* [GIT PULL net-2.6] vhost-net: access_ok fix
From: Michael S. Tsirkin @ 2010-10-19 14:59 UTC (permalink / raw)
To: David Miller; +Cc: kvm, virtualization, netdev, linux-kernel
David,
Not sure if it's too late for 2.6.36 - in case it's not, the following tree
includes a last minute bugfix for vhost-net, found by code inspection.
It is on top of net-2.6.
Thanks!
The following changes since commit b0057c51db66c5f0f38059f242c57d61c4741d89:
tg3: restore rx_dropped accounting (2010-10-11 16:06:24 -0700)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git vhost-net
Dan Carpenter (1):
vhost: fix return code for log_access_ok()
drivers/vhost/vhost.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
--
MST
^ permalink raw reply
* Re: [PATCH] Fixed race condition at ip_vs.ko module init.
From: Simon Horman @ 2010-10-19 15:23 UTC (permalink / raw)
To: Eduardo Blanco
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Patrick McHardy
In-Reply-To: <AANLkTinVcSK0Dqx4Pkp1dZskR8YmkGYkkrSNgDqX6psW@mail.gmail.com>
On Tue, Oct 19, 2010 at 10:26:47AM +0100, Eduardo Blanco wrote:
> Lists were initialized after the module was registered. Multiple ipvsadm
> processes at module load triggered a race condition that resulted in a null
> pointer dereference in do_ip_vs_get_ctl(). As a result, __ip_vs_mutex
> was left locked preventing all further ipvsadm commands.
>
> Signed-off-by: Eduardo J. Blanco <ejblanco@google.com>
Thanks Eduardo.
Patrick, please consider pulling
git://git.kernel.org/pub/scm/linux/kernel/git/horms/lvs-test-2.6.git for-patrick
to get this change.
> ---
> net/netfilter/ipvs/ip_vs_ctl.c | 19 ++++++++++---------
> 1 files changed, 10 insertions(+), 9 deletions(-)
>
> diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
> index 0f0c079..68624dc 100644
> --- a/net/netfilter/ipvs/ip_vs_ctl.c
> +++ b/net/netfilter/ipvs/ip_vs_ctl.c
> @@ -3385,6 +3385,16 @@ int __init ip_vs_control_init(void)
>
> EnterFunction(2);
>
> + /* Initialize ip_vs_svc_table, ip_vs_svc_fwm_table, ip_vs_rtable */
> + for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
> + INIT_LIST_HEAD(&ip_vs_svc_table[idx]);
> + INIT_LIST_HEAD(&ip_vs_svc_fwm_table[idx]);
> + }
> + for(idx = 0; idx < IP_VS_RTAB_SIZE; idx++) {
> + INIT_LIST_HEAD(&ip_vs_rtable[idx]);
> + }
> + smp_wmb();
> +
> ret = nf_register_sockopt(&ip_vs_sockopts);
> if (ret) {
> pr_err("cannot register sockopt.\n");
> @@ -3403,15 +3413,6 @@ int __init ip_vs_control_init(void)
>
> sysctl_header = register_sysctl_paths(net_vs_ctl_path, vs_vars);
>
> - /* Initialize ip_vs_svc_table, ip_vs_svc_fwm_table, ip_vs_rtable */
> - for(idx = 0; idx < IP_VS_SVC_TAB_SIZE; idx++) {
> - INIT_LIST_HEAD(&ip_vs_svc_table[idx]);
> - INIT_LIST_HEAD(&ip_vs_svc_fwm_table[idx]);
> - }
> - for(idx = 0; idx < IP_VS_RTAB_SIZE; idx++) {
> - INIT_LIST_HEAD(&ip_vs_rtable[idx]);
> - }
> -
> ip_vs_new_estimator(&ip_vs_stats);
>
> /* Hook the defense timer */
> --
> 1.7.1
>
^ permalink raw reply
* Re: [PATCH v13 10/16] Add a hook to intercept external buffers from NIC driver.
From: David Miller @ 2010-10-19 15:24 UTC (permalink / raw)
To: xiaohui.xin; +Cc: netdev, kvm, linux-kernel, mst, mingo, herbert, jdike
In-Reply-To: <33f51c5eddec843260451eccbfa3c5fdea13b479.1287132437.git.xiaohui.xin@intel.com>
From: xiaohui.xin@intel.com
Date: Fri, 15 Oct 2010 17:12:11 +0800
> @@ -2891,6 +2922,11 @@ static int __netif_receive_skb(struct sk_buff *skb)
> ncls:
> #endif
>
> + /* To intercept mediate passthru(zero-copy) packets here */
> + skb = handle_mpassthru(skb, &pt_prev, &ret, orig_dev);
> + if (!skb)
> + goto out;
> +
> /* Handle special case of bridge or macvlan */
> rx_handler = rcu_dereference(skb->dev->rx_handler);
> if (rx_handler) {
If you consume the packet here, devices in passthru mode cannot
be use with bonding.
But there is nothing that prevents a bond being created with such
a device.
So we have to either prevent such configurations (bad) or make
it work somehow (good) :-)
^ permalink raw reply
* [net-next 0/3] bnx2x: patch series
From: Dmitry Kravkov @ 2010-10-19 15:12 UTC (permalink / raw)
To: davem; +Cc: netdev, Vladislav Zolotarov, Eilon Greenstein
Hi Dave,
Please consider applying this patch series to net-next.
It contains fixes related to 57710 and 57711 HW.
Thanks,
Dmitry
^ permalink raw reply
* [net-next 3/3] bnx2x: update version to 1.60.00-3
From: Dmitry Kravkov @ 2010-10-19 15:13 UTC (permalink / raw)
To: davem, netdev; +Cc: Vladislav Zolotarov, Eilon Greenstein
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/bnx2x/bnx2x.h | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bnx2x/bnx2x.h b/drivers/net/bnx2x/bnx2x.h
index 01b8d85..3bf236b 100644
--- a/drivers/net/bnx2x/bnx2x.h
+++ b/drivers/net/bnx2x/bnx2x.h
@@ -20,8 +20,8 @@
* (you will need to reboot afterwards) */
/* #define BNX2X_STOP_ON_ERROR */
-#define DRV_MODULE_VERSION "1.60.00-2"
-#define DRV_MODULE_RELDATE "2010/10/18"
+#define DRV_MODULE_VERSION "1.60.00-3"
+#define DRV_MODULE_RELDATE "2010/10/19"
#define BNX2X_BC_VER 0x040200
#if defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)
--
1.7.1
^ permalink raw reply related
* [net-next 1/3] bnx2x: fix possible deadlock in HC hw block
From: Dmitry Kravkov @ 2010-10-19 15:13 UTC (permalink / raw)
To: davem, netdev; +Cc: Vladislav Zolotarov, Eilon Greenstein
The possible deadlock (on 57710 devices only) will prevent from the
device to generate interrupts.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/bnx2x/bnx2x_main.c | 37 +++++++++++++++++++++++++++++--------
1 files changed, 29 insertions(+), 8 deletions(-)
diff --git a/drivers/net/bnx2x/bnx2x_main.c b/drivers/net/bnx2x/bnx2x_main.c
index 012c093..3f49b55 100644
--- a/drivers/net/bnx2x/bnx2x_main.c
+++ b/drivers/net/bnx2x/bnx2x_main.c
@@ -1111,14 +1111,19 @@ static void bnx2x_hc_int_enable(struct bnx2x *bp)
HC_CONFIG_0_REG_INT_LINE_EN_0 |
HC_CONFIG_0_REG_ATTN_BIT_EN_0);
- DP(NETIF_MSG_INTR, "write %x to HC %d (addr 0x%x)\n",
- val, port, addr);
+ if (!CHIP_IS_E1(bp)) {
+ DP(NETIF_MSG_INTR, "write %x to HC %d (addr 0x%x)\n",
+ val, port, addr);
- REG_WR(bp, addr, val);
+ REG_WR(bp, addr, val);
- val &= ~HC_CONFIG_0_REG_MSI_MSIX_INT_EN_0;
+ val &= ~HC_CONFIG_0_REG_MSI_MSIX_INT_EN_0;
+ }
}
+ if (CHIP_IS_E1(bp))
+ REG_WR(bp, HC_REG_INT_MASK + port*4, 0x1FFFF);
+
DP(NETIF_MSG_INTR, "write %x to HC %d (addr 0x%x) mode %s\n",
val, port, addr, (msix ? "MSI-X" : (msi ? "MSI" : "INTx")));
@@ -1212,10 +1217,26 @@ static void bnx2x_hc_int_disable(struct bnx2x *bp)
u32 addr = port ? HC_REG_CONFIG_1 : HC_REG_CONFIG_0;
u32 val = REG_RD(bp, addr);
- val &= ~(HC_CONFIG_0_REG_SINGLE_ISR_EN_0 |
- HC_CONFIG_0_REG_MSI_MSIX_INT_EN_0 |
- HC_CONFIG_0_REG_INT_LINE_EN_0 |
- HC_CONFIG_0_REG_ATTN_BIT_EN_0);
+ /*
+ * in E1 we must use only PCI configuration space to disable
+ * MSI/MSIX capablility
+ * It's forbitten to disable IGU_PF_CONF_MSI_MSIX_EN in HC block
+ */
+ if (CHIP_IS_E1(bp)) {
+ /* Since IGU_PF_CONF_MSI_MSIX_EN still always on
+ * Use mask register to prevent from HC sending interrupts
+ * after we exit the function
+ */
+ REG_WR(bp, HC_REG_INT_MASK + port*4, 0);
+
+ val &= ~(HC_CONFIG_0_REG_SINGLE_ISR_EN_0 |
+ HC_CONFIG_0_REG_INT_LINE_EN_0 |
+ HC_CONFIG_0_REG_ATTN_BIT_EN_0);
+ } else
+ val &= ~(HC_CONFIG_0_REG_SINGLE_ISR_EN_0 |
+ HC_CONFIG_0_REG_MSI_MSIX_INT_EN_0 |
+ HC_CONFIG_0_REG_INT_LINE_EN_0 |
+ HC_CONFIG_0_REG_ATTN_BIT_EN_0);
DP(NETIF_MSG_INTR, "write %x to HC %d (addr 0x%x)\n",
val, port, addr);
--
1.7.1
^ permalink raw reply related
* [net-next 2/3] bnx2x: prevent false parity error in MSI-X memory of HC block
From: Dmitry Kravkov @ 2010-10-19 15:13 UTC (permalink / raw)
To: davem, netdev; +Cc: Vladislav Zolotarov, Eilon Greenstein
From: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/bnx2x/bnx2x_main.c | 28 +++++++++++++++++++++++++++-
drivers/net/bnx2x/bnx2x_reg.h | 6 +++++-
2 files changed, 32 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bnx2x/bnx2x_main.c b/drivers/net/bnx2x/bnx2x_main.c
index 3f49b55..f22e283 100644
--- a/drivers/net/bnx2x/bnx2x_main.c
+++ b/drivers/net/bnx2x/bnx2x_main.c
@@ -5457,7 +5457,8 @@ static int bnx2x_init_hw_func(struct bnx2x *bp)
struct bnx2x_ilt *ilt = BP_ILT(bp);
u16 cdu_ilt_start;
u32 addr, val;
- int i;
+ u32 main_mem_base, main_mem_size, main_mem_prty_clr;
+ int i, main_mem_width;
DP(BNX2X_MSG_MCP, "starting func init func %d\n", func);
@@ -5706,6 +5707,31 @@ static int bnx2x_init_hw_func(struct bnx2x *bp)
bnx2x_init_block(bp, MCP_BLOCK, FUNC0_STAGE + func);
bnx2x_init_block(bp, DMAE_BLOCK, FUNC0_STAGE + func);
+ if (CHIP_IS_E1x(bp)) {
+ main_mem_size = HC_REG_MAIN_MEMORY_SIZE / 2; /*dwords*/
+ main_mem_base = HC_REG_MAIN_MEMORY +
+ BP_PORT(bp) * (main_mem_size * 4);
+ main_mem_prty_clr = HC_REG_HC_PRTY_STS_CLR;
+ main_mem_width = 8;
+
+ val = REG_RD(bp, main_mem_prty_clr);
+ if (val)
+ DP(BNX2X_MSG_MCP, "Hmmm... Parity errors in HC "
+ "block during "
+ "function init (0x%x)!\n", val);
+
+ /* Clear "false" parity errors in MSI-X table */
+ for (i = main_mem_base;
+ i < main_mem_base + main_mem_size * 4;
+ i += main_mem_width) {
+ bnx2x_read_dmae(bp, i, main_mem_width / 4);
+ bnx2x_write_dmae(bp, bnx2x_sp_mapping(bp, wb_data),
+ i, main_mem_width / 4);
+ }
+ /* Clear HC parity attention */
+ REG_RD(bp, main_mem_prty_clr);
+ }
+
bnx2x_phy_probe(&bp->link_params);
return 0;
diff --git a/drivers/net/bnx2x/bnx2x_reg.h b/drivers/net/bnx2x/bnx2x_reg.h
index 18a8628..1cefe48 100644
--- a/drivers/net/bnx2x/bnx2x_reg.h
+++ b/drivers/net/bnx2x/bnx2x_reg.h
@@ -800,9 +800,13 @@
#define HC_REG_HC_PRTY_MASK 0x1080a0
/* [R 3] Parity register #0 read */
#define HC_REG_HC_PRTY_STS 0x108094
-#define HC_REG_INT_MASK 0x108108
+/* [RC 3] Parity register #0 read clear */
+#define HC_REG_HC_PRTY_STS_CLR 0x108098
+#define HC_REG_INT_MASK 0x108108
#define HC_REG_LEADING_EDGE_0 0x108040
#define HC_REG_LEADING_EDGE_1 0x108048
+#define HC_REG_MAIN_MEMORY 0x108800
+#define HC_REG_MAIN_MEMORY_SIZE 152
#define HC_REG_P0_PROD_CONS 0x108200
#define HC_REG_P1_PROD_CONS 0x108400
#define HC_REG_PBA_COMMAND 0x108140
--
1.7.1
^ permalink raw reply related
* Re: [PATCH] SIW: Documentation (initial)
From: Bernard Metzler @ 2010-10-19 15:36 UTC (permalink / raw)
To: Randy Dunlap
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, netdev-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <20101014155703.3d4b5d71.randy.dunlap-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Randy,
...back from vacation.
