* Re: [RFC 5/5] [powerpc] Implement a p1010rdb clock source.
From: Robin Holt @ 2011-08-08 9:15 UTC (permalink / raw)
To: Wolfgang Grandegger
Cc: socketcan-core-0fE9KPoRgkgATYTw5x5z8w, U Bhaskar-B22300,
Marc Kleine-Budde, netdev-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <4E3FA066.3020301-5Yr1BZd7O62+XT7JhA+gdA@public.gmane.org>
On Mon, Aug 08, 2011 at 10:37:58AM +0200, Wolfgang Grandegger wrote:
> On 08/06/2011 04:34 PM, Robin Holt wrote:
> > flexcan driver needs the clk_get, clk_get_rate, etc functions
> > to work. This patch provides the minimum functionality.
>
> This needs some more general thoughts... apart from the question where
> the code should go.
>
> Like for the MSCAN on the MPC5200, the user should be *able* to select
> an appropriate clock source and divider via DTS node properties.
> Currently it seems, that the DTS properties must match some
> pre-configured values, most likely set by the boot loader. Please
> correct me if I'm wrong. For me this is generic and should go into the
> Flexcan driver. From there, a platform specific function, e.g.
> flexcan_set_clock() might be called.
The P1010 really only supports the system bus clock for a source. I was
wrong last week when I said it supported either. That was a confusion
I have because of a task I was assigned a couple months ago.
It can select different divider values.
Robin
>
>
> > Signed-off-by: Robin Holt <holt-sJ/iWh9BUns@public.gmane.org>
> > To: Marc Kleine-Budde <mkl-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org>
> > To: Wolfgang Grandegger <wg-5Yr1BZd7O62+XT7JhA+gdA@public.gmane.org>
> > To: U Bhaskar-B22300 <B22300-KZfg59tc24xl57MIdRCFDg@public.gmane.org>
> > Cc: socketcan-core-0fE9KPoRgkgATYTw5x5z8w@public.gmane.org
> > Cc: netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> > ---
> > arch/powerpc/platforms/85xx/p1010rdb.c | 78 ++++++++++++++++++++++++++++++++
> > 1 files changed, 78 insertions(+), 0 deletions(-)
> >
> > diff --git a/arch/powerpc/platforms/85xx/p1010rdb.c b/arch/powerpc/platforms/85xx/p1010rdb.c
> > index 3540a88..8f78ddd 100644
> > --- a/arch/powerpc/platforms/85xx/p1010rdb.c
> > +++ b/arch/powerpc/platforms/85xx/p1010rdb.c
> > @@ -28,6 +28,7 @@
> > #include <asm/udbg.h>
> > #include <asm/mpic.h>
> > #include <asm/swiotlb.h>
> > +#include <asm/clk_interface.h>
> >
> > #include <sysdev/fsl_soc.h>
> > #include <sysdev/fsl_pci.h>
> > @@ -164,6 +165,82 @@ static void __init p1010_rdb_setup_arch(void)
> > printk(KERN_INFO "P1010 RDB board from Freescale Semiconductor\n");
> > }
> >
> > +/*
> > + * p1010rdb needs to provide a clock source for the flexcan driver.
> > + */
> > +struct clk {
> > + unsigned long rate;
> > +} p1010rdb_system_clk;
> > +
> > +static struct clk *p1010_rdb_clk_get(struct device *dev, const char *id)
> > +{
>
> I persoanlly don't like the
>
> > + struct clk *clk;
> > + u32 *of_property;
> > + unsigned long clock_freq, clock_divider;
> > + const char *dev_init_name;
> > +
> > + if (!dev)
> > + return ERR_PTR(-ENOENT);
> > +
> > + /*
> > + * The can devices are named ffe1c000.can0 and ffe1d000.can1 on
> > + * the p1010rdb. Check for the "can" portion of that name before
> > + * returning a clock source.
> > + */
> > + dev_init_name = dev_name(dev);
> > + if (strlen(dev_init_name) != 13)
> > + return ERR_PTR(-ENOENT);
> > + dev_init_name += 9;
> > + if (strncmp(dev_init_name, "can", 3))
> > + return ERR_PTR(-ENOENT);
>
> There are dedicated functions to find the of node. But with my above
> proposal we do not need to provide p1010_rdb_clk_get().
>
> More comments on the other patches soon...
>
> Wolfgang.
^ permalink raw reply
* [PATCH net] vmxnet3: Don't enable vlan filters in promiscuous mode.
From: Jesse Gross @ 2011-08-08 9:15 UTC (permalink / raw)
To: David Miller
Cc: netdev, Scott J. Goldman, Shreyas Bhatewara, VMware PV-Drivers
The vmxnet3 driver enables vlan filters if filtering is enabled for
any vlan. In promiscuous mode the filter table is cleared to in
order to disable filtering. However, if a vlan device is subsequently
created that vlan will be added to the filter, re-engaging it. As a
result, not only do we not see all the vlans in promiscuous mode, we
don't even see vlans for which a filter was previously created.
CC: Scott J. Goldman <scottjg@vmware.com>
CC: Shreyas Bhatewara <sbhatewara@vmware.com>
CC: VMware PV-Drivers <pv-drivers@vmware.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
drivers/net/vmxnet3/vmxnet3_drv.c | 34 ++++++++++++++++++++--------------
1 files changed, 20 insertions(+), 14 deletions(-)
diff --git a/drivers/net/vmxnet3/vmxnet3_drv.c b/drivers/net/vmxnet3/vmxnet3_drv.c
index 1cbacb3..0959583 100644
--- a/drivers/net/vmxnet3/vmxnet3_drv.c
+++ b/drivers/net/vmxnet3/vmxnet3_drv.c
@@ -1929,14 +1929,17 @@ static void
vmxnet3_vlan_rx_add_vid(struct net_device *netdev, u16 vid)
{
struct vmxnet3_adapter *adapter = netdev_priv(netdev);
- u32 *vfTable = adapter->shared->devRead.rxFilterConf.vfTable;
- unsigned long flags;
- VMXNET3_SET_VFTABLE_ENTRY(vfTable, vid);
- spin_lock_irqsave(&adapter->cmd_lock, flags);
- VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
- VMXNET3_CMD_UPDATE_VLAN_FILTERS);
- spin_unlock_irqrestore(&adapter->cmd_lock, flags);
+ if (!(netdev->flags & IFF_PROMISC)) {
+ u32 *vfTable = adapter->shared->devRead.rxFilterConf.vfTable;
+ unsigned long flags;
+
+ VMXNET3_SET_VFTABLE_ENTRY(vfTable, vid);
+ spin_lock_irqsave(&adapter->cmd_lock, flags);
+ VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
+ VMXNET3_CMD_UPDATE_VLAN_FILTERS);
+ spin_unlock_irqrestore(&adapter->cmd_lock, flags);
+ }
set_bit(vid, adapter->active_vlans);
}
@@ -1946,14 +1949,17 @@ static void
vmxnet3_vlan_rx_kill_vid(struct net_device *netdev, u16 vid)
{
struct vmxnet3_adapter *adapter = netdev_priv(netdev);
- u32 *vfTable = adapter->shared->devRead.rxFilterConf.vfTable;
- unsigned long flags;
- VMXNET3_CLEAR_VFTABLE_ENTRY(vfTable, vid);
- spin_lock_irqsave(&adapter->cmd_lock, flags);
- VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
- VMXNET3_CMD_UPDATE_VLAN_FILTERS);
- spin_unlock_irqrestore(&adapter->cmd_lock, flags);
+ if (!(netdev->flags & IFF_PROMISC)) {
+ u32 *vfTable = adapter->shared->devRead.rxFilterConf.vfTable;
+ unsigned long flags;
+
+ VMXNET3_CLEAR_VFTABLE_ENTRY(vfTable, vid);
+ spin_lock_irqsave(&adapter->cmd_lock, flags);
+ VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
+ VMXNET3_CMD_UPDATE_VLAN_FILTERS);
+ spin_unlock_irqrestore(&adapter->cmd_lock, flags);
+ }
clear_bit(vid, adapter->active_vlans);
}
--
1.7.4.1
^ permalink raw reply related
* Re: [PATCH 1/5] Define the function to write sock's security context to seq_file.
From: Rongqing Li @ 2011-08-08 9:32 UTC (permalink / raw)
To: Stephen Smalley; +Cc: netdev, selinux, lsm
In-Reply-To: <1312552618.19283.51.camel@moss-pluto>
On 08/05/2011 09:56 PM, Stephen Smalley wrote:
> On Fri, 2011-08-05 at 16:58 +0800, rongqing.li@windriver.com wrote:
>> From: Roy.Li<rongqing.li@windriver.com>
>>
>> This function will write the sock's security context to a seq_file
>> and return the error code, and the number of characters successfully
>> written is written in int pointers parameter.
>>
>> This function will be called when export socket information to proc.
>>
>> Signed-off-by: Roy.Li<rongqing.li@windriver.com>
>> ---
>> include/net/sock.h | 1 +
>> net/core/sock.c | 26 ++++++++++++++++++++++++++
>> 2 files changed, 27 insertions(+), 0 deletions(-)
>
>> diff --git a/net/core/sock.c b/net/core/sock.c
>> index bc745d0..1126a49 100644
>> --- a/net/core/sock.c
>> +++ b/net/core/sock.c
>> @@ -2254,6 +2254,32 @@ void sk_common_release(struct sock *sk)
>> }
>> EXPORT_SYMBOL(sk_common_release);
>>
>> +int sock_write_secctx(struct sock *sk, struct seq_file *seq, int *len)
>> +{
>> + struct flowi fl;
>> + char *ctx = NULL;
>> + u32 ctxlen;
>> + int res = 0;
>> +
>> + *len = 0;
>> +
>> + if (sk == NULL)
>> + return -EINVAL;
>
> Is this ever possible?
>
Hi Stephen:
When output the tcp information to proc by tcp4_seq_show and
tcp state is TCP_SEQ_STATE_TIME_WAIT, the input argument v is
struct inet_timewait_sock, it seem we can not get the struct sock
from struct inet_timewait_sock, so I assume the sk is NULL in that
condition.
static int tcp4_seq_show(struct seq_file *seq, void *v)
{
case TCP_SEQ_STATE_TIME_WAIT:
get_timewait4_sock(v, seq, st->num, &len);
break;
}
}
>> + res = security_socket_getsockname(sk->sk_socket);
>> + if (res)
>> + return res;
>
> I'm not sure it is a good idea to output nothing if permission is denied
> to the socket, as opposed to some well-defined string indicating that
> condition. Particularly if someone later adds another field to
> the /proc files after the context; we don't want the contents of that
> field to be interpreted as the context if permission was denied.
>
From your review, I redesign the output information as below.
when disable SELinux, print "(none)" in proc
when enable SELinux, no error on getting security context, print the
real security context
when enable SELinux, there is error on getting security context, print
"??"
Do you think it is OK?
Thanks very much
-Roy
>> +
>> + security_sk_classify_flow(sk,&fl);
>> +
>> + res = security_secid_to_secctx(fl.flowi_secid,&ctx,&ctxlen);
>> + if (res)
>> + return res;
>
> Likewise, if we couldn't map the secid to a secctx for some reason, we
> likely ought to output some well-defined string indicating that
> condition.
>
>> +
>> + seq_printf(seq, " %s%n", ctx, len);
>> + security_release_secctx(ctx, ctxlen);
>> + return res;
>> +}
>> +
>> static DEFINE_RWLOCK(proto_list_lock);
>> static LIST_HEAD(proto_list);
>>
>
--
Best Reagrds,
Roy | RongQing Li
^ permalink raw reply
* Re: [RFC 4/4] [powerpc] Implement a p1010rdb clock source.
From: Marc Kleine-Budde @ 2011-08-08 9:32 UTC (permalink / raw)
To: Wolfgang Grandegger
Cc: Netdev, U Bhaskar-B22300, Kumar Gala,
socketcan-core-0fE9KPoRgkgATYTw5x5z8w,
linuxppc-dev-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org Development
In-Reply-To: <4E3FA301.4050005-5Yr1BZd7O62+XT7JhA+gdA@public.gmane.org>
[-- Attachment #1.1: Type: text/plain, Size: 3411 bytes --]
On 08/08/2011 10:49 AM, Wolfgang Grandegger wrote:
> On 08/06/2011 10:59 PM, Kumar Gala wrote:
>>
>> On Aug 6, 2011, at 3:50 PM, Robin Holt wrote:
>>
>>> On Sat, Aug 06, 2011 at 11:52:45AM -0500, Kumar Gala wrote:
>>>>
>>>> On Aug 6, 2011, at 8:58 AM, Marc Kleine-Budde wrote:
>>>>
>>>>> On 08/06/2011 06:05 AM, Robin Holt wrote:
>>>>>> flexcan driver needs the clk_get, clk_get_rate, etc functions
>>>>>> to work. This patch provides the minimum functionality.
>>>>>
>>>>> This patch has to go via the powerpc git tree. Added
>>>>> linuxppc-dev-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org on CC.
>>>>>
>>>>>> Signed-off-by: Robin Holt <holt-sJ/iWh9BUns@public.gmane.org>
>>>>>> To: Marc Kleine-Budde <mkl-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org>
>>>>>> To: Wolfgang Grandegger <wg-5Yr1BZd7O62+XT7JhA+gdA@public.gmane.org>
>>>>>> To: U Bhaskar-B22300 <B22300-KZfg59tc24xl57MIdRCFDg@public.gmane.org>
>>>>>> Cc: socketcan-core-0fE9KPoRgkgATYTw5x5z8w@public.gmane.org
>>>>>> Cc: netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>>>>>> ---
>>>>>> arch/powerpc/platforms/85xx/p1010rdb.c | 78 ++++++++++++++++++++++++++++++++
>>>>>> 1 files changed, 78 insertions(+), 0 deletions(-)
>>>>
>>>> NAK.
>>>>
>>>> This doesn't look right at all. We should be doing something based on the device tree node that isn't board specific.
>>>>
>>>> I believe Bhaskar has a version of flexcan support that he's been working on cleanup up for upstream.
>>>
>>> That version may be similar to what is in the freescale BSP which puts
>>> the clock functions inside flexcan.c
>>>
>>> The powerpc arch already provides a means for individual boards to provide
>>> the clock functions. I am not posting this patch here for acceptance
>>> for powerpc and I am sure I will get feedback there when I post to
>>> their mailing list. I am posting it here only to show that the flexcan
>>> developers earlier assertion that this can and should be done in the arch
>>> tree is correct and will work for the p1010 assuming we can get changes
>>> into the arch/powerpc directory to implement these clk_* functions.
>>
>> My point is that I don't think they should live in the arch code. The clk_* functions you want to implement are tied more the FlexCAN IP than anything arch specific. As such I believe they should be in the driver.
>>
>> For example when FSL has a P9999 with FlexCAN on it, we should NOT have to add any arch code to support it.
>
> The Flexcan is found on ARM and now also on PowerPC SOCs. My current
> understanding is that the ability to set the clock source and divider is
> only available on PowerPC SOCs and therefore it's clearly arch specific
> and should go to arch/powerpc/sysdev/fsl_soc.c if it's common for all
> PowerPC platforms. What do you think?
There is a bit in the CAN-Controller that selects the clock (at least on
arm). The current driver just supports the bus clock. Support for the
other, the oscillator clock, has not been implemented so far.
IIRC the oscillator clock has a frequency that results in worse standard
timing than the bus clock.
cheers, Marc
--
Pengutronix e.K. | Marc Kleine-Budde |
Industrial Linux Solutions | Phone: +49-231-2826-924 |
Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]
[-- Attachment #2: Type: text/plain, Size: 188 bytes --]
_______________________________________________
Socketcan-core mailing list
Socketcan-core-0fE9KPoRgkgATYTw5x5z8w@public.gmane.org
https://lists.berlios.de/mailman/listinfo/socketcan-core
^ permalink raw reply
* Matter of interest
From: Alfred L. Leng @ 2011-08-08 9:01 UTC (permalink / raw)
To: netdev
It is understandable for you to grow apprehensive reading from me today since there was no previous communication between us. I am Alfred L. Leng (Dr.), the managing attorney of R & L Associates Malaysia. \v\vI came across your contact via scrupulous search conducted by an IT specialist whom I have employed for the purpose of same. \v\vThe present communication is prompted by a legitimate business interest I have to share and ensue with you. I have left out the detail of the business matter until I hear from you as measure of security. \v\vPlease be reminded that I would not have incurred great efforts to contact you for a matter of nonentity or to waste your time or mine. Further, be confident that the matter in question is beneficial and legit with no ramification of any sort; details of the matter I will uncover to you in my following communication. \v\vKindly, revert to me accordingly for further exhaustive dialogue. \v\vThank you for your time as I am looking forward for yours and business relation. \v\vSincerely \v\vAlfred Leng.
^ permalink raw reply
* Nine Hundred And Fifty Thousand Pounds Has Been Given To You In The Benz Offer, Provide Your
From: jerrysmith232 @ 2011-08-08 10:08 UTC (permalink / raw)
Names
Address
Phone Number
Country
^ permalink raw reply
* Re: [PATCH] netfilter: avoid double free in nf_reinject
From: Jan Engelhardt @ 2011-08-08 10:13 UTC (permalink / raw)
To: David Miller; +Cc: ja, kaber, netfilter-devel, netdev, kswamy
In-Reply-To: <20110807.221158.636741102851927559.davem@davemloft.net>
On Monday 2011-08-08 07:11, David Miller wrote:
>From: Julian Anastasov <ja@ssi.bg>
>Date: Fri, 5 Aug 2011 13:36:28 +0300 (EEST)
>
>>
>> NF_STOLEN means skb was already freed
>>
>> Signed-off-by: Julian Anastasov <ja@ssi.bg>
>> ---
>>
>> May be fixes IPVS+ip_queue problem reported by Kumar Swamy:
>>
>> http://marc.info/?l=linux-virtual-server&m=131098073717449&w=2
>
>Since the netfilter maintainers are taking too damn long to integrate
>bug fixes (and this has been happening for months), I'm going to apply
>this directly.
I am offering myself as a candidate for a substitute.
^ permalink raw reply
* Re: [EXAMPLE CODE] Parasite thread injection and TCP connection hijacking
From: Tejun Heo @ 2011-08-08 10:20 UTC (permalink / raw)
To: Andy Lutomirski
Cc: Matt Helsley, Pavel Emelyanov, Nathan Lynch, Oren Laadan,
Daniel Lezcano, James E.J. Bottomley, David S. Miller,
linux-kernel, netdev
In-Reply-To: <20110806130037.GD23937@htj.dyndns.org>
On Sat, Aug 06, 2011 at 03:00:37PM +0200, Tejun Heo wrote:
> Hello,
>
> On Sat, Aug 06, 2011 at 08:45:28AM -0400, Andy Lutomirski wrote:
> > > 2. Decide where to inject the foreign code and save the original code
> > > with PTRACE_PEEKDATA. Tracer can poke any mapped area regardless
> > > of protection flags but it can't add execution permission to the
> > > code, so it needs to choose memory area which already has X flag
> > > set. The example code uses the page the %rip is in.
> >
> > If the process is executing from the vsyscall page, then you'll
> > probably fail. (Admittedly, this is rather unlikely, given that the
> > vsyscalls are now exactly one instruction.) Presumably you also
> > fail if executing from a read-only MAP_SHARED mapping.
>
> Heh, yeah, I originally thought about scanning /proc/PID/maps to look
> for the page to use but was lazy and just used %rip. I think that
> should work. I'll note the problem in README.
Okay, updated README.
http://code.google.com/p/ptrace-parasite/source/browse/README
Thanks.
