tail; /* Index of current slot in round */
sfq_index max_depth; /* Maximal depth */
- sfq_index ht[SFQ_HASH_DIVISOR]; /* Hash table */
+ sfq_index *ht; /* Hash table */
sfq_index *next; /* Active slots link */
short *allot; /* Current allotment per slot */
unsigned short *hash; /* Hash value indexed by slots */
@@ -120,7 +121,7 @@ static __inline__ unsigned sfq_fold_hash(struct sfq_sched_data *q, u32 h, u32 h1
/* Have we any rotation primitives? If not, WHY? */
h ^= (h1<<pert) ^ (h1>>(0x1F - pert));
h ^= h>>10;
- return h & 0x3FF;
+ return h & (q->hash_divisor-1);
}
static unsigned sfq_hash(struct sfq_sched_data *q, struct sk_buff *skb)
@@ -388,6 +389,7 @@ static void sfq_perturbation(unsigned long arg)
static void sfq_q_destroy(struct sfq_sched_data *q)
{
del_timer(&q->perturb_timer);
+ kfree(q->ht);
kfree(q->dep);
kfree(q->next);
kfree(q->allot);
@@ -409,6 +411,7 @@ sfq_default_parameters(struct Qdisc *sch)
q->quantum = psched_mtu(sch->dev);
q->perturbation = 0;
q->perturb_period = 0;
+ q->hash_divisor = SFQ_DIVISOR_DEFAULT;
q->depth = SFQ_DEPTH_DEFAULT;
q->limit = SFQ_DEPTH_DEFAULT - 2;
}
@@ -431,6 +434,8 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
q->quantum = ctl->quantum;
if (ctl->perturb_period)
q->perturb_period = ctl->perturb_period * HZ;
+ if (ctl->divisor)
+ q->hash_divisor = ctl->divisor;
if (ctl->flows)
q->depth = ctl->flows;
if (ctl->limit)
@@ -443,6 +448,9 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
q->tail = q->depth;
q->max_depth = 0;
+ q->ht = kcalloc(q->hash_divisor, sizeof(sfq_index), GFP_KERNEL);
+ if (!q->ht)
+ goto err_case;
q->dep = kcalloc(1 + q->depth*2, sizeof(struct sfq_head), GFP_KERNEL);
if (!q->dep)
goto err_case;
@@ -459,7 +467,7 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
if (!q->qs)
goto err_case;
- for (i=0; i<SFQ_HASH_DIVISOR; i++)
+ for (i=0; i<q->hash_divisor; i++)
q->ht[i] = q->depth;
for (i=0; i < q->depth; i++) {
skb_queue_head_init(&q->qs[i]);
@@ -505,7 +513,7 @@ static int sfq_dump(struct Qdisc *sch, struct sk_buff *skb)
opt.perturb_period = q->perturb_period/HZ;
opt.limit = q->limit;
- opt.divisor = SFQ_HASH_DIVISOR;
+ opt.divisor = q->hash_divisor;
opt.flows = q->depth;
RTA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
--
1.5.3
^ permalink raw reply related
* [PATCH 09/10] Change perturb_period to unsigned.
From: Corey Hickey @ 2007-09-28 22:52 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
perturb_period is currently a signed integer, but I can't see any good
reason why this is so--a negative perturbation period will add a timer
that expires in the past, causing constant perturbation, which makes
hashing useless.
if (q->perturb_period) {
q->perturb_timer.expires = jiffies + q->perturb_period;
add_timer(&q->perturb_timer);
}
Strictly speaking, this will break binary compatibility with older
versions of tc, but that ought not to be a problem because (a) there's
no valid use for a negative perturb_period, and (b) negative values
will be seen as high values (> INT_MAX), which don't work anyway.