Many thanks! I'll take it all over.
Bernard.
Randy Dunlap <randy.dunlap-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote on 10/15/2010 12:57:03 AM:
<snip>
> > +
> > +User Interface
> > +--------------
> > +All fast path operations such as posting of work requests and
> > +reaping of work completions currently involve a system call into
> > +the siw module. Kernel/user-mapped send and receive as well as
>
> I didn't find the system call(s). Are they new syscalls or just
> (socket) reads/writes? (I was probably looking for new syscalls.)
>
I will have to clarify. Currently all operations are using the
infiniband/core infrastructure (e.g. via uverbs write file
operation). There is no private interface between libsiw and
siw kernel module in place.
<snip>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* Re: [net-next 0/3] bnx2x: patch series
From: David Miller @ 2010-10-19 15:38 UTC (permalink / raw)
To: dmitry; +Cc: netdev, vladz, eilong
In-Reply-To: <1287501178.740.37.camel@lb-tlvb-dmitry>
From: "Dmitry Kravkov" <dmitry@broadcom.com>
Date: Tue, 19 Oct 2010 17:12:58 +0200
> Please consider applying this patch series to net-next.
> It contains fixes related to 57710 and 57711 HW.
All applied, thanks.
^ permalink raw reply
* Re: [PATCH] bridge: make br_parse_ip_options static
From: Bandan Das @ 2010-10-19 16:03 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: David Miller, netdev
In-Reply-To: <20101019075503.057dd8d6@nehalam>
> >
> > My main motivation behind not making this static was that
> > there would be possibly other places in the bridge code
> > (besides br_netfilter.c) where we enter the IP stack and might
> > want to call this. Not sure if it's indeed the case though..
> >
>
> I checked by doing make allmodconfig as well as looking by
> git grep 'br_parse_ip_options'
>
> --
Sorry, my wording was misleading :) What I meant was at this time,
it's only used in br_netfilter.c but may be sometime in the future,
it could possible be used in some other places where we enter the IP
stack from the bridge code.
But, I completely agree with you. It makes more sense to make it static
as of now.
Bandan
^ permalink raw reply
* [PATCH v2.6.36-rc7] net/neighbour: cancel_delayed_work() + flush_scheduled_work() -> cancel_delayed_work_sync()
From: Tejun Heo @ 2010-10-19 16:04 UTC (permalink / raw)
To: David S. Miller, netdev@vger.kernel.org, lkml
flush_scheduled_work() is going away. Prepare for it.
Signed-off-by: Tejun Heo <tj@kernel.org>
---
net/core/neighbour.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
Index: work/net/core/neighbour.c
===================================================================
--- work.orig/net/core/neighbour.c
+++ work/net/core/neighbour.c
@@ -1486,8 +1486,7 @@ int neigh_table_clear(struct neigh_table
struct neigh_table **tp;
/* It is not clean... Fix it to unload IPv6 module safely */
- cancel_delayed_work(&tbl->gc_work);
- flush_scheduled_work();
+ cancel_delayed_work_sync(&tbl->gc_work);
del_timer_sync(&tbl->proxy_timer);
pneigh_queue_purge(&tbl->proxy_queue);
neigh_ifdown(tbl, NULL);
^ permalink raw reply
* Future of the Wimedia LLC Protocol (WLP) subsystem/drivers
From: David Vrabel @ 2010-10-19 16:30 UTC (permalink / raw)
To: netdev
Hi,
I've have been nominally the maintainer of the Wimedia LLC Protocol
(WLP) subsystem and driver since it was originally submitted. I am no
longer in a position to even pretend to be a maintainer.
The only usable hardware was an Intel i1480 devices with beta firmware
that was never released as a product. Intel have since sold all there
UWB/WLP IP and I see little prospect of there ever being hardware
commercially available for WLP.
Here are a number of options:
1. Someone else maintains it. Any volunteers?
2. It gets labelled as Orphaned in MAINTAINERS.
3. It gets moved to staging.
4, It gets removed.
If no one says anything I'll submit a patch to Linus to mark it as Orphaned.
David
--
David Vrabel, Senior Software Engineer, Drivers
CSR, Churchill House, Cambridge Business Park, Tel: +44 (0)1223 692562
Cowley Road, Cambridge, CB4 0WZ http://www.csr.com/
Member of the CSR plc group of companies. CSR plc registered in England and Wales, registered number 4187346, registered office Churchill House, Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
^ permalink raw reply
* [PATCH net-next] 9p: client code cleanup
From: Stephen Hemminger @ 2010-10-19 16:48 UTC (permalink / raw)
To: Eric Van Hensbergen, Ron Minnich, Latchesar Ionkov, David Miller
Cc: v9fs-developer, netdev
Make p9_client_version static since only used in one file.
Remove p9_client_auth because it is defined but never used.
Compile tested only.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
---
include/net/9p/client.h | 3 ---
net/9p/client.c | 6 ++----
2 files changed, 2 insertions(+), 7 deletions(-)
--- a/include/net/9p/client.h 2010-10-19 09:23:32.679670429 -0700
+++ b/include/net/9p/client.h 2010-10-19 09:23:34.483850813 -0700
@@ -212,15 +212,12 @@ struct p9_dirent {
int p9_client_statfs(struct p9_fid *fid, struct p9_rstatfs *sb);
int p9_client_rename(struct p9_fid *fid, struct p9_fid *newdirfid, char *name);
-int p9_client_version(struct p9_client *);
struct p9_client *p9_client_create(const char *dev_name, char *options);
void p9_client_destroy(struct p9_client *clnt);
void p9_client_disconnect(struct p9_client *clnt);
void p9_client_begin_disconnect(struct p9_client *clnt);
struct p9_fid *p9_client_attach(struct p9_client *clnt, struct p9_fid *afid,
char *uname, u32 n_uname, char *aname);
-struct p9_fid *p9_client_auth(struct p9_client *clnt, char *uname,
- u32 n_uname, char *aname);
struct p9_fid *p9_client_walk(struct p9_fid *oldfid, int nwname, char **wnames,
int clone);
int p9_client_open(struct p9_fid *fid, int mode);
--- a/net/9p/client.c 2010-10-19 09:23:32.715674029 -0700
+++ b/net/9p/client.c 2010-10-19 09:24:16.184019417 -0700
@@ -671,7 +671,7 @@ static void p9_fid_destroy(struct p9_fid
kfree(fid);
}
-int p9_client_version(struct p9_client *c)
+static int p9_client_version(struct p9_client *c)
{
int err = 0;
struct p9_req_t *req;
@@ -730,7 +730,6 @@ error:
return err;
}
-EXPORT_SYMBOL(p9_client_version);
struct p9_client *p9_client_create(const char *dev_name, char *options)
{
@@ -887,54 +886,6 @@ error:
}
EXPORT_SYMBOL(p9_client_attach);
-struct p9_fid *
-p9_client_auth(struct p9_client *clnt, char *uname, u32 n_uname, char *aname)
-{
- int err;
- struct p9_req_t *req;
- struct p9_qid qid;
- struct p9_fid *afid;
-
- P9_DPRINTK(P9_DEBUG_9P, ">>> TAUTH uname %s aname %s\n", uname, aname);
- err = 0;
-
- afid = p9_fid_create(clnt);
- if (IS_ERR(afid)) {
- err = PTR_ERR(afid);
- afid = NULL;
- goto error;
- }
-
- req = p9_client_rpc(clnt, P9_TAUTH, "dss?d",
- afid ? afid->fid : P9_NOFID, uname, aname, n_uname);
- if (IS_ERR(req)) {
- err = PTR_ERR(req);
- goto error;
- }
-
- err = p9pdu_readf(req->rc, clnt->proto_version, "Q", &qid);
- if (err) {
- p9pdu_dump(1, req->rc);
- p9_free_req(clnt, req);
- goto error;
- }
-
- P9_DPRINTK(P9_DEBUG_9P, "<<< RAUTH qid %x.%llx.%x\n",
- qid.type,
- (unsigned long long)qid.path,
- qid.version);
-
- memmove(&afid->qid, &qid, sizeof(struct p9_qid));
- p9_free_req(clnt, req);
- return afid;
-
-error:
- if (afid)
- p9_fid_destroy(afid);
- return ERR_PTR(err);
-}
-EXPORT_SYMBOL(p9_client_auth);
-
struct p9_fid *p9_client_walk(struct p9_fid *oldfid, int nwname, char **wnames,
int clone)
{
^ permalink raw reply
* [PATCH net-next-2.6] be2net: Changes to use only priority codes allowed by f/w
From: Somnath Kotur @ 2010-10-19 8:51 UTC (permalink / raw)
To: netdev
Changes to use one of the priority codes allowed by CNA f/w for NIC traffic
from host. The driver gets the bit map of the priority codes allowed for
host traffic through a asynchronous message from the f/w that the driver
subscribes to.