--
tejun
^ permalink raw reply
* "Winner"
From: =?utf-8?q?Motorola_=C2=AE__=3Cinfo=40mail=2Ecom=3E?= @ 2011-08-09 1:56 UTC (permalink / raw)
To: Recipients
You Won £600,000.00GBP from Motorola Promotion 2011. Bee Line Courier Service UK (beeline@diploma.com)for your Check delivery with your Name,Address,Country,Phone Number
^ permalink raw reply
* [PATCH 2/9] drivers/net/wireless/wl12xx: add missing kfree
From: Julia Lawall @ 2011-08-08 11:17 UTC (permalink / raw)
To: Luciano Coelho
Cc: kernel-janitors, John W. Linville, linux-wireless, netdev,
linux-kernel
From: Julia Lawall <julia@diku.dk>
In each case, the freed data should be freed in the error handling code as
well.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@exists@
local idexpression x;
statement S,S1;
expression E;
identifier fl;
expression *ptr != NULL;
@@
x = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
<... when != x
when != if (...) { <+...kfree(x)...+> }
when any
when != true x == NULL
x->fl
...>
(
if (x == NULL) S1
|
if (...) { ... when != x
when forall
(
return \(0\|<+...x...+>\|ptr\);
|
* return ...;
)
}
)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
drivers/net/wireless/wl12xx/acx.c | 6 +-----
drivers/net/wireless/wl12xx/testmode.c | 5 ++++-
2 files changed, 5 insertions(+), 6 deletions(-)
diff --git a/drivers/net/wireless/wl12xx/testmode.c b/drivers/net/wireless/wl12xx/testmode.c
index 5d5e1ef..88add68 100644
--- a/drivers/net/wireless/wl12xx/testmode.c
+++ b/drivers/net/wireless/wl12xx/testmode.c
@@ -139,12 +139,15 @@ static int wl1271_tm_cmd_interrogate(struct wl1271 *wl, struct nlattr *tb[])
if (ret < 0) {
wl1271_warning("testmode cmd interrogate failed: %d", ret);
+ kfree(cmd);
return ret;
}
skb = cfg80211_testmode_alloc_reply_skb(wl->hw->wiphy, sizeof(*cmd));
- if (!skb)
+ if (!skb) {
+ kfree(cmd);
return -ENOMEM;
+ }
NLA_PUT(skb, WL1271_TM_ATTR_DATA, sizeof(*cmd), cmd);
diff --git a/drivers/net/wireless/wl12xx/acx.c b/drivers/net/wireless/wl12xx/acx.c
index 7e33f1f..34f6ab5 100644
--- a/drivers/net/wireless/wl12xx/acx.c
+++ b/drivers/net/wireless/wl12xx/acx.c
@@ -77,8 +77,6 @@ int wl1271_acx_sleep_auth(struct wl1271 *wl, u8 sleep_auth)
auth->sleep_auth = sleep_auth;
ret = wl1271_cmd_configure(wl, ACX_SLEEP_AUTH, auth, sizeof(*auth));
- if (ret < 0)
- return ret;
out:
kfree(auth);
@@ -624,10 +622,8 @@ int wl1271_acx_cca_threshold(struct wl1271 *wl)
ret = wl1271_cmd_configure(wl, ACX_CCA_THRESHOLD,
detection, sizeof(*detection));
- if (ret < 0) {
+ if (ret < 0)
wl1271_warning("failed to set cca threshold: %d", ret);
- return ret;
- }
out:
kfree(detection);
^ permalink raw reply related
* [PATCH 9/9] drivers/net/wireless/wl1251: add missing kfree
From: Julia Lawall @ 2011-08-08 11:18 UTC (permalink / raw)
To: Luciano Coelho
Cc: kernel-janitors, John W. Linville, linux-wireless, netdev,
linux-kernel
From: Julia Lawall <julia@diku.dk>
In each case, the kfree already at the end of the function is also needed
in the error case.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@exists@
local idexpression x;
statement S,S1;
expression E;
identifier fl;
expression *ptr != NULL;
@@
x = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
<... when != x
when != if (...) { <+...kfree(x)...+> }
when any
when != true x == NULL
x->fl
...>
(
if (x == NULL) S1
|
if (...) { ... when != x
when forall
(
return \(0\|<+...x...+>\|ptr\);
|
* return ...;
)
}
)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
drivers/net/wireless/wl1251/acx.c | 6 +-----
drivers/net/wireless/wl1251/cmd.c | 2 +-
2 files changed, 2 insertions(+), 6 deletions(-)
diff --git a/drivers/net/wireless/wl1251/acx.c b/drivers/net/wireless/wl1251/acx.c
index ef8370e..ad87a1a 100644
--- a/drivers/net/wireless/wl1251/acx.c
+++ b/drivers/net/wireless/wl1251/acx.c
@@ -140,8 +140,6 @@ int wl1251_acx_sleep_auth(struct wl1251 *wl, u8 sleep_auth)
auth->sleep_auth = sleep_auth;
ret = wl1251_cmd_configure(wl, ACX_SLEEP_AUTH, auth, sizeof(*auth));
- if (ret < 0)
- return ret;
out:
kfree(auth);
@@ -681,10 +679,8 @@ int wl1251_acx_cca_threshold(struct wl1251 *wl)
ret = wl1251_cmd_configure(wl, ACX_CCA_THRESHOLD,
detection, sizeof(*detection));
- if (ret < 0) {
+ if (ret < 0)
wl1251_warning("failed to set cca threshold: %d", ret);
- return ret;
- }
out:
kfree(detection);
diff --git a/drivers/net/wireless/wl1251/cmd.c b/drivers/net/wireless/wl1251/cmd.c
index 81f164b..d14d69d 100644
--- a/drivers/net/wireless/wl1251/cmd.c
+++ b/drivers/net/wireless/wl1251/cmd.c
@@ -241,7 +241,7 @@ int wl1251_cmd_data_path(struct wl1251 *wl, u8 channel, bool enable)
if (ret < 0) {
wl1251_error("tx %s cmd for channel %d failed",
enable ? "start" : "stop", channel);
- return ret;
+ goto out;
}
wl1251_debug(DEBUG_BOOT, "tx %s cmd channel %d",
^ permalink raw reply related
* Re: [RFC 5/5] [powerpc] Implement a p1010rdb clock source.
From: Robin Holt @ 2011-08-08 11:31 UTC (permalink / raw)
To: Wolfgang Grandegger
Cc: socketcan-core-0fE9KPoRgkgATYTw5x5z8w, U Bhaskar-B22300,
Marc Kleine-Budde, netdev-u79uwXL29TY76Z2rM5mHXA
In-Reply-To: <4E3FA066.3020301-5Yr1BZd7O62+XT7JhA+gdA@public.gmane.org>
On Mon, Aug 08, 2011 at 10:37:58AM +0200, Wolfgang Grandegger wrote:
> On 08/06/2011 04:34 PM, Robin Holt wrote:
> > flexcan driver needs the clk_get, clk_get_rate, etc functions
> > to work. This patch provides the minimum functionality.
>
> This needs some more general thoughts... apart from the question where
> the code should go.
>
> Like for the MSCAN on the MPC5200, the user should be *able* to select
> an appropriate clock source and divider via DTS node properties.
> Currently it seems, that the DTS properties must match some
> pre-configured values, most likely set by the boot loader. Please
> correct me if I'm wrong. For me this is generic and should go into the
> Flexcan driver. From there, a platform specific function, e.g.
> flexcan_set_clock() might be called.
OK. Dug a bit more. The p1010 built-in clocksource seems to be the
periphereal clock frequency which is system bus frequency divided
by 2. The clock source can not be changed, but the clock divider can
by freezing the interface and setting the CTRL register. This appears
to only be done by the boot loader. I do not see why we can not leave
that functionality in the boot loader and then go back to a variation
on my earlier flexcan_clk_* patch. Is that close to the direction you
think we should go or have I completely misunderstood your wishes?
Thanks,
Robin
^ permalink raw reply
* [patch 11/13] [PATCH] qeth: support forced signal adapter indications
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Einar Lueck
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117_qeth_siga_r.patch --]
[-- Type: text/plain, Size: 907 bytes --]
From: Einar Lueck <elelueck@de.ibm.com>
This patch ensures that signal adapter commands are issued if they are
indicated to be required.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
drivers/s390/net/qeth_core.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
--- a/drivers/s390/net/qeth_core.h
+++ b/drivers/s390/net/qeth_core.h
@@ -231,7 +231,8 @@ static inline int qeth_is_ipa_enabled(st
#define QETH_IN_BUF_COUNT_MAX 128
#define QETH_MAX_BUFFER_ELEMENTS(card) ((card)->qdio.in_buf_size >> 12)
#define QETH_IN_BUF_REQUEUE_THRESHOLD(card) \
- ((card)->qdio.in_buf_pool.buf_count / 2)
+ ((card)->ssqd.qdioac1 & AC1_SIGA_INPUT_NEEDED ? 1 : \
+ ((card)->qdio.in_buf_pool.buf_count / 2))
/* buffers we have to be behind before we get a PCI */
#define QETH_PCI_THRESHOLD_A(card) ((card)->qdio.in_buf_pool.buf_count+1)
^ permalink raw reply
* [patch 10/13] [PATCH] qdio: support forced signal adapter indications
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Einar Lueck, Jan Glauber
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117_qdio_siga_r.patch --]
[-- Type: text/plain, Size: 2357 bytes --]
This patch ensures that signal adapter commands are issued if they are
indicated to be required.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
arch/s390/include/asm/qdio.h | 10 ++++++++++
drivers/s390/cio/qdio.h | 9 ---------
drivers/s390/cio/qdio_main.c | 5 +----
3 files changed, 11 insertions(+), 13 deletions(-)
--- a/arch/s390/include/asm/qdio.h
+++ b/arch/s390/include/asm/qdio.h
@@ -279,6 +279,16 @@ struct qdio_outbuf_state {
#define CHSC_AC1_INITIATE_INPUTQ 0x80
+
+/* qdio adapter-characteristics-1 flag */
+#define AC1_SIGA_INPUT_NEEDED 0x40 /* process input queues */
+#define AC1_SIGA_OUTPUT_NEEDED 0x20 /* process output queues */
+#define AC1_SIGA_SYNC_NEEDED 0x10 /* ask hypervisor to sync */
+#define AC1_AUTOMATIC_SYNC_ON_THININT 0x08 /* set by hypervisor */
+#define AC1_AUTOMATIC_SYNC_ON_OUT_PCI 0x04 /* set by hypervisor */
+#define AC1_SC_QEBSM_AVAILABLE 0x02 /* available for subchannel */
+#define AC1_SC_QEBSM_ENABLED 0x01 /* enabled for subchannel */
+
#define CHSC_AC2_DATA_DIV_AVAILABLE 0x0010
#define CHSC_AC2_DATA_DIV_ENABLED 0x0002
--- a/drivers/s390/cio/qdio.h
+++ b/drivers/s390/cio/qdio.h
@@ -87,15 +87,6 @@ enum qdio_irq_states {
#define CHSC_FLAG_QDIO_CAPABILITY 0x80
#define CHSC_FLAG_VALIDITY 0x40
-/* qdio adapter-characteristics-1 flag */
-#define AC1_SIGA_INPUT_NEEDED 0x40 /* process input queues */
-#define AC1_SIGA_OUTPUT_NEEDED 0x20 /* process output queues */
-#define AC1_SIGA_SYNC_NEEDED 0x10 /* ask hypervisor to sync */
-#define AC1_AUTOMATIC_SYNC_ON_THININT 0x08 /* set by hypervisor */
-#define AC1_AUTOMATIC_SYNC_ON_OUT_PCI 0x04 /* set by hypervisor */
-#define AC1_SC_QEBSM_AVAILABLE 0x02 /* available for subchannel */
-#define AC1_SC_QEBSM_ENABLED 0x01 /* enabled for subchannel */
-
/* SIGA flags */
#define QDIO_SIGA_WRITE 0x00
#define QDIO_SIGA_READ 0x01
--- a/drivers/s390/cio/qdio_main.c
+++ b/drivers/s390/cio/qdio_main.c
@@ -1592,12 +1592,9 @@ set:
used = atomic_add_return(count, &q->nr_buf_used) - count;
BUG_ON(used + count > QDIO_MAX_BUFFERS_PER_Q);
- /* no need to signal as long as the adapter had free buffers */
- if (used)
- return 0;
-
if (need_siga_in(q))
return qdio_siga_input(q);
+
return 0;
}
^ permalink raw reply
* [patch 12/13] [PATCH] qeth: exploit asynchronous delivery of storage blocks
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Einar Lueck
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117_qeth_cq.patch --]
[-- Type: text/plain, Size: 28255 bytes --]
From: Einar Lueck <elelueck@de.ibm.com>
This patch exploits the QDIO support for asynchronous delivery of storage
blocks for Hipersockets. The exploitation is not configured per default and
may be enabled via the function qeth_configure_cq.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
drivers/s390/net/qeth_core.h | 28 +-
drivers/s390/net/qeth_core_main.c | 495 ++++++++++++++++++++++++++++++++++----
2 files changed, 480 insertions(+), 43 deletions(-)
--- a/drivers/s390/net/qeth_core.h
+++ b/drivers/s390/net/qeth_core.h
@@ -110,6 +110,10 @@ struct qeth_perf_stats {
unsigned int sc_dp_p;
unsigned int sc_p_dp;
+ /* qdio_cq_handler: number of times called, time spent in */
+ __u64 cq_start_time;
+ unsigned int cq_cnt;
+ unsigned int cq_time;
/* qdio_input_handler: number of times called, time spent in */
__u64 inbound_start_time;
unsigned int inbound_cnt;
@@ -376,6 +380,11 @@ enum qeth_qdio_buffer_states {
* outbound: filled by driver; owned by hardware in order to be sent
*/
QETH_QDIO_BUF_PRIMED,
+ /*
+ * inbound: not applicable
+ * outbound: handled via transfer pending / completion queue
+ */
+ QETH_QDIO_BUF_HANDLED_DELAYED,
};
enum qeth_qdio_info_states {
@@ -413,8 +422,11 @@ struct qeth_qdio_out_buffer {
atomic_t state;
int next_element_to_fill;
struct sk_buff_head skb_list;
- struct list_head ctx_list;
int is_header[16];
+
+ struct qaob *aob;
+ struct qeth_qdio_out_q *q;
+ struct qeth_qdio_out_buffer *next_pending;
};
struct qeth_card;
@@ -427,7 +439,8 @@ enum qeth_out_q_states {
struct qeth_qdio_out_q {
struct qdio_buffer qdio_bufs[QDIO_MAX_BUFFERS_PER_Q];
- struct qeth_qdio_out_buffer bufs[QDIO_MAX_BUFFERS_PER_Q];
+ struct qeth_qdio_out_buffer *bufs[QDIO_MAX_BUFFERS_PER_Q];
+ struct qdio_outbuf_state *bufstates; /* convenience pointer */
int queue_no;
struct qeth_card *card;
atomic_t state;
@@ -448,7 +461,9 @@ struct qeth_qdio_out_q {
struct qeth_qdio_info {
atomic_t state;
/* input */
+ int no_in_queues;
struct qeth_qdio_q *in_q;
+ struct qeth_qdio_q *c_q;
struct qeth_qdio_buffer_pool in_buf_pool;
struct qeth_qdio_buffer_pool init_pool;
int in_buf_size;
@@ -456,6 +471,7 @@ struct qeth_qdio_info {
/* output */
int no_out_queues;
struct qeth_qdio_out_q **out_qs;
+ struct qdio_outbuf_state *out_bufstates;
/* priority queueing */
int do_prio_queueing;
@@ -527,6 +543,12 @@ enum qeth_cmd_buffer_state {
BUF_STATE_PROCESSED,
};
+enum qeth_cq {
+ QETH_CQ_DISABLED = 0,
+ QETH_CQ_ENABLED = 1,
+ QETH_CQ_NOTAVAILABLE = 2,
+};
+
struct qeth_ipato {
int enabled;
int invert4;
@@ -651,6 +673,7 @@ struct qeth_card_options {
int rx_sg_cb;
enum qeth_ipa_isolation_modes isolation;
int sniffer;
+ enum qeth_cq cq;
};
/*
@@ -888,6 +911,7 @@ void qeth_dbf_longtext(debug_info_t *id,
int qeth_core_ethtool_get_settings(struct net_device *, struct ethtool_cmd *);
int qeth_set_access_ctrl_online(struct qeth_card *card);
int qeth_hdr_chk_and_bounce(struct sk_buff *, int);
+int qeth_configure_cq(struct qeth_card *, enum qeth_cq);
int qeth_hw_trap(struct qeth_card *, enum qeth_diags_trap_action);
int qeth_query_ipassists(struct qeth_card *, enum qeth_prot_versions prot);
--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -44,6 +44,7 @@ struct qeth_card_list_struct qeth_core_c
EXPORT_SYMBOL_GPL(qeth_core_card_list);
struct kmem_cache *qeth_core_header_cache;
EXPORT_SYMBOL_GPL(qeth_core_header_cache);
+static struct kmem_cache *qeth_qdio_outbuf_cache;
static struct device *qeth_core_root_dev;
static unsigned int known_devices[][6] = QETH_MODELLIST_ARRAY;
@@ -56,6 +57,10 @@ static struct qeth_cmd_buffer *qeth_get_
static void qeth_setup_ccw(struct qeth_channel *, unsigned char *, __u32);
static void qeth_free_buffer_pool(struct qeth_card *);
static int qeth_qdio_establish(struct qeth_card *);
+static void qeth_free_qdio_buffers(struct qeth_card *);
+static void qeth_clear_output_buffer(struct qeth_qdio_out_q *queue,
+ struct qeth_qdio_out_buffer *buf,
+ enum qeth_qdio_buffer_states newbufstate);
static inline const char *qeth_get_cardname(struct qeth_card *card)
@@ -239,6 +244,150 @@ int qeth_realloc_buffer_pool(struct qeth
}
EXPORT_SYMBOL_GPL(qeth_realloc_buffer_pool);
+static inline int qeth_cq_init(struct qeth_card *card)
+{
+ int rc;
+
+ if (card->options.cq == QETH_CQ_ENABLED) {
+ QETH_DBF_TEXT(SETUP, 2, "cqinit");
+ memset(card->qdio.c_q->qdio_bufs, 0,
+ QDIO_MAX_BUFFERS_PER_Q * sizeof(struct qdio_buffer));
+ card->qdio.c_q->next_buf_to_init = 127;
+ rc = do_QDIO(CARD_DDEV(card), QDIO_FLAG_SYNC_INPUT,
+ card->qdio.no_in_queues - 1, 0,
+ 127);
+ if (rc) {
+ QETH_DBF_TEXT_(SETUP, 2, "1err%d", rc);
+ goto out;
+ }
+ }
+ rc = 0;
+out:
+ return rc;
+}
+
+static inline int qeth_alloc_cq(struct qeth_card *card)
+{
+ int rc;
+
+ if (card->options.cq == QETH_CQ_ENABLED) {
+ int i;
+ struct qdio_outbuf_state *outbuf_states;
+
+ QETH_DBF_TEXT(SETUP, 2, "cqon");
+ card->qdio.c_q = kzalloc(sizeof(struct qeth_qdio_q),
+ GFP_KERNEL);
+ if (!card->qdio.c_q) {
+ rc = -1;
+ goto kmsg_out;
+ }
+ QETH_DBF_HEX(SETUP, 2, &card->qdio.c_q, sizeof(void *));
+
+ for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i) {
+ card->qdio.c_q->bufs[i].buffer =
+ &card->qdio.c_q->qdio_bufs[i];
+ }
+
+ card->qdio.no_in_queues = 2;
+
+ card->qdio.out_bufstates = (struct qdio_outbuf_state *)
+ kzalloc(card->qdio.no_out_queues *
+ QDIO_MAX_BUFFERS_PER_Q *
+ sizeof(struct qdio_outbuf_state), GFP_KERNEL);
+ outbuf_states = card->qdio.out_bufstates;
+ if (outbuf_states == NULL) {
+ rc = -1;
+ goto free_cq_out;
+ }
+ for (i = 0; i < card->qdio.no_out_queues; ++i) {
+ card->qdio.out_qs[i]->bufstates = outbuf_states;
+ outbuf_states += QDIO_MAX_BUFFERS_PER_Q;
+ }
+ } else {
+ QETH_DBF_TEXT(SETUP, 2, "nocq");
+ card->qdio.c_q = NULL;
+ card->qdio.no_in_queues = 1;
+ }
+ QETH_DBF_TEXT_(SETUP, 2, "iqc%d", card->qdio.no_in_queues);
+ rc = 0;
+out:
+ return rc;
+free_cq_out:
+ kfree(card->qdio.c_q);
+ card->qdio.c_q = NULL;
+kmsg_out:
+ dev_err(&card->gdev->dev, "Failed to create completion queue\n");
+ goto out;
+}
+
+static inline void qeth_free_cq(struct qeth_card *card)
+{
+ if (card->qdio.c_q) {
+ --card->qdio.no_in_queues;
+ kfree(card->qdio.c_q);
+ card->qdio.c_q = NULL;
+ }
+ kfree(card->qdio.out_bufstates);
+ card->qdio.out_bufstates = NULL;
+}
+
+static inline void qeth_cleanup_handled_pending(struct qeth_qdio_out_q *q,
+ int bidx, int forced_cleanup)
+{
+ if (q->bufs[bidx]->next_pending != NULL) {
+ struct qeth_qdio_out_buffer *head = q->bufs[bidx];
+ struct qeth_qdio_out_buffer *c = q->bufs[bidx]->next_pending;
+
+ while (c) {
+ if (forced_cleanup ||
+ atomic_read(&c->state) ==
+ QETH_QDIO_BUF_HANDLED_DELAYED) {
+ struct qeth_qdio_out_buffer *f = c;
+ QETH_CARD_TEXT(f->q->card, 5, "fp");
+ QETH_CARD_TEXT_(f->q->card, 5, "%lx", (long) f);
+ c = f->next_pending;
+ BUG_ON(head->next_pending != f);
+ head->next_pending = c;
+ kmem_cache_free(qeth_qdio_outbuf_cache, f);
+ } else {
+ head = c;
+ c = c->next_pending;
+ }
+
+ }
+ }
+}
+
+
+static inline void qeth_qdio_handle_aob(struct qeth_card *card,
+ unsigned long phys_aob_addr) {
+ struct qaob *aob;
+ struct qeth_qdio_out_buffer *buffer;
+
+ aob = (struct qaob *) phys_to_virt(phys_aob_addr);
+ QETH_CARD_TEXT(card, 5, "haob");
+ QETH_CARD_TEXT_(card, 5, "%lx", phys_aob_addr);
+ buffer = (struct qeth_qdio_out_buffer *) aob->user1;
+ QETH_CARD_TEXT_(card, 5, "%lx", aob->user1);
+
+ BUG_ON(buffer == NULL);
+
+ buffer->aob = NULL;
+ qeth_clear_output_buffer(buffer->q, buffer,
+ QETH_QDIO_BUF_HANDLED_DELAYED);
+ /* from here on: do not touch buffer anymore */
+ qdio_release_aob(aob);
+}
+
+static inline int qeth_is_cq(struct qeth_card *card, unsigned int queue)
+{
+ return card->options.cq == QETH_CQ_ENABLED &&
+ card->qdio.c_q != NULL &&
+ queue != 0 &&
+ queue == card->qdio.no_in_queues - 1;
+}
+
+
static int qeth_issue_next_read(struct qeth_card *card)
{
int rc;
@@ -681,6 +830,7 @@ EXPORT_SYMBOL_GPL(qeth_do_run_thread);
void qeth_schedule_recovery(struct qeth_card *card)
{
QETH_CARD_TEXT(card, 2, "startrec");
+ WARN_ON(1);
if (qeth_set_thread_start_bit(card, QETH_RECOVER_THREAD) == 0)
schedule_work(&card->kernel_thread_starter);
}
@@ -884,7 +1034,8 @@ out:
}
static void qeth_clear_output_buffer(struct qeth_qdio_out_q *queue,
- struct qeth_qdio_out_buffer *buf)
+ struct qeth_qdio_out_buffer *buf,
+ enum qeth_qdio_buffer_states newbufstate)
{
int i;
struct sk_buff *skb;
@@ -912,21 +1063,36 @@ static void qeth_clear_output_buffer(str
buf->buffer->element[15].eflags = 0;
buf->buffer->element[15].sflags = 0;
buf->next_element_to_fill = 0;
- atomic_set(&buf->state, QETH_QDIO_BUF_EMPTY);
+ atomic_set(&buf->state, newbufstate);
+}
+
+static void qeth_clear_outq_buffers(struct qeth_qdio_out_q *q, int free)
+{
+ int j;
+
+ for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) {
+ if (!q->bufs[j])
+ continue;
+ qeth_cleanup_handled_pending(q, j, free);
+ qeth_clear_output_buffer(q, q->bufs[j], QETH_QDIO_BUF_EMPTY);
+ if (free) {
+ kmem_cache_free(qeth_qdio_outbuf_cache, q->bufs[j]);
+ q->bufs[j] = NULL;
+ }
+ }
}
void qeth_clear_qdio_buffers(struct qeth_card *card)
{
- int i, j;
+ int i;
QETH_CARD_TEXT(card, 2, "clearqdbf");
/* clear outbound buffers to free skbs */
- for (i = 0; i < card->qdio.no_out_queues; ++i)
+ for (i = 0; i < card->qdio.no_out_queues; ++i) {
if (card->qdio.out_qs[i]) {
- for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j)
- qeth_clear_output_buffer(card->qdio.out_qs[i],
- &card->qdio.out_qs[i]->bufs[j]);
+ qeth_clear_outq_buffers(card->qdio.out_qs[i], 0);
}
+ }
}
EXPORT_SYMBOL_GPL(qeth_clear_qdio_buffers);
@@ -945,11 +1111,14 @@ static void qeth_free_buffer_pool(struct
static void qeth_free_qdio_buffers(struct qeth_card *card)
{
- int i, j;
+ int i;
if (atomic_xchg(&card->qdio.state, QETH_QDIO_UNINITIALIZED) ==
QETH_QDIO_UNINITIALIZED)
return;
+
+ qeth_free_cq(card);
+
kfree(card->qdio.in_q);
card->qdio.in_q = NULL;
/* inbound buffer pool */
@@ -957,9 +1126,7 @@ static void qeth_free_qdio_buffers(struc
/* free outbound qdio_qs */
if (card->qdio.out_qs) {
for (i = 0; i < card->qdio.no_out_queues; ++i) {
- for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j)
- qeth_clear_output_buffer(card->qdio.out_qs[i],
- &card->qdio.out_qs[i]->bufs[j]);
+ qeth_clear_outq_buffers(card->qdio.out_qs[i], 1);
kfree(card->qdio.out_qs[i]);
}
kfree(card->qdio.out_qs);
@@ -1053,6 +1220,7 @@ static void qeth_set_intial_options(stru
card->options.performance_stats = 0;
card->options.rx_sg_cb = QETH_RX_SG_CB;
card->options.isolation = ISOLATION_MODE_NONE;
+ card->options.cq = QETH_CQ_DISABLED;
}
static int qeth_do_start_thread(struct qeth_card *card, unsigned long thread)
@@ -1182,6 +1350,7 @@ static int qeth_determine_card_type(stru
card->info.type = known_devices[i][QETH_DEV_MODEL_IND];
card->qdio.no_out_queues =
known_devices[i][QETH_QUEUE_NO_IND];
+ card->qdio.no_in_queues = 1;
card->info.is_multicast_different =
known_devices[i][QETH_MULTICAST_IND];
qeth_get_channel_path_desc(card);
@@ -2029,6 +2198,37 @@ static int qeth_ulp_setup(struct qeth_ca
return rc;
}
+static int qeth_init_qdio_out_buf(struct qeth_qdio_out_q *q, int bidx)
+{
+ int rc;
+ struct qeth_qdio_out_buffer *newbuf;
+
+ rc = 0;
+ newbuf = kmem_cache_zalloc(qeth_qdio_outbuf_cache, GFP_ATOMIC);
+ if (!newbuf) {
+ rc = -ENOMEM;
+ goto out;
+ }
+ newbuf->buffer = &q->qdio_bufs[bidx];
+ skb_queue_head_init(&newbuf->skb_list);
+ lockdep_set_class(&newbuf->skb_list.lock, &qdio_out_skb_queue_key);
+ newbuf->q = q;
+ newbuf->aob = NULL;
+ newbuf->next_pending = q->bufs[bidx];
+ atomic_set(&newbuf->state, QETH_QDIO_BUF_EMPTY);
+ q->bufs[bidx] = newbuf;
+ if (q->bufstates) {
+ q->bufstates[bidx].user = newbuf;
+ QETH_CARD_TEXT_(q->card, 2, "nbs%d", bidx);
+ QETH_CARD_TEXT_(q->card, 2, "%lx", (long) newbuf);
+ QETH_CARD_TEXT_(q->card, 2, "%lx",
+ (long) newbuf->next_pending);
+ }
+out:
+ return rc;
+}
+
+
static int qeth_alloc_qdio_buffers(struct qeth_card *card)
{
int i, j;
@@ -2040,7 +2240,7 @@ static int qeth_alloc_qdio_buffers(struc
return 0;
card->qdio.in_q = kmalloc(sizeof(struct qeth_qdio_q),
- GFP_KERNEL);
+ GFP_KERNEL);
if (!card->qdio.in_q)
goto out_nomem;
QETH_DBF_TEXT(SETUP, 2, "inq");
@@ -2053,6 +2253,7 @@ static int qeth_alloc_qdio_buffers(struc
/* inbound buffer pool */
if (qeth_alloc_buffer_pool(card))
goto out_freeinq;
+
/* outbound */
card->qdio.out_qs =
kmalloc(card->qdio.no_out_queues *
@@ -2070,21 +2271,30 @@ static int qeth_alloc_qdio_buffers(struc
card->qdio.out_qs[i]->queue_no = i;
/* give outbound qeth_qdio_buffers their qdio_buffers */
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) {
- card->qdio.out_qs[i]->bufs[j].buffer =
- &card->qdio.out_qs[i]->qdio_bufs[j];
- skb_queue_head_init(&card->qdio.out_qs[i]->bufs[j].