If perturb_period is too large, (perturb_period * HZ) will overflow the
size of an unsigned int and wrap around. So, check for thet and reject
values that are too high.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 2 +-
net/sched/sch_sfq.c | 8 ++++++--
2 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 58a0ea6..8559974 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -142,7 +142,7 @@ enum
struct tc_sfq_qopt
{
unsigned quantum; /* Bytes per round allocated to flow */
- int perturb_period; /* Period of hash perturbation */
+ unsigned perturb_period; /* Period of hash perturbation */
__u32 limit; /* Maximal packets in queue */
unsigned divisor; /* Hash divisor */
unsigned flows; /* Maximal number of flows */
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 2d3cc38..170fd37 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -74,6 +74,9 @@
typedef unsigned int sfq_index;
#define SFQ_MAX_DEPTH (UINT_MAX / 2 - 1)
+/* We don't want perturb_period * HZ to overflow an unsigned int. */
+#define SFQ_MAX_PERTURB (UINT_MAX / HZ)
+
struct sfq_head
{
sfq_index next;
@@ -83,7 +86,7 @@ struct sfq_head
struct sfq_sched_data
{
/* Parameters */
- int perturb_period;
+ unsigned perturb_period;
unsigned quantum; /* Allotment per round: MUST BE >= MTU */
int limit;
unsigned depth;
@@ -441,7 +444,8 @@ sfq_q_init(struct sfq_sched_data *q, struct rtattr *opt)
if (ctl->limit)
q->limit = ctl->limit;
- if (q->depth > SFQ_MAX_DEPTH)
+ if (q->perturb_period > SFQ_MAX_PERTURB ||
+ q->depth > SFQ_MAX_DEPTH)
return -EINVAL;
}
q->limit = min_t(u32, q->limit, q->depth - 2);
--
1.5.3
^ permalink raw reply related
* sfq
From: Corey Hickey @ 2007-09-28 22:54 UTC (permalink / raw)
To: netdev
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
These patches follow the ESFQ-->SFQ kernel patches. See the kernel
patch summary for general information.
Thanks,
Corey
include/linux/pkt_sched.h | 23 ++++++++++++++---------
tc/q_sfq.c | 43 ++++++++++++++++++++++++++++++++++++++-----
2 files changed, 52 insertions(+), 14 deletions(-)
[PATCH 1/3] SFQ: Support changing depth and divisor.
[PATCH 2/3] Change perturb_period to unsigned.
[PATCH 3/3] Use nested compat attributes for passing parameters to the kernel.
^ permalink raw reply
* [PATCH 1/3] SFQ: Support changing depth and divisor.
From: Corey Hickey @ 2007-09-28 22:54 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
This can safely be applied either before or after the kernel
patches because the tc_sfq_qopt struct is unchanged:
- old kernels will ignore the parameters from new iproute2
- new kernels will use the same default parameters
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 9 ---------
tc/q_sfq.c | 21 ++++++++++++++++++++-
2 files changed, 20 insertions(+), 10 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 268c515..9d41f63 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -148,15 +148,6 @@ struct tc_sfq_qopt
unsigned flows; /* Maximal number of flows */
};
-/*
- * NOTE: limit, divisor and flows are hardwired to code at the moment.
- *
- * limit=flows=128, divisor=1024;
- *
- * The only reason for this is efficiency, it is possible
- * to change these parameters in compile time.
- */
-
/* RED section */
enum
diff --git a/tc/q_sfq.c b/tc/q_sfq.c
index 05385cf..7754db7 100644
--- a/tc/q_sfq.c
+++ b/tc/q_sfq.c
@@ -25,7 +25,7 @@
static void explain(void)
{
- fprintf(stderr, "Usage: ... sfq [ limit NUMBER ] [ perturb SECS ] [ quantum BYTES ]\n");
+ fprintf(stderr, "Usage: ... sfq [ limit NUMBER ] [ depth FLOWS ] [ divisor HASHBITS ] [ perturb SECS ] [ quantum BYTES ]\n");
}
#define usage() return(-1)
@@ -63,6 +63,25 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
return -1;
}
ok++;
+ } else if (strcmp(*argv, "depth") == 0) {
+ NEXT_ARG();
+ if (get_unsigned(&opt.flows, *argv, 0)) {
+ fprintf(stderr, "Illegal \"depth\"\n");
+ return -1;
+ }
+ ok++;
+ } else if (strcmp(*argv, "divisor") == 0) {
+ NEXT_ARG();
+ if (get_unsigned(&opt.divisor, *argv, 0)) {
+ fprintf(stderr, "Illegal \"divisor\"\n");
+ return -1;
+ }
+ if (opt.divisor >= 15) {
+ fprintf(stderr, "Illegal \"divisor\", must be < 15\n");
+ return -1;
+ }
+ opt.divisor = 1<<opt.divisor;
+ ok++;
} else if (strcmp(*argv, "help") == 0) {
explain();
return -1;
--
1.5.3
^ permalink raw reply related
* [PATCH 2/3] Change perturb_period to unsigned.