Signed-off-by: Somnath Kotur <somnath.kotur@emulex.com>
---
drivers/net/benet/be.h | 2 +
drivers/net/benet/be_cmds.c | 60 +++++++++++++++++++++++++++++++++++++++++-
drivers/net/benet/be_cmds.h | 31 ++++++++++++++++++++++
drivers/net/benet/be_main.c | 21 ++++++++++-----
4 files changed, 105 insertions(+), 9 deletions(-)
diff --git a/drivers/net/benet/be.h b/drivers/net/benet/be.h
index 1afabb1..7ff66fb 100644
--- a/drivers/net/benet/be.h
+++ b/drivers/net/benet/be.h
@@ -264,6 +264,8 @@ struct be_adapter {
u16 vlans_added;
u16 max_vlans; /* Number of vlans supported */
u8 vlan_tag[VLAN_GROUP_ARRAY_LEN];
+ u8 vlan_prio_bmap; /* Available priority BitMap */
+ u16 recommended_prio; /* Recommended Priority */
struct be_dma_mem mc_cmd_mem;
struct be_dma_mem stats_cmd;
diff --git a/drivers/net/benet/be_cmds.c b/drivers/net/benet/be_cmds.c
index bf2dc26..b1728db 100644
--- a/drivers/net/benet/be_cmds.c
+++ b/drivers/net/benet/be_cmds.c
@@ -96,6 +96,50 @@ static void be_async_link_state_process(struct be_adapter *adapter,
evt->port_link_status == ASYNC_EVENT_LINK_UP);
}
+/* Grp5 CoS Priority evt */
+static void be_async_grp5_cos_priority_process(struct be_adapter *adapter,
+ struct be_async_event_grp5_cos_priority *evt)
+{
+ if (evt->valid) {
+ adapter->vlan_prio_bmap = evt->available_priority_bmap;
+ adapter->recommended_prio =
+ evt->reco_default_priority << VLAN_PRIO_SHIFT;
+ }
+}
+
+/* Grp5 QOS Speed evt */
+static void be_async_grp5_qos_speed_process(struct be_adapter *adapter,
+ struct be_async_event_grp5_qos_link_speed *evt)
+{
+ if (evt->physical_port == adapter->port_num) {
+ /* qos_link_speed is in units of 10 Mbps */
+ adapter->link_speed = evt->qos_link_speed * 10;
+ }
+}
+
+static void be_async_grp5_evt_process(struct be_adapter *adapter,
+ u32 trailer, struct be_mcc_compl *evt)
+{
+ u8 event_type = 0;
+
+ event_type = (trailer >> ASYNC_TRAILER_EVENT_TYPE_SHIFT) &
+ ASYNC_TRAILER_EVENT_TYPE_MASK;
+
+ switch (event_type) {
+ case ASYNC_EVENT_COS_PRIORITY:
+ be_async_grp5_cos_priority_process(adapter,
+ (struct be_async_event_grp5_cos_priority *)evt);
+ break;
+ case ASYNC_EVENT_QOS_SPEED:
+ be_async_grp5_qos_speed_process(adapter,
+ (struct be_async_event_grp5_qos_link_speed *)evt);
+ break;
+ default:
+ dev_warn(&adapter->pdev->dev, "Unknown grp5 event!\n");
+ break;
+ }
+}
+
static inline bool is_link_state_evt(u32 trailer)
{
return ((trailer >> ASYNC_TRAILER_EVENT_CODE_SHIFT) &
@@ -103,6 +147,13 @@ static inline bool is_link_state_evt(u32 trailer)
ASYNC_EVENT_CODE_LINK_STATE;
}
+static inline bool is_grp5_evt(u32 trailer)
+{
+ return (((trailer >> ASYNC_TRAILER_EVENT_CODE_SHIFT) &
+ ASYNC_TRAILER_EVENT_CODE_MASK) ==
+ ASYNC_EVENT_CODE_GRP_5);
+}
+
static struct be_mcc_compl *be_mcc_compl_get(struct be_adapter *adapter)
{
struct be_queue_info *mcc_cq = &adapter->mcc_obj.cq;
@@ -143,6 +194,9 @@ int be_process_mcc(struct be_adapter *adapter, int *status)
if (is_link_state_evt(compl->flags))
be_async_link_state_process(adapter,
(struct be_async_event_link_state *) compl);
+ else if (is_grp5_evt(compl->flags))
+ be_async_grp5_evt_process(adapter,
+ compl->flags, compl);
} else if (compl->flags & CQE_FLAGS_COMPLETED_MASK) {
*status = be_mcc_compl_process(adapter, compl);
atomic_dec(&mcc_obj->q.used);
@@ -677,10 +731,10 @@ int be_cmd_mccq_create(struct be_adapter *adapter,
ctxt = &req->context;
be_wrb_hdr_prepare(wrb, sizeof(*req), true, 0,
- OPCODE_COMMON_MCC_CREATE);
+ OPCODE_COMMON_MCC_CREATE_EXT);
be_cmd_hdr_prepare(&req->hdr, CMD_SUBSYSTEM_COMMON,
- OPCODE_COMMON_MCC_CREATE, sizeof(*req));
+ OPCODE_COMMON_MCC_CREATE_EXT, sizeof(*req));
req->num_pages = cpu_to_le16(PAGES_4K_SPANNED(q_mem->va, q_mem->size));
@@ -688,6 +742,8 @@ int be_cmd_mccq_create(struct be_adapter *adapter,
AMAP_SET_BITS(struct amap_mcc_context, ring_size, ctxt,
be_encoded_q_len(mccq->len));
AMAP_SET_BITS(struct amap_mcc_context, cq_id, ctxt, cq->id);
+ /* Subscribe to Link State and Group 5 Events(bits 1 and 5 set) */
+ req->async_event_bitmap[0] |= 0x00000022;
be_dws_cpu_to_le(ctxt, sizeof(req->context));
diff --git a/drivers/net/benet/be_cmds.h b/drivers/net/benet/be_cmds.h
index b7a40b1..93c721a 100644
--- a/drivers/net/benet/be_cmds.h
+++ b/drivers/net/benet/be_cmds.h
@@ -82,7 +82,12 @@ struct be_mcc_compl {
*/
#define ASYNC_TRAILER_EVENT_CODE_SHIFT 8 /* bits 8 - 15 */
#define ASYNC_TRAILER_EVENT_CODE_MASK 0xFF
+#define ASYNC_TRAILER_EVENT_TYPE_SHIFT 16 /* bits 16 - 23 */
+#define ASYNC_TRAILER_EVENT_TYPE_MASK 0xFF
#define ASYNC_EVENT_CODE_LINK_STATE 0x1
+#define ASYNC_EVENT_CODE_GRP_5 0x5
+#define ASYNC_EVENT_QOS_SPEED 0x1
+#define ASYNC_EVENT_COS_PRIORITY 0x2
struct be_async_event_trailer {
u32 code;
};
@@ -105,6 +110,30 @@ struct be_async_event_link_state {
struct be_async_event_trailer trailer;
} __packed;
+/* When the event code of an async trailer is GRP-5 and event_type is QOS_SPEED
+ * the mcc_compl must be interpreted as follows
+ */
+struct be_async_event_grp5_qos_link_speed {
+ u8 physical_port;
+ u8 rsvd[5];
+ u16 qos_link_speed;
+ u32 event_tag;
+ struct be_async_event_trailer trailer;
+} __packed;
+
+/* When the event code of an async trailer is GRP5 and event type is
+ * CoS-Priority, the mcc_compl must be interpreted as follows
+ */
+struct be_async_event_grp5_cos_priority {
+ u8 physical_port;
+ u8 available_priority_bmap;
+ u8 reco_default_priority;
+ u8 valid;
+ u8 rsvd0;
+ u8 event_tag;
+ struct be_async_event_trailer trailer;
+} __packed;
+
struct be_mcc_mailbox {
struct be_mcc_wrb wrb;
struct be_mcc_compl compl;
@@ -125,6 +154,7 @@ struct be_mcc_mailbox {
#define OPCODE_COMMON_EQ_CREATE 13
#define OPCODE_COMMON_MCC_CREATE 21
#define OPCODE_COMMON_SET_QOS 28
+#define OPCODE_COMMON_MCC_CREATE_EXT 90
#define OPCODE_COMMON_SEEPROM_READ 30
#define OPCODE_COMMON_NTWK_RX_FILTER 34
#define OPCODE_COMMON_GET_FW_VERSION 35
@@ -338,6 +368,7 @@ struct be_cmd_req_mcc_create {
struct be_cmd_req_hdr hdr;
u16 num_pages;
u16 rsvd0;
+ u32 async_event_bitmap[1];
u8 context[sizeof(struct amap_mcc_context) / 8];
struct phys_addr pages[8];
} __packed;
diff --git a/drivers/net/benet/be_main.c b/drivers/net/benet/be_main.c
index 9a1cd28..5374be1 100644
--- a/drivers/net/benet/be_main.c
+++ b/drivers/net/benet/be_main.c
@@ -429,9 +429,12 @@ static inline void wrb_fill(struct be_eth_wrb *wrb, u64 addr, int len)
wrb->frag_len = len & ETH_WRB_FRAG_LEN_MASK;
}
-static void wrb_fill_hdr(struct be_eth_hdr_wrb *hdr, struct sk_buff *skb,
- bool vlan, u32 wrb_cnt, u32 len)
+static void wrb_fill_hdr(struct be_adapter *adapter, struct be_eth_hdr_wrb *hdr,
+ struct sk_buff *skb, u32 wrb_cnt, u32 len)
{
+ u8 vlan_prio = 0;
+ u16 vlan_tag = 0;
+
memset(hdr, 0, sizeof(*hdr));
AMAP_SET_BITS(struct amap_eth_hdr_wrb, crc, hdr, 1);
@@ -449,10 +452,15 @@ static void wrb_fill_hdr(struct be_eth_hdr_wrb *hdr, struct sk_buff *skb,
AMAP_SET_BITS(struct amap_eth_hdr_wrb, udpcs, hdr, 1);
}
- if (vlan && vlan_tx_tag_present(skb)) {
+ if (adapter->vlan_grp && vlan_tx_tag_present(skb)) {
AMAP_SET_BITS(struct amap_eth_hdr_wrb, vlan, hdr, 1);
- AMAP_SET_BITS(struct amap_eth_hdr_wrb, vlan_tag,
- hdr, vlan_tx_tag_get(skb));
+ vlan_tag = vlan_tx_tag_get(skb);
+ vlan_prio = (vlan_tag & VLAN_PRIO_MASK) >> VLAN_PRIO_SHIFT;
+ /* If vlan priority provided by OS is NOT in available bmap */
+ if (!(adapter->vlan_prio_bmap & (1 << vlan_prio)))
+ vlan_tag = (vlan_tag & ~VLAN_PRIO_MASK) |
+ adapter->recommended_prio;
+ AMAP_SET_BITS(struct amap_eth_hdr_wrb, vlan_tag, hdr, vlan_tag);
}
AMAP_SET_BITS(struct amap_eth_hdr_wrb, event, hdr, 1);
@@ -532,8 +540,7 @@ static int make_tx_wrbs(struct be_adapter *adapter,
queue_head_inc(txq);
}
- wrb_fill_hdr(hdr, first_skb, adapter->vlan_grp ? true : false,
- wrb_cnt, copied);
+ wrb_fill_hdr(adapter, hdr, first_skb, wrb_cnt, copied);
be_dws_cpu_to_le(hdr, sizeof(*hdr));
return copied;
--
1.5.6.1
^ permalink raw reply related
* [RFC net-next] caif: code cleanup
From: Stephen Hemminger @ 2010-10-19 16:55 UTC (permalink / raw)
To: Sjur Braendeland; +Cc: netdev
Cleanup of new CAIF code.
* make local functions static
* remove code that is never used
* expand get_caif_conf() since wrapper is no longer needed
* make args to comparison functions const
* rename connect_req_to_link_param to keep exported names
consistent
Compile tested only.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
---
include/net/caif/caif_dev.h | 27 +-----
include/net/caif/cfctrl.h | 12 --
include/net/caif/cfmuxl.h | 2
include/net/caif/cfpkt.h | 75 ------------------
include/net/caif/cfsrvl.h | 3
net/caif/caif_config_util.c | 6 -
net/caif/caif_dev.c | 24 +----
net/caif/cfcnfg.c | 2
net/caif/cfctrl.c | 74 +-----------------
net/caif/cfmuxl.c | 35 --------
net/caif/cfpkt_skbuff.c | 178 +-------------------------------------------
net/caif/cfsrvl.c | 7 -
12 files changed, 29 insertions(+), 416 deletions(-)
--- a/include/net/caif/caif_dev.h 2010-10-19 09:25:53.581751379 -0700
+++ b/include/net/caif/caif_dev.h 2010-10-19 09:53:00.077792212 -0700
@@ -74,19 +74,8 @@ int caif_connect_client(struct caif_conn
int caif_disconnect_client(struct cflayer *client_layer);
/**
- * caif_release_client - Release adaptation layer reference to client.
- *
- * @client_layer: Client layer.
- *
- * Releases a client/adaptation layer use of the caif stack.
- * This function must be used after caif_disconnect_client to
- * decrease the reference count of the service layer.
- */
-void caif_release_client(struct cflayer *client_layer);
-
-/**
- * connect_req_to_link_param - Translate configuration parameters
- * from socket format to internal format.
+ * caif_connect_req_to_link_param - Translate configuration parameters
+ * from socket format to internal format.
* @cnfg: Pointer to configuration handler
* @con_req: Configuration parameters supplied in function
* caif_connect_client
@@ -94,14 +83,8 @@ void caif_release_client(struct cflayer
* setting up channels.
*
*/
-int connect_req_to_link_param(struct cfcnfg *cnfg,
- struct caif_connect_request *con_req,
- struct cfctrl_link_param *channel_setup_param);
-
-/**
- * get_caif_conf() - Get the configuration handler.
- */
-struct cfcnfg *get_caif_conf(void);
-
+int caif_connect_req_to_link_param(struct cfcnfg *cnfg,
+ struct caif_connect_request *con_req,
+ struct cfctrl_link_param *channel_setup_param);
#endif /* CAIF_DEV_H_ */
--- a/include/net/caif/cfctrl.h 2010-10-19 09:25:53.709764168 -0700
+++ b/include/net/caif/cfctrl.h 2010-10-19 09:39:11.361244959 -0700
@@ -121,19 +121,9 @@ int cfctrl_linkup_request(struct cflayer
struct cflayer *user_layer);
int cfctrl_linkdown_req(struct cflayer *cfctrl, u8 linkid,
struct cflayer *client);
-void cfctrl_sleep_req(struct cflayer *cfctrl);
-void cfctrl_wake_req(struct cflayer *cfctrl);
-void cfctrl_getstartreason_req(struct cflayer *cfctrl);
+
struct cflayer *cfctrl_create(void);
-void cfctrl_set_dnlayer(struct cflayer *this, struct cflayer *dn);
-void cfctrl_set_uplayer(struct cflayer *this, struct cflayer *up);
struct cfctrl_rsp *cfctrl_get_respfuncs(struct cflayer *layer);
-bool cfctrl_req_eq(struct cfctrl_request_info *r1,
- struct cfctrl_request_info *r2);
-void cfctrl_insert_req(struct cfctrl *ctrl,
- struct cfctrl_request_info *req);
-struct cfctrl_request_info *cfctrl_remove_req(struct cfctrl *ctrl,
- struct cfctrl_request_info *req);
void cfctrl_cancel_req(struct cflayer *layr, struct cflayer *adap_layer);
#endif /* CFCTRL_H_ */
--- a/include/net/caif/cfpkt.h 2010-10-19 09:25:53.785771759 -0700
+++ b/include/net/caif/cfpkt.h 2010-10-19 09:37:32.651429118 -0700
@@ -16,12 +16,6 @@ struct cfpkt;
*/
struct cfpkt *cfpkt_create(u16 len);
-/* Create a CAIF packet.
- * data Data to copy.
- * len Length of packet to be created
- * @return New packet.
- */
-struct cfpkt *cfpkt_create_uplink(const unsigned char *data, unsigned int len);
/*
* Destroy a CAIF Packet.