- skb_list);
- lockdep_set_class(
- &card->qdio.out_qs[i]->bufs[j].skb_list.lock,
- &qdio_out_skb_queue_key);
- INIT_LIST_HEAD(&card->qdio.out_qs[i]->bufs[j].ctx_list);
+ BUG_ON(card->qdio.out_qs[i]->bufs[j] != NULL);
+ if (qeth_init_qdio_out_buf(card->qdio.out_qs[i], j))
+ goto out_freeoutqbufs;
}
}
+
+ /* completion */
+ if (qeth_alloc_cq(card))
+ goto out_freeoutq;
+
return 0;
+out_freeoutqbufs:
+ while (j > 0) {
+ --j;
+ kmem_cache_free(qeth_qdio_outbuf_cache,
+ card->qdio.out_qs[i]->bufs[j]);
+ card->qdio.out_qs[i]->bufs[j] = NULL;
+ }
out_freeoutq:
- while (i > 0)
+ while (i > 0) {
kfree(card->qdio.out_qs[--i]);
+ qeth_clear_outq_buffers(card->qdio.out_qs[i], 1);
+ }
kfree(card->qdio.out_qs);
card->qdio.out_qs = NULL;
out_freepool:
@@ -2401,13 +2611,21 @@ int qeth_init_qdio_queues(struct qeth_ca
QETH_DBF_TEXT_(SETUP, 2, "1err%d", rc);
return rc;
}
+
+ /* completion */
+ rc = qeth_cq_init(card);
+ if (rc) {
+ return rc;
+ }
+
/* outbound queue */
for (i = 0; i < card->qdio.no_out_queues; ++i) {
memset(card->qdio.out_qs[i]->qdio_bufs, 0,
QDIO_MAX_BUFFERS_PER_Q * sizeof(struct qdio_buffer));
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) {
qeth_clear_output_buffer(card->qdio.out_qs[i],
- &card->qdio.out_qs[i]->bufs[j]);
+ card->qdio.out_qs[i]->bufs[j],
+ QETH_QDIO_BUF_EMPTY);
}
card->qdio.out_qs[i]->card = card;
card->qdio.out_qs[i]->next_buf_to_fill = 0;
@@ -2789,8 +3007,6 @@ void qeth_queue_input_buffer(struct qeth
qeth_get_micros() -
card->perf_stats.inbound_do_qdio_start_time;
if (rc) {
- dev_warn(&card->gdev->dev,
- "QDIO reported an error, rc=%i\n", rc);
QETH_CARD_TEXT(card, 2, "qinberr");
}
queue->next_buf_to_init = (queue->next_buf_to_init + count) %
@@ -2864,12 +3080,12 @@ static int qeth_switch_to_nonpacking_if_
queue->card->perf_stats.sc_p_dp++;
queue->do_pack = 0;
/* flush packing buffers */
- buffer = &queue->bufs[queue->next_buf_to_fill];
+ buffer = queue->bufs[queue->next_buf_to_fill];
if ((atomic_read(&buffer->state) ==
QETH_QDIO_BUF_EMPTY) &&
(buffer->next_element_to_fill > 0)) {
atomic_set(&buffer->state,
- QETH_QDIO_BUF_PRIMED);
+ QETH_QDIO_BUF_PRIMED);
flush_count++;
queue->next_buf_to_fill =
(queue->next_buf_to_fill + 1) %
@@ -2880,6 +3096,7 @@ static int qeth_switch_to_nonpacking_if_
return flush_count;
}
+
/*
* Called to flush a packing buffer if no more pci flags are on the queue.
* Checks if there is a packing buffer and prepares it to be flushed.
@@ -2889,7 +3106,7 @@ static int qeth_flush_buffers_on_no_pci(
{
struct qeth_qdio_out_buffer *buffer;
- buffer = &queue->bufs[queue->next_buf_to_fill];
+ buffer = queue->bufs[queue->next_buf_to_fill];
if ((atomic_read(&buffer->state) == QETH_QDIO_BUF_EMPTY) &&
(buffer->next_element_to_fill > 0)) {
/* it's a packing buffer */
@@ -2910,10 +3127,14 @@ static void qeth_flush_buffers(struct qe
unsigned int qdio_flags;
for (i = index; i < index + count; ++i) {
- buf = &queue->bufs[i % QDIO_MAX_BUFFERS_PER_Q];
+ int bidx = i % QDIO_MAX_BUFFERS_PER_Q;
+ buf = queue->bufs[bidx];
buf->buffer->element[buf->next_element_to_fill - 1].eflags |=
SBAL_EFLAGS_LAST_ENTRY;
+ if (queue->bufstates)
+ queue->bufstates[bidx].user = buf;
+
if (queue->card->info.type == QETH_CARD_TYPE_IQD)
continue;
@@ -2965,6 +3186,9 @@ static void qeth_flush_buffers(struct qe
if (rc == QDIO_ERROR_SIGA_TARGET)
return;
QETH_CARD_TEXT(queue->card, 2, "flushbuf");
+ QETH_CARD_TEXT_(queue->card, 2, " q%d", queue->queue_no);
+ QETH_CARD_TEXT_(queue->card, 2, " idx%d", index);
+ QETH_CARD_TEXT_(queue->card, 2, " c%d", count);
QETH_CARD_TEXT_(queue->card, 2, " err%d", rc);
/* this must not happen under normal circumstances. if it
@@ -3026,14 +3250,120 @@ void qeth_qdio_start_poll(struct ccw_dev
}
EXPORT_SYMBOL_GPL(qeth_qdio_start_poll);
+int qeth_configure_cq(struct qeth_card *card, enum qeth_cq cq)
+{
+ int rc;
+
+ if (card->options.cq == QETH_CQ_NOTAVAILABLE) {
+ rc = -1;
+ goto out;
+ } else {
+ if (card->options.cq == cq) {
+ rc = 0;
+ goto out;
+ }
+
+ if (card->state != CARD_STATE_DOWN &&
+ card->state != CARD_STATE_RECOVER) {
+ rc = -1;
+ goto out;
+ }
+
+ qeth_free_qdio_buffers(card);
+ card->options.cq = cq;
+ rc = 0;
+ }
+out:
+ return rc;
+
+}
+EXPORT_SYMBOL_GPL(qeth_configure_cq);
+
+
+static void qeth_qdio_cq_handler(struct qeth_card *card,
+ unsigned int qdio_err,
+ unsigned int queue, int first_element, int count) {
+ struct qeth_qdio_q *cq = card->qdio.c_q;
+ int i;
+ int rc;
+
+ if (!qeth_is_cq(card, queue))
+ goto out;
+
+ QETH_CARD_TEXT_(card, 5, "qcqhe%d", first_element);
+ QETH_CARD_TEXT_(card, 5, "qcqhc%d", count);
+ QETH_CARD_TEXT_(card, 5, "qcqherr%d", qdio_err);
+
+ if (qdio_err) {
+ netif_stop_queue(card->dev);
+ qeth_schedule_recovery(card);
+ goto out;
+ }
+
+ if (card->options.performance_stats) {
+ card->perf_stats.cq_cnt++;
+ card->perf_stats.cq_start_time = qeth_get_micros();
+ }
+
+ for (i = first_element; i < first_element + count; ++i) {
+ int bidx = i % QDIO_MAX_BUFFERS_PER_Q;
+ struct qdio_buffer *buffer = &cq->qdio_bufs[bidx];
+ int e;
+
+ e = 0;
+ while (buffer->element[e].addr) {
+ unsigned long phys_aob_addr;
+
+ phys_aob_addr = (unsigned long) buffer->element[e].addr;
+ qeth_qdio_handle_aob(card, phys_aob_addr);
+ buffer->element[e].addr = NULL;
+ buffer->element[e].eflags = 0;
+ buffer->element[e].sflags = 0;
+ buffer->element[e].length = 0;
+
+ ++e;
+ }
+
+ buffer->element[15].eflags = 0;
+ buffer->element[15].sflags = 0;
+ }
+ rc = do_QDIO(CARD_DDEV(card), QDIO_FLAG_SYNC_INPUT, queue,
+ card->qdio.c_q->next_buf_to_init,
+ count);
+ if (rc) {
+ dev_warn(&card->gdev->dev,
+ "QDIO reported an error, rc=%i\n", rc);
+ QETH_CARD_TEXT(card, 2, "qcqherr");
+ }
+ card->qdio.c_q->next_buf_to_init = (card->qdio.c_q->next_buf_to_init
+ + count) % QDIO_MAX_BUFFERS_PER_Q;
+
+ netif_wake_queue(card->dev);
+
+ if (card->options.performance_stats) {
+ int delta_t = qeth_get_micros();
+ delta_t -= card->perf_stats.cq_start_time;
+ card->perf_stats.cq_time += delta_t;
+ }
+out:
+ return;
+}
+
void qeth_qdio_input_handler(struct ccw_device *ccwdev, unsigned int qdio_err,
- unsigned int queue, int first_element, int count,
+ unsigned int queue, int first_elem, int count,
unsigned long card_ptr)
{
struct qeth_card *card = (struct qeth_card *)card_ptr;
- if (qdio_err)
+ QETH_CARD_TEXT_(card, 2, "qihq%d", queue);
+ QETH_CARD_TEXT_(card, 2, "qiec%d", qdio_err);
+
+ if (qeth_is_cq(card, queue))
+ qeth_qdio_cq_handler(card, qdio_err, queue, first_elem, count);
+ else if (qdio_err)
qeth_schedule_recovery(card);
+
+
}
EXPORT_SYMBOL_GPL(qeth_qdio_input_handler);
@@ -3059,9 +3389,26 @@ void qeth_qdio_output_handler(struct ccw
qeth_get_micros();
}
for (i = first_element; i < (first_element + count); ++i) {
- buffer = &queue->bufs[i % QDIO_MAX_BUFFERS_PER_Q];
+ int bidx = i % QDIO_MAX_BUFFERS_PER_Q;
+ buffer = queue->bufs[bidx];
qeth_handle_send_error(card, buffer, qdio_error);
- qeth_clear_output_buffer(queue, buffer);
+
+ if (queue->bufstates &&
+ (queue->bufstates[bidx].flags &
+ QDIO_OUTBUF_STATE_FLAG_PENDING) != 0) {
+ buffer->aob = queue->bufstates[bidx].aob;
+ QETH_CARD_TEXT_(queue->card, 5, "pel%d", bidx);
+ QETH_CARD_TEXT_(queue->card, 5, "aob");
+ QETH_CARD_TEXT_(queue->card, 5, "%lx",
+ virt_to_phys(buffer->aob));
+ BUG_ON(bidx < 0 || bidx >= QDIO_MAX_BUFFERS_PER_Q);
+ if (qeth_init_qdio_out_buf(queue, bidx))
+ qeth_schedule_recovery(card);
+ } else {
+ qeth_clear_output_buffer(queue, buffer,
+ QETH_QDIO_BUF_EMPTY);
+ }
+ qeth_cleanup_handled_pending(queue, bidx, 0);
}
atomic_sub(count, &queue->used_buffers);
/* check if we need to do something on this outbound queue */
@@ -3293,7 +3640,7 @@ int qeth_do_send_packet_fast(struct qeth
QETH_OUT_Q_LOCKED) != QETH_OUT_Q_UNLOCKED);
/* ... now we've got the queue */
index = queue->next_buf_to_fill;
- buffer = &queue->bufs[queue->next_buf_to_fill];
+ buffer = queue->bufs[queue->next_buf_to_fill];
/*
* check if buffer is empty to make sure that we do not 'overtake'
* ourselves and try to fill a buffer that is already primed
@@ -3327,7 +3674,7 @@ int qeth_do_send_packet(struct qeth_card
while (atomic_cmpxchg(&queue->state, QETH_OUT_Q_UNLOCKED,
QETH_OUT_Q_LOCKED) != QETH_OUT_Q_UNLOCKED);
start_index = queue->next_buf_to_fill;
- buffer = &queue->bufs[queue->next_buf_to_fill];
+ buffer = queue->bufs[queue->next_buf_to_fill];
/*
* check if buffer is empty to make sure that we do not 'overtake'
* ourselves and try to fill a buffer that is already primed
@@ -3349,7 +3696,7 @@ int qeth_do_send_packet(struct qeth_card
queue->next_buf_to_fill =
(queue->next_buf_to_fill + 1) %
QDIO_MAX_BUFFERS_PER_Q;
- buffer = &queue->bufs[queue->next_buf_to_fill];
+ buffer = queue->bufs[queue->next_buf_to_fill];
/* we did a step forward, so check buffer state
* again */
if (atomic_read(&buffer->state) !=
@@ -3927,6 +4274,20 @@ static void qeth_determine_capabilities(
if (rc)
QETH_DBF_TEXT_(SETUP, 2, "6err%d", rc);
+ QETH_DBF_TEXT_(SETUP, 2, "qfmt%d", card->ssqd.qfmt);
+ QETH_DBF_TEXT_(SETUP, 2, "%d", card->ssqd.qdioac1);
+ QETH_DBF_TEXT_(SETUP, 2, "%d", card->ssqd.qdioac3);
+ QETH_DBF_TEXT_(SETUP, 2, "icnt%d", card->ssqd.icnt);
+ if (!((card->ssqd.qfmt != QDIO_IQDIO_QFMT) ||
+ ((card->ssqd.qdioac1 & CHSC_AC1_INITIATE_INPUTQ) == 0) ||
+ ((card->ssqd.qdioac3 & CHSC_AC3_FORMAT2_CQ_AVAILABLE) == 0))) {
+ dev_info(&card->gdev->dev,
+ "Completion Queueing supported\n");
+ } else {
+ card->options.cq = QETH_CQ_NOTAVAILABLE;
+ }
+
+
out_offline:
if (ddev_offline == 1)
ccw_device_set_offline(ddev);
@@ -3934,6 +4295,24 @@ out:
return;
}
+static inline void qeth_qdio_establish_cq(struct qeth_card *card,
+ struct qdio_buffer **in_sbal_ptrs,
+ void (**queue_start_poll) (struct ccw_device *, int, unsigned long)) {
+ int i;
+
+ if (card->options.cq == QETH_CQ_ENABLED) {
+ int offset = QDIO_MAX_BUFFERS_PER_Q *
+ (card->qdio.no_in_queues - 1);
+ i = QDIO_MAX_BUFFERS_PER_Q * (card->qdio.no_in_queues - 1);
+ for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i) {
+ in_sbal_ptrs[offset + i] = (struct qdio_buffer *)
+ virt_to_phys(card->qdio.c_q->bufs[i].buffer);
+ }
+
+ queue_start_poll[card->qdio.no_in_queues - 1] = NULL;
+ }
+}
+
static int qeth_qdio_establish(struct qeth_card *card)
{
struct qdio_initialize init_data;
@@ -3956,22 +4335,28 @@ static int qeth_qdio_establish(struct qe
qeth_create_qib_param_field(card, qib_param_field);
qeth_create_qib_param_field_blkt(card, qib_param_field);
- in_sbal_ptrs = kmalloc(QDIO_MAX_BUFFERS_PER_Q * sizeof(void *),
+ in_sbal_ptrs = kmalloc(card->qdio.no_in_queues *
+ QDIO_MAX_BUFFERS_PER_Q * sizeof(void *),
GFP_KERNEL);
if (!in_sbal_ptrs) {
rc = -ENOMEM;
goto out_free_qib_param;
}
- for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i)
+ for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i) {
in_sbal_ptrs[i] = (struct qdio_buffer *)
virt_to_phys(card->qdio.in_q->bufs[i].buffer);
+ }
- queue_start_poll = kmalloc(sizeof(void *) * 1, GFP_KERNEL);
+ queue_start_poll = kzalloc(sizeof(void *) * card->qdio.no_in_queues,
+ GFP_KERNEL);
if (!queue_start_poll) {
rc = -ENOMEM;
goto out_free_in_sbals;
}
- queue_start_poll[0] = card->discipline.start_poll;
+ for (i = 0; i < card->qdio.no_in_queues; ++i)
+ queue_start_poll[i] = card->discipline.start_poll;
+
+ qeth_qdio_establish_cq(card, in_sbal_ptrs, queue_start_poll);
out_sbal_ptrs =
kmalloc(card->qdio.no_out_queues * QDIO_MAX_BUFFERS_PER_Q *
@@ -3983,7 +4368,7 @@ static int qeth_qdio_establish(struct qe
for (i = 0, k = 0; i < card->qdio.no_out_queues; ++i)
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j, ++k) {
out_sbal_ptrs[k] = (struct qdio_buffer *)virt_to_phys(
- card->qdio.out_qs[i]->bufs[j].buffer);
+ card->qdio.out_qs[i]->bufs[j]->buffer);
}
memset(&init_data, 0, sizeof(struct qdio_initialize));
@@ -3991,7 +4376,7 @@ static int qeth_qdio_establish(struct qe
init_data.q_format = qeth_get_qdio_q_format(card);
init_data.qib_param_field_format = 0;
init_data.qib_param_field = qib_param_field;
- init_data.no_input_qs = 1;
+ init_data.no_input_qs = card->qdio.no_in_queues;
init_data.no_output_qs = card->qdio.no_out_queues;
init_data.input_handler = card->discipline.input_handler;
init_data.output_handler = card->discipline.output_handler;
@@ -3999,6 +4384,7 @@ static int qeth_qdio_establish(struct qe
init_data.int_parm = (unsigned long) card;
init_data.input_sbal_addr_array = (void **) in_sbal_ptrs;
init_data.output_sbal_addr_array = (void **) out_sbal_ptrs;
+ init_data.output_sbal_state_array = card->qdio.out_bufstates;
init_data.scan_threshold =
(card->info.type == QETH_CARD_TYPE_IQD) ? 8 : 32;
@@ -4015,6 +4401,17 @@ static int qeth_qdio_establish(struct qe
qdio_free(CARD_DDEV(card));
}
}
+
+ switch (card->options.cq) {
+ case QETH_CQ_ENABLED:
+ dev_info(&card->gdev->dev, "Completion Queue support enabled");
+ break;
+ case QETH_CQ_DISABLED:
+ dev_info(&card->gdev->dev, "Completion Queue support disabled");
+ break;
+ default:
+ break;
+ }
out:
kfree(out_sbal_ptrs);
out_free_queue_start_poll:
@@ -4193,6 +4590,8 @@ static inline int qeth_create_skb_frag(s
(*pskb)->truesize += data_len;
(*pfrag)++;
}
+
+
return 0;
}
@@ -4666,6 +5065,8 @@ static struct {
{"tx do_QDIO count"},
{"tx csum"},
{"tx lin"},
+ {"cq handler count"},
+ {"cq handler time"}
};
int qeth_core_get_sset_count(struct net_device *dev, int stringset)
@@ -4724,6 +5125,8 @@ void qeth_core_get_ethtool_stats(struct
data[32] = card->perf_stats.outbound_do_qdio_cnt;
data[33] = card->perf_stats.tx_csum;
data[34] = card->perf_stats.tx_lin;
+ data[35] = card->perf_stats.cq_cnt;
+ data[36] = card->perf_stats.cq_time;
}
EXPORT_SYMBOL_GPL(qeth_core_get_ethtool_stats);
@@ -4882,7 +5285,16 @@ static int __init qeth_core_init(void)
goto slab_err;
}
+ qeth_qdio_outbuf_cache = kmem_cache_create("qeth_buf",
+ sizeof(struct qeth_qdio_out_buffer), 0, 0, NULL);
+ if (!qeth_qdio_outbuf_cache) {
+ rc = -ENOMEM;
+ goto cqslab_err;
+ }
+
return 0;
+cqslab_err:
+ kmem_cache_destroy(qeth_core_header_cache);
slab_err:
root_device_unregister(qeth_core_root_dev);
register_err:
@@ -4907,6 +5319,7 @@ static void __exit qeth_core_exit(void)
&driver_attr_group);
ccwgroup_driver_unregister(&qeth_core_ccwgroup_driver);
ccw_driver_unregister(&qeth_ccw_driver);
+ kmem_cache_destroy(qeth_qdio_outbuf_cache);
kmem_cache_destroy(qeth_core_header_cache);
qeth_unregister_dbf_views();
pr_info("core functions removed\n");
^ permalink raw reply
* [patch 09/13] [PATCH] qdio: support asynchronous delivery of storage blocks
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Einar Lueck, Jan Glauber
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117_qdio_cq.patch --]
[-- Type: text/plain, Size: 29697 bytes --]
This patch introduces support for asynchronous delivery of storage blocks for
Hipersockets. Upper layers may exploit this functionality to reuse SBALs for
which the delivery status is still pending.