From: Corey Hickey @ 2007-09-28 22:54 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
This corresponds to the kernel patch doing the same.
Here, too, this will technically break binary compatibility with older
kernels, but that shouldn't be a problem because negative perturb_period
values aren't usable anyway.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 2 +-
tc/q_sfq.c | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index 9d41f63..fb04a89 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -142,7 +142,7 @@ enum
struct tc_sfq_qopt
{
unsigned quantum; /* Bytes per round allocated to flow */
- int perturb_period; /* Period of hash perturbation */
+ unsigned perturb_period; /* Period of hash perturbation */
__u32 limit; /* Maximal packets in queue */
unsigned divisor; /* Hash divisor */
unsigned flows; /* Maximal number of flows */
diff --git a/tc/q_sfq.c b/tc/q_sfq.c
index 7754db7..c9fcc53 100644
--- a/tc/q_sfq.c
+++ b/tc/q_sfq.c
@@ -47,7 +47,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
ok++;
} else if (strcmp(*argv, "perturb") == 0) {
NEXT_ARG();
- if (get_integer(&opt.perturb_period, *argv, 0)) {
+ if (get_u32(&opt.perturb_period, *argv, 0)) {
fprintf(stderr, "Illegal \"perturb\"\n");
return -1;
}
@@ -115,7 +115,7 @@ static int sfq_print_opt(struct qdisc_util *qu, FILE *f, struct rtattr *opt)
fprintf(f, "flows %u/%u ", qopt->flows, qopt->divisor);
}
if (qopt->perturb_period)
- fprintf(f, "perturb %dsec ", qopt->perturb_period);
+ fprintf(f, "perturb %usec ", qopt->perturb_period);
return 0;
}
--
1.5.3
^ permalink raw reply related
* [PATCH 3/3] Use nested compat attributes for passing parameters to the kernel.
From: Corey Hickey @ 2007-09-28 22:54 UTC (permalink / raw)
To: netdev; +Cc: Corey Hickey
In-Reply-To: <1191019977201-git-send-email-bugfood-ml@fatooh.org>
Note that I have left sfq_print_opt() alone. At this point, there
can be no difference between the data in the nested rtattrs and the
data in the compat rtattr, and I didn't want to add clutter that
isn't useful. Let me know if I should do differently.
Signed-off-by: Corey Hickey <bugfood-ml@fatooh.org>
---
include/linux/pkt_sched.h | 14 ++++++++++++++
tc/q_sfq.c | 18 ++++++++++++++++--
2 files changed, 30 insertions(+), 2 deletions(-)
diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
index fb04a89..aad04eb 100644
--- a/include/linux/pkt_sched.h
+++ b/include/linux/pkt_sched.h
@@ -148,6 +148,20 @@ struct tc_sfq_qopt
unsigned flows; /* Maximal number of flows */
};
+enum
+{
+ TCA_SFQ_UNSPEC,
+ TCA_SFQ_COMPAT,
+ TCA_SFQ_QUANTUM,
+ TCA_SFQ_PERTURB,
+ TCA_SFQ_LIMIT,
+ TCA_SFQ_DIVISOR,
+ TCA_SFQ_FLOWS,
+ __TCA_SFQ_MAX,
+};
+
+#define TCA_SFQ_MAX (__TCA_SFQ_MAX - 1)
+
/* RED section */
enum
diff --git a/tc/q_sfq.c b/tc/q_sfq.c
index c9fcc53..5bb3eb7 100644
--- a/tc/q_sfq.c
+++ b/tc/q_sfq.c
@@ -34,9 +34,13 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
{
int ok=0;
struct tc_sfq_qopt opt;
+ struct rtattr *nest;
memset(&opt, 0, sizeof(opt));
+ /* put blank data in rtattr so there is a "hole" to fill later */
+ nest = addattr_nest_compat(n, 1024, TCA_OPTIONS, &opt, sizeof(opt));
+
while (argc > 0) {
if (strcmp(*argv, "quantum") == 0) {
NEXT_ARG();
@@ -44,6 +48,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
fprintf(stderr, "Illegal \"limit\"\n");
return -1;
}
+ addattr32(n, 1024, TCA_SFQ_QUANTUM, opt.quantum);
ok++;
} else if (strcmp(*argv, "perturb") == 0) {
NEXT_ARG();
@@ -51,6 +56,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
fprintf(stderr, "Illegal \"perturb\"\n");
return -1;
}
+ addattr32(n, 1024, TCA_SFQ_PERTURB, opt.perturb_period);