* pkt Packet to be destoyed.
@@ -181,22 +175,6 @@ u16 cfpkt_iterate(struct cfpkt *pkt,
u16 (*iter_func)(u16 chks, void *buf, u16 len),
u16 data);
-/* Append by giving user access to packet buffer
- * cfpkt Packet to append to
- * buf Buffer inside pkt that user shall copy data into
- * buflen Length of buffer and number of bytes added to packet
- * @return 0 on error, 1 on success
- */
-int cfpkt_raw_append(struct cfpkt *cfpkt, void **buf, unsigned int buflen);
-
-/* Extract by giving user access to packet buffer
- * cfpkt Packet to extract from
- * buf Buffer inside pkt that user shall copy data from
- * buflen Length of buffer and number of bytes removed from packet
- * @return 0 on error, 1 on success
- */
-int cfpkt_raw_extract(struct cfpkt *cfpkt, void **buf, unsigned int buflen);
-
/* Map from a "native" packet (e.g. Linux Socket Buffer) to a CAIF packet.
* dir - Direction indicating whether this packet is to be sent or received.
* nativepkt - The native packet to be transformed to a CAIF packet
@@ -210,59 +188,6 @@ struct cfpkt *cfpkt_fromnative(enum caif
*/
void *cfpkt_tonative(struct cfpkt *pkt);
-/*
- * Insert a packet in the packet queue.
- * pktq Packet queue to insert into
- * pkt Packet to be inserted in queue
- * prio Priority of packet
- */
-void cfpkt_queue(struct cfpktq *pktq, struct cfpkt *pkt,
- unsigned short prio);
-
-/*
- * Remove a packet from the packet queue.
- * pktq Packet queue to fetch packets from.
- * @return Dequeued packet.
- */
-struct cfpkt *cfpkt_dequeue(struct cfpktq *pktq);
-
-/*
- * Peek into a packet from the packet queue.
- * pktq Packet queue to fetch packets from.
- * @return Peeked packet.
- */
-struct cfpkt *cfpkt_qpeek(struct cfpktq *pktq);
-
-/*
- * Initiates the packet queue.
- * @return Pointer to new packet queue.
- */
-struct cfpktq *cfpktq_create(void);
-
-/*
- * Get the number of packets in the queue.
- * pktq Packet queue to fetch count from.
- * @return Number of packets in queue.
- */
-int cfpkt_qcount(struct cfpktq *pktq);
-
-/*
- * Put content of packet into buffer for debuging purposes.
- * pkt Packet to copy data from
- * buf Buffer to copy data into
- * buflen Length of data to copy
- * @return Pointer to copied data
- */
-char *cfpkt_log_pkt(struct cfpkt *pkt, char *buf, int buflen);
-
-/*
- * Clones a packet and releases the original packet.
- * This is used for taking ownership of a packet e.g queueing.
- * pkt Packet to clone and release.
- * @return Cloned packet.
- */
-struct cfpkt *cfpkt_clone_release(struct cfpkt *pkt);
-
/*
* Returns packet information for a packet.
--- a/net/caif/caif_dev.c 2010-10-19 09:26:08.959287331 -0700
+++ b/net/caif/caif_dev.c 2010-10-19 09:53:28.508446823 -0700
@@ -257,7 +257,7 @@ static int caif_device_notify(struct not
break;
}
dev_hold(dev);
- cfcnfg_add_phy_layer(get_caif_conf(),
+ cfcnfg_add_phy_layer(cfg,
phy_type,
dev,
&caifd->layer,
@@ -300,7 +300,7 @@ static int caif_device_notify(struct not
if (atomic_read(&caifd->in_use))
netdev_warn(dev,
"Unregistering an active CAIF device\n");
- cfcnfg_del_phy_layer(get_caif_conf(), &caifd->layer);
+ cfcnfg_del_phy_layer(cfg, &caifd->layer);
dev_put(dev);
atomic_set(&caifd->state, what);
break;
@@ -320,24 +320,18 @@ static struct notifier_block caif_device
.priority = 0,
};
-
-struct cfcnfg *get_caif_conf(void)
-{
- return cfg;
-}
-EXPORT_SYMBOL(get_caif_conf);
-
int caif_connect_client(struct caif_connect_request *conn_req,
struct cflayer *client_layer, int *ifindex,
int *headroom, int *tailroom)
{
struct cfctrl_link_param param;
int ret;
- ret = connect_req_to_link_param(get_caif_conf(), conn_req, ¶m);
+
+ ret = caif_connect_req_to_link_param(cfg, conn_req, ¶m);
if (ret)
return ret;
/* Hook up the adaptation layer. */
- return cfcnfg_add_adaptation_layer(get_caif_conf(), ¶m,
+ return cfcnfg_add_adaptation_layer(cfg, ¶m,
client_layer, ifindex,
headroom, tailroom);
}
@@ -345,16 +339,10 @@ EXPORT_SYMBOL(caif_connect_client);
int caif_disconnect_client(struct cflayer *adap_layer)
{
- return cfcnfg_disconn_adapt_layer(get_caif_conf(), adap_layer);
+ return cfcnfg_disconn_adapt_layer(cfg, adap_layer);
}
EXPORT_SYMBOL(caif_disconnect_client);
-void caif_release_client(struct cflayer *adap_layer)
-{
- cfcnfg_release_adap_layer(adap_layer);
-}
-EXPORT_SYMBOL(caif_release_client);
-
/* Per-namespace Caif devices handling */
static int caif_init_net(struct net *net)
{
--- a/net/caif/cfcnfg.c 2010-10-19 09:24:41.354535051 -0700
+++ b/net/caif/cfcnfg.c 2010-10-19 09:25:17.390135851 -0700
@@ -257,7 +257,7 @@ static void cfcnfg_linkdestroy_rsp(struc
{
}
-int protohead[CFCTRL_SRV_MASK] = {
+static const int protohead[CFCTRL_SRV_MASK] = {
[CFCTRL_SRV_VEI] = 4,
[CFCTRL_SRV_DATAGRAM] = 7,
[CFCTRL_SRV_UTIL] = 4,
--- a/net/caif/cfctrl.c 2010-10-19 09:29:11.169475442 -0700
+++ b/net/caif/cfctrl.c 2010-10-19 09:40:24.472511862 -0700
@@ -58,7 +58,7 @@ struct cflayer *cfctrl_create(void)
return &this->serv.layer;
}
-static bool param_eq(struct cfctrl_link_param *p1, struct cfctrl_link_param *p2)
+static bool param_eq(const struct cfctrl_link_param *p1, const struct cfctrl_link_param *p2)
{
bool eq =
p1->linktype == p2->linktype &&
@@ -100,8 +100,8 @@ static bool param_eq(struct cfctrl_link_
return false;
}
-bool cfctrl_req_eq(struct cfctrl_request_info *r1,
- struct cfctrl_request_info *r2)
+static bool cfctrl_req_eq(const struct cfctrl_request_info *r1,
+ const struct cfctrl_request_info *r2)
{
if (r1->cmd != r2->cmd)
return false;
@@ -112,7 +112,7 @@ bool cfctrl_req_eq(struct cfctrl_request
}
/* Insert request at the end */
-void cfctrl_insert_req(struct cfctrl *ctrl,
+static void cfctrl_insert_req(struct cfctrl *ctrl,
struct cfctrl_request_info *req)
{
spin_lock(&ctrl->info_list_lock);
@@ -123,8 +123,8 @@ void cfctrl_insert_req(struct cfctrl *ct
}
/* Compare and remove request */
-struct cfctrl_request_info *cfctrl_remove_req(struct cfctrl *ctrl,
- struct cfctrl_request_info *req)
+static struct cfctrl_request_info *cfctrl_remove_req(struct cfctrl *ctrl,
+ struct cfctrl_request_info *req)
{
struct cfctrl_request_info *p, *tmp, *first;
@@ -154,16 +154,6 @@ struct cfctrl_rsp *cfctrl_get_respfuncs(
return &this->res;
}
-void cfctrl_set_dnlayer(struct cflayer *this, struct cflayer *dn)
-{
- this->dn = dn;
-}
-
-void cfctrl_set_uplayer(struct cflayer *this, struct cflayer *up)
-{
- this->up = up;
-}
-
static void init_info(struct caif_payload_info *info, struct cfctrl *cfctrl)
{
info->hdr_len = 0;
@@ -304,58 +294,6 @@ int cfctrl_linkdown_req(struct cflayer *
return ret;
}
-void cfctrl_sleep_req(struct cflayer *layer)
-{
- int ret;
- struct cfctrl *cfctrl = container_obj(layer);
- struct cfpkt *pkt = cfpkt_create(CFPKT_CTRL_PKT_LEN);
- if (!pkt) {
- pr_warn("Out of memory\n");
- return;
- }
- cfpkt_addbdy(pkt, CFCTRL_CMD_SLEEP);
- init_info(cfpkt_info(pkt), cfctrl);
- ret =
- cfctrl->serv.layer.dn->transmit(cfctrl->serv.layer.dn, pkt);
- if (ret < 0)
- cfpkt_destroy(pkt);
-}
-
-void cfctrl_wake_req(struct cflayer *layer)
-{
- int ret;
- struct cfctrl *cfctrl = container_obj(layer);
- struct cfpkt *pkt = cfpkt_create(CFPKT_CTRL_PKT_LEN);
- if (!pkt) {
- pr_warn("Out of memory\n");
- return;
- }
- cfpkt_addbdy(pkt, CFCTRL_CMD_WAKE);
- init_info(cfpkt_info(pkt), cfctrl);
- ret =
- cfctrl->serv.layer.dn->transmit(cfctrl->serv.layer.dn, pkt);
- if (ret < 0)
- cfpkt_destroy(pkt);
-}
-
-void cfctrl_getstartreason_req(struct cflayer *layer)
-{
- int ret;
- struct cfctrl *cfctrl = container_obj(layer);
- struct cfpkt *pkt = cfpkt_create(CFPKT_CTRL_PKT_LEN);
- if (!pkt) {
- pr_warn("Out of memory\n");
- return;
- }
- cfpkt_addbdy(pkt, CFCTRL_CMD_START_REASON);
- init_info(cfpkt_info(pkt), cfctrl);
- ret =
- cfctrl->serv.layer.dn->transmit(cfctrl->serv.layer.dn, pkt);
- if (ret < 0)
- cfpkt_destroy(pkt);
-}
-
-
void cfctrl_cancel_req(struct cflayer *layr, struct cflayer *adap_layer)
{
struct cfctrl_request_info *p, *tmp;
--- a/net/caif/cfmuxl.c 2010-10-19 09:30:45.802913489 -0700
+++ b/net/caif/cfmuxl.c 2010-10-19 09:40:33.573416250 -0700
@@ -71,41 +71,6 @@ int cfmuxl_set_uplayer(struct cflayer *l
return 0;
}
-bool cfmuxl_is_phy_inuse(struct cflayer *layr, u8 phyid)
-{
- struct list_head *node;
- struct cflayer *layer;
- struct cfmuxl *muxl = container_obj(layr);
- bool match = false;
- spin_lock(&muxl->receive_lock);
-
- list_for_each(node, &muxl->srvl_list) {
- layer = list_entry(node, struct cflayer, node);
- if (cfsrvl_phyid_match(layer, phyid)) {
- match = true;
- break;
- }
-
- }
- spin_unlock(&muxl->receive_lock);
- return match;
-}
-
-u8 cfmuxl_get_phyid(struct cflayer *layr, u8 channel_id)
-{
- struct cflayer *up;
- int phyid;
- struct cfmuxl *muxl = container_obj(layr);
- spin_lock(&muxl->receive_lock);
- up = get_up(muxl, channel_id);
- if (up != NULL)
- phyid = cfsrvl_getphyid(up);
- else
- phyid = 0;
- spin_unlock(&muxl->receive_lock);
- return phyid;
-}
-
int cfmuxl_set_dnlayer(struct cflayer *layr, struct cflayer *dn, u8 phyid)
{
struct cfmuxl *muxl = (struct cfmuxl *) layr;
--- a/net/caif/cfpkt_skbuff.c 2010-10-19 09:31:38.804197063 -0700
+++ b/net/caif/cfpkt_skbuff.c 2010-10-19 09:42:03.710371120 -0700
@@ -42,22 +42,22 @@ struct cfpkt_priv_data {
bool erronous;
};
-inline struct cfpkt_priv_data *cfpkt_priv(struct cfpkt *pkt)
+static inline struct cfpkt_priv_data *cfpkt_priv(struct cfpkt *pkt)
{
return (struct cfpkt_priv_data *) pkt->skb.cb;
}
-inline bool is_erronous(struct cfpkt *pkt)
+static inline bool is_erronous(struct cfpkt *pkt)
{
return cfpkt_priv(pkt)->erronous;
}
-inline struct sk_buff *pkt_to_skb(struct cfpkt *pkt)
+static inline struct sk_buff *pkt_to_skb(struct cfpkt *pkt)
{
return &pkt->skb;
}
-inline struct cfpkt *skb_to_pkt(struct sk_buff *skb)
+static inline struct cfpkt *skb_to_pkt(struct sk_buff *skb)
{
return (struct cfpkt *) skb;
}
@@ -317,17 +317,6 @@ int cfpkt_setlen(struct cfpkt *pkt, u16
}
EXPORT_SYMBOL(cfpkt_setlen);
-struct cfpkt *cfpkt_create_uplink(const unsigned char *data, unsigned int len)
-{
- struct cfpkt *pkt = cfpkt_create_pfx(len + PKT_POSTFIX, PKT_PREFIX);
- if (!pkt)
- return NULL;
- if (unlikely(data != NULL))
- cfpkt_add_body(pkt, data, len);
- return pkt;
-}
-EXPORT_SYMBOL(cfpkt_create_uplink);
-
struct cfpkt *cfpkt_append(struct cfpkt *dstpkt,
struct cfpkt *addpkt,
u16 expectlen)
@@ -408,169 +397,12 @@ struct cfpkt *cfpkt_split(struct cfpkt *
}
EXPORT_SYMBOL(cfpkt_split);
-char *cfpkt_log_pkt(struct cfpkt *pkt, char *buf, int buflen)
-{
- struct sk_buff *skb = pkt_to_skb(pkt);
- char *p = buf;
- int i;
-
- /*
- * Sanity check buffer length, it needs to be at least as large as
- * the header info: ~=50+ bytes
- */
- if (buflen < 50)
- return NULL;
-
- snprintf(buf, buflen, "%s: pkt:%p len:%ld(%ld+%ld) {%ld,%ld} data: [",
- is_erronous(pkt) ? "ERRONOUS-SKB" :
- (skb->data_len != 0 ? "COMPLEX-SKB" : "SKB"),
- skb,
- (long) skb->len,
- (long) (skb_tail_pointer(skb) - skb->data),
- (long) skb->data_len,
- (long) (skb->data - skb->head),
- (long) (skb_tail_pointer(skb) - skb->head));
- p = buf + strlen(buf);
-
- for (i = 0; i < skb_tail_pointer(skb) - skb->data && i < 300; i++) {
- if (p > buf + buflen - 10) {
- sprintf(p, "...");
- p = buf + strlen(buf);
- break;
- }
- sprintf(p, "%02x,", skb->data[i]);
- p = buf + strlen(buf);
- }
- sprintf(p, "]\n");
- return buf;
-}
-EXPORT_SYMBOL(cfpkt_log_pkt);
-
-int cfpkt_raw_append(struct cfpkt *pkt, void **buf, unsigned int buflen)
-{
- struct sk_buff *skb = pkt_to_skb(pkt);
- struct sk_buff *lastskb;
-
- caif_assert(buf != NULL);
- if (unlikely(is_erronous(pkt)))
- return -EPROTO;
- /* Make sure SKB is writable */
- if (unlikely(skb_cow_data(skb, 0, &lastskb) < 0)) {
- PKT_ERROR(pkt, "skb_cow_data failed\n");
- return -EPROTO;
- }
-
- if (unlikely(skb_linearize(skb) != 0)) {
- PKT_ERROR(pkt, "linearize failed\n");
- return -EPROTO;
- }
-
- if (unlikely(skb_tailroom(skb) < buflen)) {
- PKT_ERROR(pkt, "buffer too short - failed\n");
- return -EPROTO;
- }
-
- *buf = skb_put(skb, buflen);
- return 1;
-}
-EXPORT_SYMBOL(cfpkt_raw_append);
-
-int cfpkt_raw_extract(struct cfpkt *pkt, void **buf, unsigned int buflen)
-{
- struct sk_buff *skb = pkt_to_skb(pkt);
-
- caif_assert(buf != NULL);
- if (unlikely(is_erronous(pkt)))
- return -EPROTO;
-
- if (unlikely(buflen > skb->len)) {
- PKT_ERROR(pkt, "buflen too large - failed\n");
- return -EPROTO;
- }
-
- if (unlikely(buflen > skb_headlen(skb))) {
- if (unlikely(skb_linearize(skb) != 0)) {
- PKT_ERROR(pkt, "linearize failed\n");
- return -EPROTO;
- }
- }
-
- *buf = skb->data;
- skb_pull(skb, buflen);
-
- return 1;
-}
-EXPORT_SYMBOL(cfpkt_raw_extract);
-
-inline bool cfpkt_erroneous(struct cfpkt *pkt)
+bool cfpkt_erroneous(struct cfpkt *pkt)
{
return cfpkt_priv(pkt)->erronous;
}
EXPORT_SYMBOL(cfpkt_erroneous);
-struct cfpktq *cfpktq_create(void)
-{
- struct cfpktq *q = kmalloc(sizeof(struct cfpktq), GFP_ATOMIC);
- if (!q)
- return NULL;
- skb_queue_head_init(&q->head);
- atomic_set(&q->count, 0);
- spin_lock_init(&q->lock);
- return q;
-}
-EXPORT_SYMBOL(cfpktq_create);
-
-void cfpkt_queue(struct cfpktq *pktq, struct cfpkt *pkt, unsigned short prio)
-{
- atomic_inc(&pktq->count);
- spin_lock(&pktq->lock);
- skb_queue_tail(&pktq->head, pkt_to_skb(pkt));
- spin_unlock(&pktq->lock);
-
-}
-EXPORT_SYMBOL(cfpkt_queue);
-
-struct cfpkt *cfpkt_qpeek(struct cfpktq *pktq)
-{
- struct cfpkt *tmp;
- spin_lock(&pktq->lock);
- tmp = skb_to_pkt(skb_peek(&pktq->head));
- spin_unlock(&pktq->lock);
- return tmp;
-}
-EXPORT_SYMBOL(cfpkt_qpeek);
-
-struct cfpkt *cfpkt_dequeue(struct cfpktq *pktq)
-{
- struct cfpkt *pkt;
- spin_lock(&pktq->lock);
- pkt = skb_to_pkt(skb_dequeue(&pktq->head));
- if (pkt) {
- atomic_dec(&pktq->count);
- caif_assert(atomic_read(&pktq->count) >= 0);
- }
- spin_unlock(&pktq->lock);
- return pkt;
-}
-EXPORT_SYMBOL(cfpkt_dequeue);
-
-int cfpkt_qcount(struct cfpktq *pktq)
-{
- return atomic_read(&pktq->count);
-}
-EXPORT_SYMBOL(cfpkt_qcount);
-
-struct cfpkt *cfpkt_clone_release(struct cfpkt *pkt)
-{
- struct cfpkt *clone;
- clone = skb_to_pkt(skb_clone(pkt_to_skb(pkt), GFP_ATOMIC));
- /* Free original packet. */
- cfpkt_destroy(pkt);
- if (!clone)
- return NULL;
- return clone;
-}
-EXPORT_SYMBOL(cfpkt_clone_release);
struct caif_payload_info *cfpkt_info(struct cfpkt *pkt)
{
--- a/net/caif/cfsrvl.c 2010-10-19 09:38:07.842929217 -0700
+++ b/net/caif/cfsrvl.c 2010-10-19 09:40:31.817241735 -0700
@@ -151,12 +151,7 @@ static int cfservl_modemcmd(struct cflay
return -EINVAL;
}
-void cfservl_destroy(struct cflayer *layer)
-{
- kfree(layer);
-}
-
-void cfsrvl_release(struct kref *kref)
+static void cfsrvl_release(struct kref *kref)
{
struct cfsrvl *service = container_of(kref, struct cfsrvl, ref);
kfree(service);
--- a/include/net/caif/cfmuxl.h 2010-10-19 09:25:53.761769361 -0700
+++ b/include/net/caif/cfmuxl.h 2010-10-19 09:39:23.178419722 -0700
@@ -16,7 +16,5 @@ int cfmuxl_set_uplayer(struct cflayer *l
struct cflayer *cfmuxl_remove_dnlayer(struct cflayer *layr, u8 phyid);
int cfmuxl_set_dnlayer(struct cflayer *layr, struct cflayer *up, u8 phyid);
struct cflayer *cfmuxl_remove_uplayer(struct cflayer *layr, u8 linkid);
-bool cfmuxl_is_phy_inuse(struct cflayer *layr, u8 phyid);
-u8 cfmuxl_get_phyid(struct cflayer *layr, u8 channel_id);
#endif /* CFMUXL_H_ */
--- a/include/net/caif/cfsrvl.h 2010-10-19 09:25:53.837776953 -0700
+++ b/include/net/caif/cfsrvl.h 2010-10-19 09:39:21.410243950 -0700
@@ -22,7 +22,6 @@ struct cfsrvl {
struct kref ref;
};
-void cfsrvl_release(struct kref *kref);
struct cflayer *cfvei_create(u8 linkid, struct dev_info *dev_info);
struct cflayer *cfdgml_create(u8 linkid, struct dev_info *dev_info);
struct cflayer *cfutill_create(u8 linkid, struct dev_info *dev_info);
@@ -31,7 +30,7 @@ struct cflayer *cfrfml_create(u8 linkid,
int mtu_size);
struct cflayer *cfdbgl_create(u8 linkid, struct dev_info *dev_info);
bool cfsrvl_phyid_match(struct cflayer *layer, int phyid);
-void cfservl_destroy(struct cflayer *layer);
+
void cfsrvl_init(struct cfsrvl *service,
u8 channel_id,
struct dev_info *dev_info,
--- a/net/caif/caif_config_util.c 2010-10-19 09:53:45.182003880 -0700
+++ b/net/caif/caif_config_util.c 2010-10-19 09:53:58.887283867 -0700
@@ -10,9 +10,9 @@
#include <net/caif/cfcnfg.h>
#include <net/caif/caif_dev.h>
-int connect_req_to_link_param(struct cfcnfg *cnfg,
- struct caif_connect_request *s,
- struct cfctrl_link_param *l)
+int caif_connect_req_to_link_param(struct cfcnfg *cnfg,
+ struct caif_connect_request *s,
+ struct cfctrl_link_param *l)
{
struct dev_info *dev_info;
enum cfcnfg_phy_preference pref;
^ permalink raw reply
* [PATCH 2/2] Revert napi_poll fix for bonding driver
From: nhorman @ 2010-10-19 17:04 UTC (permalink / raw)
To: netdev; +Cc: bonding-devel, fubar, davem, andy, amwang, nhorman
In-Reply-To: <1287507866-25156-1-git-send-email-nhorman@tuxdriver.com>
From: Neil Horman <nhorman@tuxdriver.com>
In an erlier patch I modified napi_poll so that devices with IFF_MASTER polled
the per_cpu list instead of the device list for napi. I did this because the
bonding driver has no napi instances to poll, it instead expects to check the
slave devices napi instances, which napi_poll was unaware of. Looking at this
more closely however, I now see this isn't strictly needed. As the bond driver
poll_controller calls the slaves poll_controller via netpoll_poll_dev, which
recursively calls poll_napi on each slave, allowing those napi instances to get
serviced. The earlier patch isn't at all harmfull, its just not needed, so lets
revert it to make the code cleaner. Sorry for the noise,
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
---
net/core/netpoll.c | 9 +--------
1 files changed, 1 insertions(+), 8 deletions(-)
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index d79d221..4e98ffa 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -156,15 +156,8 @@ static void poll_napi(struct net_device *dev)
{
struct napi_struct *napi;
int budget = 16;
- struct softnet_data *sd = &__get_cpu_var(softnet_data);
- struct list_head *nlist;
- if (dev->flags & IFF_MASTER)
- nlist = &sd->poll_list;
- else
- nlist = &dev->napi_list;
-
- list_for_each_entry(napi, nlist, dev_list) {
+ list_for_each_entry(napi, &dev->napi_list, dev_list) {
if (napi->poll_owner != smp_processor_id() &&
spin_trylock(&napi->poll_lock)) {
budget = poll_one_napi(dev->npinfo, napi, budget);
--
1.7.2.3
^ permalink raw reply related
* [PATCH 1/2] Remove netpoll blocking from uninit path
From: nhorman @ 2010-10-19 17:04 UTC (permalink / raw)
To: netdev; +Cc: bonding-devel, fubar, davem, andy, amwang, nhorman
In-Reply-To: <1287507866-25156-1-git-send-email-nhorman@tuxdriver.com>
From: Neil Horman <nhorman@tuxdriver.com>
Some recent testing in netpoll with bonding showed this backtrace
------------[ cut here ]------------
kernel BUG at drivers/net/bonding/bonding.h:134!