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
arch/s390/include/asm/qdio.h | 68 ++++++++++++
drivers/s390/cio/qdio.h | 29 +++++
drivers/s390/cio/qdio_debug.c | 3
drivers/s390/cio/qdio_main.c | 203 +++++++++++++++++++++++++++++++++-----
drivers/s390/cio/qdio_setup.c | 83 ++++++++++++++-
drivers/s390/cio/qdio_thinint.c | 88 +++++++++-------
drivers/s390/net/qeth_core_main.c | 30 ++++-
7 files changed, 424 insertions(+), 80 deletions(-)
--- a/arch/s390/include/asm/qdio.h
+++ b/arch/s390/include/asm/qdio.h
@@ -123,6 +123,40 @@ struct slibe {
};
/**
+ * struct qaob - queue asynchronous operation block
+ * @res0: reserved parameters
+ * @res1: reserved parameter
+ * @res2: reserved parameter
+ * @res3: reserved parameter
+ * @aorc: asynchronous operation return code
+ * @flags: internal flags
+ * @cbtbs: control block type
+ * @sb_count: number of storage blocks
+ * @sba: storage block element addresses
+ * @dcount: size of storage block elements
+ * @user0: user defineable value
+ * @res4: reserved paramater
+ * @user1: user defineable value
+ * @user2: user defineable value
+ */
+struct qaob {
+ u64 res0[6];
+ u8 res1;
+ u8 res2;
+ u8 res3;
+ u8 aorc;
+ u8 flags;
+ u16 cbtbs;
+ u8 sb_count;
+ u64 sba[QDIO_MAX_ELEMENTS_PER_BUFFER];
+ u16 dcount[QDIO_MAX_ELEMENTS_PER_BUFFER];
+ u64 user0;
+ u64 res4[2];
+ u64 user1;
+ u64 user2;
+} __attribute__ ((packed, aligned(256)));
+
+/**
* struct slib - storage list information block (SLIB)
* @nsliba: next SLIB address (if any)
* @sla: SL address
@@ -225,6 +259,31 @@ struct slsb {
#define CHSC_AC2_DATA_DIV_AVAILABLE 0x0010
#define CHSC_AC2_DATA_DIV_ENABLED 0x0002
+/**
+ * struct qdio_outbuf_state - SBAL related asynchronous operation information
+ * (for communication with upper layer programs)
+ * (only required for use with completion queues)
+ * @flags: flags indicating state of buffer
+ * @aob: pointer to QAOB used for the particular SBAL
+ * @user: pointer to upper layer program's state information related to SBAL
+ * (stored in user1 data of QAOB)
+ */
+struct qdio_outbuf_state {
+ u8 flags;
+ struct qaob *aob;
+ void *user;
+};
+
+#define QDIO_OUTBUF_STATE_FLAG_NONE 0x00
+#define QDIO_OUTBUF_STATE_FLAG_PENDING 0x01
+
+#define CHSC_AC1_INITIATE_INPUTQ 0x80
+
+#define CHSC_AC2_DATA_DIV_AVAILABLE 0x0010
+#define CHSC_AC2_DATA_DIV_ENABLED 0x0002
+
+#define CHSC_AC3_FORMAT2_CQ_AVAILABLE 0x8000
+
struct qdio_ssqd_desc {
u8 flags;
u8:8;
@@ -243,8 +302,7 @@ struct qdio_ssqd_desc {
u64 sch_token;
u8 mro;
u8 mri;
- u8:8;
- u8 sbalic;
+ u16 qdioac3;
u16:16;
u8:8;
u8 mmwc;
@@ -280,9 +338,11 @@ typedef void qdio_handler_t(struct ccw_d
* @no_output_qs: number of output queues
* @input_handler: handler to be called for input queues
* @output_handler: handler to be called for output queues
+ * @queue_start_poll: polling handlers (one per input queue or NULL)
* @int_parm: interruption parameter
* @input_sbal_addr_array: address of no_input_qs * 128 pointers
* @output_sbal_addr_array: address of no_output_qs * 128 pointers
+ * @output_sbal_state_array: no_output_qs * 128 state info (for CQ or NULL)
*/
struct qdio_initialize {
struct ccw_device *cdev;
@@ -297,11 +357,12 @@ struct qdio_initialize {
unsigned int no_output_qs;
qdio_handler_t *input_handler;
qdio_handler_t *output_handler;
- void (*queue_start_poll) (struct ccw_device *, int, unsigned long);
+ void (**queue_start_poll) (struct ccw_device *, int, unsigned long);
int scan_threshold;
unsigned long int_parm;
void **input_sbal_addr_array;
void **output_sbal_addr_array;
+ struct qdio_outbuf_state *output_sbal_state_array;
};
#define QDIO_STATE_INACTIVE 0x00000002 /* after qdio_cleanup */
@@ -316,6 +377,7 @@ struct qdio_initialize {
extern int qdio_allocate(struct qdio_initialize *);
extern int qdio_establish(struct qdio_initialize *);
extern int qdio_activate(struct ccw_device *);
+extern void qdio_release_aob(struct qaob *);
extern int do_QDIO(struct ccw_device *, unsigned int, int, unsigned int,
unsigned int);
extern int qdio_start_irq(struct ccw_device *, int);
--- a/drivers/s390/cio/qdio.h
+++ b/drivers/s390/cio/qdio.h
@@ -44,6 +44,7 @@ enum qdio_irq_states {
#define SLSB_STATE_NOT_INIT 0x0
#define SLSB_STATE_EMPTY 0x1
#define SLSB_STATE_PRIMED 0x2
+#define SLSB_STATE_PENDING 0x3
#define SLSB_STATE_HALTED 0xe
#define SLSB_STATE_ERROR 0xf
#define SLSB_TYPE_INPUT 0x0
@@ -67,6 +68,8 @@ enum qdio_irq_states {
(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_NOT_INIT) /* 0xa0 */
#define SLSB_P_OUTPUT_EMPTY \
(SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_EMPTY) /* 0xa1 */
+#define SLSB_P_OUTPUT_PENDING \
+ (SLSB_OWNER_PROG | SLSB_TYPE_OUTPUT | SLSB_STATE_PENDING) /* 0xa3 */
#define SLSB_CU_OUTPUT_PRIMED \
(SLSB_OWNER_CU | SLSB_TYPE_OUTPUT | SLSB_STATE_PRIMED) /* 0x62 */
#define SLSB_P_OUTPUT_HALTED \
@@ -97,6 +100,7 @@ enum qdio_irq_states {
#define QDIO_SIGA_WRITE 0x00
#define QDIO_SIGA_READ 0x01
#define QDIO_SIGA_SYNC 0x02
+#define QDIO_SIGA_WRITEQ 0x04
#define QDIO_SIGA_QEBSM_FLAG 0x80
#ifdef CONFIG_64BIT
@@ -253,6 +257,12 @@ struct qdio_input_q {
struct qdio_output_q {
/* PCIs are enabled for the queue */
int pci_out_enabled;
+ /* cq: use asynchronous output buffers */
+ int use_cq;
+ /* cq: aobs used for particual SBAL */
+ struct qaob **aobs;
+ /* cq: sbal state related to asynchronous operation */
+ struct qdio_outbuf_state *sbal_state;
/* timer to check for more outbound work */
struct timer_list timer;
/* used SBALs before tasklet schedule */
@@ -432,9 +442,20 @@ struct indicator_t {
extern struct indicator_t *q_indicators;
-static inline int shared_ind(u32 *dsci)
+static inline int has_multiple_inq_on_dsci(struct qdio_irq *irq)
{
- return dsci == &q_indicators[TIQDIO_SHARED_IND].ind;
+ return irq->nr_input_qs > 1;
+}
+
+static inline int references_shared_dsci(struct qdio_irq *irq)
+{
+ return irq->dsci == &q_indicators[TIQDIO_SHARED_IND].ind;
+}
+
+static inline int shared_ind(struct qdio_q *q)
+{
+ struct qdio_irq *i = q->irq_ptr;
+ return references_shared_dsci(i) || has_multiple_inq_on_dsci(i);
}
/* prototypes for thin interrupt */
@@ -449,6 +470,7 @@ void tiqdio_free_memory(void);
int tiqdio_register_thinints(void);
void tiqdio_unregister_thinints(void);
+
/* prototypes for setup */
void qdio_inbound_processing(unsigned long data);
void qdio_outbound_processing(unsigned long data);
@@ -469,6 +491,9 @@ int qdio_setup_create_sysfs(struct ccw_d
void qdio_setup_destroy_sysfs(struct ccw_device *cdev);
int qdio_setup_init(void);
void qdio_setup_exit(void);
+int qdio_enable_async_operation(struct qdio_output_q *q);
+void qdio_disable_async_operation(struct qdio_output_q *q);
+struct qaob *qdio_allocate_aob(void);
int debug_get_buf_state(struct qdio_q *q, unsigned int bufnr,
unsigned char *state);
--- a/drivers/s390/cio/qdio_debug.c
+++ b/drivers/s390/cio/qdio_debug.c
@@ -76,6 +76,9 @@ static int qstat_show(struct seq_file *m
case SLSB_P_OUTPUT_NOT_INIT:
seq_printf(m, "N");
break;
+ case SLSB_P_OUTPUT_PENDING:
+ seq_printf(m, "P");
+ break;
case SLSB_P_INPUT_PRIMED:
case SLSB_CU_OUTPUT_PRIMED:
seq_printf(m, "+");
--- a/drivers/s390/cio/qdio_main.c
+++ b/drivers/s390/cio/qdio_main.c
@@ -14,6 +14,7 @@
#include <linux/timer.h>
#include <linux/delay.h>
#include <linux/gfp.h>
+#include <linux/io.h>
#include <linux/kernel_stat.h>
#include <linux/atomic.h>
#include <asm/debug.h>
@@ -77,11 +78,13 @@ static inline int do_siga_input(unsigned
* Note: For IQDC unicast queues only the highest priority queue is processed.
*/
static inline int do_siga_output(unsigned long schid, unsigned long mask,
- unsigned int *bb, unsigned int fc)
+ unsigned int *bb, unsigned int fc,
+ unsigned long aob)
{
register unsigned long __fc asm("0") = fc;
register unsigned long __schid asm("1") = schid;
register unsigned long __mask asm("2") = mask;
+ register unsigned long __aob asm("3") = aob;
int cc = QDIO_ERROR_SIGA_ACCESS_EXCEPTION;
asm volatile(
@@ -90,7 +93,8 @@ static inline int do_siga_output(unsigne
" srl %0,28\n"
"1:\n"
EX_TABLE(0b, 1b)
- : "+d" (cc), "+d" (__fc), "+d" (__schid), "+d" (__mask)
+ : "+d" (cc), "+d" (__fc), "+d" (__schid), "+d" (__mask),
+ "+d" (__aob)
: : "cc", "memory");
*bb = ((unsigned int) __fc) >> 31;
return cc;
@@ -212,7 +216,7 @@ again:
/* returns number of examined buffers and their common state in *state */
static inline int get_buf_states(struct qdio_q *q, unsigned int bufnr,
unsigned char *state, unsigned int count,
- int auto_ack)
+ int auto_ack, int merge_pending)
{
unsigned char __state = 0;
int i;
@@ -224,9 +228,14 @@ static inline int get_buf_states(struct
return qdio_do_eqbs(q, state, bufnr, count, auto_ack);
for (i = 0; i < count; i++) {
- if (!__state)
+ if (!__state) {
__state = q->slsb.val[bufnr];
- else if (q->slsb.val[bufnr] != __state)
+ if (merge_pending && __state == SLSB_P_OUTPUT_PENDING)
+ __state = SLSB_P_OUTPUT_EMPTY;
+ } else if (merge_pending) {
+ if ((q->slsb.val[bufnr] & __state) != __state)
+ break;
+ } else if (q->slsb.val[bufnr] != __state)
break;
bufnr = next_buf(bufnr);
}
@@ -237,7 +246,7 @@ static inline int get_buf_states(struct
static inline int get_buf_state(struct qdio_q *q, unsigned int bufnr,
unsigned char *state, int auto_ack)
{
- return get_buf_states(q, bufnr, state, 1, auto_ack);
+ return get_buf_states(q, bufnr, state, 1, auto_ack, 0);
}
/* wrap-around safe setting of slsb states, returns number of changed buffers */
@@ -308,19 +317,28 @@ static inline int qdio_siga_sync_q(struc
return qdio_siga_sync(q, q->mask, 0);
}
-static int qdio_siga_output(struct qdio_q *q, unsigned int *busy_bit)
+static int qdio_siga_output(struct qdio_q *q, unsigned int *busy_bit,
+ unsigned long aob)
{
unsigned long schid = *((u32 *) &q->irq_ptr->schid);
unsigned int fc = QDIO_SIGA_WRITE;
u64 start_time = 0;
int retries = 0, cc;
+ unsigned long laob = 0;
+
+ if (q->u.out.use_cq && aob != 0) {
+ fc = QDIO_SIGA_WRITEQ;
+ laob = aob;
+ }
if (is_qebsm(q)) {
schid = q->irq_ptr->sch_token;
fc |= QDIO_SIGA_QEBSM_FLAG;
}
again:
- cc = do_siga_output(schid, q->mask, busy_bit, fc);
+ WARN_ON_ONCE((aob && queue_type(q) != QDIO_IQDIO_QFMT) ||
+ (aob && fc != QDIO_SIGA_WRITEQ));
+ cc = do_siga_output(schid, q->mask, busy_bit, fc, laob);
/* hipersocket busy condition */
if (unlikely(*busy_bit)) {
@@ -379,7 +397,7 @@ int debug_get_buf_state(struct qdio_q *q
{
if (need_siga_sync(q))
qdio_siga_sync_q(q);
- return get_buf_states(q, bufnr, state, 1, 0);
+ return get_buf_states(q, bufnr, state, 1, 0, 0);
}
static inline void qdio_stop_polling(struct qdio_q *q)
@@ -507,7 +525,7 @@ static int get_inbound_buffer_frontier(s
* No siga sync here, as a PCI or we after a thin interrupt
* already sync'ed the queues.
*/
- count = get_buf_states(q, q->first_to_check, &state, count, 1);
+ count = get_buf_states(q, q->first_to_check, &state, count, 1, 0);
if (!count)
goto out;
@@ -590,6 +608,107 @@ static inline int qdio_inbound_q_done(st
return 0;
}
+static inline int contains_aobs(struct qdio_q *q)
+{
+ return !q->is_input_q && q->u.out.use_cq;
+}
+
+static inline void qdio_trace_aob(struct qdio_irq *irq, struct qdio_q *q,
+ int i, struct qaob *aob)
+{
+ int tmp;
+
+ DBF_DEV_EVENT(DBF_INFO, irq, "AOB%d:%lx", i,
+ (unsigned long) virt_to_phys(aob));
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES00:%lx",
+ (unsigned long) aob->res0[0]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES01:%lx",
+ (unsigned long) aob->res0[1]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES02:%lx",
+ (unsigned long) aob->res0[2]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES03:%lx",
+ (unsigned long) aob->res0[3]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES04:%lx",
+ (unsigned long) aob->res0[4]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES05:%lx",
+ (unsigned long) aob->res0[5]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES1:%x", aob->res1);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES2:%x", aob->res2);
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES3:%x", aob->res3);
+ DBF_DEV_EVENT(DBF_INFO, irq, "AORC:%u", aob->aorc);
+ DBF_DEV_EVENT(DBF_INFO, irq, "FLAGS:%u", aob->flags);
+ DBF_DEV_EVENT(DBF_INFO, irq, "CBTBS:%u", aob->cbtbs);
+ DBF_DEV_EVENT(DBF_INFO, irq, "SBC:%u", aob->sb_count);
+ for (tmp = 0; tmp < QDIO_MAX_ELEMENTS_PER_BUFFER; ++tmp) {
+ DBF_DEV_EVENT(DBF_INFO, irq, "SBA%d:%lx", tmp,
+ (unsigned long) aob->sba[tmp]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "rSBA%d:%lx", tmp,
+ (unsigned long) q->sbal[i]->element[tmp].addr);
+ DBF_DEV_EVENT(DBF_INFO, irq, "DC%d:%u", tmp, aob->dcount[tmp]);
+ DBF_DEV_EVENT(DBF_INFO, irq, "rDC%d:%u", tmp,
+ q->sbal[i]->element[tmp].length);
+ }
+ DBF_DEV_EVENT(DBF_INFO, irq, "USER0:%lx", (unsigned long) aob->user0);
+ for (tmp = 0; tmp < 2; ++tmp) {
+ DBF_DEV_EVENT(DBF_INFO, irq, "RES4%d:%lx", tmp,
+ (unsigned long) aob->res4[tmp]);
+ }
+ DBF_DEV_EVENT(DBF_INFO, irq, "USER1:%lx", (unsigned long) aob->user1);
+ DBF_DEV_EVENT(DBF_INFO, irq, "USER2:%lx", (unsigned long) aob->user2);
+}
+
+static inline void qdio_handle_aobs(struct qdio_q *q, int start, int count)
+{
+ unsigned char state = 0;
+ int j, b = start;
+
+ if (!contains_aobs(q))
+ return;
+
+ for (j = 0; j < count; ++j) {
+ get_buf_state(q, b, &state, 0);
+ if (state == SLSB_P_OUTPUT_PENDING) {
+ struct qaob *aob = q->u.out.aobs[b];
+ if (aob == NULL)
+ continue;
+
+ BUG_ON(q->u.out.sbal_state == NULL);
+ q->u.out.sbal_state[b].flags |=
+ QDIO_OUTBUF_STATE_FLAG_PENDING;
+ q->u.out.aobs[b] = NULL;
+ } else if (state == SLSB_P_OUTPUT_EMPTY) {
+ BUG_ON(q->u.out.sbal_state == NULL);
+ q->u.out.sbal_state[b].aob = NULL;
+ }
+ b = next_buf(b);
+ }
+}
+
+static inline unsigned long qdio_aob_for_buffer(struct qdio_output_q *q,
+ int bufnr)
+{
+ unsigned long phys_aob = 0;
+
+ if (!q->use_cq)
+ goto out;
+
+ if (!q->aobs[bufnr]) {
+ struct qaob *aob = qdio_allocate_aob();
+ q->aobs[bufnr] = aob;
+ }
+ if (q->aobs[bufnr]) {
+ BUG_ON(q->sbal_state == NULL);
+ q->sbal_state[bufnr].flags = QDIO_OUTBUF_STATE_FLAG_NONE;
+ q->sbal_state[bufnr].aob = q->aobs[bufnr];
+ q->aobs[bufnr]->user1 = (u64) q->sbal_state[bufnr].user;
+ phys_aob = virt_to_phys(q->aobs[bufnr]);
+ BUG_ON(phys_aob & 0xFF);
+ }
+
+out:
+ return phys_aob;
+}
+
static void qdio_kick_handler(struct qdio_q *q)
{
int start = q->first_to_kick;
@@ -610,6 +729,8 @@ static void qdio_kick_handler(struct qdi
start, count);
}
+ qdio_handle_aobs(q, start, count);
+
q->handler(q->irq_ptr->cdev, q->qdio_error, q->nr, start, count,
q->irq_ptr->int_parm);
@@ -672,23 +793,26 @@ static int get_outbound_buffer_frontier(
*/
count = min(atomic_read(&q->nr_buf_used), QDIO_MAX_BUFFERS_MASK);
stop = add_buf(q->first_to_check, count);
-
if (q->first_to_check == stop)
- return q->first_to_check;
+ goto out;
- count = get_buf_states(q, q->first_to_check, &state, count, 0);
+ count = get_buf_states(q, q->first_to_check, &state, count, 0, 1);
if (!count)
- return q->first_to_check;
+ goto out;
switch (state) {
+ case SLSB_P_OUTPUT_PENDING:
+ BUG();
case SLSB_P_OUTPUT_EMPTY:
/* the adapter got it */
- DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "out empty:%1d %02x", q->nr, count);
+ DBF_DEV_EVENT(DBF_INFO, q->irq_ptr,
+ "out empty:%1d %02x", q->nr, count);
atomic_sub(count, &q->nr_buf_used);
q->first_to_check = add_buf(q->first_to_check, count);
if (q->irq_ptr->perf_stat_enabled)
account_sbals(q, count);
+
break;
case SLSB_P_OUTPUT_ERROR:
process_buffer_error(q, count);
@@ -701,7 +825,8 @@ static int get_outbound_buffer_frontier(
/* the adapter has not fetched the output yet */
if (q->irq_ptr->perf_stat_enabled)
q->q_stats.nr_sbal_nop++;
- DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "out primed:%1d", q->nr);
+ DBF_DEV_EVENT(DBF_INFO, q->irq_ptr, "out primed:%1d",
+ q->nr);
break;
case SLSB_P_OUTPUT_NOT_INIT:
case SLSB_P_OUTPUT_HALTED:
@@ -709,6 +834,8 @@ static int get_outbound_buffer_frontier(
default:
BUG();
}
+
+out:
return q->first_to_check;
}
@@ -732,7 +859,7 @@ static inline int qdio_outbound_q_moved(
return 0;
}
-static int qdio_kick_outbound_q(struct qdio_q *q)
+static int qdio_kick_outbound_q(struct qdio_q *q, unsigned long aob)
{
int retries = 0, cc;
unsigned int busy_bit;
@@ -744,7 +871,7 @@ static int qdio_kick_outbound_q(struct q
retry:
qperf_inc(q, siga_write);
- cc = qdio_siga_output(q, &busy_bit);
+ cc = qdio_siga_output(q, &busy_bit, aob);
switch (cc) {
case 0:
break;
@@ -921,8 +1048,9 @@ static void qdio_int_handler_pci(struct
}
q->u.in.queue_start_poll(q->irq_ptr->cdev, q->nr,
q->irq_ptr->int_parm);
- } else
+ } else {
tasklet_schedule(&q->tasklet);
+ }
}
if (!pci_out_supported(q))
@@ -1236,6 +1364,26 @@ out_err:
}
EXPORT_SYMBOL_GPL(qdio_allocate);
+static void qdio_detect_hsicq(struct qdio_irq *irq_ptr)
+{
+ struct qdio_q *q = irq_ptr->input_qs[0];
+ int i, use_cq = 0;
+
+ if (irq_ptr->nr_input_qs > 1 && queue_type(q) == QDIO_IQDIO_QFMT)
+ use_cq = 1;
+
+ for_each_output_queue(irq_ptr, q, i) {
+ if (use_cq) {
+ if (qdio_enable_async_operation(&q->u.out) < 0) {
+ use_cq = 0;
+ continue;
+ }
+ } else
+ qdio_disable_async_operation(&q->u.out);
+ }
+ DBF_EVENT("use_cq:%d", use_cq);
+}
+
/**
* qdio_establish - establish queues on a qdio subchannel
* @init_data: initialization data
@@ -1301,6 +1449,8 @@ int qdio_establish(struct qdio_initializ
qdio_setup_ssqd_info(irq_ptr);
DBF_EVENT("qib ac:%4x", irq_ptr->qib.ac);
+ qdio_detect_hsicq(irq_ptr);
+
/* qebsm is now setup if available, initialize buffer states */
qdio_init_buf_states(irq_ptr);
@@ -1480,17 +1630,21 @@ static int handle_outbound(struct qdio_q
q->u.out.pci_out_enabled = 0;
if (queue_type(q) == QDIO_IQDIO_QFMT) {
- /* One SIGA-W per buffer required for unicast HiperSockets. */
+ unsigned long phys_aob = 0;
+
+ /* One SIGA-W per buffer required for unicast HSI */
WARN_ON_ONCE(count > 1 && !multicast_outbound(q));
- rc = qdio_kick_outbound_q(q);
+ phys_aob = qdio_aob_for_buffer(&q->u.out, bufnr);
+
+ rc = qdio_kick_outbound_q(q, phys_aob);
} else if (need_siga_sync(q)) {
rc = qdio_siga_sync_q(q);
} else {
/* try to fast requeue buffers */
get_buf_state(q, prev_buf(bufnr), &state, 0);
if (state != SLSB_CU_OUTPUT_PRIMED)
- rc = qdio_kick_outbound_q(q);
+ rc = qdio_kick_outbound_q(q, 0);
else
qperf_inc(q, fast_requeue);
}
@@ -1518,6 +1672,7 @@ int do_QDIO(struct ccw_device *cdev, uns
{
struct qdio_irq *irq_ptr;
+
if (bufnr >= QDIO_MAX_BUFFERS_PER_Q || count > QDIO_MAX_BUFFERS_PER_Q)
return -EINVAL;
@@ -1562,7 +1717,7 @@ int qdio_start_irq(struct ccw_device *cd
WARN_ON(queue_irqs_enabled(q));
- if (!shared_ind(q->irq_ptr->dsci))
+ if (!shared_ind(q))
xchg(q->irq_ptr->dsci, 0);
qdio_stop_polling(q);
@@ -1572,7 +1727,7 @@ int qdio_start_irq(struct ccw_device *cd
* We need to check again to not lose initiative after
* resetting the ACK state.
*/
- if (!shared_ind(q->irq_ptr->dsci) && *q->irq_ptr->dsci)
+ if (!shared_ind(q) && *q->irq_ptr->dsci)
goto rescan;
if (!qdio_inbound_q_done(q))
goto rescan;
--- a/drivers/s390/cio/qdio_setup.c
+++ b/drivers/s390/cio/qdio_setup.c
@@ -19,6 +19,22 @@
#include "qdio_debug.h"
static struct kmem_cache *qdio_q_cache;
+static struct kmem_cache *qdio_aob_cache;
+
+struct qaob *qdio_allocate_aob()
+{
+ struct qaob *aob;
+
+ aob = kmem_cache_zalloc(qdio_aob_cache, GFP_ATOMIC);
+ return aob;
+}
+EXPORT_SYMBOL_GPL(qdio_allocate_aob);
+
+void qdio_release_aob(struct qaob *aob)
+{
+ kmem_cache_free(qdio_aob_cache, aob);
+}
+EXPORT_SYMBOL_GPL(qdio_release_aob);
/*
* qebsm is only available under 64bit but the adapter sets the feature
@@ -154,29 +170,36 @@ static void setup_queues(struct qdio_irq
struct qdio_q *q;
void **input_sbal_array = qdio_init->input_sbal_addr_array;
void **output_sbal_array = qdio_init->output_sbal_addr_array;
+ struct qdio_outbuf_state *output_sbal_state_array =
+ qdio_init->output_sbal_state_array;
int i;
for_each_input_queue(irq_ptr, q, i) {
- DBF_EVENT("in-q:%1d", i);
+ DBF_EVENT("inq:%1d", i);
setup_queues_misc(q, irq_ptr, qdio_init->input_handler, i);
q->is_input_q = 1;
- q->u.in.queue_start_poll = qdio_init->queue_start_poll;
+ q->u.in.queue_start_poll = qdio_init->queue_start_poll[i];
+
setup_storage_lists(q, irq_ptr, input_sbal_array, i);
input_sbal_array += QDIO_MAX_BUFFERS_PER_Q;
- if (is_thinint_irq(irq_ptr))
+ if (is_thinint_irq(irq_ptr)) {
tasklet_init(&q->tasklet, tiqdio_inbound_processing,
(unsigned long) q);
- else
+ } else {
tasklet_init(&q->tasklet, qdio_inbound_processing,
(unsigned long) q);
+ }
}
for_each_output_queue(irq_ptr, q, i) {
DBF_EVENT("outq:%1d", i);
setup_queues_misc(q, irq_ptr, qdio_init->output_handler, i);
+ q->u.out.sbal_state = output_sbal_state_array;
+ output_sbal_state_array += QDIO_MAX_BUFFERS_PER_Q;
+
q->is_input_q = 0;
q->u.out.scan_threshold = qdio_init->scan_threshold;
setup_storage_lists(q, irq_ptr, output_sbal_array, i);
@@ -311,6 +334,19 @@ void qdio_release_memory(struct qdio_irq
for (i = 0; i < QDIO_MAX_QUEUES_PER_IRQ; i++) {
q = irq_ptr->output_qs[i];
if (q) {
+ if (q->u.out.use_cq) {
+ int n;
+
+ for (n = 0; n < QDIO_MAX_BUFFERS_PER_Q; ++n) {
+ struct qaob *aob = q->u.out.aobs[n];
+ if (aob) {
+ qdio_release_aob(aob);
+ q->u.out.aobs[n] = NULL;
+ }
+ }
+
+ qdio_disable_async_operation(&q->u.out);
+ }
free_page((unsigned long) q->slib);
kmem_cache_free(qdio_q_cache, q);
}
@@ -465,23 +501,60 @@ void qdio_print_subchannel_info(struct q
printk(KERN_INFO "%s", s);
}
+int qdio_enable_async_operation(struct qdio_output_q *outq)
+{
+ outq->aobs = kzalloc(sizeof(struct qaob *) * QDIO_MAX_BUFFERS_PER_Q,
+ GFP_ATOMIC);
+ if (!outq->aobs) {
+ outq->use_cq = 0;
+ return -ENOMEM;
+ }
+ outq->use_cq = 1;
+ return 0;
+}
+
+void qdio_disable_async_operation(struct qdio_output_q *q)
+{
+ kfree(q->aobs);
+ q->aobs = NULL;
+ q->use_cq = 0;
+}
+
int __init qdio_setup_init(void)
{
+ int rc;
+
qdio_q_cache = kmem_cache_create("qdio_q", sizeof(struct qdio_q),
256, 0, NULL);
if (!qdio_q_cache)
return -ENOMEM;
+ qdio_aob_cache = kmem_cache_create("qdio_aob",
+ sizeof(struct qaob),
+ sizeof(struct qaob),
+ 0,
+ NULL);
+ if (!qdio_aob_cache) {
+ rc = -ENOMEM;
+ goto free_qdio_q_cache;
+ }
+
/* Check for OSA/FCP thin interrupts (bit 67). */
DBF_EVENT("thinint:%1d",
(css_general_characteristics.aif_osa) ? 1 : 0);
/* Check for QEBSM support in general (bit 58). */
DBF_EVENT("cssQEBSM:%1d", (qebsm_possible()) ? 1 : 0);
- return 0;
+ rc = 0;
+out:
+ return rc;
+free_qdio_q_cache:
+ kmem_cache_destroy(qdio_q_cache);
+ goto out;
}
void qdio_setup_exit(void)
{
+ kmem_cache_destroy(qdio_aob_cache);
kmem_cache_destroy(qdio_q_cache);
}
--- a/drivers/s390/cio/qdio_thinint.c
+++ b/drivers/s390/cio/qdio_thinint.c
@@ -67,12 +67,9 @@ static void put_indicator(u32 *addr)
void tiqdio_add_input_queues(struct qdio_irq *irq_ptr)
{
- struct qdio_q *q;
- int i;
-
mutex_lock(&tiq_list_lock);
- for_each_input_queue(irq_ptr, q, i)
- list_add_rcu(&q->entry, &tiq_list);
+ BUG_ON(irq_ptr->nr_input_qs < 1);
+ list_add_rcu(&irq_ptr->input_qs[0]->entry, &tiq_list);
mutex_unlock(&tiq_list_lock);
xchg(irq_ptr->dsci, 1 << 7);
}
@@ -80,19 +77,17 @@ void tiqdio_add_input_queues(struct qdio
void tiqdio_remove_input_queues(struct qdio_irq *irq_ptr)
{
struct qdio_q *q;
- int i;
- for (i = 0; i < irq_ptr->nr_input_qs; i++) {
- q = irq_ptr->input_qs[i];
- /* if establish triggered an error */
- if (!q || !q->entry.prev || !q->entry.next)
- continue;
+ BUG_ON(irq_ptr->nr_input_qs < 1);
+ q = irq_ptr->input_qs[0];
+ /* if establish triggered an error */
+ if (!q || !q->entry.prev || !q->entry.next)
+ return;
- mutex_lock(&tiq_list_lock);
- list_del_rcu(&q->entry);
- mutex_unlock(&tiq_list_lock);
- synchronize_rcu();
- }
+ mutex_lock(&tiq_list_lock);
+ list_del_rcu(&q->entry);
+ mutex_unlock(&tiq_list_lock);
+ synchronize_rcu();
}
static inline u32 clear_shared_ind(void)
@@ -102,6 +97,40 @@ static inline u32 clear_shared_ind(void)
return xchg(&q_indicators[TIQDIO_SHARED_IND].ind, 0);
}
+static inline void tiqdio_call_inq_handlers(struct qdio_irq *irq)
+{
+ struct qdio_q *q;
+ int i;
+
+ for_each_input_queue(irq, q, i) {
+ if (!references_shared_dsci(irq) &&
+ has_multiple_inq_on_dsci(irq))
+ xchg(q->irq_ptr->dsci, 0);
+
+ if (q->u.in.queue_start_poll) {
+ /* skip if polling is enabled or already in work */
+ if (test_and_set_bit(QDIO_QUEUE_IRQS_DISABLED,
+ &q->u.in.queue_irq_state)) {
+ qperf_inc(q, int_discarded);
+ continue;
+ }
+
+ /* avoid dsci clear here, done after processing */
+ q->u.in.queue_start_poll(q->irq_ptr->cdev, q->nr,
+ q->irq_ptr->int_parm);
+ } else {
+ if (!shared_ind(q))
+ xchg(q->irq_ptr->dsci, 0);
+
+ /*
+ * Call inbound processing but not directly
+ * since that could starve other thinint queues.
+ */
+ tasklet_schedule(&q->tasklet);
+ }
+ }
+}
+
/**
* tiqdio_thinint_handler - thin interrupt handler for qdio
* @alsi: pointer to adapter local summary indicator
@@ -120,35 +149,18 @@ static void tiqdio_thinint_handler(void
/* check for work on all inbound thinint queues */
list_for_each_entry_rcu(q, &tiq_list, entry) {
+ struct qdio_irq *irq;
/* only process queues from changed sets */
- if (unlikely(shared_ind(q->irq_ptr->dsci))) {
+ irq = q->irq_ptr;
+ if (unlikely(references_shared_dsci(irq))) {
if (!si_used)
continue;
- } else if (!*q->irq_ptr->dsci)
+ } else if (!*irq->dsci)
continue;
- if (q->u.in.queue_start_poll) {
- /* skip if polling is enabled or already in work */
- if (test_and_set_bit(QDIO_QUEUE_IRQS_DISABLED,
- &q->u.in.queue_irq_state)) {
- qperf_inc(q, int_discarded);
- continue;
- }
+ tiqdio_call_inq_handlers(irq);
- /* avoid dsci clear here, done after processing */
- q->u.in.queue_start_poll(q->irq_ptr->cdev, q->nr,
- q->irq_ptr->int_parm);
- } else {
- /* only clear it if the indicator is non-shared */
- if (!shared_ind(q->irq_ptr->dsci))
- xchg(q->irq_ptr->dsci, 0);
- /*
- * Call inbound processing but not directly
- * since that could starve other thinint queues.
- */
- tasklet_schedule(&q->tasklet);
- }
qperf_inc(q, adapter_int);
}
rcu_read_unlock();
--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -3939,6 +3939,7 @@ static int qeth_qdio_establish(struct qe
struct qdio_initialize init_data;
char *qib_param_field;
struct qdio_buffer **in_sbal_ptrs;
+ void (**queue_start_poll) (struct ccw_device *, int, unsigned long);
struct qdio_buffer **out_sbal_ptrs;
int i, j, k;
int rc = 0;
@@ -3947,8 +3948,10 @@ static int qeth_qdio_establish(struct qe
qib_param_field = kzalloc(QDIO_MAX_BUFFERS_PER_Q * sizeof(char),
GFP_KERNEL);
- if (!qib_param_field)
- return -ENOMEM;
+ if (!qib_param_field) {
+ rc = -ENOMEM;
+ goto out_free_nothing;
+ }
qeth_create_qib_param_field(card, qib_param_field);
qeth_create_qib_param_field_blkt(card, qib_param_field);
@@ -3956,20 +3959,26 @@ static int qeth_qdio_establish(struct qe
in_sbal_ptrs = kmalloc(QDIO_MAX_BUFFERS_PER_Q * sizeof(void *),
GFP_KERNEL);
if (!in_sbal_ptrs) {
- kfree(qib_param_field);
- return -ENOMEM;
+ rc = -ENOMEM;
+ goto out_free_qib_param;
}
for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i)
in_sbal_ptrs[i] = (struct qdio_buffer *)
virt_to_phys(card->qdio.in_q->bufs[i].buffer);
+ queue_start_poll = kmalloc(sizeof(void *) * 1, GFP_KERNEL);
+ if (!queue_start_poll) {
+ rc = -ENOMEM;
+ goto out_free_in_sbals;
+ }
+ queue_start_poll[0] = card->discipline.start_poll;
+
out_sbal_ptrs =
kmalloc(card->qdio.no_out_queues * QDIO_MAX_BUFFERS_PER_Q *
sizeof(void *), GFP_KERNEL);
if (!out_sbal_ptrs) {
- kfree(in_sbal_ptrs);
- kfree(qib_param_field);
- return -ENOMEM;
+ rc = -ENOMEM;
+ goto out_free_queue_start_poll;
}
for (i = 0, k = 0; i < card->qdio.no_out_queues; ++i)
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j, ++k) {
@@ -3986,7 +3995,7 @@ static int qeth_qdio_establish(struct qe
init_data.no_output_qs = card->qdio.no_out_queues;
init_data.input_handler = card->discipline.input_handler;
init_data.output_handler = card->discipline.output_handler;
- init_data.queue_start_poll = card->discipline.start_poll;
+ init_data.queue_start_poll = queue_start_poll;
init_data.int_parm = (unsigned long) card;
init_data.input_sbal_addr_array = (void **) in_sbal_ptrs;
init_data.output_sbal_addr_array = (void **) out_sbal_ptrs;
@@ -4008,8 +4017,13 @@ static int qeth_qdio_establish(struct qe
}
out:
kfree(out_sbal_ptrs);
+out_free_queue_start_poll:
+ kfree(queue_start_poll);
+out_free_in_sbals:
kfree(in_sbal_ptrs);
+out_free_qib_param:
kfree(qib_param_field);
+out_free_nothing:
return rc;
}
^ permalink raw reply
* [patch 13/13] [PATCH] qeth: add support for af_iucv HiperSockets transport
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Einar Lueck, Ursula Braun
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: netnext_qeth_iucv.patch --]
[-- Type: text/plain, Size: 29304 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
This patch extends the HiperSockets device driver to send and receive
af_iucv traffic over HiperSockets transport.
TX: Driver uses new asynchronous delivery of storage blocks to pass
flow control/congestion information from the HiperSockets microcode
to the af_iucv socket.