ok++;
} else if (strcmp(*argv, "limit") == 0) {
NEXT_ARG();
@@ -62,6 +68,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
fprintf(stderr, "Illegal \"limit\", must be > 1\n");
return -1;
}
+ addattr32(n, 1024, TCA_SFQ_LIMIT, opt.limit);
ok++;
} else if (strcmp(*argv, "depth") == 0) {
NEXT_ARG();
@@ -69,6 +76,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
fprintf(stderr, "Illegal \"depth\"\n");
return -1;
}
+ addattr32(n, 1024, TCA_SFQ_FLOWS, opt.flows);
ok++;
} else if (strcmp(*argv, "divisor") == 0) {
NEXT_ARG();
@@ -81,6 +89,7 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
return -1;
}
opt.divisor = 1<<opt.divisor;
+ addattr32(n, 1024, TCA_SFQ_DIVISOR, opt.divisor);
ok++;
} else if (strcmp(*argv, "help") == 0) {
explain();
@@ -93,8 +102,13 @@ static int sfq_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nl
argc--; argv++;
}
- if (ok)
- addattr_l(n, 1024, TCA_OPTIONS, &opt, sizeof(opt));
+ if (ok) {
+ /* fill the "hole" we left earlier with real compat data */
+ memcpy(RTA_DATA(nest), &opt, sizeof(opt));
+ addattr_nest_compat_end(n, nest);
+ }
+ else
+ nest->rta_len = 0;
return 0;
}
--
1.5.3
^ permalink raw reply related
* Re: bind and O_NONBLOCK
From: David Miller @ 2007-09-28 22:56 UTC (permalink / raw)
To: drepper; +Cc: netdev, alan
In-Reply-To: <46F35DD9.4000909@redhat.com>
From: Ulrich Drepper <drepper@redhat.com>
Date: Thu, 20 Sep 2007 22:59:53 -0700
> It would be ideal if we'd have such an implementation in the next few
> months so that we, in theory, can check whether the text in the
> specification makes sense.
It seems this functionality is only relevant for really ancient
protocols such as netbeui, as Alan mentioned.
The implementation would go in those protocols, the code for
which hasn't been touched substantially for years. It is
unclear whether those protocols are even functional at this
time. :-)
So expecting an implementation in the next few months depends
upon someone standing up and actively working on those rotting
code bases.
^ permalink raw reply
* Re: sfq (iproute2 patches)
From: Corey Hickey @ 2007-09-28 22:57 UTC (permalink / raw)
To: netdev
In-Reply-To: <11910200972768-git-send-email-bugfood-ml@fatooh.org>
Corey Hickey wrote:
> These patches follow the ESFQ-->SFQ kernel patches. See the kernel
> patch summary for general information.
Dang, I forgot to set the subject; these are the iproute2 patches.
-Corey
^ permalink raw reply
* Null dereference in socket.c
From: Chuck Ebbert @ 2007-09-28 22:58 UTC (permalink / raw)
To: netdev
After debugging an oops (https://bugzilla.redhat.com/attachment.cgi?id=209231)
I find it happens here in socket.c::sock_ioctl() line 902:
default:
=============> err = sock->ops->ioctl(sock, cmd, arg);
/*
* If this ioctl is unknown try to hand it down
* to the NIC driver.
*/
if (err == -ENOIOCTLCMD)
err = dev_ioctl(cmd, argp);
break;
ioctl is NULL and the kernel jumps to address 0. Should we add a check
for that?
Bug report:
https://bugzilla.redhat.com/show_bug.cgi?id=306801
^ permalink raw reply
* Re: Null dereference in socket.c
From: David Miller @ 2007-09-28 23:00 UTC (permalink / raw)
To: cebbert; +Cc: netdev
In-Reply-To: <46FD871C.2080004@redhat.com>
From: Chuck Ebbert <cebbert@redhat.com>
Date: Fri, 28 Sep 2007 18:58:36 -0400
> After debugging an oops (https://bugzilla.redhat.com/attachment.cgi?id=209231)
> I find it happens here in socket.c::sock_ioctl() line 902:
>
> default:
> =============> err = sock->ops->ioctl(sock, cmd, arg);
>
> /*
> * If this ioctl is unknown try to hand it down
> * to the NIC driver.