invalid opcode: 0000 [#1] SMP
last sysfs file: /sys/devices/pci0000:00/0000:00:1d.2/usb7/devnum
CPU 0
Pid: 1876, comm: rmmod Not tainted 2.6.36-rc3+ #10 D26928/
RIP: 0010:[<ffffffffa0514ba4>] [<ffffffffa0514ba4>] bond_uninit+0x6f4/0x7a0
RSP: 0018:ffff88003b1b5d58 EFLAGS: 00010296
RAX: ffff88003b9b6200 RBX: ffff8800373e8e00 RCX: 00000000000f4240
RDX: 00000000ffffffff RSI: 0000000000000286 RDI: 0000000000000286
RBP: ffff88003b1b5dc8 R08: 0000000000000000 R09: 00000001af7de920
R10: 0000000000000000 R11: ffff880002495e98 R12: ffff880037922700
R13: ffff880038c31000 R14: ffff880037922730 R15: 0000000000000286
FS: 00007f90e6d72700(0000) GS:ffff880002400000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 000000346f0d9ad0 CR3: 000000003b263000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process rmmod (pid: 1876, threadinfo ffff88003b1b4000, task ffff88003b36aa80)
Stack:
00000000ffffffff ffff88003b1b5d7a ffff8800379221e8 ffff880037922000
<0> ffff88003b1b5dc8 ffffffff813eb5fb ffff88003b1b5da8 0000000031b177a3
<0> ffff88003b1b5da8 ffff880037922000 ffff88003b1b5e48 ffff88003b1b5e48
Call Trace:
[<ffffffff813eb5fb>] ? rtmsg_ifinfo+0xcb/0xf0
[<ffffffff813daad8>] rollback_registered_many+0x168/0x280
[<ffffffff813dac09>] unregister_netdevice_many+0x19/0x80
[<ffffffff813e97b3>] __rtnl_kill_links+0x63/0x90
[<ffffffff813e980b>] __rtnl_link_unregister+0x2b/0x60
[<ffffffff813e9bde>] rtnl_link_unregister+0x1e/0x30
[<ffffffffa052124b>] bonding_exit+0x37/0x51 [bonding]
[<ffffffff81098b2e>] sys_delete_module+0x19e/0x270
[<ffffffff810bb2b2>] ? audit_syscall_entry+0x252/0x280
[<ffffffff8100b0b2>] system_call_fastpath+0x16/0x1b
RIP [<ffffffffa0514ba4>] bond_uninit+0x6f4/0x7a0 [bonding]
RSP <ffff88003b1b5d58>
---[ end trace 1395ad691cea24d1 ]---
It occurs because of my recent netpoll blocking patches, which I added to avoid
recursive deadlock in the bonding driver. It relies on some per cpu bits, but
the shutdown path forces some rescheduling as we cancel workqueues for the
driver and wait for some device refcounts. If after the forced reschedule, we
wind up on a different cpu we trigger the bughalt in unblock_netpoll_tx.
The fix is to remove the netpoll block/unblock calls from bond_release_all.
This is safe to do because bond_uninit, which is called via ndo_uninit in
rollback_registered_many, doesn't occur until we send a NETDEV_UNREGISTER event,
which triggers netconsole to remove us as a netpoll client, so we are guaranteed
not to recurse into our own tx path here.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
---
drivers/net/bonding/bond_main.c | 3 ---
1 files changed, 0 insertions(+), 3 deletions(-)
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 38d4ca0..99cb891 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -2082,7 +2082,6 @@ static int bond_release_all(struct net_device *bond_dev)
struct net_device *slave_dev;
struct sockaddr addr;
- block_netpoll_tx();
write_lock_bh(&bond->lock);
netif_carrier_off(bond_dev);
@@ -2181,8 +2180,6 @@ static int bond_release_all(struct net_device *bond_dev)
out:
write_unlock_bh(&bond->lock);
- unblock_netpoll_tx();
-
return 0;
}
--
1.7.2.3
^ permalink raw reply related
* [PATCH] bonding: minor cleanups to bond + netpoll
From: nhorman @ 2010-10-19 17:04 UTC (permalink / raw)
To: netdev; +Cc: bonding-devel, fubar, davem, andy, amwang, nhorman
Testing that was onging when the the reset patchset to enable netpoll over
bonding revealed some minor corner cases that are worth correcting. In summary:
1) Remove netpoll tx blocking from bond_release_all. Its not needed and causes
some uglyness when removing the bonding module, in the form of a backtrace that
gets logged. blocking isn't needed in this path anyway as the netconsole is
already unregistered from us at this point
2) Remove my changes to napi_poll. Closer inspection of the bonding
poll_controller show that we wind up recursively calling the napi poll routines
for the slaves through sucsessive calls to netpoll_poll_dev. My origional
change is harmless, but its not really needed, so lets make the code simpler.
Further details available in the individual commit messages
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
^ permalink raw reply
* Re: [PATCH net-next] bnx2: Increase max rx ring size from 1K to 2K
From: Michael Chan @ 2010-10-19 17:03 UTC (permalink / raw)
To: David Miller
Cc: andy@greyhouse.net, jfeeney@redhat.com, netdev@vger.kernel.org
In-Reply-To: <20101019.011434.226774173.davem@davemloft.net>
On Tue, 2010-10-19 at 01:14 -0700, David Miller wrote:
> From: "Michael Chan" <mchan@broadcom.com>
> Date: Mon, 18 Oct 2010 17:30:54 -0700
>
> > A number of customers are reporting packet loss under certain workloads
> > (e.g. heavy bursts of small packets) with flow control disabled. A larger
> > rx ring helps to prevent these losses.
> >
> > No change in default rx ring size and memory consumption.
> >
> > Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
> > Acked-by: John Feeney <jfeeney@redhat.com>
> > Signed-off-by: Michael Chan <mchan@broadcom.com>
>
> I don't see how it's any better to queue things more deeply in
> hardware, compared to simply using hardware flow control since that's
> what it's for and makes the queuing happen in the networking stack of
> the sender, in software, which in the end performs better and gives
> better feedback to the source of the data.
There are situations where flow control is not desirable. For example,
if there are many multicast receivers in the network, you may not want a
few slow receivers to slow down the entire network with flow control.
>
> These huge RX queue sizes are absolutely rediculious, and I've
> complained about this before.
Yes you have and I was initially hesitant to post this patch. But the
customer sees that many other 1G drivers in the tree can have bigger
ring sizes and as a result, they have fewer packet drops using these
other devices. In fact, 2K is still much smaller than many other 1G
drivers in the tree. Please also note that this does not add any extra
memory to the default configuration. Thanks.
>
> And instead of seeing less of this, I keep seeing more of this stuff.
> Please exert some pushback on these folks who are doing such insane
> things.
>
> Thanks.
>
^ permalink raw reply
* [PATCH] macb: Don't re-enable interrupts while in polling mode
From: Joshua Hoke @ 2010-10-19 16:48 UTC (permalink / raw)
To: Nicolas Ferre
Cc: David S. Miller, Jiri Pirko, Peter Korsgaard, Eric Dumazet,
Haavard Skinnemoen, netdev, linux-kernel, Andrew Morton
From: Joshua Hoke <joshua.hoke@sixnet.com>
[PATCH] macb: Don't re-enable interrupts while in polling mode
On a busy network, the macb driver could get stuck in the interrupt
handler, quickly triggering the watchdog, due to a confluence of
factors:
1. macb_poll re-enables interrupts unconditionally, even when it will
be called again because it exhausted its rx budget
2. macb_interrupt only disables interrupts after scheduling
macb_poll, but scheduling fails when macb_poll is already scheduled
because it didn't call napi_complete
3. macb_interrupt loops until the interrupt status register is clear,
which will never happen in this case if the driver doesn't disable
the RX interrupt
Since macb_interrupt runs in interrupt context, this effectively locks
up the machine, triggering the hardware watchdog.
This issue was readily reproducible on a flooded network with a
modified 2.6.27.48 kernel. The same problem appears to still be in the
2.6.36-rc8 driver code, so I am submitting this patch against that
version. I have not tested this version of the patch except to make
sure the kernel compiles.
Signed-off-by: Joshua Hoke <joshua.hoke@sixnet.com>
---
I'm submitting this at the request of Andrew Morton in this bug:
https://bugzilla.kernel.org/show_bug.cgi?id=20732
This version of the patch applies to 2.6.36-rc8 but has not been
tested. In particular I am assuming that napi_schedule_prep() behaves
the same as netif_rx_schedule_prep() did by failing when the macb_poll
callback is already scheduled.
diff --git a/drivers/net/macb.c b/drivers/net/macb.c
index ff2f158..36cf594 100644
--- a/drivers/net/macb.c
+++ b/drivers/net/macb.c
@@ -515,14 +515,15 @@ static int macb_poll(struct napi_struct *napi, int
budget)
(unsigned long)status, budget);
work_done = macb_rx(bp, budget);
- if (work_done < budget)
+ if (work_done < budget) {
napi_complete(napi);
- /*
- * We've done what we can to clean the buffers. Make sure we
- * get notified when new packets arrive.
- */
- macb_writel(bp, IER, MACB_RX_INT_FLAGS);
+ /*
+ * We've done what we can to clean the buffers. Make
sure we
+ * get notified when new packets arrive.
+ */
+ macb_writel(bp, IER, MACB_RX_INT_FLAGS);
+ }
/* TODO: Handle errors */
@@ -550,12 +551,16 @@ static irqreturn_t macb_interrupt(int irq, void
*dev_id)
}
if (status & MACB_RX_INT_FLAGS) {
+ /*
+ * There's no point taking any more interrupts
+ * until we have processed the buffers. The
+ * scheduling call may fail if the poll routine
+ * is already scheduled, so disable interrupts
+ * now.
+ */
+ macb_writel(bp, IDR, MACB_RX_INT_FLAGS);
+
if (napi_schedule_prep(&bp->napi)) {
- /*
- * There's no point taking any more
interrupts
- * until we have processed the buffers
- */
- macb_writel(bp, IDR, MACB_RX_INT_FLAGS);
dev_dbg(&bp->pdev->dev,
"scheduling RX softirq\n");
__napi_schedule(&bp->napi);
^ permalink raw reply related
* [PATCH net-next] napi: unexport napi_reuse_skb
From: Stephen Hemminger @ 2010-10-19 17:12 UTC (permalink / raw)
To: David Miller; +Cc: netdev
The function napi_reuse_skb is only used inside core.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
---
include/linux/netdevice.h | 2 --
net/core/dev.c | 3 +--
2 files changed, 1 insertion(+), 4 deletions(-)
--- a/include/linux/netdevice.h 2010-10-19 09:03:13.211106810 -0700
+++ b/include/linux/netdevice.h 2010-10-19 09:03:24.568216914 -0700
@@ -1745,8 +1745,6 @@ extern gro_result_t napi_skb_finish(gro_
extern gro_result_t napi_gro_receive(struct napi_struct *napi,
struct sk_buff *skb);
extern void napi_gro_flush(struct napi_struct *napi);
-extern void napi_reuse_skb(struct napi_struct *napi,
- struct sk_buff *skb);
extern struct sk_buff * napi_get_frags(struct napi_struct *napi);
extern gro_result_t napi_frags_finish(struct napi_struct *napi,
struct sk_buff *skb,
--- a/net/core/dev.c 2010-10-19 09:03:13.235109156 -0700
+++ b/net/core/dev.c 2010-10-19 09:03:58.839566666 -0700
@@ -3287,14 +3287,13 @@ gro_result_t napi_gro_receive(struct nap
}
EXPORT_SYMBOL(napi_gro_receive);
-void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
+static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
{
__skb_pull(skb, skb_headlen(skb));
skb_reserve(skb, NET_IP_ALIGN - skb_headroom(skb));
napi->skb = skb;
}
-EXPORT_SYMBOL(napi_reuse_skb);
struct sk_buff *napi_get_frags(struct napi_struct *napi)
{
^ permalink raw reply
* [PATCH net-next] rds: make local functions/variables static
From: Stephen Hemminger @ 2010-10-19 18:08 UTC (permalink / raw)
To: Andy Grover, Zach Brown, Chris Mason, David Miller; +Cc: rds-devel, netdev
The RDS protocol has lots of functions that should be
declared static. rds_message_get/add_version_extension is
removed since it defined but never used.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
---
net/rds/connection.c | 2 +-
net/rds/ib.c | 10 +++++-----
net/rds/ib.h | 4 ----
net/rds/ib_rdma.c | 2 +-
net/rds/ib_sysctl.c | 2 +-
net/rds/iw.c | 4 ++--
net/rds/iw.h | 4 ----
net/rds/iw_rdma.c | 3 ++-
net/rds/iw_sysctl.c | 2 +-
net/rds/message.c | 24 ++----------------------
net/rds/page.c | 3 ++-
net/rds/rdma_transport.c | 4 ++--
net/rds/rdma_transport.h | 4 ----
net/rds/rds.h | 5 -----
net/rds/recv.c | 3 +--
net/rds/send.c | 4 +++-
net/rds/tcp.c | 6 +++---
net/rds/tcp.h | 2 --
net/rds/tcp_recv.c | 3 ++-
net/rds/tcp_send.c | 2 +-
20 files changed, 29 insertions(+), 64 deletions(-)
--- a/net/rds/connection.c 2010-10-19 10:36:00.346844139 -0700
+++ b/net/rds/connection.c 2010-10-19 10:50:24.678156846 -0700
@@ -88,7 +88,7 @@ static struct rds_connection *rds_conn_l
* and receiving over this connection again in the future. It is up to
* the transport to have serialized this call with its send and recv.
*/
-void rds_conn_reset(struct rds_connection *conn)
+static void rds_conn_reset(struct rds_connection *conn)
{
rdsdebug("connection %pI4 to %pI4 reset\n",
&conn->c_laddr, &conn->c_faddr);
--- a/net/rds/ib.c 2010-10-19 10:36:24.489145829 -0700
+++ b/net/rds/ib.c 2010-10-19 10:38:05.034732462 -0700
@@ -42,7 +42,7 @@
#include "rds.h"
#include "ib.h"
-unsigned int fmr_pool_size = RDS_FMR_POOL_SIZE;
+static unsigned int fmr_pool_size = RDS_FMR_POOL_SIZE;
unsigned int fmr_message_size = RDS_FMR_SIZE + 1; /* +1 allows for unaligned MRs */
unsigned int rds_ib_retry_count = RDS_IB_DEFAULT_RETRY_COUNT;
@@ -65,7 +65,7 @@ struct list_head rds_ib_devices;
DEFINE_SPINLOCK(ib_nodev_conns_lock);
LIST_HEAD(ib_nodev_conns);
-void rds_ib_nodev_connect(void)
+static void rds_ib_nodev_connect(void)
{
struct rds_ib_connection *ic;
@@ -75,7 +75,7 @@ void rds_ib_nodev_connect(void)
spin_unlock(&ib_nodev_conns_lock);
}
-void rds_ib_dev_shutdown(struct rds_ib_device *rds_ibdev)
+static void rds_ib_dev_shutdown(struct rds_ib_device *rds_ibdev)
{
struct rds_ib_connection *ic;
unsigned long flags;
@@ -118,7 +118,7 @@ void rds_ib_dev_put(struct rds_ib_device
queue_work(rds_wq, &rds_ibdev->free_work);
}
-void rds_ib_add_one(struct ib_device *device)
+static void rds_ib_add_one(struct ib_device *device)
{
struct rds_ib_device *rds_ibdev;
struct ib_device_attr *dev_attr;
@@ -229,7 +229,7 @@ struct rds_ib_device *rds_ib_get_client_
*
* This can be called at any time and can be racing with any other RDS path.
*/
-void rds_ib_remove_one(struct ib_device *device)
+static void rds_ib_remove_one(struct ib_device *device)
{
struct rds_ib_device *rds_ibdev;
--- a/net/rds/ib.h 2010-10-19 10:37:19.938432563 -0700
+++ b/net/rds/ib.h 2010-10-19 10:53:22.118325615 -0700
@@ -265,13 +265,10 @@ static inline void rds_ib_dma_sync_sg_fo
/* ib.c */
extern struct rds_transport rds_ib_transport;
-extern void rds_ib_add_one(struct ib_device *device);
-extern void rds_ib_remove_one(struct ib_device *device);
struct rds_ib_device *rds_ib_get_client_data(struct ib_device *device);
void rds_ib_dev_put(struct rds_ib_device *rds_ibdev);
extern struct ib_client rds_ib_client;
-extern unsigned int fmr_pool_size;
extern unsigned int fmr_message_size;
extern unsigned int rds_ib_retry_count;
@@ -374,6 +371,5 @@ extern unsigned long rds_ib_sysctl_max_u
extern unsigned long rds_ib_sysctl_max_unsig_bytes;
extern unsigned long rds_ib_sysctl_max_recv_allocation;
extern unsigned int rds_ib_sysctl_flow_control;
-extern ctl_table rds_ib_sysctl_table[];
#endif
--- a/net/rds/ib_rdma.c 2010-10-19 10:38:36.109695644 -0700
+++ b/net/rds/ib_rdma.c 2010-10-19 10:38:43.474397946 -0700
@@ -38,7 +38,7 @@
#include "ib.h"
#include "xlist.h"
-struct workqueue_struct *rds_ib_fmr_wq;
+static struct workqueue_struct *rds_ib_fmr_wq;
static DEFINE_PER_CPU(unsigned long, clean_list_grace);
#define CLEAN_LIST_BUSY_BIT 0
--- a/net/rds/ib_sysctl.c 2010-10-19 10:39:22.998166930 -0700
+++ b/net/rds/ib_sysctl.c 2010-10-19 10:39:45.780252357 -0700
@@ -61,7 +61,7 @@ static unsigned long rds_ib_sysctl_max_u
*/
unsigned int rds_ib_sysctl_flow_control = 0;
-ctl_table rds_ib_sysctl_table[] = {
+static ctl_table rds_ib_sysctl_table[] = {
{
.procname = "max_send_wr",
.data = &rds_ib_sysctl_max_send_wr,
--- a/net/rds/iw.c 2010-10-19 10:40:14.130811167 -0700
+++ b/net/rds/iw.c 2010-10-19 10:40:40.601201008 -0700
@@ -56,7 +56,7 @@ struct list_head rds_iw_devices;
DEFINE_SPINLOCK(iw_nodev_conns_lock);
LIST_HEAD(iw_nodev_conns);
-void rds_iw_add_one(struct ib_device *device)
+static void rds_iw_add_one(struct ib_device *device)
{
struct rds_iw_device *rds_iwdev;
struct ib_device_attr *dev_attr;
@@ -124,7 +124,7 @@ free_attr:
kfree(dev_attr);
}
-void rds_iw_remove_one(struct ib_device *device)
+static void rds_iw_remove_one(struct ib_device *device)
{
struct rds_iw_device *rds_iwdev;
struct rds_iw_cm_id *i_cm_id, *next;
--- a/net/rds/iw.h 2010-10-19 10:53:48.052691754 -0700
+++ b/net/rds/iw.h 2010-10-19 10:54:09.078609969 -0700
@@ -268,8 +268,6 @@ static inline u32 rds_iw_local_dma_lkey(
/* ib.c */
extern struct rds_transport rds_iw_transport;
-extern void rds_iw_add_one(struct ib_device *device);
-extern void rds_iw_remove_one(struct ib_device *device);
extern struct ib_client rds_iw_client;
extern unsigned int fastreg_pool_size;
@@ -318,7 +316,6 @@ void *rds_iw_get_mr(struct scatterlist *
void rds_iw_sync_mr(void *trans_private, int dir);
void rds_iw_free_mr(void *trans_private, int invalidate);
void rds_iw_flush_mrs(void);
-void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id);
/* ib_recv.c */
int rds_iw_recv_init(void);
@@ -378,7 +375,6 @@ extern unsigned long rds_iw_sysctl_max_u
extern unsigned long rds_iw_sysctl_max_unsig_bytes;
extern unsigned long rds_iw_sysctl_max_recv_allocation;
extern unsigned int rds_iw_sysctl_flow_control;
-extern ctl_table rds_iw_sysctl_table[];
/*
* Helper functions for getting/setting the header and data SGEs in
--- a/net/rds/iw_rdma.c 2010-10-19 10:40:49.237980581 -0700
+++ b/net/rds/iw_rdma.c 2010-10-19 10:40:58.058777701 -0700
@@ -157,7 +157,8 @@ static int rds_iw_add_cm_id(struct rds_i
return 0;
}
-void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev, struct rdma_cm_id *cm_id)
+static void rds_iw_remove_cm_id(struct rds_iw_device *rds_iwdev,
+ struct rdma_cm_id *cm_id)
{
struct rds_iw_cm_id *i_cm_id;
--- a/net/rds/iw_sysctl.c 2010-10-19 10:41:15.516354119 -0700
+++ b/net/rds/iw_sysctl.c 2010-10-19 10:43:10.922788525 -0700
@@ -55,7 +55,7 @@ static unsigned long rds_iw_sysctl_max_u
unsigned int rds_iw_sysctl_flow_control = 1;
-ctl_table rds_iw_sysctl_table[] = {
+static ctl_table rds_iw_sysctl_table[] = {
{
.procname = "max_send_wr",
.data = &rds_iw_sysctl_max_send_wr,
--- a/net/rds/message.c 2010-10-19 10:43:34.584930068 -0700
+++ b/net/rds/message.c 2010-10-19 10:59:47.121527590 -0700
@@ -106,8 +106,8 @@ void rds_message_populate_header(struct
}
EXPORT_SYMBOL_GPL(rds_message_populate_header);
-int rds_message_add_extension(struct rds_header *hdr,
- unsigned int type, const void *data, unsigned int len)
+int rds_message_add_extension(struct rds_header *hdr, unsigned int type,
+ const void *data, unsigned int len)
{
unsigned int ext_len = sizeof(u8) + len;
unsigned char *dst;
@@ -177,26 +177,6 @@ none:
return RDS_EXTHDR_NONE;
}
-int rds_message_add_version_extension(struct rds_header *hdr, unsigned int version)
-{
- struct rds_ext_header_version ext_hdr;
-
- ext_hdr.h_version = cpu_to_be32(version);
- return rds_message_add_extension(hdr, RDS_EXTHDR_VERSION, &ext_hdr, sizeof(ext_hdr));
-}
-
-int rds_message_get_version_extension(struct rds_header *hdr, unsigned int *version)
-{
- struct rds_ext_header_version ext_hdr;
- unsigned int pos = 0, len = sizeof(ext_hdr);
-
- /* We assume the version extension is the only one present */
- if (rds_message_next_extension(hdr, &pos, &ext_hdr, &len) != RDS_EXTHDR_VERSION)
- return 0;
- *version = be32_to_cpu(ext_hdr.h_version);
- return 1;
-}
-
int rds_message_add_rdma_dest_extension(struct rds_header *hdr, u32 r_key, u32 offset)
{
struct rds_ext_header_rdma_dest ext_hdr;
--- a/net/rds/rdma_transport.c 2010-10-19 10:44:45.687369510 -0700
+++ b/net/rds/rdma_transport.c 2010-10-19 10:45:03.729004485 -0700
@@ -207,7 +207,7 @@ static void rds_rdma_listen_stop(void)
}
}
-int rds_rdma_init(void)
+static int rds_rdma_init(void)
{
int ret;
@@ -234,7 +234,7 @@ out:
}
module_init(rds_rdma_init);
-void rds_rdma_exit(void)
+static void rds_rdma_exit(void)
{
/* stop listening first to ensure no new connections are attempted */
rds_rdma_listen_stop();
--- a/net/rds/rdma_transport.h 2010-10-19 10:53:05.556814850 -0700
+++ b/net/rds/rdma_transport.h 2010-10-19 10:53:17.429897281 -0700
@@ -11,10 +11,6 @@ int rds_rdma_conn_connect(struct rds_con
int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
struct rdma_cm_event *event);
-/* from rdma_transport.c */
-int rds_rdma_init(void);
-void rds_rdma_exit(void);
-
/* from ib.c */
extern struct rds_transport rds_ib_transport;
int rds_ib_init(void);
--- a/net/rds/rds.h 2010-10-19 10:46:16.483601811 -0700
+++ b/net/rds/rds.h 2010-10-19 10:59:20.875122353 -0700
@@ -619,7 +619,6 @@ struct rds_connection *rds_conn_create_o
struct rds_transport *trans, gfp_t gfp);
void rds_conn_shutdown(struct rds_connection *conn);
void rds_conn_destroy(struct rds_connection *conn);
-void rds_conn_reset(struct rds_connection *conn);
void rds_conn_drop(struct rds_connection *conn);
void rds_conn_connect_if_down(struct rds_connection *conn);
void rds_for_each_conn_info(struct socket *sock, unsigned int len,
@@ -668,8 +667,6 @@ int rds_message_add_extension(struct rds
unsigned int type, const void *data, unsigned int len);
int rds_message_next_extension(struct rds_header *hdr,
unsigned int *pos, void *buf, unsigned int *buflen);
-int rds_message_add_version_extension(struct rds_header *hdr, unsigned int version);
-int rds_message_get_version_extension(struct rds_header *hdr, unsigned int *version);
int rds_message_add_rdma_dest_extension(struct rds_header *hdr, u32 r_key, u32 offset);
int rds_message_inc_copy_to_user(struct rds_incoming *inc,
struct iovec *first_iov, size_t size);
@@ -706,7 +703,6 @@ void rds_page_exit(void);
/* recv.c */
void rds_inc_init(struct rds_incoming *inc, struct rds_connection *conn,
__be32 saddr);
-void rds_inc_addref(struct rds_incoming *inc);
void rds_inc_put(struct rds_incoming *inc);
void rds_recv_incoming(struct rds_connection *conn, __be32 saddr, __be32 daddr,
struct rds_incoming *inc, gfp_t gfp, enum km_type km);
@@ -728,7 +724,6 @@ void rds_send_drop_to(struct rds_sock *r
typedef int (*is_acked_func)(struct rds_message *rm, uint64_t ack);
void rds_send_drop_acked(struct rds_connection *conn, u64 ack,
is_acked_func is_acked);
-void rds_send_remove_from_sock(struct list_head *messages, int status);
int rds_send_pong(struct rds_connection *conn, __be16 dport);
struct rds_message *rds_send_get_message(struct rds_connection *,
struct rm_rdma_op *);
--- a/net/rds/recv.c 2010-10-19 10:45:14.810008886 -0700
+++ b/net/rds/recv.c 2010-10-19 10:45:35.707903271 -0700
@@ -48,12 +48,11 @@ void rds_inc_init(struct rds_incoming *i
}
EXPORT_SYMBOL_GPL(rds_inc_init);
-void rds_inc_addref(struct rds_incoming *inc)
+static void rds_inc_addref(struct rds_incoming *inc)
{
rdsdebug("addref inc %p ref %d\n", inc, atomic_read(&inc->i_refcount));
atomic_inc(&inc->i_refcount);
}
-EXPORT_SYMBOL_GPL(rds_inc_addref);
void rds_inc_put(struct rds_incoming *inc)
{
--- a/net/rds/send.c 2010-10-19 10:45:43.272589570 -0700
+++ b/net/rds/send.c 2010-10-19 10:46:36.589425686 -0700
@@ -52,6 +52,8 @@ static int send_batch_count = 64;
module_param(send_batch_count, int, 0444);
MODULE_PARM_DESC(send_batch_count, " batch factor when working the send queue");
+static void rds_send_remove_from_sock(struct list_head *messages, int status);
+
/*
* Reset the send state. Callers must ensure that this doesn't race with
* rds_send_xmit().
@@ -555,7 +557,7 @@ EXPORT_SYMBOL_GPL(rds_send_get_message);
* removing the messages from the 'messages' list regardless of if it found
* the messages on the socket list or not.
*/
-void rds_send_remove_from_sock(struct list_head *messages, int status)
+static void rds_send_remove_from_sock(struct list_head *messages, int status)
{
unsigned long flags;
struct rds_sock *rs = NULL;
--- a/net/rds/tcp.c 2010-10-19 10:46:52.450865841 -0700
+++ b/net/rds/tcp.c 2010-10-19 10:47:54.492499537 -0700
@@ -41,7 +41,7 @@
/* only for info exporting */
static DEFINE_SPINLOCK(rds_tcp_tc_list_lock);
static LIST_HEAD(rds_tcp_tc_list);
-unsigned int rds_tcp_tc_count;
+static unsigned int rds_tcp_tc_count;
/* Track rds_tcp_connection structs so they can be cleaned up */
static DEFINE_SPINLOCK(rds_tcp_conn_lock);
@@ -243,7 +243,7 @@ static void rds_tcp_destroy_conns(void)
}
}
-void rds_tcp_exit(void)
+static void rds_tcp_exit(void)
{
rds_info_deregister_func(RDS_INFO_TCP_SOCKETS, rds_tcp_tc_info);
rds_tcp_listen_stop();
@@ -274,7 +274,7 @@ struct rds_transport rds_tcp_transport =
.t_prefer_loopback = 1,
};
-int rds_tcp_init(void)
+static int rds_tcp_init(void)
{
int ret;
--- a/net/rds/tcp.h 2010-10-19 10:54:15.559202132 -0700
+++ b/net/rds/tcp.h 2010-10-19 10:55:08.212010108 -0700
@@ -43,8 +43,6 @@ struct rds_tcp_statistics {
};
/* tcp.c */
-int rds_tcp_init(void);
-void rds_tcp_exit(void);
void rds_tcp_tune(struct socket *sock);
void rds_tcp_nonagle(struct socket *sock);
void rds_tcp_set_callbacks(struct socket *sock, struct rds_connection *conn);
--- a/net/rds/tcp_recv.c 2010-10-19 10:48:47.077278176 -0700
+++ b/net/rds/tcp_recv.c 2010-10-19 10:48:59.814436186 -0700
@@ -272,7 +272,8 @@ out:
}
/* the caller has to hold the sock lock */
-int rds_tcp_read_sock(struct rds_connection *conn, gfp_t gfp, enum km_type km)
+static int rds_tcp_read_sock(struct rds_connection *conn, gfp_t gfp,
+ enum km_type km)
{
struct rds_tcp_connection *tc = conn->c_transport_data;
struct socket *sock = tc->t_sock;
--- a/net/rds/tcp_send.c 2010-10-19 10:49:25.756795362 -0700
+++ b/net/rds/tcp_send.c 2010-10-19 10:49:40.974179535 -0700
@@ -63,7 +63,7 @@ void rds_tcp_xmit_complete(struct rds_co
}
/* the core send_sem serializes this with other xmit and shutdown */
-int rds_tcp_sendmsg(struct socket *sock, void *data, unsigned int len)
+static int rds_tcp_sendmsg(struct socket *sock, void *data, unsigned int len)
{
struct kvec vec = {
.iov_base = data,
--- a/net/rds/page.c 2010-10-19 11:00:18.352389347 -0700
+++ b/net/rds/page.c 2010-10-19 11:01:04.412612535 -0700
@@ -40,7 +40,8 @@ struct rds_page_remainder {
unsigned long r_offset;
};
-DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_page_remainder, rds_page_remainders);
+static DEFINE_PER_CPU_SHARED_ALIGNED(struct rds_page_remainder,
+ rds_page_remainders);
/*
* returns 0 on success or -errno on failure.
^ permalink raw reply
* Re: [Ksummit-2010-discuss] [v2] Remaining BKL users, what to do
From: Valdis.Kletnieks @ 2010-10-19 18:24 UTC (permalink / raw)
To: Greg KH
Cc: Dave Airlie, Arnd Bergmann, codalist, ksummit-2010-discuss,
autofs, Jan Harkes, Samuel Ortiz, Jan Kara,
Arnaldo Carvalho de Melo, netdev, Anders Larsen, linux-kernel,
dri-devel, Bryan Schumaker, Christoph Hellwig, Petr Vandrovec,
Mikulas Patocka, linux-fsdevel, Evgeniy Dushistov, Ingo Molnar,
Andrew Hendry, linux-media
In-Reply-To: <20101019004004.GB28380@kroah.com>
[-- Attachment #1: Type: text/plain, Size: 939 bytes --]
On Mon, 18 Oct 2010 17:40:04 PDT, Greg KH said:
> I do have access to this hardware, but its on an old single processor
> laptop, so any work that it would take to help do this development,
> really wouldn't be able to be tested to be valid at all.
The i810 is a graphics chipset embedded on the memory controller, which
was designed for the Intel Pentium II, Pentium III, and Celeron CPUs. Page 8
of the datasheet specifically says:
Processor/Host Bus Support
- Optimized for the Intel Pentium II processor, Intel Pentium III processor, and Intel
CeleronTM processor
- Supports processor 370-Pin Socket and SC242
connectors
- Supports 32-Bit System Bus Addressing
- 4 deep in-order queue; 4 or 1 deep request queue
- Supports Uni-processor systems only
So no need to clean it up for multiprocessor support.
http://download.intel.com/design/chipsets/datashts/29067602.pdf
http://www.intel.com/design/chipsets/specupdt/29069403.pdf
[-- Attachment #2: Type: application/pgp-signature, Size: 227 bytes --]
^ permalink raw reply
* Re: [RFC PATCH 1/9] ipvs network name space aware
From: Paul E. McKenney @ 2010-10-19 18:44 UTC (permalink / raw)
To: Hans Schillstrom
Cc: Daniel Lezcano, lvs-devel@vger.kernel.org, netdev@vger.kernel.org,
netfilter-devel@vger.kernel.org, horms@verge.net.au, ja@ssi.bg,
wensong@linux-vs.org
In-Reply-To: <201010181523.49568.hans.schillstrom@ericsson.com>
On Mon, Oct 18, 2010 at 03:23:48PM +0200, Hans Schillstrom wrote:
> On Monday 18 October 2010 13:37:38 Daniel Lezcano wrote:
> > On 10/18/2010 11:54 AM, Hans Schillstrom wrote:
> > > On Monday 18 October 2010 10:59:25 Daniel Lezcano wrote:
> > >
> > >> On 10/08/2010 01:16 PM, Hans Schillstrom wrote:
> > >>
> > >>> This part contains the include files
> > >>> where include/net/netns/ip_vs.h is new and contains all moved vars.
> > >>>
> > >>> SUMMARY
> > >>>
> > >>> include/net/ip_vs.h | 136 ++++---
> > >>> include/net/net_namespace.h | 2 +
> > >>> include/net/netns/ip_vs.h | 112 +++++
> > >>>
> > >>> Signed-off-by:Hans Schillstrom<hans.schillstrom@ericsson.com>
> > >>> ---
> > >>>
> > >>>
> > >>>
> > >> [ ... ]
> > >>
> > >>
> > >>> #ifdef CONFIG_IP_VS_IPV6
> > >>> diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
> > >>> index bd10a79..b59cdc5 100644
> > >>> --- a/include/net/net_namespace.h
> > >>> +++ b/include/net/net_namespace.h
> > >>> @@ -15,6 +15,7 @@
> > >>> #include<net/netns/ipv4.h>
> > >>> #include<net/netns/ipv6.h>
> > >>> #include<net/netns/dccp.h>
> > >>> +#include<net/netns/ip_vs.h>
> > >>> #include<net/netns/x_tables.h>
> > >>> #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
> > >>> #include<net/netns/conntrack.h>
> > >>> @@ -91,6 +92,7 @@ struct net {
> > >>> struct sk_buff_head wext_nlevents;
> > >>> #endif
> > >>> struct net_generic *gen;
> > >>> + struct netns_ipvs *ipvs;
> > >>> };
> > >>>
> > >>>
> > >> IMHO, it would be better to use the net_generic infra-structure instead
> > >> of adding a new field in the netns structure.
> > >>
> > >>
> > >>
> > > I realized that to, but the performance penalty is quite high with net_generic :-(
> > > But on the other hand if you are going to backport it, (without recompiling the kernel)
> > > you gonna need it!
> > >
> >
> > Hmm, yes. We don't want to have the init_net_ns performances to be impacted.
> >
> > You use here a pointer which will be dereferenced like the net_generic,
> > I don't think there will be
> > a big difference between using net_generic and using a pointer in the
> > net namespace structure.
> >
> > The difference is the id usage, but this one is based on the idr which
> > is quite fast.
> >
>
> I'm not so sure about that, have a look at net_generic and rcu_read_lock
> and compare
> ipvs = net->ipvs;
> vs.
> ipvs = net_generic(net, id)
>
> static inline void *net_generic(struct net *net, int id)
> {
> struct net_generic *ng;
> void *ptr;
>
> rcu_read_lock();
> ng = rcu_dereference(net->gen);
> BUG_ON(id == 0 || id > ng->len);
> ptr = ng->ptr[id - 1];
> rcu_read_unlock();
>
> return ptr;
> }
> ...
> static inline void rcu_read_lock(void)
> {
> __rcu_read_lock();
> __acquire(RCU);
> rcu_read_acquire();
> }
>
> Another way of doing it is to pass the ipvs ptr instead of the net ptr,
> and add *net to the ipvs struct.
>
> > We should experiment a bit here to compare both solutions.
> Agre
> >
> I single stepped through the rcu_read_lock() on a x86_64
> and it's quite many "stepi" that you need to enter :-(
Was this by chance with lockdep enabled? If not, could you please send
your .config?
Thanx, Paul
> > IMHO, we can (1) create a non-pointer netns_ipvs field in the net
> > namespace structure or (2) use a pointer [with net_generic].
> >
> > (1) is the faster but with the drawback of having a bigger memory
> > footprint even if the ipvs module is not loaded.
> > In this case we have to take care of what we store in the netns_ipvs
> > structure, that is reduce the per namespace table and so. At the first
> > glance, I think we can reduce this to the sysctls and a very few data,
> > for example using global tables tagged with the namespace and we don't
> > break the cacheline alignment optimization.
> >
> > (2) is slower but as the memory is allocated and freed when the module
> > is loaded/unloaded. What I don't like with this approach is we add some
> > overhead even if the netns is not compiled in the kernel.
> >
> or (3)
> Like (1) with data that needs to be cache aligned in "struct net"
> and the rest in an ipvs struct.
> Global hash tables or not ?
>
> > > My sugestion, take both with a configuration switch like:
> > > (i.e. avoid the rcu locking)
> > >
> > > --- net/ip_vs.h ---
> > > ...
> > > extern int ip_vs_net_id; /* net id for ip_vs */
> > >
> > >
> > > static inline struct netns_ipvs * net_ipvs(struct net* net, int id) {
> > > #ifdef CONFIG_IP_VS_FAST_NETNS
> > > return net->ipvs;
> > > #else
> > > return (struct netns_ipvs *)net_generic(net, id);
> > > #endif
> > > }
> > > ...
> > >
> > > and where you need the netns_ipvs struct ptr,
> > > [snip]
> > > struct ip_vs_conn *ip_vs_conn_in_get(struct net *net, ....
> > > {
> > > struct netns_ipvs *ipvs = net_ipvs(net, ip_vs_net_id);
> > > ...
> > >
> >
> > It is a nice way to wrap both solutions but at this point I don't think
> > it is worth to add a 3rd option to compile ipvs.
> >
> >
>
> --
> Regards
> Hans Schillstrom <hans.schillstrom@ericsson.com>
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* You have just exceeded your Email Quota limit
From: Webmail service @ 2010-10-19 18:29 UTC (permalink / raw)
This is to inform you that you have just exceeded your Email
Quota limit of 350MB. Before you can continue using your
email account, you are required to expand your Email Quota
because in less than 48 hours your email will be disable. Increase
your Email Quota and continue to use your email account.
To expand your Email Quota to 2.2GB, use the below web link:
http://do.my/gra/
Thank you for your understanding.
Copyright ©2010 Webmail service.
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
^ 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