RX: Memory for incoming traffic is preallocated and passed to
HiperSockets layer. If receiver is not capable to clean its buffers
shared with HiperSockets and pass new memory to the HiperSockets
layer this will cause flow control/congestion events on the
sender.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: Einar Lueck <elelueck@de.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
---
drivers/s390/net/qeth_core.h | 19 ++-
drivers/s390/net/qeth_core_main.c | 230 +++++++++++++++++++++++++++++++-------
drivers/s390/net/qeth_l2_main.c | 2
drivers/s390/net/qeth_l3.h | 4
drivers/s390/net/qeth_l3_main.c | 87 +++++++++++---
drivers/s390/net/qeth_l3_sys.c | 110 ++++++++++++++++++
6 files changed, 391 insertions(+), 61 deletions(-)
--- a/drivers/s390/net/qeth_core.h
+++ b/drivers/s390/net/qeth_core.h
@@ -217,6 +217,7 @@ static inline int qeth_is_ipa_enabled(st
*/
#define QETH_TX_TIMEOUT 100 * HZ
#define QETH_RCD_TIMEOUT 60 * HZ
+#define QETH_RECLAIM_WORK_TIME HZ
#define QETH_HEADER_SIZE 32
#define QETH_MAX_PORTNO 15
@@ -265,6 +266,7 @@ static inline int qeth_is_ipa_enabled(st
/* large receive scatter gather copy break */
#define QETH_RX_SG_CB (PAGE_SIZE >> 1)
+#define QETH_RX_PULL_LEN 256
struct qeth_hdr_layer3 {
__u8 id;
@@ -382,6 +384,16 @@ enum qeth_qdio_buffer_states {
QETH_QDIO_BUF_PRIMED,
/*
* inbound: not applicable
+ * outbound: identified to be pending in TPQ
+ */
+ QETH_QDIO_BUF_PENDING,
+ /*
+ * inbound: not applicable
+ * outbound: found in completion queue
+ */
+ QETH_QDIO_BUF_IN_CQ,
+ /*
+ * inbound: not applicable
* outbound: handled via transfer pending / completion queue
*/
QETH_QDIO_BUF_HANDLED_DELAYED,
@@ -409,6 +421,7 @@ struct qeth_qdio_buffer {
struct qdio_buffer *buffer;
/* the buffer pool entry currently associated to this buffer */
struct qeth_buffer_pool_entry *pool_entry;
+ struct sk_buff *rx_skb;
};
struct qeth_qdio_q {
@@ -674,6 +687,7 @@ struct qeth_card_options {
enum qeth_ipa_isolation_modes isolation;
int sniffer;
enum qeth_cq cq;
+ char hsuid[9];
};
/*
@@ -771,6 +785,8 @@ struct qeth_card {
struct mutex discipline_mutex;
struct napi_struct napi;
struct qeth_rx rx;
+ struct delayed_work buffer_reclaim_work;
+ int reclaim_index;
};
struct qeth_card_list_struct {
@@ -836,6 +852,7 @@ int qeth_core_create_device_attributes(s
void qeth_core_remove_device_attributes(struct device *);
int qeth_core_create_osn_attributes(struct device *);
void qeth_core_remove_osn_attributes(struct device *);
+void qeth_buffer_reclaim_work(struct work_struct *);
/* exports for qeth discipline device drivers */
extern struct qeth_card_list_struct qeth_core_card_list;
@@ -864,7 +881,7 @@ int qeth_check_qdio_errors(struct qeth_c
unsigned int, const char *);
void qeth_queue_input_buffer(struct qeth_card *, int);
struct sk_buff *qeth_core_get_next_skb(struct qeth_card *,
- struct qdio_buffer *, struct qdio_buffer_element **, int *,
+ struct qeth_qdio_buffer *, struct qdio_buffer_element **, int *,
struct qeth_hdr **);
void qeth_schedule_recovery(struct qeth_card *);
void qeth_qdio_start_poll(struct ccw_device *, int, unsigned long);
--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -21,6 +21,7 @@
#include <linux/mii.h>
#include <linux/kthread.h>
#include <linux/slab.h>
+#include <net/iucv/af_iucv.h>
#include <asm/ebcdic.h>
#include <asm/io.h>
@@ -58,6 +59,10 @@ static void qeth_setup_ccw(struct qeth_c
static void qeth_free_buffer_pool(struct qeth_card *);
static int qeth_qdio_establish(struct qeth_card *);
static void qeth_free_qdio_buffers(struct qeth_card *);
+static void qeth_notify_skbs(struct qeth_qdio_out_q *queue,
+ struct qeth_qdio_out_buffer *buf,
+ enum iucv_tx_notify notification);
+static void qeth_release_skbs(struct qeth_qdio_out_buffer *buf);
static void qeth_clear_output_buffer(struct qeth_qdio_out_q *queue,
struct qeth_qdio_out_buffer *buf,
enum qeth_qdio_buffer_states newbufstate);
@@ -204,7 +209,7 @@ static int qeth_alloc_buffer_pool(struct
QETH_CARD_TEXT(card, 5, "alocpool");
for (i = 0; i < card->qdio.init_pool.buf_count; ++i) {
- pool_entry = kmalloc(sizeof(*pool_entry), GFP_KERNEL);
+ pool_entry = kzalloc(sizeof(*pool_entry), GFP_KERNEL);
if (!pool_entry) {
qeth_free_buffer_pool(card);
return -ENOMEM;
@@ -331,6 +336,30 @@ static inline void qeth_free_cq(struct q
card->qdio.out_bufstates = NULL;
}
+static inline enum iucv_tx_notify qeth_compute_cq_notification(int sbalf15,
+ int delayed) {
+ enum iucv_tx_notify n;
+
+ switch (sbalf15) {
+ case 0:
+ n = delayed ? TX_NOTIFY_DELAYED_OK : TX_NOTIFY_OK;
+ break;
+ case 4:
+ case 16:
+ case 17:
+ case 18:
+ n = delayed ? TX_NOTIFY_DELAYED_UNREACHABLE :
+ TX_NOTIFY_UNREACHABLE;
+ break;
+ default:
+ n = delayed ? TX_NOTIFY_DELAYED_GENERALERROR :
+ TX_NOTIFY_GENERALERROR;
+ break;
+ }
+
+ return n;
+}
+
static inline void qeth_cleanup_handled_pending(struct qeth_qdio_out_q *q,
int bidx, int forced_cleanup)
{
@@ -345,6 +374,11 @@ static inline void qeth_cleanup_handled_
struct qeth_qdio_out_buffer *f = c;
QETH_CARD_TEXT(f->q->card, 5, "fp");
QETH_CARD_TEXT_(f->q->card, 5, "%lx", (long) f);
+ /* release here to avoid interleaving between
+ outbound tasklet and inbound tasklet
+ regarding notifications and lifecycle */
+ qeth_release_skbs(c);
+
c = f->next_pending;
BUG_ON(head->next_pending != f);
head->next_pending = c;
@@ -363,6 +397,7 @@ static inline void qeth_qdio_handle_aob(
unsigned long phys_aob_addr) {
struct qaob *aob;
struct qeth_qdio_out_buffer *buffer;
+ enum iucv_tx_notify notification;
aob = (struct qaob *) phys_to_virt(phys_aob_addr);
QETH_CARD_TEXT(card, 5, "haob");
@@ -372,6 +407,22 @@ static inline void qeth_qdio_handle_aob(
BUG_ON(buffer == NULL);
+ if (atomic_cmpxchg(&buffer->state, QETH_QDIO_BUF_PRIMED,
+ QETH_QDIO_BUF_IN_CQ) == QETH_QDIO_BUF_PRIMED) {
+ notification = TX_NOTIFY_OK;
+ } else {
+ BUG_ON(atomic_read(&buffer->state) != QETH_QDIO_BUF_PENDING);
+
+ atomic_set(&buffer->state, QETH_QDIO_BUF_IN_CQ);
+ notification = TX_NOTIFY_DELAYED_OK;
+ }
+
+ if (aob->aorc != 0) {
+ QETH_CARD_TEXT_(card, 2, "aorc%02X", aob->aorc);
+ notification = qeth_compute_cq_notification(aob->aorc, 1);
+ }
+ qeth_notify_skbs(buffer->q, buffer, notification);
+
buffer->aob = NULL;
qeth_clear_output_buffer(buffer->q, buffer,
QETH_QDIO_BUF_HANDLED_DELAYED);
@@ -738,7 +789,7 @@ static int qeth_setup_channel(struct qet
QETH_DBF_TEXT(SETUP, 2, "setupch");
for (cnt = 0; cnt < QETH_CMD_BUFFER_NO; cnt++) {
channel->iob[cnt].data =
- kmalloc(QETH_BUFSIZE, GFP_DMA|GFP_KERNEL);
+ kzalloc(QETH_BUFSIZE, GFP_DMA|GFP_KERNEL);
if (channel->iob[cnt].data == NULL)
break;
channel->iob[cnt].state = BUF_STATE_FREE;
@@ -1033,23 +1084,60 @@ out:
return;
}
-static void qeth_clear_output_buffer(struct qeth_qdio_out_q *queue,
+static void qeth_notify_skbs(struct qeth_qdio_out_q *q,
struct qeth_qdio_out_buffer *buf,
- enum qeth_qdio_buffer_states newbufstate)
+ enum iucv_tx_notify notification)
{
- int i;
struct sk_buff *skb;
- /* is PCI flag set on buffer? */
- if (buf->buffer->element[0].sflags & SBAL_SFLAGS0_PCI_REQ)
- atomic_dec(&queue->set_pci_flags_count);
+ if (skb_queue_empty(&buf->skb_list))
+ goto out;
+ skb = skb_peek(&buf->skb_list);
+ while (skb) {
+ QETH_CARD_TEXT_(q->card, 5, "skbn%d", notification);
+ QETH_CARD_TEXT_(q->card, 5, "%lx", (long) skb);
+ if (skb->protocol == ETH_P_AF_IUCV) {
+ if (skb->sk) {
+ struct iucv_sock *iucv = iucv_sk(skb->sk);
+ iucv->sk_txnotify(skb, notification);
+ }
+ }
+ if (skb_queue_is_last(&buf->skb_list, skb))
+ skb = NULL;
+ else
+ skb = skb_queue_next(&buf->skb_list, skb);
+ }
+out:
+ return;
+}
+
+static void qeth_release_skbs(struct qeth_qdio_out_buffer *buf)
+{
+ struct sk_buff *skb;
skb = skb_dequeue(&buf->skb_list);
while (skb) {
+ QETH_CARD_TEXT(buf->q->card, 5, "skbr");
+ QETH_CARD_TEXT_(buf->q->card, 5, "%lx", (long) skb);
atomic_dec(&skb->users);
dev_kfree_skb_any(skb);
skb = skb_dequeue(&buf->skb_list);
}
+}
+
+static void qeth_clear_output_buffer(struct qeth_qdio_out_q *queue,
+ struct qeth_qdio_out_buffer *buf,
+ enum qeth_qdio_buffer_states newbufstate)
+{
+ int i;
+
+ /* is PCI flag set on buffer? */
+ if (buf->buffer->element[0].sflags & SBAL_SFLAGS0_PCI_REQ)
+ atomic_dec(&queue->set_pci_flags_count);
+
+ if (newbufstate == QETH_QDIO_BUF_EMPTY) {
+ qeth_release_skbs(buf);
+ }
for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(queue->card); ++i) {
if (buf->buffer->element[i].addr && buf->is_header[i])
kmem_cache_free(qeth_core_header_cache,
@@ -1111,14 +1199,16 @@ static void qeth_free_buffer_pool(struct
static void qeth_free_qdio_buffers(struct qeth_card *card)
{
- int i;
+ int i, j;
if (atomic_xchg(&card->qdio.state, QETH_QDIO_UNINITIALIZED) ==
QETH_QDIO_UNINITIALIZED)
return;
qeth_free_cq(card);
-
+ cancel_delayed_work_sync(&card->buffer_reclaim_work);
+ for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j)
+ kfree_skb(card->qdio.in_q->bufs[j].rx_skb);
kfree(card->qdio.in_q);
card->qdio.in_q = NULL;
/* inbound buffer pool */
@@ -1289,6 +1379,7 @@ static int qeth_setup_card(struct qeth_c
card->ipato.invert6 = 0;
/* init QDIO stuff */
qeth_init_qdio_info(card);
+ INIT_DELAYED_WORK(&card->buffer_reclaim_work, qeth_buffer_reclaim_work);
return 0;
}
@@ -1310,7 +1401,7 @@ static struct qeth_card *qeth_alloc_card
if (!card)
goto out;
QETH_DBF_HEX(SETUP, 2, &card, sizeof(void *));
- card->ip_tbd_list = kmalloc(sizeof(struct list_head), GFP_KERNEL);
+ card->ip_tbd_list = kzalloc(sizeof(struct list_head), GFP_KERNEL);
if (!card->ip_tbd_list) {
QETH_DBF_TEXT(SETUP, 0, "iptbdnom");
goto out_card;
@@ -2239,7 +2330,7 @@ static int qeth_alloc_qdio_buffers(struc
QETH_QDIO_ALLOCATED) != QETH_QDIO_UNINITIALIZED)
return 0;
- card->qdio.in_q = kmalloc(sizeof(struct qeth_qdio_q),
+ card->qdio.in_q = kzalloc(sizeof(struct qeth_qdio_q),
GFP_KERNEL);
if (!card->qdio.in_q)
goto out_nomem;
@@ -2247,27 +2338,28 @@ static int qeth_alloc_qdio_buffers(struc
QETH_DBF_HEX(SETUP, 2, &card->qdio.in_q, sizeof(void *));
memset(card->qdio.in_q, 0, sizeof(struct qeth_qdio_q));
/* give inbound qeth_qdio_buffers their qdio_buffers */
- for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i)
+ for (i = 0; i < QDIO_MAX_BUFFERS_PER_Q; ++i) {
card->qdio.in_q->bufs[i].buffer =
&card->qdio.in_q->qdio_bufs[i];
+ card->qdio.in_q->bufs[i].rx_skb = NULL;
+ }
/* inbound buffer pool */
if (qeth_alloc_buffer_pool(card))
goto out_freeinq;
/* outbound */
card->qdio.out_qs =
- kmalloc(card->qdio.no_out_queues *
+ kzalloc(card->qdio.no_out_queues *
sizeof(struct qeth_qdio_out_q *), GFP_KERNEL);
if (!card->qdio.out_qs)
goto out_freepool;
for (i = 0; i < card->qdio.no_out_queues; ++i) {
- card->qdio.out_qs[i] = kmalloc(sizeof(struct qeth_qdio_out_q),
+ card->qdio.out_qs[i] = kzalloc(sizeof(struct qeth_qdio_out_q),
GFP_KERNEL);
if (!card->qdio.out_qs[i])
goto out_freeoutq;
QETH_DBF_TEXT_(SETUP, 2, "outq %i", i);
QETH_DBF_HEX(SETUP, 2, &card->qdio.out_qs[i], sizeof(void *));
- memset(card->qdio.out_qs[i], 0, sizeof(struct qeth_qdio_out_q));
card->qdio.out_qs[i]->queue_no = i;
/* give outbound qeth_qdio_buffers their qdio_buffers */
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) {
@@ -2565,6 +2657,12 @@ static int qeth_init_input_buffer(struct
struct qeth_buffer_pool_entry *pool_entry;
int i;
+ if ((card->options.cq == QETH_CQ_ENABLED) && (!buf->rx_skb)) {
+ buf->rx_skb = dev_alloc_skb(QETH_RX_PULL_LEN + ETH_HLEN);
+ if (!buf->rx_skb)
+ return 1;
+ }
+
pool_entry = qeth_find_free_buffer_pool_entry(card);
if (!pool_entry)
return 1;
@@ -2954,9 +3052,19 @@ int qeth_check_qdio_errors(struct qeth_c
}
EXPORT_SYMBOL_GPL(qeth_check_qdio_errors);
+void qeth_buffer_reclaim_work(struct work_struct *work)
+{
+ struct qeth_card *card = container_of(work, struct qeth_card,
+ buffer_reclaim_work.work);
+
+ QETH_CARD_TEXT_(card, 2, "brw:%x", card->reclaim_index);
+ qeth_queue_input_buffer(card, card->reclaim_index);
+}
+
void qeth_queue_input_buffer(struct qeth_card *card, int index)
{
struct qeth_qdio_q *queue = card->qdio.in_q;
+ struct list_head *lh;
int count;
int i;
int rc;
@@ -2988,6 +3096,20 @@ void qeth_queue_input_buffer(struct qeth
atomic_add_unless(&card->force_alloc_skb, -1, 0);
}
+ if (!count) {
+ i = 0;
+ list_for_each(lh, &card->qdio.in_buf_pool.entry_list)
+ i++;
+ if (i == card->qdio.in_buf_pool.buf_count) {
+ QETH_CARD_TEXT(card, 2, "qsarbw");
+ card->reclaim_index = index;
+ schedule_delayed_work(
+ &card->buffer_reclaim_work,
+ QETH_RECLAIM_WORK_TIME);
+ }
+ return;
+ }
+
/*
* according to old code it should be avoided to requeue all
* 128 buffers in order to benefit from PCI avoidance.
@@ -3396,15 +3518,34 @@ void qeth_qdio_output_handler(struct ccw
if (queue->bufstates &&
(queue->bufstates[bidx].flags &
QDIO_OUTBUF_STATE_FLAG_PENDING) != 0) {
+ BUG_ON(card->options.cq != QETH_CQ_ENABLED);
+
+ if (atomic_cmpxchg(&buffer->state,
+ QETH_QDIO_BUF_PRIMED,
+ QETH_QDIO_BUF_PENDING) ==
+ QETH_QDIO_BUF_PRIMED) {
+ qeth_notify_skbs(queue, buffer,
+ TX_NOTIFY_PENDING);
+ }
buffer->aob = queue->bufstates[bidx].aob;
QETH_CARD_TEXT_(queue->card, 5, "pel%d", bidx);
- QETH_CARD_TEXT_(queue->card, 5, "aob");
+ QETH_CARD_TEXT(queue->card, 5, "aob");
QETH_CARD_TEXT_(queue->card, 5, "%lx",
virt_to_phys(buffer->aob));
BUG_ON(bidx < 0 || bidx >= QDIO_MAX_BUFFERS_PER_Q);
- if (qeth_init_qdio_out_buf(queue, bidx))
+ if (qeth_init_qdio_out_buf(queue, bidx)) {
+ QETH_CARD_TEXT(card, 2, "outofbuf");
qeth_schedule_recovery(card);
+ }
} else {
+ if (card->options.cq == QETH_CQ_ENABLED) {
+ enum iucv_tx_notify n;
+
+ n = qeth_compute_cq_notification(
+ buffer->buffer->element[15].sflags, 0);
+ qeth_notify_skbs(queue, buffer, n);
+ }
+
qeth_clear_output_buffer(queue, buffer,
QETH_QDIO_BUF_EMPTY);
}
@@ -4335,7 +4476,7 @@ static int qeth_qdio_establish(struct qe
qeth_create_qib_param_field(card, qib_param_field);
qeth_create_qib_param_field_blkt(card, qib_param_field);
- in_sbal_ptrs = kmalloc(card->qdio.no_in_queues *
+ in_sbal_ptrs = kzalloc(card->qdio.no_in_queues *
QDIO_MAX_BUFFERS_PER_Q * sizeof(void *),
GFP_KERNEL);
if (!in_sbal_ptrs) {
@@ -4359,7 +4500,7 @@ static int qeth_qdio_establish(struct qe
qeth_qdio_establish_cq(card, in_sbal_ptrs, queue_start_poll);
out_sbal_ptrs =
- kmalloc(card->qdio.no_out_queues * QDIO_MAX_BUFFERS_PER_Q *
+ kzalloc(card->qdio.no_out_queues * QDIO_MAX_BUFFERS_PER_Q *
sizeof(void *), GFP_KERNEL);
if (!out_sbal_ptrs) {
rc = -ENOMEM;
@@ -4557,29 +4698,36 @@ out:
}
EXPORT_SYMBOL_GPL(qeth_core_hardsetup_card);
-static inline int qeth_create_skb_frag(struct qdio_buffer_element *element,
+static inline int qeth_create_skb_frag(struct qeth_qdio_buffer *qethbuffer,
+ struct qdio_buffer_element *element,
struct sk_buff **pskb, int offset, int *pfrag, int data_len)
{
struct page *page = virt_to_page(element->addr);
if (*pskb == NULL) {
- /* the upper protocol layers assume that there is data in the
- * skb itself. Copy a small amount (64 bytes) to make them
- * happy. */
- *pskb = dev_alloc_skb(64 + ETH_HLEN);
- if (!(*pskb))
- return -ENOMEM;
+ if (qethbuffer->rx_skb) {
+ /* only if qeth_card.options.cq == QETH_CQ_ENABLED */
+ *pskb = qethbuffer->rx_skb;
+ qethbuffer->rx_skb = NULL;
+ } else {
+ *pskb = dev_alloc_skb(QETH_RX_PULL_LEN + ETH_HLEN);
+ if (!(*pskb))
+ return -ENOMEM;
+ }
+
skb_reserve(*pskb, ETH_HLEN);
- if (data_len <= 64) {
+ if (data_len <= QETH_RX_PULL_LEN) {
memcpy(skb_put(*pskb, data_len), element->addr + offset,
data_len);
} else {
get_page(page);
- memcpy(skb_put(*pskb, 64), element->addr + offset, 64);
- skb_fill_page_desc(*pskb, *pfrag, page, offset + 64,
- data_len - 64);
- (*pskb)->data_len += data_len - 64;
- (*pskb)->len += data_len - 64;
- (*pskb)->truesize += data_len - 64;
+ memcpy(skb_put(*pskb, QETH_RX_PULL_LEN),
+ element->addr + offset, QETH_RX_PULL_LEN);
+ skb_fill_page_desc(*pskb, *pfrag, page,
+ offset + QETH_RX_PULL_LEN,
+ data_len - QETH_RX_PULL_LEN);
+ (*pskb)->data_len += data_len - QETH_RX_PULL_LEN;
+ (*pskb)->len += data_len - QETH_RX_PULL_LEN;
+ (*pskb)->truesize += data_len - QETH_RX_PULL_LEN;
(*pfrag)++;
}
} else {
@@ -4596,11 +4744,12 @@ static inline int qeth_create_skb_frag(s
}
struct sk_buff *qeth_core_get_next_skb(struct qeth_card *card,
- struct qdio_buffer *buffer,
+ struct qeth_qdio_buffer *qethbuffer,
struct qdio_buffer_element **__element, int *__offset,
struct qeth_hdr **hdr)
{
struct qdio_buffer_element *element = *__element;
+ struct qdio_buffer *buffer = qethbuffer->buffer;
int offset = *__offset;
struct sk_buff *skb = NULL;
int skb_len = 0;
@@ -4645,9 +4794,10 @@ struct sk_buff *qeth_core_get_next_skb(s
if (!skb_len)
return NULL;
- if ((skb_len >= card->options.rx_sg_cb) &&
- (!(card->info.type == QETH_CARD_TYPE_OSN)) &&
- (!atomic_read(&card->force_alloc_skb))) {
+ if (((skb_len >= card->options.rx_sg_cb) &&
+ (!(card->info.type == QETH_CARD_TYPE_OSN)) &&
+ (!atomic_read(&card->force_alloc_skb))) ||
+ (card->options.cq == QETH_CQ_ENABLED)) {
use_rx_sg = 1;
} else {
skb = dev_alloc_skb(skb_len + headroom);
@@ -4662,8 +4812,8 @@ struct sk_buff *qeth_core_get_next_skb(s
data_len = min(skb_len, (int)(element->length - offset));
if (data_len) {
if (use_rx_sg) {
- if (qeth_create_skb_frag(element, &skb, offset,
- &frag, data_len))
+ if (qeth_create_skb_frag(qethbuffer, element,
+ &skb, offset, &frag, data_len))
goto no_mem;
} else {
memcpy(skb_put(skb, data_len), data_ptr,
--- a/drivers/s390/net/qeth_l2_main.c
+++ b/drivers/s390/net/qeth_l2_main.c
@@ -409,7 +409,7 @@ static int qeth_l2_process_inbound_buffe
BUG_ON(!budget);
while (budget) {
skb = qeth_core_get_next_skb(card,
- card->qdio.in_q->bufs[card->rx.b_index].buffer,
+ &card->qdio.in_q->bufs[card->rx.b_index],
&card->rx.b_element, &card->rx.e_offset, &hdr);
if (!skb) {
*done = 1;
--- a/drivers/s390/net/qeth_l3.h
+++ b/drivers/s390/net/qeth_l3.h
@@ -63,5 +63,9 @@ int qeth_l3_add_rxip(struct qeth_card *,
void qeth_l3_del_rxip(struct qeth_card *card, enum qeth_prot_versions,
const u8 *);
int qeth_l3_is_addr_covered_by_ipato(struct qeth_card *, struct qeth_ipaddr *);
+struct qeth_ipaddr *qeth_l3_get_addr_buffer(enum qeth_prot_versions);
+int qeth_l3_add_ip(struct qeth_card *, struct qeth_ipaddr *);
+int qeth_l3_delete_ip(struct qeth_card *, struct qeth_ipaddr *);
+void qeth_l3_set_ip_addr_list(struct qeth_card *);
#endif /* __QETH_L3_H__ */
--- a/drivers/s390/net/qeth_l3_main.c
+++ b/drivers/s390/net/qeth_l3_main.c
@@ -29,6 +29,7 @@
#include <net/ip.h>
#include <net/arp.h>
#include <net/ip6_checksum.h>
+#include <net/iucv/af_iucv.h>
#include "qeth_l3.h"
@@ -267,7 +268,7 @@ static int __qeth_l3_insert_ip_todo(stru
}
}
-static int qeth_l3_delete_ip(struct qeth_card *card, struct qeth_ipaddr *addr)
+int qeth_l3_delete_ip(struct qeth_card *card, struct qeth_ipaddr *addr)
{
unsigned long flags;
int rc = 0;
@@ -286,7 +287,7 @@ static int qeth_l3_delete_ip(struct qeth
return rc;
}
-static int qeth_l3_add_ip(struct qeth_card *card, struct qeth_ipaddr *addr)
+int qeth_l3_add_ip(struct qeth_card *card, struct qeth_ipaddr *addr)
{
unsigned long flags;
int rc = 0;
@@ -305,7 +306,7 @@ static int qeth_l3_add_ip(struct qeth_ca
}
-static struct qeth_ipaddr *qeth_l3_get_addr_buffer(
+struct qeth_ipaddr *qeth_l3_get_addr_buffer(
enum qeth_prot_versions prot)
{
struct qeth_ipaddr *addr;
@@ -421,7 +422,7 @@ again:
list_splice(&fail_list, &card->ip_list);
}
-static void qeth_l3_set_ip_addr_list(struct qeth_card *card)
+void qeth_l3_set_ip_addr_list(struct qeth_card *card)
{
struct list_head *tbd_list;
struct qeth_ipaddr *todo, *addr;
@@ -438,7 +439,7 @@ static void qeth_l3_set_ip_addr_list(str
spin_lock_irqsave(&card->ip_lock, flags);
tbd_list = card->ip_tbd_list;
- card->ip_tbd_list = kmalloc(sizeof(struct list_head), GFP_ATOMIC);
+ card->ip_tbd_list = kzalloc(sizeof(struct list_head), GFP_ATOMIC);
if (!card->ip_tbd_list) {
QETH_CARD_TEXT(card, 0, "silnomem");
card->ip_tbd_list = tbd_list;
@@ -1993,12 +1994,13 @@ static int qeth_l3_process_inbound_buffe
__u16 vlan_tag = 0;
int is_vlan;
unsigned int len;
+ __u16 magic;
*done = 0;
BUG_ON(!budget);
while (budget) {
skb = qeth_core_get_next_skb(card,
- card->qdio.in_q->bufs[card->rx.b_index].buffer,
+ &card->qdio.in_q->bufs[card->rx.b_index],
&card->rx.b_element, &card->rx.e_offset, &hdr);
if (!skb) {
*done = 1;
@@ -2007,12 +2009,26 @@ static int qeth_l3_process_inbound_buffe
skb->dev = card->dev;
switch (hdr->hdr.l3.id) {
case QETH_HEADER_TYPE_LAYER3:
- is_vlan = qeth_l3_rebuild_skb(card, skb, hdr,
+ magic = *(__u16 *)skb->data;
+ if ((card->info.type == QETH_CARD_TYPE_IQD) &&
+ (magic == ETH_P_AF_IUCV)) {
+ skb->protocol = ETH_P_AF_IUCV;
+ skb->pkt_type = PACKET_HOST;
+ skb->mac_header = NET_SKB_PAD;
+ skb->dev = card->dev;
+ len = skb->len;
+ card->dev->header_ops->create(skb, card->dev, 0,
+ card->dev->dev_addr, "FAKELL",
+ card->dev->addr_len);
+ netif_receive_skb(skb);
+ } else {
+ is_vlan = qeth_l3_rebuild_skb(card, skb, hdr,
&vlan_tag);
- len = skb->len;
- if (is_vlan && !card->options.sniffer)
- __vlan_hwaccel_put_tag(skb, vlan_tag);
- napi_gro_receive(&card->napi, skb);
+ len = skb->len;
+ if (is_vlan && !card->options.sniffer)
+ __vlan_hwaccel_put_tag(skb, vlan_tag);
+ napi_gro_receive(&card->napi, skb);
+ }
break;
case QETH_HEADER_TYPE_LAYER2: /* for HiperSockets sniffer */
skb->pkt_type = PACKET_HOST;
@@ -2784,6 +2800,30 @@ int inline qeth_l3_get_cast_type(struct
return cast_type;
}
+static void qeth_l3_fill_af_iucv_hdr(struct qeth_card *card,
+ struct qeth_hdr *hdr, struct sk_buff *skb)
+{
+ char daddr[16];
+ struct af_iucv_trans_hdr *iucv_hdr;
+
+ skb_pull(skb, 14);
+ card->dev->header_ops->create(skb, card->dev, 0,
+ card->dev->dev_addr, card->dev->dev_addr,
+ card->dev->addr_len);
+ skb_pull(skb, 14);
+ iucv_hdr = (struct af_iucv_trans_hdr *)skb->data;
+ memset(hdr, 0, sizeof(struct qeth_hdr));
+ hdr->hdr.l3.id = QETH_HEADER_TYPE_LAYER3;
+ hdr->hdr.l3.ext_flags = 0;
+ hdr->hdr.l3.length = skb->len;
+ hdr->hdr.l3.flags = QETH_HDR_IPV6 | QETH_CAST_UNICAST;
+ memset(daddr, 0, sizeof(daddr));
+ daddr[0] = 0xfe;
+ daddr[1] = 0x80;
+ memcpy(&daddr[8], iucv_hdr->destUserID, 8);
+ memcpy(hdr->hdr.l3.dest_addr, daddr, 16);
+}
+
static void qeth_l3_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
struct sk_buff *skb, int ipv, int cast_type)
{
@@ -2936,8 +2976,11 @@ static int qeth_l3_hard_start_xmit(struc
int data_offset = -1;
int nr_frags;
- if (((card->info.type == QETH_CARD_TYPE_IQD) && (!ipv)) ||
- card->options.sniffer)
+ if (((card->info.type == QETH_CARD_TYPE_IQD) &&
+ (((card->options.cq != QETH_CQ_ENABLED) && !ipv) ||
+ ((card->options.cq == QETH_CQ_ENABLED) &&
+ (skb->protocol != ETH_P_AF_IUCV)))) ||
+ card->options.sniffer)
goto tx_drop;
if ((card->state != CARD_STATE_UP) || !card->lan_online) {
@@ -2959,7 +3002,10 @@ static int qeth_l3_hard_start_xmit(struc
if ((card->info.type == QETH_CARD_TYPE_IQD) && (!large_send) &&
(skb_shinfo(skb)->nr_frags == 0)) {
new_skb = skb;
- data_offset = ETH_HLEN;
+ if (new_skb->protocol == ETH_P_AF_IUCV)
+ data_offset = 0;
+ else
+ data_offset = ETH_HLEN;
hdr = kmem_cache_alloc(qeth_core_header_cache, GFP_ATOMIC);
if (!hdr)
goto tx_drop;
@@ -3024,9 +3070,13 @@ static int qeth_l3_hard_start_xmit(struc
qeth_l3_fill_header(card, hdr, new_skb, ipv,
cast_type);
} else {
- qeth_l3_fill_header(card, hdr, new_skb, ipv,
- cast_type);
- hdr->hdr.l3.length = new_skb->len - data_offset;
+ if (new_skb->protocol == ETH_P_AF_IUCV)
+ qeth_l3_fill_af_iucv_hdr(card, hdr, new_skb);
+ else {
+ qeth_l3_fill_header(card, hdr, new_skb, ipv,
+ cast_type);
+ hdr->hdr.l3.length = new_skb->len - data_offset;
+ }
}
if (skb->ip_summed == CHECKSUM_PARTIAL)
@@ -3289,6 +3339,8 @@ static int qeth_l3_setup_netdev(struct q
card->dev->flags |= IFF_NOARP;
card->dev->netdev_ops = &qeth_l3_netdev_ops;
qeth_l3_iqd_read_initial_mac(card);
+ if (card->options.hsuid[0])
+ memcpy(card->dev->perm_addr, card->options.hsuid, 9);
} else
return -ENODEV;
@@ -3659,7 +3711,6 @@ static int qeth_l3_ip6_event(struct noti
struct qeth_ipaddr *addr;
struct qeth_card *card;
-
card = qeth_l3_get_card_from_dev(dev);
if (!card)
return NOTIFY_DONE;
--- a/drivers/s390/net/qeth_l3_sys.c
+++ b/drivers/s390/net/qeth_l3_sys.c
@@ -9,7 +9,7 @@
*/
#include <linux/slab.h>
-
+#include <asm/ebcdic.h>
#include "qeth_l3.h"
#define QETH_DEVICE_ATTR(_id, _name, _mode, _show, _store) \
@@ -308,6 +308,8 @@ static ssize_t qeth_l3_dev_sniffer_store
if (card->info.type != QETH_CARD_TYPE_IQD)
return -EPERM;
+ if (card->options.cq == QETH_CQ_ENABLED)
+ return -EPERM;
mutex_lock(&card->conf_mutex);
if ((card->state != CARD_STATE_DOWN) &&
@@ -347,6 +349,111 @@ out:
static DEVICE_ATTR(sniffer, 0644, qeth_l3_dev_sniffer_show,
qeth_l3_dev_sniffer_store);
+
+static ssize_t qeth_l3_dev_hsuid_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct qeth_card *card = dev_get_drvdata(dev);
+ char tmp_hsuid[9];
+
+ if (!card)
+ return -EINVAL;
+
+ if (card->info.type != QETH_CARD_TYPE_IQD)
+ return -EPERM;
+
+ if (card->state == CARD_STATE_DOWN)
+ return -EPERM;
+
+ memcpy(tmp_hsuid, card->options.hsuid, sizeof(tmp_hsuid));
+ EBCASC(tmp_hsuid, 8);
+ return sprintf(buf, "%s\n", tmp_hsuid);
+}
+
+static ssize_t qeth_l3_dev_hsuid_store(struct device *dev,
+ struct device_attribute *attr, const char *buf, size_t count)
+{
+ struct qeth_card *card = dev_get_drvdata(dev);
+ struct qeth_ipaddr *addr;
+ char *tmp;
+ int i;
+
+ if (!card)
+ return -EINVAL;
+
+ if (card->info.type != QETH_CARD_TYPE_IQD)
+ return -EPERM;
+ if (card->state != CARD_STATE_DOWN &&
+ card->state != CARD_STATE_RECOVER)
+ return -EPERM;
+ if (card->options.sniffer)
+ return -EPERM;
+ if (card->options.cq == QETH_CQ_NOTAVAILABLE)
+ return -EPERM;
+
+ tmp = strsep((char **)&buf, "\n");
+ if (strlen(tmp) > 8)
+ return -EINVAL;
+
+ if (card->options.hsuid[0]) {
+ /* delete old ip address */
+ addr = qeth_l3_get_addr_buffer(QETH_PROT_IPV6);
+ if (addr != NULL) {
+ addr->u.a6.addr.s6_addr32[0] = 0xfe800000;
+ addr->u.a6.addr.s6_addr32[1] = 0x00000000;
+ for (i = 8; i < 16; i++)
+ addr->u.a6.addr.s6_addr[i] =
+ card->options.hsuid[i - 8];
+ addr->u.a6.pfxlen = 0;
+ addr->type = QETH_IP_TYPE_NORMAL;
+ } else
+ return -ENOMEM;
+ if (!qeth_l3_delete_ip(card, addr))
+ kfree(addr);
+ qeth_l3_set_ip_addr_list(card);
+ }
+
+ if (strlen(tmp) == 0) {
+ /* delete ip address only */
+ card->options.hsuid[0] = '\0';
+ if (card->dev)
+ memcpy(card->dev->perm_addr, card->options.hsuid, 9);
+ qeth_configure_cq(card, QETH_CQ_DISABLED);
+ return count;
+ }
+
+ if (qeth_configure_cq(card, QETH_CQ_ENABLED))
+ return -EPERM;
+
+ for (i = 0; i < 8; i++)
+ card->options.hsuid[i] = ' ';
+ card->options.hsuid[8] = '\0';
+ strncpy(card->options.hsuid, tmp, strlen(tmp));
+ ASCEBC(card->options.hsuid, 8);
+ if (card->dev)
+ memcpy(card->dev->perm_addr, card->options.hsuid, 9);
+
+ addr = qeth_l3_get_addr_buffer(QETH_PROT_IPV6);
+ if (addr != NULL) {
+ addr->u.a6.addr.s6_addr32[0] = 0xfe800000;
+ addr->u.a6.addr.s6_addr32[1] = 0x00000000;
+ for (i = 8; i < 16; i++)
+ addr->u.a6.addr.s6_addr[i] = card->options.hsuid[i - 8];
+ addr->u.a6.pfxlen = 0;
+ addr->type = QETH_IP_TYPE_NORMAL;
+ } else
+ return -ENOMEM;
+ if (!qeth_l3_add_ip(card, addr))
+ kfree(addr);
+ qeth_l3_set_ip_addr_list(card);
+
+ return count;
+}
+
+static DEVICE_ATTR(hsuid, 0644, qeth_l3_dev_hsuid_show,
+ qeth_l3_dev_hsuid_store);
+
+
static struct attribute *qeth_l3_device_attrs[] = {
&dev_attr_route4.attr,
&dev_attr_route6.attr,
@@ -354,6 +461,7 @@ static struct attribute *qeth_l3_device_
&dev_attr_broadcast_mode.attr,
&dev_attr_canonical_macaddr.attr,
&dev_attr_sniffer.attr,
+ &dev_attr_hsuid.attr,
NULL,
};
^ permalink raw reply
* [patch 00/13] s390: network patches for net-next
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390
Hi Dave,
this patch set contains s390 network patches for net-next.
Non of the patches touch drivers/net.
There are 2 qeth patches:
shortlog:
Ursula Braun (1)
qeth: do not apply priority queuing to HiperSockets
Frank Blaschka (1)
qeth: l3 ipv6 vlan not working on shared OSA chpid
And the respin of the af_iucv HiperSockets transport feature
(I sent out two week ago for review):
Following patch set implements a new feature to use HiperSockets
as a transport mechanism for the af_iucv address family. Basic
design idea is: The current transport mechanism for af_iucv (iucv)
is only available on VM. HiperSockets provide similar capabilities
as iucv and are available on LPAR. There is guaranty for data
integrity, in order delivery and delivery acknowledgment. A new
HiperSockets feature is exploited to support
flow control/congestion management. For this it is nearby to extend
af_iucv address family to use HiperSockets transport.
shortlog:
Einar Lueck (4)
qdio: support asynchronous delivery of storage blocks
qdio: support forced signal adapter indications
qeth: exploit asynchronous delivery of storage blocks
qeth: support forced signal adapter indications
Ursula Braun (3)
iucv: kernel option for z/VM IUCV and HiperSockets
af_iucv: cleanup - use iucv_sk(sk) early
af_iucv: add HiperSockets transport
Frank Blaschka (4)
iucv: introduce loadable iucv interface
af_iucv: use loadable iucv interface
if_ether: add new Ethernet Protocol ID for af_iucv
qeth: add support for af_iucv HiperSockets transport
Thanks,
Frank
^ permalink raw reply
* [patch 03/13] [PATCH] iucv: introduce loadable iucv interface
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117-iucv-introduce-loadable-iucv-interface.patch --]
[-- Type: text/plain, Size: 3456 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
This patch adds a symbol to dynamically load iucv functions.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
include/net/iucv/iucv.h | 36 +++++++++++++++++++++++++++++++++++-
net/iucv/iucv.c | 23 +++++++++++++++++++++++
2 files changed, 58 insertions(+), 1 deletion(-)
--- a/include/net/iucv/iucv.h
+++ b/include/net/iucv/iucv.h
@@ -120,7 +120,7 @@ struct iucv_message {
u32 reply_size;
u8 rmmsg[8];
u8 flags;
-};
+} __packed;
/*
* struct iucv_handler
@@ -459,3 +459,37 @@ int __iucv_message_send(struct iucv_path
int iucv_message_send2way(struct iucv_path *path, struct iucv_message *msg,
u8 flags, u32 srccls, void *buffer, size_t size,
void *answer, size_t asize, size_t *residual);
+
+struct iucv_interface {
+ int (*message_receive)(struct iucv_path *path, struct iucv_message *msg,
+ u8 flags, void *buffer, size_t size, size_t *residual);
+ int (*__message_receive)(struct iucv_path *path,
+ struct iucv_message *msg, u8 flags, void *buffer, size_t size,
+ size_t *residual);
+ int (*message_reply)(struct iucv_path *path, struct iucv_message *msg,
+ u8 flags, void *reply, size_t size);
+ int (*message_reject)(struct iucv_path *path, struct iucv_message *msg);
+ int (*message_send)(struct iucv_path *path, struct iucv_message *msg,
+ u8 flags, u32 srccls, void *buffer, size_t size);
+ int (*__message_send)(struct iucv_path *path, struct iucv_message *msg,
+ u8 flags, u32 srccls, void *buffer, size_t size);
+ int (*message_send2way)(struct iucv_path *path,
+ struct iucv_message *msg, u8 flags, u32 srccls, void *buffer,
+ size_t size, void *answer, size_t asize, size_t *residual);
+ int (*message_purge)(struct iucv_path *path, struct iucv_message *msg,
+ u32 srccls);
+ int (*path_accept)(struct iucv_path *path, struct iucv_handler *handler,
+ u8 userdata[16], void *private);
+ int (*path_connect)(struct iucv_path *path,
+ struct iucv_handler *handler,
+ u8 userid[8], u8 system[8], u8 userdata[16], void *private);
+ int (*path_quiesce)(struct iucv_path *path, u8 userdata[16]);
+ int (*path_resume)(struct iucv_path *path, u8 userdata[16]);
+ int (*path_sever)(struct iucv_path *path, u8 userdata[16]);
+ int (*iucv_register)(struct iucv_handler *handler, int smp);
+ void (*iucv_unregister)(struct iucv_handler *handler, int smp);
+ struct bus_type *bus;
+ struct device *root;
+};
+
+extern struct iucv_interface iucv_if;
--- a/net/iucv/iucv.c
+++ b/net/iucv/iucv.c
@@ -1974,6 +1974,27 @@ out:
return rc;
}
+struct iucv_interface iucv_if = {
+ .message_receive = iucv_message_receive,
+ .__message_receive = __iucv_message_receive,
+ .message_reply = iucv_message_reply,
+ .message_reject = iucv_message_reject,
+ .message_send = iucv_message_send,
+ .__message_send = __iucv_message_send,
+ .message_send2way = iucv_message_send2way,
+ .message_purge = iucv_message_purge,
+ .path_accept = iucv_path_accept,
+ .path_connect = iucv_path_connect,
+ .path_quiesce = iucv_path_quiesce,
+ .path_resume = iucv_path_resume,
+ .path_sever = iucv_path_sever,
+ .iucv_register = iucv_register,
+ .iucv_unregister = iucv_unregister,
+ .bus = NULL,
+ .root = NULL,
+};
+EXPORT_SYMBOL(iucv_if);
+
/**
* iucv_init
*
@@ -2038,6 +2059,8 @@ static int __init iucv_init(void)
rc = bus_register(&iucv_bus);
if (rc)
goto out_reboot;
+ iucv_if.root = iucv_root;
+ iucv_if.bus = &iucv_bus;
return 0;
out_reboot:
^ permalink raw reply
* [patch 01/13] [PATCH] qeth: l3 ipv6 vlan not working on shared OSA chpid
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: qeth_l3_ipv6_vlan.patch --]
[-- Type: text/plain, Size: 715 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
In layer 3 mode IPv6 over VLAN does not work on newer OSA levels in
case the sender and receiver run on the same (shared) OSA adapter.
Keep vlan info in the skb so the qdio header is filled with the
required vlan tag.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
drivers/s390/net/qeth_l3_main.c | 1 -
1 file changed, 1 deletion(-)
--- a/drivers/s390/net/qeth_l3_main.c
+++ b/drivers/s390/net/qeth_l3_main.c
@@ -2993,7 +2993,6 @@ static int qeth_l3_hard_start_xmit(struc
tag = (u16 *)(new_skb->data + 12);
*tag = __constant_htons(ETH_P_8021Q);
*(tag + 1) = htons(vlan_tx_tag_get(new_skb));
- new_skb->vlan_tci = 0;
}
}
^ permalink raw reply
* [patch 06/13] [PATCH] af_iucv: cleanup - use iucv_sk(sk) early
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Ursula Braun
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117-af_iucv-cleanup-use-iucv_sk-early.patch --]
[-- Type: text/plain, Size: 3567 bytes --]
From: Ursula Braun <ursula.braun@de.ibm.com>
Code cleanup making make use of local variable for struct iucv_sock.
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
net/iucv/af_iucv.c | 44 +++++++++++++++++++++++---------------------
1 file changed, 23 insertions(+), 21 deletions(-)
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c
@@ -446,23 +446,25 @@ static void iucv_sock_init(struct sock *
static struct sock *iucv_sock_alloc(struct socket *sock, int proto, gfp_t prio)
{
struct sock *sk;
+ struct iucv_sock *iucv;
sk = sk_alloc(&init_net, PF_IUCV, prio, &iucv_proto);
if (!sk)
return NULL;
+ iucv = iucv_sk(sk);
sock_init_data(sock, sk);
- INIT_LIST_HEAD(&iucv_sk(sk)->accept_q);
- spin_lock_init(&iucv_sk(sk)->accept_q_lock);
- skb_queue_head_init(&iucv_sk(sk)->send_skb_q);
- INIT_LIST_HEAD(&iucv_sk(sk)->message_q.list);
- spin_lock_init(&iucv_sk(sk)->message_q.lock);
- skb_queue_head_init(&iucv_sk(sk)->backlog_skb_q);
- iucv_sk(sk)->send_tag = 0;
- iucv_sk(sk)->flags = 0;
- iucv_sk(sk)->msglimit = IUCV_QUEUELEN_DEFAULT;
- iucv_sk(sk)->path = NULL;
- memset(&iucv_sk(sk)->src_user_id , 0, 32);
+ INIT_LIST_HEAD(&iucv->accept_q);
+ spin_lock_init(&iucv->accept_q_lock);
+ skb_queue_head_init(&iucv->send_skb_q);
+ INIT_LIST_HEAD(&iucv->message_q.list);
+ spin_lock_init(&iucv->message_q.lock);
+ skb_queue_head_init(&iucv->backlog_skb_q);
+ iucv->send_tag = 0;
+ iucv->flags = 0;
+ iucv->msglimit = IUCV_QUEUELEN_DEFAULT;
+ iucv->path = NULL;
+ memset(&iucv->src_user_id , 0, 32);
sk->sk_destruct = iucv_sock_destruct;
sk->sk_sndtimeo = IUCV_CONN_TIMEOUT;
@@ -669,7 +671,7 @@ static int iucv_sock_connect(struct sock
{
struct sockaddr_iucv *sa = (struct sockaddr_iucv *) addr;
struct sock *sk = sock->sk;
- struct iucv_sock *iucv;
+ struct iucv_sock *iucv = iucv_sk(sk);
unsigned char user_data[16];
int err;
@@ -691,14 +693,13 @@ static int iucv_sock_connect(struct sock
lock_sock(sk);
/* Set the destination information */
- memcpy(iucv_sk(sk)->dst_user_id, sa->siucv_user_id, 8);
- memcpy(iucv_sk(sk)->dst_name, sa->siucv_name, 8);
+ memcpy(iucv->dst_user_id, sa->siucv_user_id, 8);
+ memcpy(iucv->dst_name, sa->siucv_name, 8);
high_nmcpy(user_data, sa->siucv_name);
- low_nmcpy(user_data, iucv_sk(sk)->src_name);
+ low_nmcpy(user_data, iucv->src_name);
ASCEBC(user_data, sizeof(user_data));
- iucv = iucv_sk(sk);
/* Create path. */
iucv->path = iucv_path_alloc(iucv->msglimit,
IUCV_IPRMDATA, GFP_KERNEL);
@@ -836,20 +837,21 @@ static int iucv_sock_getname(struct sock
{
struct sockaddr_iucv *siucv = (struct sockaddr_iucv *) addr;
struct sock *sk = sock->sk;
+ struct iucv_sock *iucv = iucv_sk(sk);
addr->sa_family = AF_IUCV;
*len = sizeof(struct sockaddr_iucv);
if (peer) {
- memcpy(siucv->siucv_user_id, iucv_sk(sk)->dst_user_id, 8);
- memcpy(siucv->siucv_name, &iucv_sk(sk)->dst_name, 8);
+ memcpy(siucv->siucv_user_id, iucv->dst_user_id, 8);
+ memcpy(siucv->siucv_name, iucv->dst_name, 8);
} else {
- memcpy(siucv->siucv_user_id, iucv_sk(sk)->src_user_id, 8);
- memcpy(siucv->siucv_name, iucv_sk(sk)->src_name, 8);
+ memcpy(siucv->siucv_user_id, iucv->src_user_id, 8);
+ memcpy(siucv->siucv_name, iucv->src_name, 8);
}
memset(&siucv->siucv_port, 0, sizeof(siucv->siucv_port));
memset(&siucv->siucv_addr, 0, sizeof(siucv->siucv_addr));
- memset(siucv->siucv_nodeid, 0, sizeof(siucv->siucv_nodeid));
+ memset(&siucv->siucv_nodeid, 0, sizeof(siucv->siucv_nodeid));
return 0;
}
^ permalink raw reply
* [patch 07/13] [PATCH] if_ether: add new Ethernet Protocol ID for af_iucv
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117-af_iucv-new-protocol-id.patch --]
[-- Type: text/plain, Size: 823 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
Add a new ethertype for af_iucv over s/390 HiperSockets transport. Since
HiperSockets is not a real ethernet hw this is not an officially registered ID.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
include/linux/if_ether.h | 1 +
1 file changed, 1 insertion(+)
--- a/include/linux/if_ether.h
+++ b/include/linux/if_ether.h
@@ -88,6 +88,7 @@
#define ETH_P_QINQ2 0x9200 /* deprecated QinQ VLAN [ NOT AN OFFICIALLY REGISTERED ID ] */
#define ETH_P_QINQ3 0x9300 /* deprecated QinQ VLAN [ NOT AN OFFICIALLY REGISTERED ID ] */
#define ETH_P_EDSA 0xDADA /* Ethertype DSA [ NOT AN OFFICIALLY REGISTERED ID ] */
+#define ETH_P_AF_IUCV 0xFBFB /* IBM af_iucv [ NOT AN OFFICIALLY REGISTERED ID ] */
/*
* Non DIX types. Won't clash for 1500 types.
^ permalink raw reply
* [patch 04/13] [PATCH] iucv: kernel option for z/VM IUCV and HiperSockets
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Ursula Braun
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117-af_iucv-kernel-option-for-z-VM-IUCV-and-HiperSockets.patch transports --]
[-- Type: text/plain, Size: 1375 bytes --]
From: Ursula Braun <ursula.braun@de.ibm.com>
When adding HiperSockets transport to AF_IUCV Sockets, af_iucv either
depends on IUCV or QETH_L3 (or both). This patch introduces the
necessary changes for kernel configuration.
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
net/iucv/Kconfig | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)
--- a/net/iucv/Kconfig
+++ b/net/iucv/Kconfig
@@ -1,15 +1,17 @@
config IUCV
- tristate "IUCV support (S390 - z/VM only)"
depends on S390
+ def_tristate y if S390
+ prompt "IUCV support (S390 - z/VM only)"
help
Select this option if you want to use inter-user communication
under VM or VIF. If you run on z/VM, say "Y" to enable a fast
communication link between VM guests.
config AFIUCV
- tristate "AF_IUCV support (S390 - z/VM only)"
- depends on IUCV
+ depends on S390
+ def_tristate m if QETH_L3 || IUCV
+ prompt "AF_IUCV Socket support (S390 - z/VM and HiperSockets transport)"
help
- Select this option if you want to use inter-user communication under
- VM or VIF sockets. If you run on z/VM, say "Y" to enable a fast
- communication link between VM guests.
+ Select this option if you want to use AF_IUCV socket applications
+ based on z/VM inter-user communication vehicle or based on
+ HiperSockets.
^ permalink raw reply
* [patch 05/13] [PATCH] af_iucv: use loadable iucv interface
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: net1117-af_iucv-use-loadable-iucv-interface.patch --]
[-- Type: text/plain, Size: 8680 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
For future af_iucv extensions the module should be able to run in LPAR
mode too. For this we use the new dynamic loading iucv interface.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
net/iucv/af_iucv.c | 119 ++++++++++++++++++++++++++++++++---------------------
1 file changed, 74 insertions(+), 45 deletions(-)
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c
@@ -42,6 +42,8 @@ static struct proto iucv_proto = {
.obj_size = sizeof(struct iucv_sock),
};
+static struct iucv_interface *pr_iucv;
+
/* special AF_IUCV IPRM messages */
static const u8 iprm_shutdown[8] =
{0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01};
@@ -165,7 +167,7 @@ static int afiucv_pm_freeze(struct devic
case IUCV_CLOSING:
case IUCV_CONNECTED:
if (iucv->path) {
- err = iucv_path_sever(iucv->path, NULL);
+ err = pr_iucv->path_sever(iucv->path, NULL);
iucv_path_free(iucv->path);
iucv->path = NULL;
}
@@ -229,7 +231,7 @@ static const struct dev_pm_ops afiucv_pm
static struct device_driver af_iucv_driver = {
.owner = THIS_MODULE,
.name = "afiucv",
- .bus = &iucv_bus,
+ .bus = NULL,
.pm = &afiucv_pm_ops,
};
@@ -412,7 +414,7 @@ static void iucv_sock_close(struct sock
low_nmcpy(user_data, iucv->src_name);
high_nmcpy(user_data, iucv->dst_name);
ASCEBC(user_data, sizeof(user_data));
- iucv_path_sever(iucv->path, user_data);
+ pr_iucv->path_sever(iucv->path, user_data);
iucv_path_free(iucv->path);
iucv->path = NULL;
}
@@ -704,8 +706,9 @@ static int iucv_sock_connect(struct sock
err = -ENOMEM;
goto done;
}
- err = iucv_path_connect(iucv->path, &af_iucv_handler,
- sa->siucv_user_id, NULL, user_data, sk);
+ err = pr_iucv->path_connect(iucv->path, &af_iucv_handler,
+ sa->siucv_user_id, NULL, user_data,
+ sk);
if (err) {
iucv_path_free(iucv->path);
iucv->path = NULL;
@@ -738,7 +741,7 @@ static int iucv_sock_connect(struct sock
}
if (err) {
- iucv_path_sever(iucv->path, NULL);
+ pr_iucv->path_sever(iucv->path, NULL);
iucv_path_free(iucv->path);
iucv->path = NULL;
}
@@ -871,7 +874,7 @@ static int iucv_send_iprm(struct iucv_pa
memcpy(prmdata, (void *) skb->data, skb->len);
prmdata[7] = 0xff - (u8) skb->len;
- return iucv_message_send(path, msg, IUCV_IPRMDATA, 0,
+ return pr_iucv->message_send(path, msg, IUCV_IPRMDATA, 0,
(void *) prmdata, 8);
}
@@ -999,13 +1002,13 @@ static int iucv_sock_sendmsg(struct kioc
/* this error should never happen since the
* IUCV_IPRMDATA path flag is set... sever path */
if (err == 0x15) {
- iucv_path_sever(iucv->path, NULL);
+ pr_iucv->path_sever(iucv->path, NULL);
skb_unlink(skb, &iucv->send_skb_q);
err = -EPIPE;
goto fail;
}
} else
- err = iucv_message_send(iucv->path, &txmsg, 0, 0,
+ err = pr_iucv->message_send(iucv->path, &txmsg, 0, 0,
(void *) skb->data, skb->len);
if (err) {
if (err == 3) {
@@ -1095,8 +1098,9 @@ static void iucv_process_message(struct
skb->len = 0;
}
} else {
- rc = iucv_message_receive(path, msg, msg->flags & IUCV_IPRMDATA,
- skb->data, len, NULL);
+ rc = pr_iucv->message_receive(path, msg,
+ msg->flags & IUCV_IPRMDATA,
+ skb->data, len, NULL);
if (rc) {
kfree_skb(skb);
return;
@@ -1110,7 +1114,7 @@ static void iucv_process_message(struct
kfree_skb(skb);
skb = NULL;
if (rc) {
- iucv_path_sever(path, NULL);
+ pr_iucv->path_sever(path, NULL);
return;
}
skb = skb_dequeue(&iucv_sk(sk)->backlog_skb_q);
@@ -1327,8 +1331,8 @@ static int iucv_sock_shutdown(struct soc
if (how == SEND_SHUTDOWN || how == SHUTDOWN_MASK) {
txmsg.class = 0;
txmsg.tag = 0;
- err = iucv_message_send(iucv->path, &txmsg, IUCV_IPRMDATA, 0,
- (void *) iprm_shutdown, 8);
+ err = pr_iucv->message_send(iucv->path, &txmsg, IUCV_IPRMDATA,
+ 0, (void *) iprm_shutdown, 8);
if (err) {
switch (err) {
case 1:
@@ -1345,7 +1349,7 @@ static int iucv_sock_shutdown(struct soc
}
if (how == RCV_SHUTDOWN || how == SHUTDOWN_MASK) {
- err = iucv_path_quiesce(iucv_sk(sk)->path, NULL);
+ err = pr_iucv->path_quiesce(iucv->path, NULL);
if (err)
err = -ENOTCONN;
@@ -1372,7 +1376,7 @@ static int iucv_sock_release(struct sock
/* Unregister with IUCV base support */
if (iucv_sk(sk)->path) {
- iucv_path_sever(iucv_sk(sk)->path, NULL);
+ pr_iucv->path_sever(iucv_sk(sk)->path, NULL);
iucv_path_free(iucv_sk(sk)->path);
iucv_sk(sk)->path = NULL;
}
@@ -1514,14 +1518,14 @@ static int iucv_callback_connreq(struct
high_nmcpy(user_data, iucv->dst_name);
ASCEBC(user_data, sizeof(user_data));
if (sk->sk_state != IUCV_LISTEN) {
- err = iucv_path_sever(path, user_data);
+ err = pr_iucv->path_sever(path, user_data);
iucv_path_free(path);
goto fail;
}
/* Check for backlog size */
if (sk_acceptq_is_full(sk)) {
- err = iucv_path_sever(path, user_data);
+ err = pr_iucv->path_sever(path, user_data);
iucv_path_free(path);
goto fail;
}
@@ -1529,7 +1533,7 @@ static int iucv_callback_connreq(struct
/* Create the new socket */
nsk = iucv_sock_alloc(NULL, sk->sk_type, GFP_ATOMIC);
if (!nsk) {
- err = iucv_path_sever(path, user_data);
+ err = pr_iucv->path_sever(path, user_data);
iucv_path_free(path);
goto fail;
}
@@ -1553,9 +1557,9 @@ static int iucv_callback_connreq(struct
/* set message limit for path based on msglimit of accepting socket */
niucv->msglimit = iucv->msglimit;
path->msglim = iucv->msglimit;
- err = iucv_path_accept(path, &af_iucv_handler, nuser_data, nsk);
+ err = pr_iucv->path_accept(path, &af_iucv_handler, nuser_data, nsk);
if (err) {
- err = iucv_path_sever(path, user_data);
+ err = pr_iucv->path_sever(path, user_data);
iucv_path_free(path);
iucv_sock_kill(nsk);
goto fail;
@@ -1589,7 +1593,7 @@ static void iucv_callback_rx(struct iucv
int len;
if (sk->sk_shutdown & RCV_SHUTDOWN) {
- iucv_message_reject(path, msg);
+ pr_iucv->message_reject(path, msg);
return;
}
@@ -1718,6 +1722,41 @@ static const struct net_proto_family iuc
.create = iucv_sock_create,
};
+static int __init afiucv_iucv_init(void)
+{
+ int err;
+
+ err = pr_iucv->iucv_register(&af_iucv_handler, 0);
+ if (err)
+ goto out;
+ /* establish dummy device */
+ af_iucv_driver.bus = pr_iucv->bus;
+ err = driver_register(&af_iucv_driver);
+ if (err)
+ goto out_iucv;
+ af_iucv_dev = kzalloc(sizeof(struct device), GFP_KERNEL);
+ if (!af_iucv_dev) {
+ err = -ENOMEM;
+ goto out_driver;
+ }
+ dev_set_name(af_iucv_dev, "af_iucv");
+ af_iucv_dev->bus = pr_iucv->bus;
+ af_iucv_dev->parent = pr_iucv->root;
+ af_iucv_dev->release = (void (*)(struct device *))kfree;
+ af_iucv_dev->driver = &af_iucv_driver;
+ err = device_register(af_iucv_dev);
+ if (err)
+ goto out_driver;
+ return 0;
+
+out_driver:
+ driver_unregister(&af_iucv_driver);
+out_iucv:
+ pr_iucv->iucv_unregister(&af_iucv_handler, 0);
+out:
+ return err;
+}
+
static int __init afiucv_init(void)
{
int err;
@@ -1735,44 +1774,33 @@ static int __init afiucv_init(void)
goto out;
}
- err = iucv_register(&af_iucv_handler, 0);
- if (err)
+ pr_iucv = try_then_request_module(symbol_get(iucv_if), "iucv");
+ if (!pr_iucv) {
+ printk(KERN_WARNING "iucv_if lookup failed\n");
+ err = -EPROTONOSUPPORT;
goto out;
+ }
+
err = proto_register(&iucv_proto, 0);
if (err)
- goto out_iucv;
+ goto out;
err = sock_register(&iucv_sock_family_ops);
if (err)
goto out_proto;
- /* establish dummy device */
- err = driver_register(&af_iucv_driver);
+
+ err = afiucv_iucv_init();
if (err)
goto out_sock;
- af_iucv_dev = kzalloc(sizeof(struct device), GFP_KERNEL);
- if (!af_iucv_dev) {
- err = -ENOMEM;
- goto out_driver;
- }
- dev_set_name(af_iucv_dev, "af_iucv");
- af_iucv_dev->bus = &iucv_bus;
- af_iucv_dev->parent = iucv_root;
- af_iucv_dev->release = (void (*)(struct device *))kfree;
- af_iucv_dev->driver = &af_iucv_driver;
- err = device_register(af_iucv_dev);
- if (err)
- goto out_driver;
return 0;
-out_driver:
- driver_unregister(&af_iucv_driver);
out_sock:
sock_unregister(PF_IUCV);
out_proto:
proto_unregister(&iucv_proto);
-out_iucv:
- iucv_unregister(&af_iucv_handler, 0);
out:
+ if (pr_iucv)
+ symbol_put(iucv_if);
return err;
}
@@ -1780,9 +1808,10 @@ static void __exit afiucv_exit(void)
{
device_unregister(af_iucv_dev);
driver_unregister(&af_iucv_driver);
+ pr_iucv->iucv_unregister(&af_iucv_handler, 0);
+ symbol_put(iucv_if);
sock_unregister(PF_IUCV);
proto_unregister(&iucv_proto);
- iucv_unregister(&af_iucv_handler, 0);
}
module_init(afiucv_init);
^ permalink raw reply
* [patch 02/13] [PATCH] qeth: do not apply priority queuing to HiperSockets
From: frank.blaschka @ 2011-08-08 11:33 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, Ursula Braun
In-Reply-To: <20110808113346.516293249@de.ibm.com>
[-- Attachment #1: qeth-nr-outq.patch --]
[-- Type: text/plain, Size: 2489 bytes --]
From: Ursula Braun <ursula.braun@de.ibm.com>
OSA cards can be configured to support 1 or 4 output queues. This
does not apply to HiperSockets. This patch limits determination of
the configured number of output queues to OSA cards only, but excludes
HiperSockets.
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
qeth_core_main.c | 40 +++++++++++++++++++++-------------------
1 file changed, 21 insertions(+), 19 deletions(-)
--- linux-2.5/drivers/s390/net/qeth_core_main.c 22 Jul 2011 08:38:02 -0000 1.119
+++ linux-2.5/drivers/s390/net/qeth_core_main.c 4 Aug 2011 07:02:55 -0000 1.120
@@ -995,27 +995,29 @@ static void qeth_get_channel_path_desc(s
ccwdev = card->data.ccwdev;
chp_dsc = (struct channelPath_dsc *)ccw_device_get_chp_desc(ccwdev, 0);
if (chp_dsc != NULL) {
- /* CHPP field bit 6 == 1 -> single queue */
- if ((chp_dsc->chpp & 0x02) == 0x02) {
- if ((atomic_read(&card->qdio.state) !=
- QETH_QDIO_UNINITIALIZED) &&
- (card->qdio.no_out_queues == 4))
- /* change from 4 to 1 outbound queues */
- qeth_free_qdio_buffers(card);
- card->qdio.no_out_queues = 1;
- if (card->qdio.default_out_queue != 0)
- dev_info(&card->gdev->dev,
+ if (card->info.type != QETH_CARD_TYPE_IQD) {
+ /* CHPP field bit 6 == 1 -> single queue */
+ if ((chp_dsc->chpp & 0x02) == 0x02) {
+ if ((atomic_read(&card->qdio.state) !=
+ QETH_QDIO_UNINITIALIZED) &&
+ (card->qdio.no_out_queues == 4))
+ /* change from 4 to 1 outbound queues */
+ qeth_free_qdio_buffers(card);
+ card->qdio.no_out_queues = 1;
+ if (card->qdio.default_out_queue != 0)
+ dev_info(&card->gdev->dev,
"Priority Queueing not supported\n");
- card->qdio.default_out_queue = 0;
- } else {
- if ((atomic_read(&card->qdio.state) !=
- QETH_QDIO_UNINITIALIZED) &&
- (card->qdio.no_out_queues == 1)) {
- /* change from 1 to 4 outbound queues */
- qeth_free_qdio_buffers(card);
- card->qdio.default_out_queue = 2;
+ card->qdio.default_out_queue = 0;
+ } else {
+ if ((atomic_read(&card->qdio.state) !=
+ QETH_QDIO_UNINITIALIZED) &&
+ (card->qdio.no_out_queues == 1)) {
+ /* change from 1 to 4 outbound queues */
+ qeth_free_qdio_buffers(card);
+ card->qdio.default_out_queue = 2;
+ }
+ card->qdio.no_out_queues = 4;
}
- card->qdio.no_out_queues = 4;
}
card->info.func_level = 0x4100 + chp_dsc->desc;
kfree(chp_dsc);
^ 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