> */
> if (err == -ENOIOCTLCMD)
> err = dev_ioctl(cmd, argp);
> break;
>
>
> ioctl is NULL and the kernel jumps to address 0. Should we add a check
> for that?
>
> Bug report:
> https://bugzilla.redhat.com/show_bug.cgi?id=306801
Every protocol should provide a non-NULL ->ioctl() method,
find out which one isn't and fix it.
^ permalink raw reply
* Re: [PATCH] Update get_net_ns_by_pid
From: Serge E. Hallyn @ 2007-09-28 23:13 UTC (permalink / raw)
To: Eric W. Biederman; +Cc: Andrew Morton, Linux Containers, netdev
In-Reply-To: <m1k5qbfo1n.fsf@ebiederm.dsl.xmission.com>
Quoting Eric W. Biederman (ebiederm@xmission.com):
>
> In the -mm tree the rules for access an nsproxy have changed,
> and in get_net_ns_by_pid we access the nsproxy, so update
> it to follow the new rules.
>
> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Yup, looks right.
I assume Pavel's Acked-by would actually matter, but still
Acked-by: Serge Hallyn <serue@us.ibm.com>
thanks,
-serge
> ---
>
> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
> index 739fbad..1caba10 100644
> --- a/net/core/rtnetlink.c
> +++ b/net/core/rtnetlink.c
> @@ -746,10 +746,10 @@ static struct net *get_net_ns_by_pid(pid_t pid)
> rcu_read_lock();
> tsk = find_task_by_pid(pid);
> if (tsk) {
> - task_lock(tsk);
> - if (tsk->nsproxy)
> - net = get_net(tsk->nsproxy->net_ns);
> - task_unlock(tsk);
> + struct nsproxy *nsproxy;
> + nsproxy = task_nsproxy(tsk);
> + if (nsproxy)
> + net = get_net(nsproxy->net_ns);
> }
> rcu_read_unlock();
> return net;
> _______________________________________________
> Containers mailing list
> Containers@lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/containers
^ permalink raw reply
* Re: Null dereference in socket.c
From: Stephen Hemminger @ 2007-09-28 23:42 UTC (permalink / raw)
To: David Miller; +Cc: cebbert, netdev
In-Reply-To: <20070928.160057.132434706.davem@davemloft.net>
On Fri, 28 Sep 2007 16:00:57 -0700 (PDT)
David Miller <davem@davemloft.net> wrote:
> From: Chuck Ebbert <cebbert@redhat.com>
> Date: Fri, 28 Sep 2007 18:58:36 -0400
>
> > After debugging an oops (https://bugzilla.redhat.com/attachment.cgi?id=209231)
> > I find it happens here in socket.c::sock_ioctl() line 902:
> >
> > default:
> > =============> err = sock->ops->ioctl(sock, cmd, arg);
> >
> > /*
> > * If this ioctl is unknown try to hand it down
> > * to the NIC driver.
> > */
> > if (err == -ENOIOCTLCMD)
> > err = dev_ioctl(cmd, argp);
> > break;
> >
> >
> > ioctl is NULL and the kernel jumps to address 0. Should we add a check
> > for that?
> >
> > Bug report:
> > https://bugzilla.redhat.com/show_bug.cgi?id=306801
>
> Every protocol should provide a non-NULL ->ioctl() method,
> find out which one isn't and fix it
Auditing the net-2.6.24 tree all instances found by cscope are safe.
--
Stephen Hemminger <shemminger@linux-foundation.org>
^ permalink raw reply
* Re: [IPV6] Fix ICMPv6 redirect handling with target multicast address
From: David Stevens @ 2007-09-29 0:50 UTC (permalink / raw)
To: Brian Haley
Cc: David Miller, netdev@vger.kernel.org, netdev-owner,
YOSHIFUJI Hideaki
In-Reply-To: <46FD2B37.6040502@hp.com>
Brian,
A multicast address should never be the target of a neighbor
discovery request; the sender should use the mapping function for all
multicasts. So, I'm not sure that your example can ever happen, and it
certainly is ok to send ICMPv6 errors to multicast addresses in general.
But I don't see that it hurts anything. either (since it should never
happen :-)),
so I don't particularly object, either.
I think it'd also be better if you add the check to be:
if (ipv6_addr_type(target) &
(IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST))
or something along those lines, rather than reproducing ipv6_addr_type()
code
separately in a new ipv6_addr_linklocal() function.
+-DLS
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox