* [PATCH] sky2: suspend / resume fix
From: Stephen Hemminger @ 2006-06-13 4:45 UTC (permalink / raw)
To: Linus Torvalds; +Cc: Jeff Garzik, netdev
In-Reply-To: <Pine.LNX.4.64.0606121537420.5498@g5.osdl.org>
The resume bug was cause not by an early interrupt but because
the idle timeout was not being stopped on suspend. Also disable hardware
IRQ's on suspend. Will need to revisit this when wake-on-lan is added.
Patch against 2.6.17-rc latest which includes Linus's attempt at
fixing this.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -2164,6 +2164,13 @@ static void sky2_descriptor_error(struct
/* If idle then force a fake soft NAPI poll once a second
* to work around cases where sharing an edge triggered interrupt.
*/
+static inline void sky2_idle_start(struct sky2_hw *hw)
+{
+ if (idle_timeout > 0)
+ mod_timer(&hw->idle_timer,
+ jiffies + msecs_to_jiffies(idle_timeout));
+}
+
static void sky2_idle(unsigned long arg)
{
struct sky2_hw *hw = (struct sky2_hw *) arg;
@@ -2183,9 +2190,6 @@ static int sky2_poll(struct net_device *
int work_done = 0;
u32 status = sky2_read32(hw, B0_Y2_SP_EISR);
- if (!~status)
- return 0;
-
if (status & Y2_IS_HW_ERR)
sky2_hw_intr(hw);
@@ -3353,9 +3357,7 @@ static int __devinit sky2_probe(struct p
sky2_write32(hw, B0_IMSK, Y2_IS_BASE);
setup_timer(&hw->idle_timer, sky2_idle, (unsigned long) hw);
- if (idle_timeout > 0)
- mod_timer(&hw->idle_timer,
- jiffies + msecs_to_jiffies(idle_timeout));
+ sky2_idle_start(hw);
pci_set_drvdata(pdev, hw);
@@ -3427,9 +3429,14 @@ static void __devexit sky2_remove(struct
static int sky2_suspend(struct pci_dev *pdev, pm_message_t state)
{
struct sky2_hw *hw = pci_get_drvdata(pdev);
- int i;
+ int i, err;
+ u32 imask;
+
+ pci_save_state(pdev);
- for (i = 0; i < 2; i++) {
+ del_timer_sync(&hw->idle_timer);
+
+ for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
if (dev) {
@@ -3441,8 +3448,17 @@ static int sky2_suspend(struct pci_dev *
}
}
- pci_save_state(pdev);
- return sky2_set_power_state(hw, pci_choose_state(pdev, state));
+ imask = sky2_read32(hw, B0_IMSK);
+ sky2_write32(hw, B0_IMSK, 0);
+ synchronize_irq(hw->pdev->irq);
+
+ err = sky2_set_power_state(hw, pci_choose_state(pdev, state));
+ if (err) {
+ sky2_write32(hw, B0_IMSK, imask);
+ sky2_idle_start(hw);
+ }
+
+ return err;
}
static int sky2_resume(struct pci_dev *pdev)
@@ -3460,7 +3476,7 @@ static int sky2_resume(struct pci_dev *p
if (err)
goto out;
- for (i = 0; i < 2; i++) {
+ for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
if (dev && netif_running(dev)) {
netif_device_attach(dev);
@@ -3473,7 +3489,10 @@ static int sky2_resume(struct pci_dev *p
}
}
}
-out:
+
+ sky2_idle_start(hw);
+
+ out:
return err;
}
#endif
^ permalink raw reply
* Re: Using netconsole for debugging suspend/resume
From: David Miller @ 2006-06-13 4:49 UTC (permalink / raw)
To: ak; +Cc: jeremy, lkml, mpm, linux-kernel, netdev
In-Reply-To: <200606130547.49624.ak@suse.de>
From: Andi Kleen <ak@suse.de>
Date: Tue, 13 Jun 2006 05:47:49 +0200
> I've been playing with the idea of writing "early1394" that just
> turns the DMA controller on as early as possible similar to earlyprintk
> on the target. Then it would be possible to use it for early
> debugging too. But so far it's not done yet.
Does this raw1394 thing with firescope just assume DMA address ==
physical address? How would it work to access all of physical
memory properly on IOMMU platforms?
^ permalink raw reply
* Re: [PATCH 2.6.18 5.2/6 resend] bnx2: Use compressed firmware 2/6
From: David Miller @ 2006-06-13 4:54 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150163780.26368.26.camel@rh4>
Michael, you just sent out what appears to be 6 copies of the
bnx2_fw.h patch :-) I think your patch sending script messed
something up.
^ permalink raw reply
* Re: Using netconsole for debugging suspend/resume
From: Andi Kleen @ 2006-06-13 4:54 UTC (permalink / raw)
To: David Miller; +Cc: jeremy, lkml, mpm, linux-kernel, netdev
In-Reply-To: <20060612.214948.124554804.davem@davemloft.net>
On Tuesday 13 June 2006 06:49, David Miller wrote:
> From: Andi Kleen <ak@suse.de>
> Date: Tue, 13 Jun 2006 05:47:49 +0200
>
> > I've been playing with the idea of writing "early1394" that just
> > turns the DMA controller on as early as possible similar to earlyprintk
> > on the target. Then it would be possible to use it for early
> > debugging too. But so far it's not done yet.
>
> Does this raw1394 thing with firescope just assume DMA address ==
> physical address?
Yes.
> How would it work to access all of physical
> memory properly on IOMMU platforms?
It assumes you don't have an IOMMU - relies on all memory
being accessible by ohci1394. On x86-64 it can't access > 4GB
also, but that's normally ok because the kernel log buffer
is below that.
I guess if you use 1394 with remote DMA for other protocols (like
video etc.) there must be some way for the subsystem to map
the memory even on IOMMU systems. I admit I haven't dived that
deeply into the 1394 subsystem so I don't know how that works.
-Andi
^ permalink raw reply
* Re: [PATCH] sky2: suspend / resume fix
From: Jeff Garzik @ 2006-06-13 5:00 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: Linus Torvalds, netdev
In-Reply-To: <20060613134552.331377c3@localhost.localdomain>
Stephen Hemminger wrote:
> @@ -2183,9 +2190,6 @@ static int sky2_poll(struct net_device *
> int work_done = 0;
> u32 status = sky2_read32(hw, B0_Y2_SP_EISR);
>
> - if (!~status)
> - return 0;
> -
> if (status & Y2_IS_HW_ERR)
> sky2_hw_intr(hw);
>
It's probably better to do "goto end_of_function" than simply return,
but I disagree that this check should be removed.
As long as you are reading values from the hardware, you should include
this 0xffffffff check. It's certainly possible that the controller
could have been unplugged from the time the interrupt handler last ran,
and the time that NAPI poll runs.
> @@ -3427,9 +3429,14 @@ static void __devexit sky2_remove(struct
> static int sky2_suspend(struct pci_dev *pdev, pm_message_t state)
> {
> struct sky2_hw *hw = pci_get_drvdata(pdev);
> - int i;
> + int i, err;
> + u32 imask;
> +
> + pci_save_state(pdev);
>
> - for (i = 0; i < 2; i++) {
> + del_timer_sync(&hw->idle_timer);
It would seem smarter to put the timer removal before pci-save-state.
> + for (i = 0; i < hw->ports; i++) {
> struct net_device *dev = hw->dev[i];
>
> if (dev) {
> @@ -3441,8 +3448,17 @@ static int sky2_suspend(struct pci_dev *
> }
> }
>
> - pci_save_state(pdev);
> - return sky2_set_power_state(hw, pci_choose_state(pdev, state));
> + imask = sky2_read32(hw, B0_IMSK);
> + sky2_write32(hw, B0_IMSK, 0);
> + synchronize_irq(hw->pdev->irq);
This seems "obviously racy" to me.
> + err = sky2_set_power_state(hw, pci_choose_state(pdev, state));
> + if (err) {
> + sky2_write32(hw, B0_IMSK, imask);
> + sky2_idle_start(hw);
> + }
> +
> + return err;
> }
>
> static int sky2_resume(struct pci_dev *pdev)
> @@ -3460,7 +3476,7 @@ static int sky2_resume(struct pci_dev *p
> if (err)
> goto out;
>
> - for (i = 0; i < 2; i++) {
> + for (i = 0; i < hw->ports; i++) {
> struct net_device *dev = hw->dev[i];
> if (dev && netif_running(dev)) {
> netif_device_attach(dev);
> @@ -3473,7 +3489,10 @@ static int sky2_resume(struct pci_dev *p
> }
> }
> }
> -out:
> +
> + sky2_idle_start(hw);
> +
> + out:
> return err;
> }
> #endif
>
>
^ permalink raw reply
* Re: Using netconsole for debugging suspend/resume
From: David Miller @ 2006-06-13 5:03 UTC (permalink / raw)
To: ak; +Cc: jeremy, lkml, mpm, linux-kernel, netdev
In-Reply-To: <200606130654.14477.ak@suse.de>
From: Andi Kleen <ak@suse.de>
Date: Tue, 13 Jun 2006 06:54:14 +0200
> I guess if you use 1394 with remote DMA for other protocols (like
> video etc.) there must be some way for the subsystem to map
> the memory even on IOMMU systems. I admit I haven't dived that
> deeply into the 1394 subsystem so I don't know how that works.
Video-1394 has it's own driver, which does a consistent DMA
allocation, and then maps that into userspace using remap_pfn_range().
Entirely portable.
Strangely I don't even see any bus_to_virt() etc. calls in
the raw1394 driver, just these ptr2int() things...
^ permalink raw reply
* Re: [RFC/PATCH 1/2] in-kernel sockets API
From: Stephen Hemminger @ 2006-06-13 5:07 UTC (permalink / raw)
To: Sridhar Samudrala; +Cc: netdev, linux-kernel
In-Reply-To: <1150156562.19929.32.camel@w-sridhar2.beaverton.ibm.com>
On Mon, 12 Jun 2006 16:56:01 -0700
Sridhar Samudrala <sri@us.ibm.com> wrote:
> This patch makes it convenient to use the sockets API by the in-kernel
> users like sunrpc, cifs & ocfs2 etc and any future users.
> Currently they get to this API by directly accesing the function pointers
> in the sock structure.
>
> Most of these functions are pretty simple and can be made inline and moved
> to linux/net.h.
...
> @@ -2176,3 +2279,13 @@ EXPORT_SYMBOL(sock_wake_async);
> EXPORT_SYMBOL(sockfd_lookup);
> EXPORT_SYMBOL(kernel_sendmsg);
> EXPORT_SYMBOL(kernel_recvmsg);
> +EXPORT_SYMBOL(kernel_bind);
> +EXPORT_SYMBOL(kernel_listen);
> +EXPORT_SYMBOL(kernel_accept);
> +EXPORT_SYMBOL(kernel_connect);
> +EXPORT_SYMBOL(kernel_getsockname);
> +EXPORT_SYMBOL(kernel_getpeername);
> +EXPORT_SYMBOL(kernel_getsockopt);
> +EXPORT_SYMBOL(kernel_setsockopt);
> +EXPORT_SYMBOL(kernel_sendpage);
> +EXPORT_SYMBOL(kernel_ioctl);
Don't we want to restrict this to GPL code with EXPORT_SYMBOL_GPL?
^ permalink raw reply
* Re: [PATCH 2.6.18 1/6] bnx2: Add an rx drop counter
From: David Miller @ 2006-06-13 5:16 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150157824.26173.6.camel@rh4>
From: "Michael Chan" <mchan@broadcom.com>
Date: Mon, 12 Jun 2006 17:17:03 -0700
> Add a counter for packets dropped by firmware.
>
> Signed-off-by: Michael Chan <mchan@broadcom.com>
Applied to net-2.6.18, thanks.
^ permalink raw reply
* Re: [PATCH 2.6.18 2/6] bnx2: Allow WoL settings on new 5708 chips
From: David Miller @ 2006-06-13 5:16 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150157828.26173.7.camel@rh4>
From: "Michael Chan" <mchan@broadcom.com>
Date: Mon, 12 Jun 2006 17:17:08 -0700
> Allow WOL settings on 5708 B2 and newer chips that have the problem
> fixed.
>
> Signed-off-by: Michael Chan <mchan@broadcom.com>
Applied to net-2.6.18, thanks.
^ permalink raw reply
* Re: [PATCH 2.6.18 3/6] bnx2: Use CPU native page size
From: David Miller @ 2006-06-13 5:20 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150157833.26173.8.camel@rh4>
From: "Michael Chan" <mchan@broadcom.com>
Date: Mon, 12 Jun 2006 17:17:13 -0700
> Use CPU native page size to determine various ring sizes. This allows
> order-0 memory allocations on all systems.
>
> Signed-off-by: Michael Chan <mchan@broadcom.com>
Are you sure you want to do this when the base page size is up to 64K
on some IA64/PowerPC64/Sparc64 configurations?
Maybe you can use a limit with some ifdef tests, something like:
#if PAGE_SHIFT > 13
#define BCM_PAGE_BITS 13
#else
#define BCM_PAGE_BITS PAGE_SHIFT
#endif
which would limit it to 8K.
^ permalink raw reply
* Re: [PATCH 2.6.18 4/6 revised] bnx2: Add firmware decompression
From: David Miller @ 2006-06-13 5:21 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150162949.26368.5.camel@rh4>
From: "Michael Chan" <mchan@broadcom.com>
Date: Mon, 12 Jun 2006 18:42:29 -0700
> On Mon, 2006-06-12 at 17:38 -0700, David Miller wrote:
> > From: "Michael Chan" <mchan@broadcom.com>
> > Date: Mon, 12 Jun 2006 17:17:18 -0700
> >
> > > Add functions to decompress firmware before loading to the internal
> > > CPUs. Compressing the firmware reduces the driver size significantly.
> > >
> > > Signed-off-by: Michael Chan <mchan@broadcom.com>
> > ...
> > > +#define FNAME 0x8
> > > + if (zbuf[3] & FNAME)
> > > + while (zbuf[n++] != 0);
> >
> > This can potentially loop past the end of zbuf[], so maybe
> > put a sanity check against 'len' and return non-zero if we
> > traverse past the end?
>
> Add functions to decompress firmware before loading to the internal
> CPUs. Compressing the firmware reduces the driver size significantly.
>
> Added file name length sanity check in the gzip header to prevent
> going past the end of buffer [suggested by DaveM].
>
> Signed-off-by: Michael Chan <mchan@broadcom.com>
This looks great, applied.
Thanks a lot.
^ permalink raw reply
* Re: [PATCH 2.6.18 5/6] bnx2: Use compressed firmware
From: David Miller @ 2006-06-13 5:22 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150158970.26368.1.camel@rh4>
Applied the firmware update patch, thanks a lot.
^ permalink raw reply
* Re: [PATCH 2.6.18 6/6] bnx2: Update version and reldate
From: David Miller @ 2006-06-13 5:23 UTC (permalink / raw)
To: mchan; +Cc: netdev
In-Reply-To: <1150157842.26173.10.camel@rh4>
From: "Michael Chan" <mchan@broadcom.com>
Date: Mon, 12 Jun 2006 17:17:22 -0700
> Update driver version to 1.4.42.
>
> Signed-off-by: Michael Chan <mchan@broadcom.com>
I'll apply this once we decide what is going to happen
with the BCM_PAGE_BITS change.
^ permalink raw reply
* Re: [PATCH] [2.6.17-rc6] Section mismatch in drivers/net/ne.o during modpost
From: Keith Owens @ 2006-06-13 5:35 UTC (permalink / raw)
To: Mikael Pettersson; +Cc: rdunlap, akpm, jgarzik, linux-kernel, netdev
In-Reply-To: <200606110051.k5B0pLBI010621@harpo.it.uu.se>
Mikael Pettersson (on Sun, 11 Jun 2006 02:51:21 +0200 (MEST)) wrote:
>On Sat, 10 Jun 2006 12:13:35 -0700, Randy.Dunlap wrote:
>>On Sat, 10 Jun 2006 14:11:42 +0200 (MEST) Mikael Pettersson wrote:
>>
>>> While compiling 2.6.17-rc6 for a 486 with an NE2000 ISA ethernet card, I got:
>>>
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.data:isapnp_clone_list from .text between 'init_module' (at offset 0x158) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.data:isapnp_clone_list from .text between 'init_module' (at offset 0x176) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.data:isapnp_clone_list from .text between 'init_module' (at offset 0x183) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.data:isapnp_clone_list from .text between 'init_module' (at offset 0x1ea) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.data:isapnp_clone_list from .text between 'init_module' (at offset 0x251) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.text: from .text between 'init_module' (at offset 0x266) and 'ne_block_input'
>>> WARNING: drivers/net/ne.o - Section mismatch: reference to .init.text: from .text between 'init_module' (at offset 0x29b) and 'ne_block_input'
>>>
>>> Not sure how serious this is; the driver seems to work fine later on.
>>
>>Doesn't look serious. init_module() is not __init, but it calls
>>some __init functions and touches some __initdata.
>>
>>BTW, I would be happy to see some consistent results from modpost
>>section checking. I don't see all of these warnings (I see only 1)
>>when using gcc 3.3.6. What gcc version are you using?
>>Does that matter? (not directed at anyone in particular)
>
>The messages above are from when I used gcc-4.1.1.
>With gcc-3.2.3 I only see a single warning.
Probably over enthusiastic gcc inlining. gcc 4 will inline functions
that are not declared as inline. That is not so bad, except that some
versions of gcc will ignore a mismatch in function attributes and
inline a __init function into normal text, generating additional
section mismatches. For a specific example, see
http://marc.theaimsgroup.com/?l=linux-kernel&m=113824309203482&w=2
^ permalink raw reply
* IPsec and EAGAIN
From: James Ring @ 2006-06-13 5:51 UTC (permalink / raw)
To: netdev
Hi all,
I have recently run into the issue where connect() returns -EAGAIN
if the remote host is on the other end of an IPsec tunnel for which no
SAD exists yet.
I have read a few threads on the topic, and it seems that the idea is to
implement a scheme similar to ARP for queueing outgoing packets until the
SAD has been negotiated by the IKE process.
I have a couple of questions: are there any patches out which solve this
problem?
Also, what is the purpose of the code in the xfrm_lookup function
(in net/xfrm_policy.c) when `flags' is true? I have applied a patch
similar to
http://sourceforge.net/mailarchive/message.php?msg_id=8621194
which gives desirable behaviour (ie. telnet appears to pause while the
racoon negotiation is performed, after which it connects just fine, no
EAGAIN).
Please CC me to any responses, I am not subscribed to the list.
Thanks for your time,
James
--
James Ring
^ permalink raw reply
* [PATCH/2.6.17-rc6] net: Add tsi108 Ethernet device driver suppo rt
From: Zang Roy-r61911 @ 2006-06-13 7:07 UTC (permalink / raw)
To: jgarzik
Cc: netdev, Yang Xin-Xin-r48390, linux-kernel, Alexandre.Bounine,
linuxppc-dev list
This patch adds a net device driver and configuration options for
Tundra Semiconductor Tsi108 integrated dual port Gigabit
Ethernet controller
Signed-off-by: Alexandre Bounine <alexandreb@tundra.com>
Signed-off-by: Roy Zang <tie-fei.zang@freescale.com>
---
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index bdaaad8..8a4ef29 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2177,6 +2177,14 @@ config SPIDER_NET
This driver supports the Gigabit Ethernet chips present on the
Cell Processor-Based Blades from IBM.
+config TSI108_ETH
+ tristate "Tundra TSI108 gigabit Ethernet support"
+ depends on TSI108_BRIDGE
+ help
+ This driver supports Tundra TSI108 gigabit Ethernet ports.
+ To compile this driver as a module, choose M here: the module
+ will be called tsi108_eth.
+
config GIANFAR
tristate "Gianfar Ethernet"
depends on 85xx || 83xx
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index b90468a..5f90b30 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -109,6 +109,7 @@ obj-$(CONFIG_B44) += b44.o
obj-$(CONFIG_FORCEDETH) += forcedeth.o
obj-$(CONFIG_NE_H8300) += ne-h8300.o 8390.o
+obj-$(CONFIG_TSI108_ETH) += tsi108_eth.o
obj-$(CONFIG_MV643XX_ETH) += mv643xx_eth.o
obj-$(CONFIG_PPP) += ppp_generic.o slhc.o
diff --git a/drivers/net/tsi108_eth.c b/drivers/net/tsi108_eth.c
new file mode 100644
index 0000000..77d5a11
--- /dev/null
+++ b/drivers/net/tsi108_eth.c
@@ -0,0 +1,1755 @@
+/*******************************************************************************
+
+ Copyright(c) 2005 Tundra Semiconductor Corporation.
+
+ This program is free software; you can redistribute it and/or modify it
+ under the terms of the GNU General Public License as published by the Free
+ Software Foundation; either version 2 of the License, or (at your option)
+ any later version.
+
+ This program is distributed in the hope that it will be useful, but WITHOUT
+ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ more details.
+
+ You should have received a copy of the GNU General Public License along with
+ this program; if not, write to the Free Software Foundation, Inc., 59
+ Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+
+*******************************************************************************/
+
+/* This driver is based on the driver code originally developed
+ * for the Intel IOC80314 (ForestLake) Gigabit Ethernet by
+ * scott.wood@timesys.com * Copyright (C) 2003 TimeSys Corporation
+ *
+ * Currently changes from original version are:
+ * - portig to Tsi108-based platform and kernel 2.6 (kong.lai@tundra.com)
+ * - modifications to handle two ports independently and support for
+ * additional PHY devices (alexandre.bounine@tundra.com)
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/net.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/skbuff.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/crc32.h>
+#include <linux/mii.h>
+#include <linux/device.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <linux/pci.h>
+#include <linux/rtnetlink.h>
+#include <linux/timer.h>
+#include <linux/platform_device.h>
+
+#include <asm/tsi108_irq.h>
+#include <asm/tsi108.h>
+#include "tsi108_eth.h"
+
+#undef DEBUG
+#ifdef DEBUG
+#define DBG(fmt...) do { printk(fmt); } while(0)
+#else
+#define DBG(fmt...) do { } while(0)
+#endif
+
+typedef struct net_device net_device;
+typedef struct sk_buff sk_buff;
+
+#define MII_READ_DELAY 10000 /* max link wait time in msec */
+
+#define TSI108_RXRING_LEN 256
+
+/* NOTE: The driver currently does not support receiving packets
+ * larger than the buffer size, so don't decrease this (unless you
+ * want to add such support).
+ */
+#define TSI108_RXBUF_SIZE 1536
+
+#define TSI108_TXRING_LEN 256
+
+#define TSI108_TX_INT_FREQ 64
+
+/* Check the phy status every half a second. */
+#define CHECK_PHY_INTERVAL (HZ/2)
+
+static int tsi108_init_one(struct platform_device *pdev);
+static int tsi108_ether_remove(struct platform_device *pdev);
+
+typedef struct {
+ volatile u32 regs; /* Base of normal regs */
+ volatile u32 phyregs; /* Base of register bank used for PHY access */
+ int phy; /* Index of PHY for this interface */
+ int irq_num;
+
+ struct timer_list timer;/* Timer that triggers the check phy function */
+ int rxtail; /* Next entry in rxring to read */
+ int rxhead; /* Next entry in rxring to give a new buffer */
+ int rxfree; /* Number of free, allocated RX buffers */
+
+ int rxpending; /* Non-zero if there are still descriptors
+ * to be processed from a previous descriptor
+ * interrupt condition that has been cleared */
+
+ int txtail; /* Next TX descriptor to check status on */
+ int txhead; /* Next TX descriptor to use */
+
+ /* Number of free TX descriptors. This could be calculated from
+ * rxhead and rxtail if one descriptor were left unused to disambiguate
+ * full and empty conditions, but it's simpler to just keep track
+ * explicitly. */
+
+ int txfree;
+
+ int phy_ok; /* The PHY is currently powered on. */
+
+ /* PHY status (duplex is 1 for half, 2 for full,
+ * so that the default 0 indicates that neither has
+ * yet been configured). */
+
+ int link_up;
+ int speed;
+ int duplex;
+
+ tx_desc *txring;
+ rx_desc *rxring;
+ sk_buff *txskbs[TSI108_TXRING_LEN];
+ sk_buff *rxskbs[TSI108_RXRING_LEN];
+
+ dma_addr_t txdma, rxdma;
+
+ /* txlock nests in misclock and phy_lock */
+
+ spinlock_t txlock, misclock;
+
+ /* stats is used to hold the upper bits of each hardware counter,
+ * and tmpstats is used to hold the full values for returning
+ * to the caller of get_stats(). They must be separate in case
+ * an overflow interrupt occurs before the stats are consumed.
+ */
+
+ struct net_device_stats stats;
+ struct net_device_stats tmpstats;
+
+ /* These stats are kept separate in hardware, thus require individual
+ * fields for handling carry. They are combined in get_stats.
+ */
+
+ unsigned long rx_fcs; /* Add to rx_frame_errors */
+ unsigned long rx_short_fcs; /* Add to rx_frame_errors */
+ unsigned long rx_long_fcs; /* Add to rx_frame_errors */
+ unsigned long rx_underruns; /* Add to rx_length_errors */
+ unsigned long rx_overruns; /* Add to rx_length_errors */
+
+ unsigned long tx_coll_abort; /* Add to tx_aborted_errors/collisions */
+ unsigned long tx_pause_drop; /* Add to tx_aborted_errors */
+
+ unsigned long mc_hash[16];
+} tsi108_prv_data;
+
+/* Structure for a device driver */
+
+static struct platform_driver tsi_eth_driver = {
+ .probe = tsi108_init_one,
+ .remove = tsi108_ether_remove,
+ .driver = {
+ .name = "tsi-ethernet",
+ },
+};
+
+static void tsi108_timed_checker(unsigned long dev_ptr);
+
+static void dump_eth_one(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ printk("Dumping %s...\n", dev->name);
+ printk("intstat %x intmask %x phy_ok %d"
+ " link %d speed %d duplex %d\n",
+ TSI108_ETH_READ_REG(TSI108_EC_INTSTAT),
+ TSI108_ETH_READ_REG(TSI108_EC_INTMASK), data->phy_ok,
+ data->link_up, data->speed, data->duplex);
+
+ printk("TX: head %d, tail %d, free %d, stat %x, estat %x, err %x\n",
+ data->txhead, data->txtail, data->txfree,
+ TSI108_ETH_READ_REG(TSI108_EC_TXSTAT),
+ TSI108_ETH_READ_REG(TSI108_EC_TXESTAT),
+ TSI108_ETH_READ_REG(TSI108_EC_TXERR));
+
+ printk("RX: head %d, tail %d, free %d, stat %x,"
+ " estat %x, err %x, pending %d\n\n",
+ data->rxhead, data->rxtail, data->rxfree,
+ TSI108_ETH_READ_REG(TSI108_EC_RXSTAT),
+ TSI108_ETH_READ_REG(TSI108_EC_RXESTAT),
+ TSI108_ETH_READ_REG(TSI108_EC_RXERR), data->rxpending);
+}
+
+/* Synchronization is needed between the thread and up/down events.
+ * Note that the PHY is accessed through the same registers for both
+ * interfaces, so this can't be made interface-specific.
+ */
+
+static spinlock_t phy_lock = SPIN_LOCK_UNLOCKED;
+
+static inline u16 tsi108_read_mii(tsi108_prv_data * data, int reg, int *status)
+{
+ int i;
+ u16 ret;
+
+ TSI108_ETH_WRITE_PHYREG(TSI108_MAC_MII_ADDR,
+ (data->phy << TSI108_MAC_MII_ADDR_PHY) |
+ (reg << TSI108_MAC_MII_ADDR_REG));
+ mb();
+ TSI108_ETH_WRITE_PHYREG(TSI108_MAC_MII_CMD, 0);
+ mb();
+ TSI108_ETH_WRITE_PHYREG(TSI108_MAC_MII_CMD, TSI108_MAC_MII_CMD_READ);
+ mb();
+ for (i = 0; i < 100; i++) {
+ if (!(TSI108_ETH_READ_PHYREG(TSI108_MAC_MII_IND) &
+ (TSI108_MAC_MII_IND_NOTVALID | TSI108_MAC_MII_IND_BUSY)))
+ break;
+ udelay(10);
+ }
+
+ if (i == 100) {
+ if (status)
+ *status = -EBUSY;
+
+ ret = 0xffff;
+ } else {
+ if (status)
+ *status = 0;
+
+ ret = TSI108_ETH_READ_PHYREG(TSI108_MAC_MII_DATAIN);
+ }
+
+ return ret;
+}
+
+static inline void tsi108_write_mii(tsi108_prv_data * data, int reg, u16 val)
+{
+ TSI108_ETH_WRITE_PHYREG(TSI108_MAC_MII_ADDR,
+ (data->phy << TSI108_MAC_MII_ADDR_PHY) |
+ (reg << TSI108_MAC_MII_ADDR_REG));
+ mb();
+ TSI108_ETH_WRITE_PHYREG(TSI108_MAC_MII_DATAOUT, val);
+ mb();
+ while (TSI108_ETH_READ_PHYREG(TSI108_MAC_MII_IND) &
+ TSI108_MAC_MII_IND_BUSY) ;
+}
+
+static inline void tsi108_write_tbi(tsi108_prv_data * data, int reg, u16 val)
+{
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_MII_ADDR,
+ (0x1e << TSI108_MAC_MII_ADDR_PHY)
+ | (reg << TSI108_MAC_MII_ADDR_REG));
+ mb();
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_MII_DATAOUT, val);
+ mb();
+ while (TSI108_ETH_READ_REG(TSI108_MAC_MII_IND) &
+ TSI108_MAC_MII_IND_BUSY) ;
+}
+
+static void tsi108_check_phy(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u16 sumstat;
+ u32 mac_cfg2_reg, portctrl_reg;
+ u32 fdx_flag = 0, reg_update = 0;
+
+ /* Do a dummy read, as for some reason the first read
+ * after a link becomes up returns link down, even if
+ * it's been a while since the link came up.
+ */
+
+ spin_lock(&phy_lock);
+
+ if (!data->phy_ok)
+ goto out;
+
+ tsi108_read_mii(data, PHY_STAT, NULL);
+
+ if (!(tsi108_read_mii(data, PHY_STAT, NULL) & PHY_STAT_LINKUP)) {
+ if (data->link_up == 1) {
+ netif_stop_queue(dev);
+ data->link_up = 0;
+ printk(KERN_NOTICE "%s : link is down\n", dev->name);
+ netif_carrier_off(dev);
+ }
+
+ goto out;
+ }
+
+ {
+ mac_cfg2_reg = TSI108_ETH_READ_REG(TSI108_MAC_CFG2);
+ portctrl_reg = TSI108_ETH_READ_REG(TSI108_EC_PORTCTRL);
+
+ sumstat = tsi108_read_mii(data, PHY_SUM_STAT, NULL);
+
+ switch (sumstat & PHY_SUM_STAT_SPEED_MASK) {
+ case PHY_SUM_STAT_1000T_FD:
+ fdx_flag++;
+ case PHY_SUM_STAT_1000T_HD:
+ if (data->speed != 1000) {
+ mac_cfg2_reg &= ~TSI108_MAC_CFG2_IFACE_MASK;
+ mac_cfg2_reg |= TSI108_MAC_CFG2_GIG;
+ portctrl_reg &= ~TSI108_EC_PORTCTRL_NOGIG;
+ data->speed = 1000;
+ reg_update++;
+ }
+ break;
+ case PHY_SUM_STAT_100TX_FD:
+ fdx_flag++;
+ case PHY_SUM_STAT_100TX_HD:
+ if (data->speed != 100) {
+ mac_cfg2_reg &= ~TSI108_MAC_CFG2_IFACE_MASK;
+ mac_cfg2_reg |= TSI108_MAC_CFG2_NOGIG;
+ portctrl_reg |= TSI108_EC_PORTCTRL_NOGIG;
+ data->speed = 100;
+ reg_update++;
+ }
+ break;
+
+ case PHY_SUM_STAT_10T_FD:
+ fdx_flag++;
+ case PHY_SUM_STAT_10T_HD:
+ if (data->speed != 10) {
+ mac_cfg2_reg &= ~TSI108_MAC_CFG2_IFACE_MASK;
+ mac_cfg2_reg |= TSI108_MAC_CFG2_NOGIG;
+ portctrl_reg |= TSI108_EC_PORTCTRL_NOGIG;
+ data->speed = 10;
+ reg_update++;
+ }
+ break;
+
+ default:
+ if (net_ratelimit())
+ printk(KERN_ERR "PHY reported invalid speed,"
+ KERN_ERR " summary status %x\n",
+ sumstat);
+ goto out;
+ }
+
+ if (fdx_flag || (sumstat & PHY_SUM_STAT_FULLDUPLEX)) {
+ if (data->duplex != 2) {
+ mac_cfg2_reg |= TSI108_MAC_CFG2_FULLDUPLEX;
+ portctrl_reg &= ~TSI108_EC_PORTCTRL_HALFDUPLEX;
+ data->duplex = 2;
+ reg_update++;
+ }
+ } else {
+ if (data->duplex != 1) {
+ mac_cfg2_reg &= ~TSI108_MAC_CFG2_FULLDUPLEX;
+ portctrl_reg |= TSI108_EC_PORTCTRL_HALFDUPLEX;
+ data->duplex = 1;
+ reg_update++;
+ }
+ }
+
+ if (reg_update) {
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG2, mac_cfg2_reg);
+ mb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_PORTCTRL, portctrl_reg);
+ mb();
+ }
+
+ }
+
+ if (data->link_up == 0) {
+ /* The manual says it can take 3-4 usecs for the speed change
+ * to take effect.
+ */
+ udelay(5);
+
+ spin_lock(&data->txlock);
+ if (netif_queue_stopped(dev)
+ && is_valid_ether_addr(dev->dev_addr) && data->txfree)
+ netif_wake_queue(dev);
+
+ data->link_up = 1;
+ spin_unlock(&data->txlock);
+ printk("%s : link is up: %dMb %s-duplex\n",
+ dev->name, data->speed,
+ (data->duplex == 2) ? "full" : "half");
+ netif_carrier_on(dev);
+ }
+
+ out:
+ spin_unlock(&phy_lock);
+}
+
+static inline void
+tsi108_stat_carry_one(int carry, int carry_bit, int carry_shift,
+ unsigned long *upper)
+{
+ if (carry & carry_bit)
+ *upper += carry_shift;
+}
+
+static void tsi108_stat_carry(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 carry1, carry2;
+
+ spin_lock_irq(&data->misclock);
+
+ carry1 = TSI108_ETH_READ_REG(TSI108_STAT_CARRY1);
+ carry2 = TSI108_ETH_READ_REG(TSI108_STAT_CARRY2);
+
+ TSI108_ETH_WRITE_REG(TSI108_STAT_CARRY1, carry1);
+ TSI108_ETH_WRITE_REG(TSI108_STAT_CARRY2, carry2);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXBYTES,
+ TSI108_STAT_RXBYTES_CARRY, &data->stats.rx_bytes);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXPKTS,
+ TSI108_STAT_RXPKTS_CARRY,
+ &data->stats.rx_packets);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXFCS,
+ TSI108_STAT_RXFCS_CARRY, &data->rx_fcs);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXMCAST,
+ TSI108_STAT_RXMCAST_CARRY,
+ &data->stats.multicast);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXALIGN,
+ TSI108_STAT_RXALIGN_CARRY,
+ &data->stats.rx_frame_errors);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXLENGTH,
+ TSI108_STAT_RXLENGTH_CARRY,
+ &data->stats.rx_length_errors);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXRUNT,
+ TSI108_STAT_RXRUNT_CARRY, &data->rx_underruns);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXJUMBO,
+ TSI108_STAT_RXJUMBO_CARRY, &data->rx_overruns);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXFRAG,
+ TSI108_STAT_RXFRAG_CARRY, &data->rx_short_fcs);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXJABBER,
+ TSI108_STAT_RXJABBER_CARRY, &data->rx_long_fcs);
+
+ tsi108_stat_carry_one(carry1, TSI108_STAT_CARRY1_RXDROP,
+ TSI108_STAT_RXDROP_CARRY,
+ &data->stats.rx_missed_errors);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXBYTES,
+ TSI108_STAT_TXBYTES_CARRY, &data->stats.tx_bytes);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXPKTS,
+ TSI108_STAT_TXPKTS_CARRY,
+ &data->stats.tx_packets);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXEXDEF,
+ TSI108_STAT_TXEXDEF_CARRY,
+ &data->stats.tx_aborted_errors);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXEXCOL,
+ TSI108_STAT_TXEXCOL_CARRY, &data->tx_coll_abort);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXTCOL,
+ TSI108_STAT_TXTCOL_CARRY,
+ &data->stats.collisions);
+
+ tsi108_stat_carry_one(carry2, TSI108_STAT_CARRY2_TXPAUSE,
+ TSI108_STAT_TXPAUSEDROP_CARRY,
+ &data->tx_pause_drop);
+
+ spin_unlock_irq(&data->misclock);
+}
+
+/* Read a stat counter atomically with respect to carries.
+ * data->misclock must be held.
+ */
+static inline unsigned long
+tsi108_read_stat(tsi108_prv_data * data, int reg, int carry_bit,
+ int carry_shift, unsigned long *upper)
+{
+ int carryreg;
+ unsigned long val;
+
+ if (reg < 0xb0)
+ carryreg = TSI108_STAT_CARRY1;
+ else
+ carryreg = TSI108_STAT_CARRY2;
+
+ again:
+ val = TSI108_ETH_READ_REG(reg) | *upper;
+
+ rmb();
+
+ /* Check to see if it overflowed, but the interrupt hasn't
+ * been serviced yet. If so, handle the carry here, and
+ * try again.
+ */
+
+ if (unlikely(TSI108_ETH_READ_REG(carryreg) & carry_bit)) {
+ *upper += carry_shift;
+ TSI108_ETH_WRITE_REG(carryreg, carry_bit);
+ mb();
+
+ goto again;
+ }
+
+ return val;
+}
+
+static struct net_device_stats *tsi108_get_stats(net_device * dev)
+{
+ unsigned long excol;
+
+ tsi108_prv_data *data = netdev_priv(dev);
+ spin_lock_irq(&data->misclock);
+
+ data->tmpstats.rx_packets =
+ tsi108_read_stat(data, TSI108_STAT_RXPKTS,
+ TSI108_STAT_CARRY1_RXPKTS,
+ TSI108_STAT_RXPKTS_CARRY, &data->stats.rx_packets);
+
+ data->tmpstats.tx_packets =
+ tsi108_read_stat(data, TSI108_STAT_TXPKTS,
+ TSI108_STAT_CARRY2_TXPKTS,
+ TSI108_STAT_TXPKTS_CARRY, &data->stats.tx_packets);
+
+ data->tmpstats.rx_bytes =
+ tsi108_read_stat(data, TSI108_STAT_RXBYTES,
+ TSI108_STAT_CARRY1_RXBYTES,
+ TSI108_STAT_RXBYTES_CARRY, &data->stats.rx_bytes);
+
+ data->tmpstats.tx_bytes =
+ tsi108_read_stat(data, TSI108_STAT_TXBYTES,
+ TSI108_STAT_CARRY2_TXBYTES,
+ TSI108_STAT_TXBYTES_CARRY, &data->stats.tx_bytes);
+
+ data->tmpstats.multicast =
+ tsi108_read_stat(data, TSI108_STAT_RXMCAST,
+ TSI108_STAT_CARRY1_RXMCAST,
+ TSI108_STAT_RXMCAST_CARRY, &data->stats.multicast);
+
+ excol = tsi108_read_stat(data, TSI108_STAT_TXEXCOL,
+ TSI108_STAT_CARRY2_TXEXCOL,
+ TSI108_STAT_TXEXCOL_CARRY,
+ &data->tx_coll_abort);
+
+ data->tmpstats.collisions =
+ tsi108_read_stat(data, TSI108_STAT_TXTCOL,
+ TSI108_STAT_CARRY2_TXTCOL,
+ TSI108_STAT_TXTCOL_CARRY, &data->stats.collisions);
+
+ data->tmpstats.collisions += excol;
+
+ data->tmpstats.rx_length_errors =
+ tsi108_read_stat(data, TSI108_STAT_RXLENGTH,
+ TSI108_STAT_CARRY1_RXLENGTH,
+ TSI108_STAT_RXLENGTH_CARRY,
+ &data->stats.rx_length_errors);
+
+ data->tmpstats.rx_length_errors +=
+ tsi108_read_stat(data, TSI108_STAT_RXRUNT,
+ TSI108_STAT_CARRY1_RXRUNT,
+ TSI108_STAT_RXRUNT_CARRY, &data->rx_underruns);
+
+ data->tmpstats.rx_length_errors +=
+ tsi108_read_stat(data, TSI108_STAT_RXJUMBO,
+ TSI108_STAT_CARRY1_RXJUMBO,
+ TSI108_STAT_RXJUMBO_CARRY, &data->rx_overruns);
+
+ data->tmpstats.rx_frame_errors =
+ tsi108_read_stat(data, TSI108_STAT_RXALIGN,
+ TSI108_STAT_CARRY1_RXALIGN,
+ TSI108_STAT_RXALIGN_CARRY,
+ &data->stats.rx_frame_errors);
+
+ data->tmpstats.rx_frame_errors +=
+ tsi108_read_stat(data, TSI108_STAT_RXFCS,
+ TSI108_STAT_CARRY1_RXFCS, TSI108_STAT_RXFCS_CARRY,
+ &data->rx_fcs);
+
+ data->tmpstats.rx_frame_errors +=
+ tsi108_read_stat(data, TSI108_STAT_RXFRAG,
+ TSI108_STAT_CARRY1_RXFRAG,
+ TSI108_STAT_RXFRAG_CARRY, &data->rx_short_fcs);
+
+ data->tmpstats.rx_missed_errors =
+ tsi108_read_stat(data, TSI108_STAT_RXDROP,
+ TSI108_STAT_CARRY1_RXDROP,
+ TSI108_STAT_RXDROP_CARRY,
+ &data->stats.rx_missed_errors);
+
+ /* These three are maintained by software. */
+ data->tmpstats.rx_fifo_errors = data->stats.rx_fifo_errors;
+ data->tmpstats.rx_crc_errors = data->stats.rx_crc_errors;
+
+ data->tmpstats.tx_aborted_errors =
+ tsi108_read_stat(data, TSI108_STAT_TXEXDEF,
+ TSI108_STAT_CARRY2_TXEXDEF,
+ TSI108_STAT_TXEXDEF_CARRY,
+ &data->stats.tx_aborted_errors);
+
+ data->tmpstats.tx_aborted_errors +=
+ tsi108_read_stat(data, TSI108_STAT_TXPAUSEDROP,
+ TSI108_STAT_CARRY2_TXPAUSE,
+ TSI108_STAT_TXPAUSEDROP_CARRY,
+ &data->tx_pause_drop);
+
+ data->tmpstats.tx_aborted_errors += excol;
+
+ data->tmpstats.tx_errors = data->tmpstats.tx_aborted_errors;
+ data->tmpstats.rx_errors = data->tmpstats.rx_length_errors +
+ data->tmpstats.rx_crc_errors +
+ data->tmpstats.rx_frame_errors +
+ data->tmpstats.rx_fifo_errors + data->tmpstats.rx_missed_errors;
+
+ spin_unlock_irq(&data->misclock);
+ return &data->tmpstats;
+}
+
+static void tsi108_restart_rx(tsi108_prv_data * data, net_device * dev)
+{
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXQ_PTRHIGH,
+ TSI108_EC_RXQ_PTRHIGH_VALID);
+
+ wmb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCTRL, TSI108_EC_RXCTRL_GO
+ | TSI108_EC_RXCTRL_QUEUE0);
+}
+
+static void tsi108_restart_tx(tsi108_prv_data * data)
+{
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXQ_PTRHIGH,
+ TSI108_EC_TXQ_PTRHIGH_VALID);
+
+ wmb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXCTRL, TSI108_EC_TXCTRL_IDLEINT |
+ TSI108_EC_TXCTRL_GO | TSI108_EC_TXCTRL_QUEUE0);
+}
+
+/* txlock must be held by caller, with IRQs disabled, and
+ * with permission to re-enable them when the lock is dropped.
+ */
+static void tsi108_check_for_completed_tx(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ int tx;
+ struct sk_buff *skb;
+ int release = 0;
+
+ while (!data->txfree || data->txhead != data->txtail) {
+ tx = data->txtail;
+
+ if (data->txring[tx].misc & TSI108_TX_OWN)
+ break;
+
+ skb = data->txskbs[tx];
+
+ if (!(data->txring[tx].misc & TSI108_TX_OK))
+ printk("%s: bad tx packet, misc %x\n",
+ dev->name, data->txring[tx].misc);
+
+ data->txtail = (data->txtail + 1) % TSI108_TXRING_LEN;
+ data->txfree++;
+
+ if (data->txring[tx].misc & TSI108_TX_EOF) {
+ dev_kfree_skb_any(skb);
+ release++;
+ }
+ }
+
+ if (release) {
+
+ if (netif_queue_stopped(dev)
+ && is_valid_ether_addr(dev->dev_addr) && data->link_up)
+ netif_wake_queue(dev);
+ }
+}
+
+static int tsi108_send_packet(sk_buff * skb, net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ int frags = skb_shinfo(skb)->nr_frags + 1;
+ int i;
+#ifdef FIXME_SG_CSUM_NOT_TESTED /* FIXME: Not supported now. */
+ long csstart;
+ long csum;
+
+ csstart = skb->len - skb->data_len;
+ if (csstart > skb->len - skb->data_len)
+ BUG();
+ csum = 0;
+ if (csstart != skb->len)
+ csum = skb_checksum(skb, csstart, skb->len - csstart, 0);
+#endif
+
+ if (!data->phy_ok && net_ratelimit())
+ printk(KERN_ERR "%s: Transmit while PHY is down!\n", dev->name);
+
+ if (!data->link_up) {
+ printk(KERN_ERR "%s: Transmit while link is down!\n",
+ dev->name);
+ netif_stop_queue(dev);
+ return 1;
+ }
+
+ if (data->txfree < MAX_SKB_FRAGS + 1) {
+ netif_stop_queue(dev);
+
+ if (net_ratelimit())
+ printk(KERN_ERR "%s: Transmit with full tx ring!\n",
+ dev->name);
+ return 1;
+ }
+
+ if (data->txfree - frags < MAX_SKB_FRAGS + 1) {
+ netif_stop_queue(dev);
+ }
+
+ spin_lock_irq(&data->txlock);
+
+ for (i = 0; i < frags; i++) {
+ int misc = 0;
+ int tx = data->txhead;
+
+ /* This is done to mark every TSI108_TX_INT_FREQ tx buffers with
+ * the interrupt bit. TX descriptor-complete interrupts are
+ * enabled when the queue fills up, and masked when there is
+ * still free space. This way, when saturating the outbound
+ * link, the tx interrupts are kept to a reasonable level.
+ * When the queue is not full, reclamation of skbs still occurs
+ * as new packets are transmitted, or on a queue-empty
+ * interrupt.
+ */
+
+ if ((tx % TSI108_TX_INT_FREQ == 0) &&
+ ((TSI108_TXRING_LEN - data->txfree) >= TSI108_TX_INT_FREQ)
+ )
+ misc = TSI108_TX_INT;
+
+ data->txskbs[tx] = skb;
+
+ if (i == 0) {
+ data->txring[tx].buf0 = virt_to_phys(skb->data);
+ data->txring[tx].len = skb->len - skb->data_len;
+ misc |= TSI108_TX_SOF;
+ } else {
+ skb_frag_t *frag = &skb_shinfo(skb)->frags[i - 1];
+
+ data->txring[tx].buf0 =
+ page_to_phys(frag->page) + frag->page_offset;
+ data->txring[tx].len = frag->size;
+ }
+
+ if (i == frags - 1)
+ misc |= TSI108_TX_EOF;
+
+#ifdef TSI108_PRINT_TX_FRAME
+ {
+ int i;
+ printk("%s: Tx Frame contents (%d)\n", dev->name,
+ skb->len);
+ for (i = 0; i < skb->len; i++)
+ printk(" %2.2x", skb->data[i]);
+ printk(".\n");
+ }
+#endif /* TSI108_PRINT_TX_FRAME */
+
+ mb();
+ data->txring[tx].misc = misc | TSI108_TX_OWN;
+
+ data->txhead = (data->txhead + 1) % TSI108_TXRING_LEN;
+ data->txfree--;
+ }
+
+ tsi108_check_for_completed_tx(dev);
+
+ /* This must be done after the check for completed tx descriptors,
+ * so that the tail pointer is correct.
+ */
+
+ if (!(TSI108_ETH_READ_REG(TSI108_EC_TXSTAT) & TSI108_EC_TXSTAT_QUEUE0))
+ tsi108_restart_tx(data);
+
+ spin_unlock_irq(&data->txlock);
+ return 0;
+}
+
+static int tsi108_check_for_completed_rx(net_device * dev, int budget)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ int done = 0;
+
+ while (data->rxfree && done != budget) {
+ int rx = data->rxtail;
+ struct sk_buff *skb;
+
+ if (data->rxring[rx].misc & TSI108_RX_OWN)
+ break;
+
+ skb = data->rxskbs[rx];
+ data->rxtail = (data->rxtail + 1) % TSI108_RXRING_LEN;
+ data->rxfree--;
+ done++;
+
+ if (data->rxring[rx].misc & TSI108_RX_BAD) {
+ spin_lock_irq(&data->misclock);
+
+ if (data->rxring[rx].misc & TSI108_RX_CRC)
+ data->stats.rx_crc_errors++;
+ if (data->rxring[rx].misc & TSI108_RX_OVER)
+ data->stats.rx_fifo_errors++;
+
+ spin_unlock_irq(&data->misclock);
+
+ dev_kfree_skb_any(skb);
+ continue;
+ }
+#ifdef TSI108_PRINT_RX_FRAME
+ {
+ int i;
+ printk("%s: Rx Frame contents (%d)\n",
+ dev->name, data->rxring[rx].len);
+ for (i = 0; i < data->rxring[rx].len; i++)
+ printk(" %2.2x", skb->data[i]);
+ printk(".\n");
+ }
+#endif /* TSI108_PRINT_RX_FRAME */
+
+ skb->dev = dev;
+ skb_put(skb, data->rxring[rx].len);
+ skb->protocol = eth_type_trans(skb, dev);
+ netif_receive_skb(skb);
+ dev->last_rx = jiffies;
+ }
+
+ return done;
+}
+
+static int tsi108_refill_rx(net_device * dev, int budget)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ int done = 0;
+
+ while (data->rxfree != TSI108_RXRING_LEN && done != budget) {
+ int rx = data->rxhead;
+ sk_buff *skb;
+
+ data->rxskbs[rx] = skb = dev_alloc_skb(TSI108_RXBUF_SIZE + 2);
+ if (!skb)
+ break;
+
+ skb_reserve(skb, 2); /* Align the data on a 4-byte boundary. */
+
+ data->rxring[rx].buf0 = virt_to_phys(skb->data);
+
+ /* Sometimes the hardware sets blen to zero after packet
+ * reception, even though the manual says that it's only ever
+ * modified by the driver.
+ */
+
+ data->rxring[rx].blen = 1536;
+ mb();
+ data->rxring[rx].misc = TSI108_RX_OWN | TSI108_RX_INT;
+
+ data->rxhead = (data->rxhead + 1) % TSI108_RXRING_LEN;
+ data->rxfree++;
+ done++;
+ }
+
+ mb();
+
+ if (done != 0 && !(TSI108_ETH_READ_REG(TSI108_EC_RXSTAT) &
+ TSI108_EC_RXSTAT_QUEUE0))
+ tsi108_restart_rx(data, dev);
+
+ return done;
+}
+
+static int tsi108_poll(net_device * dev, int *budget)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 estat = TSI108_ETH_READ_REG(TSI108_EC_RXESTAT);
+ u32 intstat = TSI108_ETH_READ_REG(TSI108_EC_INTSTAT);
+ int total_budget = min(*budget, dev->quota);
+ int num_received = 0, num_filled = 0, budget_used;
+
+ intstat &= TSI108_INT_RXQUEUE0 | TSI108_INT_RXTHRESH |
+ TSI108_INT_RXOVERRUN | TSI108_INT_RXERROR | TSI108_INT_RXWAIT;
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXESTAT, estat);
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTSTAT, intstat);
+
+ if (data->rxpending || (estat & TSI108_EC_RXESTAT_Q0_DESCINT))
+ num_received = tsi108_check_for_completed_rx(dev, total_budget);
+
+ /* This should normally fill no more slots than the number of
+ * packets received in tsi108_check_for_completed_rx(). The exception
+ * is when we previously ran out of memory for RX SKBs. In that
+ * case, it's helpful to obey the budget, not only so that the
+ * CPU isn't hogged, but so that memory (which may still be low)
+ * is not hogged by one device.
+ *
+ * A work unit is considered to be two SKBs to allow us to catch
+ * up when the ring has shrunk due to out-of-memory but we're
+ * still removing the full budget's worth of packets each time.
+ */
+
+ if (data->rxfree < TSI108_RXRING_LEN)
+ num_filled = tsi108_refill_rx(dev, total_budget * 2);
+
+ if (intstat & TSI108_INT_RXERROR) {
+ u32 err = TSI108_ETH_READ_REG(TSI108_EC_RXERR);
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXERR, err);
+
+ if (err) {
+ if (net_ratelimit())
+ printk(KERN_DEBUG "%s: RX error %x\n",
+ dev->name, err);
+
+ if (!(TSI108_ETH_READ_REG(TSI108_EC_RXSTAT) &
+ TSI108_EC_RXSTAT_QUEUE0))
+ tsi108_restart_rx(data, dev);
+ }
+ }
+
+ if (intstat & TSI108_INT_RXOVERRUN) {
+ spin_lock_irq(&data->misclock);
+ data->stats.rx_fifo_errors++;
+ spin_unlock_irq(&data->misclock);
+ }
+
+ budget_used = max(num_received, num_filled / 2);
+
+ *budget -= budget_used;
+ dev->quota -= budget_used;
+
+ if (budget_used != total_budget) {
+ data->rxpending = 0;
+ netif_rx_complete(dev);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK,
+ TSI108_ETH_READ_REG(TSI108_EC_INTMASK)
+ & ~(TSI108_INT_RXQUEUE0
+ | TSI108_INT_RXTHRESH |
+ TSI108_INT_RXOVERRUN |
+ TSI108_INT_RXERROR |
+ TSI108_INT_RXWAIT));
+
+ mb();
+
+ /* IRQs are level-triggered, so no need to re-check */
+ return 0;
+ } else {
+ data->rxpending = 1;
+ }
+
+ return 1;
+}
+
+static void tsi108_rx_int(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ /* A race could cause dev to already be scheduled, so it's not an
+ * error if that happens (and interrupts shouldn't be re-masked,
+ * because that can cause harmful races, if poll has already
+ * unmasked them but not cleared LINK_STATE_SCHED).
+ *
+ * This can happen if this code races with tsi108_poll(), which masks
+ * the interrupts after tsi108_irq_one() read the mask, but before
+ * netif_rx_schedule is called. It could also happen due to calls
+ * from tsi108_check_rxring().
+ */
+
+ if (netif_rx_schedule_prep(dev)) {
+ /* Mask, rather than ack, the receive interrupts. The ack
+ * will happen in tsi108_poll().
+ */
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK,
+ TSI108_ETH_READ_REG(TSI108_EC_INTMASK) |
+ TSI108_INT_RXQUEUE0
+ | TSI108_INT_RXTHRESH |
+ TSI108_INT_RXOVERRUN | TSI108_INT_RXERROR |
+ TSI108_INT_RXWAIT);
+ mb();
+ __netif_rx_schedule(dev);
+ } else {
+ if (!netif_running(dev)) {
+ /* This can happen if an interrupt occurs while the
+ * interface is being brought down, as the START
+ * bit is cleared before the stop function is called.
+ *
+ * In this case, the interrupts must be masked, or
+ * they will continue indefinitely.
+ *
+ * There's a race here if the interface is brought down
+ * and then up in rapid succession, as the device could
+ * be made running after the above check and before
+ * the masking below. This will only happen if the IRQ
+ * thread has a lower priority than the task brining
+ * up the interface. Fixing this race would likely
+ * require changes in generic code.
+ */
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK,
+ TSI108_ETH_READ_REG
+ (TSI108_EC_INTMASK) |
+ TSI108_INT_RXQUEUE0 |
+ TSI108_INT_RXTHRESH |
+ TSI108_INT_RXOVERRUN |
+ TSI108_INT_RXERROR |
+ TSI108_INT_RXWAIT);
+ mb();
+ }
+ }
+}
+
+/* If the RX ring has run out of memory, try periodically
+ * to allocate some more, as otherwise poll would never
+ * get called (apart from the initial end-of-queue condition).
+ *
+ * This is called once per second (by default) from the thread.
+ */
+
+static void tsi108_check_rxring(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ /* A poll is scheduled, as opposed to caling tsi108_refill_rx
+ * directly, so as to keep the receive path single-threaded
+ * (and thus not needing a lock).
+ */
+
+ if (netif_running(dev) && data->rxfree < TSI108_RXRING_LEN / 4)
+ tsi108_rx_int(dev);
+}
+
+static void tsi108_tx_int(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 estat = TSI108_ETH_READ_REG(TSI108_EC_TXESTAT);
+
+ mb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXESTAT, estat);
+ mb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTSTAT, TSI108_INT_TXQUEUE0 |
+ TSI108_INT_TXIDLE | TSI108_INT_TXERROR);
+ mb();
+ if (estat & TSI108_EC_TXESTAT_Q0_ERR) {
+ u32 err = TSI108_ETH_READ_REG(TSI108_EC_TXERR);
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXERR, err);
+
+ if (err && net_ratelimit())
+ printk(KERN_ERR "%s: TX error %x\n", dev->name, err);
+ }
+
+ if (estat & (TSI108_EC_TXESTAT_Q0_DESCINT | TSI108_EC_TXESTAT_Q0_EOQ)) {
+ spin_lock(&data->txlock);
+ tsi108_check_for_completed_tx(dev);
+ spin_unlock(&data->txlock);
+ }
+}
+
+static irqreturn_t tsi108_irq_one(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 stat = TSI108_ETH_READ_REG(TSI108_EC_INTSTAT);
+
+ if (!(stat & TSI108_INT_ANY))
+ return IRQ_NONE; /* Not our interrupt */
+
+ stat &= ~TSI108_ETH_READ_REG(TSI108_EC_INTMASK);
+
+ if (stat & (TSI108_INT_TXQUEUE0 | TSI108_INT_TXIDLE |
+ TSI108_INT_TXERROR))
+ tsi108_tx_int(dev);
+ if (stat & (TSI108_INT_RXQUEUE0 | TSI108_INT_RXTHRESH |
+ TSI108_INT_RXWAIT | TSI108_INT_RXOVERRUN |
+ TSI108_INT_RXERROR))
+ tsi108_rx_int(dev);
+
+ if (stat & TSI108_INT_SFN) {
+ if (net_ratelimit())
+ printk(KERN_DEBUG "%s: SFN error\n", dev->name);
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTSTAT, TSI108_INT_SFN);
+ }
+
+ if (stat & TSI108_INT_STATCARRY) {
+ tsi108_stat_carry(dev);
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTSTAT, TSI108_INT_STATCARRY);
+ }
+
+ return IRQ_HANDLED;
+}
+
+static irqreturn_t tsi108_irq(int irq, void *dev_id, struct pt_regs *regs)
+{
+ if ((IRQ_TSI108_GIGE0 != irq) && (IRQ_TSI108_GIGE1 != irq))
+ return IRQ_NONE; /* Not our interrupt */
+
+ return tsi108_irq_one((struct net_device *)dev_id);
+}
+
+static void tsi108_stop_ethernet(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ /* Disable all TX and RX queues ... */
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXCTRL, 0);
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCTRL, 0);
+
+ /* ...and wait for them to become idle */
+ mb();
+ while (TSI108_ETH_READ_REG(TSI108_EC_TXSTAT) &
+ TSI108_EC_TXSTAT_ACTIVE) ;
+ while (TSI108_ETH_READ_REG(TSI108_EC_RXSTAT) &
+ TSI108_EC_RXSTAT_ACTIVE) ;
+}
+
+static void tsi108_reset_ether(tsi108_prv_data * data)
+{
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG1, TSI108_MAC_CFG1_SOFTRST);
+ udelay(100);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG1, 0);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_PORTCTRL, TSI108_EC_PORTCTRL_STATRST);
+ udelay(100);
+ TSI108_ETH_WRITE_REG(TSI108_EC_PORTCTRL,
+ TSI108_ETH_READ_REG(TSI108_EC_PORTCTRL) &
+ ~TSI108_EC_PORTCTRL_STATRST);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXCFG, TSI108_EC_TXCFG_RST);
+ udelay(100);
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXCFG,
+ TSI108_ETH_READ_REG(TSI108_EC_TXCFG) &
+ ~TSI108_EC_TXCFG_RST);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCFG, TSI108_EC_RXCFG_RST);
+ udelay(100);
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCFG,
+ TSI108_ETH_READ_REG(TSI108_EC_RXCFG) &
+ ~TSI108_EC_RXCFG_RST);
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_MII_MGMT_CFG,
+ TSI108_ETH_READ_REG(TSI108_MAC_MII_MGMT_CFG) |
+ TSI108_MAC_MII_MGMT_RST);
+ udelay(100);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_MII_MGMT_CFG,
+ TSI108_ETH_READ_REG(TSI108_MAC_MII_MGMT_CFG) &
+ ~(TSI108_MAC_MII_MGMT_RST |
+ TSI108_MAC_MII_MGMT_CLK));
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_MII_MGMT_CFG,
+ TSI108_ETH_READ_REG(TSI108_MAC_MII_MGMT_CFG) |
+ TSI108_MAC_MII_MGMT_CLK);
+}
+
+static int tsi108_get_mac(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ u32 word1 = TSI108_ETH_READ_REG(TSI108_MAC_ADDR1);
+ u32 word2 = TSI108_ETH_READ_REG(TSI108_MAC_ADDR2);
+
+ /* Note that the octets are reversed from what the manual says,
+ * producing an even weirder ordering...
+ */
+ if (word2 == 0 && word1 == 0) {
+ dev->dev_addr[0] = 0x00;
+ dev->dev_addr[1] = 0x06;
+ dev->dev_addr[2] = 0xd2;
+ dev->dev_addr[3] = 0x00;
+ dev->dev_addr[4] = 0x00;
+ if (0x8 == data->phy)
+ dev->dev_addr[5] = 0x01;
+ else
+ dev->dev_addr[5] = 0x02;
+
+ word2 = (dev->dev_addr[0] << 16) | (dev->dev_addr[1] << 24);
+
+ word1 = (dev->dev_addr[2] << 0) | (dev->dev_addr[3] << 8) |
+ (dev->dev_addr[4] << 16) | (dev->dev_addr[5] << 24);
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_ADDR1, word1);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_ADDR2, word2);
+ } else {
+ dev->dev_addr[0] = (word2 >> 16) & 0xff;
+ dev->dev_addr[1] = (word2 >> 24) & 0xff;
+ dev->dev_addr[2] = (word1 >> 0) & 0xff;
+ dev->dev_addr[3] = (word1 >> 8) & 0xff;
+ dev->dev_addr[4] = (word1 >> 16) & 0xff;
+ dev->dev_addr[5] = (word1 >> 24) & 0xff;
+ }
+
+ if (!is_valid_ether_addr(dev->dev_addr)) {
+ printk("KERN_ERR: word1: %08x, word2: %08x\n", word1, word2);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int tsi108_set_mac(net_device * dev, void *addr)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 word1, word2;
+ int i;
+
+ if (!is_valid_ether_addr(addr))
+ return -EINVAL;
+
+ for (i = 0; i < 6; i++)
+ /* +2 is for the offset of the HW addr type */
+ dev->dev_addr[i] = ((unsigned char *)addr)[i + 2];
+
+ word2 = (dev->dev_addr[0] << 16) | (dev->dev_addr[1] << 24);
+
+ word1 = (dev->dev_addr[2] << 0) | (dev->dev_addr[3] << 8) |
+ (dev->dev_addr[4] << 16) | (dev->dev_addr[5] << 24);
+
+ spin_lock_irq(&data->misclock);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_ADDR1, word1);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_ADDR2, word2);
+ spin_lock(&data->txlock);
+
+ if (netif_queue_stopped(dev) && data->txfree && data->link_up)
+ netif_wake_queue(dev);
+
+ spin_unlock(&data->txlock);
+ spin_unlock_irq(&data->misclock);
+ return 0;
+}
+
+/* Protected by dev->xmit_lock. */
+static void tsi108_set_rx_mode(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 rxcfg = TSI108_ETH_READ_REG(TSI108_EC_RXCFG);
+
+ if (dev->flags & IFF_PROMISC) {
+ rxcfg &= ~(TSI108_EC_RXCFG_UC_HASH | TSI108_EC_RXCFG_MC_HASH);
+ rxcfg |= TSI108_EC_RXCFG_UFE | TSI108_EC_RXCFG_MFE;
+ goto out;
+ }
+
+ rxcfg &= ~(TSI108_EC_RXCFG_UFE | TSI108_EC_RXCFG_MFE);
+
+ if (dev->mc_count) {
+ int i;
+ struct dev_mc_list *mc = dev->mc_list;
+ rxcfg |= TSI108_EC_RXCFG_MFE | TSI108_EC_RXCFG_MC_HASH;
+
+ memset(data->mc_hash, 0, sizeof(data->mc_hash));
+
+ while (mc) {
+ u32 hash, crc;
+
+ if (mc->dmi_addrlen == 6) {
+ crc = ether_crc(6, mc->dmi_addr);
+ hash = crc >> 23;
+
+ __set_bit(hash, &data->mc_hash[0]);
+ } else {
+ printk(KERN_ERR
+ "%s: got multicast address of length %d "
+ "instead of 6.\n", dev->name,
+ mc->dmi_addrlen);
+ }
+
+ mc = mc->next;
+ }
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_HASHADDR,
+ TSI108_EC_HASHADDR_AUTOINC |
+ TSI108_EC_HASHADDR_MCAST);
+
+ for (i = 0; i < 16; i++) {
+ /* The manual says that the hardware may drop
+ * back-to-back writes to the data register.
+ */
+ udelay(1);
+ mb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_HASHDATA,
+ data->mc_hash[i]);
+ mb();
+ }
+ }
+
+ out:
+ mb();
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCFG, rxcfg);
+}
+
+static void tsi108_init_phy(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ u32 i = 0;
+ u16 phyVal = 0;
+
+ spin_lock_irq(&phy_lock);
+
+ tsi108_write_mii(data, PHY_CTRL, PHY_CTRL_RESET);
+ mb();
+ while (tsi108_read_mii(data, PHY_CTRL, NULL) & PHY_CTRL_RESET) ;
+
+#if (TSI108_PHY_TYPE == PHY_BCM54XX) /* Broadcom BCM54xx PHY */
+ tsi108_write_mii(data, 0x09, 0x0300);
+ tsi108_write_mii(data, 0x10, 0x1020);
+ tsi108_write_mii(data, 0x1c, 0x8c00);
+ mb();
+#endif
+
+ tsi108_write_mii(data,
+ PHY_CTRL,
+ PHY_CTRL_AUTONEG_EN | PHY_CTRL_AUTONEG_START);
+ mb();
+ while (tsi108_read_mii(data, PHY_CTRL, NULL) & PHY_CTRL_AUTONEG_START) ;
+
+ /* Set G/MII mode and receive clock select in TBI control #2. The
+ * second port won't work if this isn't done, even though we don't
+ * use TBI mode.
+ */
+
+ tsi108_write_tbi(data, 0x11, 0x30);
+
+ /* FIXME: It seems to take more than 2 back-to-back reads to the
+ * PHY_STAT register before the link up status bit is set.
+ */
+
+ data->link_up = 1;
+
+ while (!((phyVal = tsi108_read_mii(data, PHY_STAT, NULL)) &
+ PHY_STAT_LINKUP)) {
+ if (i++ > (MII_READ_DELAY / 10)) {
+ data->link_up = 0;
+ break;
+ }
+ mdelay(10);
+ }
+
+ printk(KERN_DEBUG "PHY_STAT reg contains %08x\n", phyVal);
+ data->phy_ok = 1;
+ spin_unlock_irq(&phy_lock);
+}
+
+static void tsi108_kill_phy(struct net_device *dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ spin_lock_irq(&phy_lock);
+ tsi108_write_mii(data, PHY_CTRL, PHY_CTRL_POWERDOWN);
+ data->phy_ok = 0;
+ spin_unlock_irq(&phy_lock);
+}
+
+static int tsi108_open(struct net_device *dev)
+{
+ int i;
+ tsi108_prv_data *data = netdev_priv(dev);
+ unsigned int rxring_size = TSI108_RXRING_LEN * sizeof(rx_desc);
+ unsigned int txring_size = TSI108_TXRING_LEN * sizeof(tx_desc);
+
+ printk(KERN_DEBUG "Inside tsi108_open()!\n");
+
+ i = request_irq(data->irq_num, tsi108_irq, 0, dev->name, dev);
+ if (i != 0) {
+ printk(KERN_ERR "tsi108_eth%d: Could not allocate IRQ%d.\n",
+ data->irq_num % IRQ_TSI108_GIGE0, data->irq_num);
+ return i;
+ } else {
+ dev->irq = data->irq_num;
+ printk(KERN_NOTICE
+ "tsi108_open : Port %d Assigned IRQ %d to %s\n",
+ data->irq_num % IRQ_TSI108_GIGE0, dev->irq, dev->name);
+ }
+
+ data->rxring = pci_alloc_consistent(NULL, rxring_size, &data->rxdma);
+
+ if (!data->rxring) {
+ printk(KERN_DEBUG
+ "TSI108_ETH: failed to allocate memory for rxring!\n");
+ return -ENOMEM;
+ } else {
+ memset(data->rxring, 0, rxring_size);
+ }
+
+ data->txring = pci_alloc_consistent(NULL, txring_size, &data->txdma);
+
+ if (!data->txring) {
+ printk(KERN_DEBUG
+ "TSI108_ETH: failed to allocate memory for txring!\n");
+ pci_free_consistent(0, rxring_size, data->rxring, data->rxdma);
+ return -ENOMEM;
+ } else {
+ memset(data->txring, 0, txring_size);
+ }
+
+ for (i = 0; i < TSI108_RXRING_LEN; i++) {
+ data->rxring[i].next0 = data->rxdma + (i + 1) * sizeof(rx_desc);
+ data->rxring[i].blen = TSI108_RXBUF_SIZE;
+ data->rxring[i].vlan = 0;
+ }
+
+ data->rxring[TSI108_RXRING_LEN - 1].next0 = data->rxdma;
+
+ data->rxtail = 0;
+ data->rxhead = 0;
+
+ for (i = 0; i < TSI108_RXRING_LEN; i++) {
+ sk_buff *skb = dev_alloc_skb(TSI108_RXBUF_SIZE + NET_IP_ALIGN);
+
+ if (!skb) {
+ /* Bah. No memory for now, but maybe we'll get
+ * some more later.
+ * For now, we'll live with the smaller ring.
+ */
+ printk(KERN_WARNING
+ "%s: Could only allocate %d receive skb(s).\n",
+ dev->name, i);
+ data->rxhead = i;
+ break;
+ }
+
+ data->rxskbs[i] = skb;
+ /* Align the payload on a 4-byte boundary */
+ skb_reserve(skb, 2);
+ data->rxskbs[i] = skb;
+ data->rxring[i].buf0 = virt_to_phys(data->rxskbs[i]->data);
+ data->rxring[i].misc = TSI108_RX_OWN | TSI108_RX_INT;
+ }
+
+ data->rxfree = i;
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXQ_PTRLOW, data->rxdma);
+
+ for (i = 0; i < TSI108_TXRING_LEN; i++) {
+ data->txring[i].next0 = data->txdma + (i + 1) * sizeof(tx_desc);
+ data->txring[i].misc = 0;
+ }
+
+ data->txring[TSI108_TXRING_LEN - 1].next0 = data->txdma;
+ data->txtail = 0;
+ data->txhead = 0;
+ data->txfree = TSI108_TXRING_LEN;
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXQ_PTRLOW, data->txdma);
+ tsi108_init_phy(dev);
+
+ init_timer(&data->timer);
+ data->timer.expires = jiffies + 1;
+ data->timer.data = (unsigned long)dev;
+ data->timer.function = &tsi108_timed_checker; /* timer handler */
+ add_timer(&data->timer);
+
+ tsi108_restart_rx(data, dev);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTSTAT, ~0);
+ mb();
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK,
+ ~(TSI108_INT_TXQUEUE0 | TSI108_INT_RXERROR |
+ TSI108_INT_RXTHRESH | TSI108_INT_RXQUEUE0 |
+ TSI108_INT_RXOVERRUN | TSI108_INT_RXWAIT |
+ TSI108_INT_SFN | TSI108_INT_STATCARRY));
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG1,
+ TSI108_MAC_CFG1_RXEN | TSI108_MAC_CFG1_TXEN);
+ netif_start_queue(dev);
+ return 0;
+}
+
+static int tsi108_close(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ netif_stop_queue(dev);
+
+ del_timer_sync(&data->timer);
+
+ printk(KERN_DEBUG "Inside tsi108_ifdown!\n");
+
+ tsi108_stop_ethernet(dev);
+ tsi108_kill_phy(dev);
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK, ~0);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG1, 0);
+
+ /* Check for any pending TX packets, and drop them. */
+
+ while (!data->txfree || data->txhead != data->txtail) {
+ int tx = data->txtail;
+ struct sk_buff *skb;
+ skb = data->txskbs[tx];
+ data->txtail = (data->txtail + 1) % TSI108_TXRING_LEN;
+ data->txfree++;
+ dev_kfree_skb(skb);
+ }
+
+ synchronize_irq(data->irq_num);
+ free_irq(data->irq_num, dev);
+
+ /* Discard the RX ring. */
+
+ while (data->rxfree) {
+ int rx = data->rxtail;
+ struct sk_buff *skb;
+
+ skb = data->rxskbs[rx];
+ data->rxtail = (data->rxtail + 1) % TSI108_RXRING_LEN;
+ data->rxfree--;
+ dev_kfree_skb(skb);
+ }
+
+ pci_free_consistent(0,
+ TSI108_RXRING_LEN * sizeof(rx_desc),
+ data->rxring, data->rxdma);
+ pci_free_consistent(0,
+ TSI108_TXRING_LEN * sizeof(tx_desc),
+ data->txring, data->txdma);
+
+ return 0;
+}
+
+static void tsi108_init_mac(net_device * dev)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG2, TSI108_MAC_CFG2_DFLT_PREAMBLE |
+ TSI108_MAC_CFG2_PADCRC);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXTHRESH,
+ (192 << TSI108_EC_TXTHRESH_STARTFILL) |
+ (192 << TSI108_EC_TXTHRESH_STOPFILL));
+
+ TSI108_ETH_WRITE_REG(TSI108_STAT_CARRYMASK1,
+ ~(TSI108_STAT_CARRY1_RXBYTES |
+ TSI108_STAT_CARRY1_RXPKTS |
+ TSI108_STAT_CARRY1_RXFCS |
+ TSI108_STAT_CARRY1_RXMCAST |
+ TSI108_STAT_CARRY1_RXALIGN |
+ TSI108_STAT_CARRY1_RXLENGTH |
+ TSI108_STAT_CARRY1_RXRUNT |
+ TSI108_STAT_CARRY1_RXJUMBO |
+ TSI108_STAT_CARRY1_RXFRAG |
+ TSI108_STAT_CARRY1_RXJABBER |
+ TSI108_STAT_CARRY1_RXDROP));
+
+ TSI108_ETH_WRITE_REG(TSI108_STAT_CARRYMASK2,
+ ~(TSI108_STAT_CARRY2_TXBYTES |
+ TSI108_STAT_CARRY2_TXPKTS |
+ TSI108_STAT_CARRY2_TXEXDEF |
+ TSI108_STAT_CARRY2_TXEXCOL |
+ TSI108_STAT_CARRY2_TXTCOL |
+ TSI108_STAT_CARRY2_TXPAUSE));
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_PORTCTRL, TSI108_EC_PORTCTRL_STATEN);
+ TSI108_ETH_WRITE_REG(TSI108_MAC_CFG1, 0);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXCFG,
+ TSI108_EC_RXCFG_SE | TSI108_EC_RXCFG_BFE);
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXQ_CFG, TSI108_EC_TXQ_CFG_DESC_INT |
+ TSI108_EC_TXQ_CFG_EOQ_OWN_INT |
+ TSI108_EC_TXQ_CFG_WSWP | (TSI108_PBM_PORT <<
+ TSI108_EC_TXQ_CFG_SFNPORT));
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXQ_CFG, TSI108_EC_RXQ_CFG_DESC_INT |
+ TSI108_EC_RXQ_CFG_EOQ_OWN_INT |
+ TSI108_EC_RXQ_CFG_WSWP | (TSI108_PBM_PORT <<
+ TSI108_EC_RXQ_CFG_SFNPORT));
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_TXQ_BUFCFG,
+ TSI108_EC_TXQ_BUFCFG_BURST256 |
+ TSI108_EC_TXQ_BUFCFG_BSWP | (TSI108_PBM_PORT <<
+ TSI108_EC_TXQ_BUFCFG_SFNPORT));
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_RXQ_BUFCFG,
+ TSI108_EC_RXQ_BUFCFG_BURST256 |
+ TSI108_EC_RXQ_BUFCFG_BSWP | (TSI108_PBM_PORT <<
+ TSI108_EC_RXQ_BUFCFG_SFNPORT));
+
+ TSI108_ETH_WRITE_REG(TSI108_EC_INTMASK, ~0);
+}
+
+static int tsi108_ioctl(net_device * dev, struct ifreq *rq, int cmd)
+{
+ tsi108_prv_data *data = netdev_priv(dev);
+ struct mii_ioctl_data *mii_data =
+ (struct mii_ioctl_data *)&rq->ifr_data;
+ int ret;
+
+ switch (cmd) {
+ case SIOCGMIIPHY:
+ mii_data->phy_id = data->phy;
+ ret = 0;
+ break;
+
+ case SIOCGMIIREG:
+ spin_lock_irq(&phy_lock);
+ mii_data->val_out =
+ tsi108_read_mii(data, mii_data->reg_num, &ret);
+ spin_unlock_irq(&phy_lock);
+ break;
+
+ default:
+ ret = -EOPNOTSUPP;
+ }
+
+ return ret;
+}
+
+static int
+tsi108_init_one(struct platform_device *pdev)
+{
+ struct net_device *dev = NULL;
+ tsi108_prv_data *data = NULL;
+ hw_info *einfo;
+ int ret;
+
+ einfo = ( hw_info *) pdev->dev.platform_data;
+
+ if (NULL == einfo) {
+ printk(KERN_ERR "tsi-eth %d: Missing additional data!\n",
+ pdev->id);
+ return -ENODEV;
+ }
+
+ /* Create an ethernet device instance */
+
+ dev = alloc_etherdev(sizeof(tsi108_prv_data));
+ if (!dev) {
+ printk("tsi108_eth: Could not allocate a device structure\n");
+ return -ENOMEM;
+ }
+
+ printk("tsi108_eth%d: probe...\n", pdev->id);
+ data = netdev_priv(dev);
+ memset(data, 0, sizeof(tsi108_prv_data));
+
+ DBG("tsi108_eth%d: regs:phyresgs:phy:irq_num=0x%x:0x%x:0x%x:0x%x\n",
+ pdev->id, einfo->regs, einfo->phyregs,
+ einfo->phy, einfo->irq_num);
+
+ data->regs = (u32)ioremap(einfo->regs, 0x400); /*FIX ME */
+ data->phyregs = (u32)ioremap(einfo->phyregs, 0x400); /*FIX ME */
+ data->phy = einfo->phy;
+ data->irq_num = einfo->irq_num;
+ dev->open = tsi108_open;
+ dev->stop = tsi108_close;
+ dev->hard_start_xmit = tsi108_send_packet;
+ dev->set_mac_address = tsi108_set_mac;
+ dev->set_multicast_list = tsi108_set_rx_mode;
+ dev->get_stats = tsi108_get_stats;
+ dev->poll = tsi108_poll;
+ dev->do_ioctl = tsi108_ioctl;
+ dev->weight = 64; /* 64 is more suitable for GigE interface - klai */
+
+ /* Apparently, the Linux networking code won't use scatter-gather
+ * if the hardware doesn't do checksums. However, it's faster
+ * to checksum in place and use SG, as (among other reasons)
+ * the cache won't be dirtied (which then has to be flushed
+ * before DMA). The checksumming is done by the driver (via
+ * a new function skb_csum_dev() in net/core/skbuff.c).
+ */
+
+#ifdef FIXME_SG_CSUM_NOT_TESTED /* FIXME: Not supported now. */
+ dev->features = NETIF_F_HIGHDMA | NETIF_F_SG | NETIF_F_HW_CSUM;
+#else
+ dev->features = NETIF_F_HIGHDMA;
+#endif
+ SET_MODULE_OWNER(dev);
+
+ spin_lock_init(&data->txlock);
+ spin_lock_init(&data->misclock);
+
+ tsi108_reset_ether(data);
+ tsi108_kill_phy(dev);
+
+ if (tsi108_get_mac(dev) != 0)
+ printk(KERN_ERR "%s: Invalid MAC address. Please correct.\n",
+ dev->name);
+
+ tsi108_init_mac(dev);
+ ret = register_netdev(dev);
+ if (ret < 0) {
+ kfree(dev);
+ return ret;
+ }
+
+ printk(KERN_INFO "%s: Tsi108 Gigabit Ethernet, MAC: "
+ "%02x:%02x:%02x:%02x:%02x:%02x\n", dev->name,
+ dev->dev_addr[0], dev->dev_addr[1], dev->dev_addr[2],
+ dev->dev_addr[3], dev->dev_addr[4], dev->dev_addr[5]);
+#ifdef DEBUG
+ dump_eth_one(dev);
+#endif
+
+ return 0;
+}
+
+/* There's no way to either get interrupts from the PHY when
+ * something changes, or to have the Tsi108 automatically communicate
+ * with the PHY to reconfigure itself.
+ *
+ * Thus, we have to do it using a timer.
+ */
+
+static void tsi108_timed_checker(unsigned long dev_ptr)
+{
+ struct net_device *dev = (struct net_device *)dev_ptr;
+ tsi108_prv_data *data = netdev_priv(dev);
+
+ tsi108_check_phy(dev);
+ tsi108_check_rxring(dev);
+ data->timer.expires = jiffies + CHECK_PHY_INTERVAL;
+ add_timer(&data->timer);
+}
+
+static int tsi108_ether_init(void)
+{
+ int ret;
+ ret = platform_driver_register (&tsi_eth_driver);
+ if (ret < 0){
+ printk("tsi108_ether_init: error initializing ethernet "
+ "device\n");
+ return ret;
+ }
+ return 0;
+}
+
+static int tsi108_ether_remove(struct platform_device *pdev)
+{
+ struct net_device *dev = platform_get_drvdata(pdev);
+ tsi108_prv_data *priv = netdev_priv(dev);
+
+ unregister_netdev(dev);
+ tsi108_stop_ethernet(dev);
+ platform_set_drvdata(pdev, NULL);
+ iounmap((void __iomem *)priv->regs);
+ iounmap((void __iomem *)priv->phyregs);
+ free_netdev(dev);
+
+ return 0;
+}
+static void tsi108_ether_exit(void)
+{
+ platform_driver_unregister(&tsi_eth_driver);
+}
+
+module_init(tsi108_ether_init);
+module_exit(tsi108_ether_exit);
+
+MODULE_AUTHOR("Tundra Semiconductor Corporation");
+MODULE_DESCRIPTION("Tsi108 Gigabit Ethernet driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/net/tsi108_eth.h b/drivers/net/tsi108_eth.h
new file mode 100644
index 0000000..cb54f92
--- /dev/null
+++ b/drivers/net/tsi108_eth.h
@@ -0,0 +1,404 @@
+/*
+ * (C) Copyright 2005 Tundra Semiconductor Corp.
+ * Kong Lai, <kong.lai@tundra.com).
+ *
+ * See file CREDITS for list of people who contributed to this
+ * project.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ */
+
+/*
+ * net/tsi108_eth.h - definitions for Tsi108 GIGE network controller.
+ */
+
+#ifndef __TSI108_ETH_H
+#define __TSI108_ETH_H
+
+#include <linux/config.h>
+#include <linux/types.h>
+
+#define TSI108_ETH_WRITE_REG(offset, val) \
+ out_be32((volatile u32 *)(data->regs + offset), val)
+
+#define TSI108_ETH_READ_REG(offset) \
+ in_be32((volatile u32 *)(data->regs + offset))
+
+#define TSI108_ETH_WRITE_PHYREG(offset, val) \
+ out_be32((volatile u32 *)(data->phyregs + offset), val)
+
+#define TSI108_ETH_READ_PHYREG(offset) \
+ in_be32((volatile u32 *)(data->phyregs + offset))
+
+/*
+ * PHY Configuration Options
+ *
+ * NOTE: Enable set of definitions corresponding to your board type
+ */
+#define PHY_MV88E 1 /* Marvel 88Exxxx PHY */
+#define PHY_BCM54XX 2 /* Broardcom BCM54xx PHY */
+#define TSI108_PHY_TYPE PHY_MV88E
+
+/*
+ * TSI108 GIGE port registers
+ */
+
+#define TSI108_ETH_PORT_NUM 2
+#define TSI108_PBM_PORT 2
+#define TSI108_SDRAM_PORT 4
+
+#define TSI108_MAC_CFG1 (0x000)
+#define TSI108_MAC_CFG1_SOFTRST (1 << 31)
+#define TSI108_MAC_CFG1_LOOPBACK (1 << 8)
+#define TSI108_MAC_CFG1_RXEN (1 << 2)
+#define TSI108_MAC_CFG1_TXEN (1 << 0)
+
+#define TSI108_MAC_CFG2 (0x004)
+#define TSI108_MAC_CFG2_DFLT_PREAMBLE (7 << 12)
+#define TSI108_MAC_CFG2_IFACE_MASK (3 << 8)
+#define TSI108_MAC_CFG2_NOGIG (1 << 8)
+#define TSI108_MAC_CFG2_GIG (2 << 8)
+#define TSI108_MAC_CFG2_PADCRC (1 << 2)
+#define TSI108_MAC_CFG2_FULLDUPLEX (1 << 0)
+
+#define TSI108_MAC_MII_MGMT_CFG (0x020)
+#define TSI108_MAC_MII_MGMT_CLK (7 << 0)
+#define TSI108_MAC_MII_MGMT_RST (1 << 31)
+
+#define TSI108_MAC_MII_CMD (0x024)
+#define TSI108_MAC_MII_CMD_READ (1 << 0)
+
+#define TSI108_MAC_MII_ADDR (0x028)
+#define TSI108_MAC_MII_ADDR_REG 0
+#define TSI108_MAC_MII_ADDR_PHY 8
+
+#define TSI108_MAC_MII_DATAOUT (0x02c)
+#define TSI108_MAC_MII_DATAIN (0x030)
+
+#define TSI108_MAC_MII_IND (0x034)
+#define TSI108_MAC_MII_IND_NOTVALID (1 << 2)
+#define TSI108_MAC_MII_IND_SCANNING (1 << 1)
+#define TSI108_MAC_MII_IND_BUSY (1 << 0)
+
+#define TSI108_MAC_IFCTRL (0x038)
+#define TSI108_MAC_IFCTRL_PHYMODE (1 << 24)
+
+#define TSI108_MAC_ADDR1 (0x040)
+#define TSI108_MAC_ADDR2 (0x044)
+
+#define TSI108_STAT_RXBYTES (0x06c)
+#define TSI108_STAT_RXBYTES_CARRY (1 << 24)
+
+#define TSI108_STAT_RXPKTS (0x070)
+#define TSI108_STAT_RXPKTS_CARRY (1 << 18)
+
+#define TSI108_STAT_RXFCS (0x074)
+#define TSI108_STAT_RXFCS_CARRY (1 << 12)
+
+#define TSI108_STAT_RXMCAST (0x078)
+#define TSI108_STAT_RXMCAST_CARRY (1 << 18)
+
+#define TSI108_STAT_RXALIGN (0x08c)
+#define TSI108_STAT_RXALIGN_CARRY (1 << 12)
+
+#define TSI108_STAT_RXLENGTH (0x090)
+#define TSI108_STAT_RXLENGTH_CARRY (1 << 12)
+
+#define TSI108_STAT_RXRUNT (0x09c)
+#define TSI108_STAT_RXRUNT_CARRY (1 << 12)
+
+#define TSI108_STAT_RXJUMBO (0x0a0)
+#define TSI108_STAT_RXJUMBO_CARRY (1 << 12)
+
+#define TSI108_STAT_RXFRAG (0x0a4)
+#define TSI108_STAT_RXFRAG_CARRY (1 << 12)
+
+#define TSI108_STAT_RXJABBER (0x0a8)
+#define TSI108_STAT_RXJABBER_CARRY (1 << 12)
+
+#define TSI108_STAT_RXDROP (0x0ac)
+#define TSI108_STAT_RXDROP_CARRY (1 << 12)
+
+#define TSI108_STAT_TXBYTES (0x0b0)
+#define TSI108_STAT_TXBYTES_CARRY (1 << 24)
+
+#define TSI108_STAT_TXPKTS (0x0b4)
+#define TSI108_STAT_TXPKTS_CARRY (1 << 18)
+
+#define TSI108_STAT_TXEXDEF (0x0c8)
+#define TSI108_STAT_TXEXDEF_CARRY (1 << 12)
+
+#define TSI108_STAT_TXEXCOL (0x0d8)
+#define TSI108_STAT_TXEXCOL_CARRY (1 << 12)
+
+#define TSI108_STAT_TXTCOL (0x0dc)
+#define TSI108_STAT_TXTCOL_CARRY (1 << 13)
+
+#define TSI108_STAT_TXPAUSEDROP (0x0e4)
+#define TSI108_STAT_TXPAUSEDROP_CARRY (1 << 12)
+
+#define TSI108_STAT_CARRY1 (0x100)
+#define TSI108_STAT_CARRY1_RXBYTES (1 << 16)
+#define TSI108_STAT_CARRY1_RXPKTS (1 << 15)
+#define TSI108_STAT_CARRY1_RXFCS (1 << 14)
+#define TSI108_STAT_CARRY1_RXMCAST (1 << 13)
+#define TSI108_STAT_CARRY1_RXALIGN (1 << 8)
+#define TSI108_STAT_CARRY1_RXLENGTH (1 << 7)
+#define TSI108_STAT_CARRY1_RXRUNT (1 << 4)
+#define TSI108_STAT_CARRY1_RXJUMBO (1 << 3)
+#define TSI108_STAT_CARRY1_RXFRAG (1 << 2)
+#define TSI108_STAT_CARRY1_RXJABBER (1 << 1)
+#define TSI108_STAT_CARRY1_RXDROP (1 << 0)
+
+#define TSI108_STAT_CARRY2 (0x104)
+#define TSI108_STAT_CARRY2_TXBYTES (1 << 13)
+#define TSI108_STAT_CARRY2_TXPKTS (1 << 12)
+#define TSI108_STAT_CARRY2_TXEXDEF (1 << 7)
+#define TSI108_STAT_CARRY2_TXEXCOL (1 << 3)
+#define TSI108_STAT_CARRY2_TXTCOL (1 << 2)
+#define TSI108_STAT_CARRY2_TXPAUSE (1 << 0)
+
+#define TSI108_STAT_CARRYMASK1 (0x108)
+#define TSI108_STAT_CARRYMASK2 (0x10c)
+
+#define TSI108_EC_PORTCTRL (0x200)
+#define TSI108_EC_PORTCTRL_STATRST (1 << 31)
+#define TSI108_EC_PORTCTRL_STATEN (1 << 28)
+#define TSI108_EC_PORTCTRL_NOGIG (1 << 18)
+#define TSI108_EC_PORTCTRL_HALFDUPLEX (1 << 16)
+
+#define TSI108_EC_INTSTAT (0x204)
+#define TSI108_EC_INTMASK (0x208)
+
+#define TSI108_INT_ANY (1 << 31)
+#define TSI108_INT_SFN (1 << 30)
+#define TSI108_INT_RXIDLE (1 << 29)
+#define TSI108_INT_RXABORT (1 << 28)
+#define TSI108_INT_RXERROR (1 << 27)
+#define TSI108_INT_RXOVERRUN (1 << 26)
+#define TSI108_INT_RXTHRESH (1 << 25)
+#define TSI108_INT_RXWAIT (1 << 24)
+#define TSI108_INT_RXQUEUE0 (1 << 16)
+#define TSI108_INT_STATCARRY (1 << 15)
+#define TSI108_INT_TXIDLE (1 << 13)
+#define TSI108_INT_TXABORT (1 << 12)
+#define TSI108_INT_TXERROR (1 << 11)
+#define TSI108_INT_TXUNDERRUN (1 << 10)
+#define TSI108_INT_TXTHRESH (1 << 9)
+#define TSI108_INT_TXWAIT (1 << 8)
+#define TSI108_INT_TXQUEUE0 (1 << 0)
+
+#define TSI108_EC_TXCFG (0x220)
+#define TSI108_EC_TXCFG_RST (1 << 31)
+
+#define TSI108_EC_TXCTRL (0x224)
+#define TSI108_EC_TXCTRL_IDLEINT (1 << 31)
+#define TSI108_EC_TXCTRL_ABORT (1 << 30)
+#define TSI108_EC_TXCTRL_GO (1 << 15)
+#define TSI108_EC_TXCTRL_QUEUE0 (1 << 0)
+
+#define TSI108_EC_TXSTAT (0x228)
+#define TSI108_EC_TXSTAT_ACTIVE (1 << 15)
+#define TSI108_EC_TXSTAT_QUEUE0 (1 << 0)
+
+#define TSI108_EC_TXESTAT (0x22c)
+#define TSI108_EC_TXESTAT_Q0_ERR (1 << 24)
+#define TSI108_EC_TXESTAT_Q0_DESCINT (1 << 16)
+#define TSI108_EC_TXESTAT_Q0_EOF (1 << 8)
+#define TSI108_EC_TXESTAT_Q0_EOQ (1 << 0)
+
+#define TSI108_EC_TXERR (0x278)
+
+#define TSI108_EC_TXQ_CFG (0x280)
+#define TSI108_EC_TXQ_CFG_DESC_INT (1 << 20)
+#define TSI108_EC_TXQ_CFG_EOQ_OWN_INT (1 << 19)
+#define TSI108_EC_TXQ_CFG_WSWP (1 << 11)
+#define TSI108_EC_TXQ_CFG_BSWP (1 << 10)
+#define TSI108_EC_TXQ_CFG_SFNPORT 0
+
+#define TSI108_EC_TXQ_BUFCFG (0x284)
+#define TSI108_EC_TXQ_BUFCFG_BURST8 (0 << 8)
+#define TSI108_EC_TXQ_BUFCFG_BURST32 (1 << 8)
+#define TSI108_EC_TXQ_BUFCFG_BURST128 (2 << 8)
+#define TSI108_EC_TXQ_BUFCFG_BURST256 (3 << 8)
+#define TSI108_EC_TXQ_BUFCFG_WSWP (1 << 11)
+#define TSI108_EC_TXQ_BUFCFG_BSWP (1 << 10)
+#define TSI108_EC_TXQ_BUFCFG_SFNPORT 0
+
+#define TSI108_EC_TXQ_PTRLOW (0x288)
+
+#define TSI108_EC_TXQ_PTRHIGH (0x28c)
+#define TSI108_EC_TXQ_PTRHIGH_VALID (1 << 31)
+
+#define TSI108_EC_TXTHRESH (0x230)
+#define TSI108_EC_TXTHRESH_STARTFILL 0
+#define TSI108_EC_TXTHRESH_STOPFILL 16
+
+#define TSI108_EC_RXCFG (0x320)
+#define TSI108_EC_RXCFG_RST (1 << 31)
+
+#define TSI108_EC_RXSTAT (0x328)
+#define TSI108_EC_RXSTAT_ACTIVE (1 << 15)
+#define TSI108_EC_RXSTAT_QUEUE0 (1 << 0)
+
+#define TSI108_EC_RXESTAT (0x32c)
+#define TSI108_EC_RXESTAT_Q0_ERR (1 << 24)
+#define TSI108_EC_RXESTAT_Q0_DESCINT (1 << 16)
+#define TSI108_EC_RXESTAT_Q0_EOF (1 << 8)
+#define TSI108_EC_RXESTAT_Q0_EOQ (1 << 0)
+
+#define TSI108_EC_HASHADDR (0x360)
+#define TSI108_EC_HASHADDR_AUTOINC (1 << 31)
+#define TSI108_EC_HASHADDR_DO1STREAD (1 << 30)
+#define TSI108_EC_HASHADDR_UNICAST (0 << 4)
+#define TSI108_EC_HASHADDR_MCAST (1 << 4)
+
+#define TSI108_EC_HASHDATA (0x364)
+
+#define TSI108_EC_RXQ_PTRLOW (0x388)
+
+#define TSI108_EC_RXQ_PTRHIGH (0x38c)
+#define TSI108_EC_RXQ_PTRHIGH_VALID (1 << 31)
+
+/* Station Enable -- accept packets destined for us */
+#define TSI108_EC_RXCFG_SE (1 << 13)
+/* Unicast Frame Enable -- for packets not destined for us */
+#define TSI108_EC_RXCFG_UFE (1 << 12)
+/* Multicast Frame Enable */
+#define TSI108_EC_RXCFG_MFE (1 << 11)
+/* Broadcast Frame Enable */
+#define TSI108_EC_RXCFG_BFE (1 << 10)
+#define TSI108_EC_RXCFG_UC_HASH (1 << 9)
+#define TSI108_EC_RXCFG_MC_HASH (1 << 8)
+
+#define TSI108_EC_RXQ_CFG (0x380)
+#define TSI108_EC_RXQ_CFG_DESC_INT (1 << 20)
+#define TSI108_EC_RXQ_CFG_EOQ_OWN_INT (1 << 19)
+#define TSI108_EC_RXQ_CFG_WSWP (1 << 11)
+#define TSI108_EC_RXQ_CFG_BSWP (1 << 10)
+#define TSI108_EC_RXQ_CFG_SFNPORT 0
+
+#define TSI108_EC_RXQ_BUFCFG (0x384)
+#define TSI108_EC_RXQ_BUFCFG_BURST8 (0 << 8)
+#define TSI108_EC_RXQ_BUFCFG_BURST32 (1 << 8)
+#define TSI108_EC_RXQ_BUFCFG_BURST128 (2 << 8)
+#define TSI108_EC_RXQ_BUFCFG_BURST256 (3 << 8)
+#define TSI108_EC_RXQ_BUFCFG_WSWP (1 << 11)
+#define TSI108_EC_RXQ_BUFCFG_BSWP (1 << 10)
+#define TSI108_EC_RXQ_BUFCFG_SFNPORT 0
+
+#define TSI108_EC_RXCTRL (0x324)
+#define TSI108_EC_RXCTRL_ABORT (1 << 30)
+#define TSI108_EC_RXCTRL_GO (1 << 15)
+#define TSI108_EC_RXCTRL_QUEUE0 (1 << 0)
+
+#define TSI108_EC_RXERR (0x378)
+
+#define PHY_CTRL 0
+#define PHY_CTRL_RESET (1 << 15)
+#define PHY_CTRL_AUTONEG_EN (1 << 12)
+#define PHY_CTRL_POWERDOWN (1 << 11)
+#define PHY_CTRL_AUTONEG_START (1 << 9)
+
+#define PHY_STAT 1
+#define PHY_STAT_LINKUP (1 << 2)
+
+#if (TSI108_PHY_TYPE == PHY_MV88E)
+/* Marvel 88E1xxx-specific */
+#define PHY_SUM_STAT 0x11
+#define PHY_SUM_STAT_SPEED_MASK (3 << 14)
+#define PHY_SUM_STAT_SPEED_10 (0 << 14)
+#define PHY_SUM_STAT_SPEED_100 (1 << 14)
+#define PHY_SUM_STAT_SPEED_1000 (2 << 14)
+#define PHY_SUM_STAT_FULLDUPLEX (1 << 13)
+
+#define PHY_SUM_STAT_1000T_FD (PHY_SUM_STAT_SPEED_1000 | PHY_SUM_STAT_FULLDUPLEX)
+#define PHY_SUM_STAT_1000T_HD (PHY_SUM_STAT_SPEED_1000)
+#define PHY_SUM_STAT_100TX_FD (PHY_SUM_STAT_SPEED_100 | PHY_SUM_STAT_FULLDUPLEX)
+#define PHY_SUM_STAT_100TX_HD (PHY_SUM_STAT_SPEED_100)
+#define PHY_SUM_STAT_10T_FD (PHY_SUM_STAT_SPEED_10 | PHY_SUM_STAT_FULLDUPLEX)
+#define PHY_SUM_STAT_10T_HD (PHY_SUM_STAT_SPEED_10)
+#elif (TSI108_PHY_TYPE == PHY_BCM54XX)
+/*Broadcom BCM54xx */
+#define PHY_SUM_STAT 0x19
+#define PHY_SUM_STAT_SPEED_MASK (7 << 8) /* Auto Negotiation HCD */
+#define PHY_SUM_STAT_1000T_FD (7 << 8) /* 1000BASE-T Full-Duplex */
+#define PHY_SUM_STAT_1000T_HD (6 << 8) /* 1000BASE-T Half-Duplex */
+#define PHY_SUM_STAT_100TX_FD (5 << 8) /* 100BASE-TX Full-Duplex */
+#define PHY_SUM_STAT_100T4 (4 << 8) /* 100BASE-T4 */
+#define PHY_SUM_STAT_100TX_HD (3 << 8) /* 100BASE-TX Half-Duplex */
+#define PHY_SUM_STAT_10T_FD (2 << 8) /* 10BASE-T Full-Duplex */
+#define PHY_SUM_STAT_10T_HD (1 << 8) /* 10BASE-T Half-Duplex */
+#define PHY_SUM_STAT_AN_FAIL (0 << 8) /* 1000BASE-T Half-Duplex */
+#else
+#error "PHY Device not specified"
+#endif /* MV88_PHY */
+
+#define TSI108_TX_EOF (1 << 0) /* End of frame; last fragment of packet */
+#define TSI108_TX_SOF (1 << 1) /* Start of frame; first frag. of packet */
+#define TSI108_TX_VLAN (1 << 2) /* Per-frame VLAN: enables VLAN override */
+#define TSI108_TX_HUGE (1 << 3) /* Huge frame enable */
+#define TSI108_TX_PAD (1 << 4) /* Pad the packet if too short */
+#define TSI108_TX_CRC (1 << 5) /* Generate CRC for this packet */
+#define TSI108_TX_INT (1 << 14) /* Generate an IRQ after frag. processed */
+#define TSI108_TX_RETRY 16 /* 4 bit field indicating num. of retries */
+#define TSI108_TX_COL (1 << 20) /* Set if a collision occured */
+#define TSI108_TX_LCOL (1 << 24) /* Set if a late collision occured */
+#define TSI108_TX_UNDER (1 << 25) /* Set if a FIFO underrun occured */
+#define TSI108_TX_RLIM (1 << 26) /* Set if the retry limit was reached */
+#define TSI108_TX_OK (1 << 30) /* Set if the frame TX was successful */
+#define TSI108_TX_OWN (1 << 31) /* Set if the device owns the descriptor */
+
+/* Note: the descriptor layouts assume big-endian byte order. */
+typedef struct {
+ u32 buf0;
+ u32 buf1; /* Base address of buffer */
+ u32 next0; /* Address of next descriptor, if any */
+ u32 next1;
+ u16 vlan; /* VLAN, if override enabled for this packet */
+ u16 len; /* Length of buffer in bytes */
+ u32 misc; /* See TSI108_TX_* above */
+ u32 reserved0; /*reserved0 and reserved1 are added to make the desc */
+ u32 reserved1; /* 32-byte aligned */
+} __attribute__ ((aligned(32))) tx_desc;
+
+#define TSI108_RX_EOF (1 << 0) /* End of frame; last fragment of packet */
+#define TSI108_RX_SOF (1 << 1) /* Start of frame; first frag. of packet */
+#define TSI108_RX_VLAN (1 << 2) /* Set on SOF if packet has a VLAN */
+#define TSI108_RX_FTYPE (1 << 3) /* Length/Type field is type, not length */
+#define TSI108_RX_RUNT (1 << 4)/* Packet is less than minimum size */
+#define TSI108_RX_HASH (1 << 7)/* Hash table match */
+#define TSI108_RX_BAD (1 << 8) /* Bad frame */
+#define TSI108_RX_OVER (1 << 9) /* FIFO overrun occured */
+#define TSI108_RX_TRUNC (1 << 11) /* Packet truncated due to excess length */
+#define TSI108_RX_CRC (1 << 12) /* Packet had a CRC error */
+#define TSI108_RX_INT (1 << 13) /* Generate an IRQ after frag. processed */
+#define TSI108_RX_OWN (1 << 15) /* Set if the device owns the descriptor */
+
+typedef struct {
+ u32 buf0; /* Base address of buffer */
+ u32 buf1; /* Base address of buffer */
+ u32 next0; /* Address of next descriptor, if any */
+ u32 next1; /* Address of next descriptor, if any */
+ u16 vlan; /* VLAN of received packet, first frag only */
+ u16 len; /* Length of received fragment in bytes */
+ u16 blen; /* Length of buffer in bytes */
+ u16 misc; /* See TSI108_RX_* above */
+ u32 reserved0; /* reserved0 and reserved1 are added to make the desc */
+ u32 reserved1; /* 32-byte aligned */
+} __attribute__ ((aligned(32))) rx_desc;
+
+#endif /* __TSI108_ETH_H */
^ permalink raw reply related
* Re: Using netconsole for debugging suspend/resume
From: Christoph Hellwig @ 2006-06-13 7:18 UTC (permalink / raw)
To: David Miller; +Cc: ak, jeremy, lkml, mpm, linux-kernel, netdev
In-Reply-To: <20060612.220346.71553967.davem@davemloft.net>
On Mon, Jun 12, 2006 at 10:03:46PM -0700, David Miller wrote:
> From: Andi Kleen <ak@suse.de>
> Date: Tue, 13 Jun 2006 06:54:14 +0200
>
> > I guess if you use 1394 with remote DMA for other protocols (like
> > video etc.) there must be some way for the subsystem to map
> > the memory even on IOMMU systems. I admit I haven't dived that
> > deeply into the 1394 subsystem so I don't know how that works.
>
> Video-1394 has it's own driver, which does a consistent DMA
> allocation, and then maps that into userspace using remap_pfn_range().
> Entirely portable.
That's actually not portable to certain arm platforms, but that's
a different story.
^ permalink raw reply
* Re: Using netconsole for debugging suspend/resume
From: David Miller @ 2006-06-13 7:31 UTC (permalink / raw)
To: hch; +Cc: ak, jeremy, lkml, mpm, linux-kernel, netdev
In-Reply-To: <20060613071819.GB16358@infradead.org>
From: Christoph Hellwig <hch@infradead.org>
Date: Tue, 13 Jun 2006 08:18:19 +0100
> That's actually not portable to certain arm platforms, but that's
> a different story.
Yes, cache issues :-/
^ permalink raw reply
* [RFT 1/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
In-Reply-To: <20060613081726.695812000@localhost.localdomain>
[-- Attachment #1: sky2-set-power-state.patch --]
[-- Type: text/plain, Size: 1932 bytes --]
Subject: [RFT 1/5] sky2: set_power_state should be void
The set power state function is cleaner if it doesn't return anything.
The only caller that could fail is in suspend() and it can check the argument
there.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -187,12 +187,11 @@ static u16 gm_phy_read(struct sky2_hw *h
return v;
}
-static int sky2_set_power_state(struct sky2_hw *hw, pci_power_t state)
+static void sky2_set_power_state(struct sky2_hw *hw, pci_power_t state)
{
u16 power_control;
u32 reg1;
int vaux;
- int ret = 0;
pr_debug("sky2_set_power_state %d\n", state);
sky2_write8(hw, B2_TST_CTRL1, TST_CFG_WRITE_ON);
@@ -275,12 +274,10 @@ static int sky2_set_power_state(struct s
break;
default:
printk(KERN_ERR PFX "Unknown power state %d\n", state);
- ret = -1;
}
sky2_pci_write16(hw, hw->pm_cap + PCI_PM_CTRL, power_control);
sky2_write8(hw, B2_TST_CTRL1, TST_CFG_WRITE_OFF);
- return ret;
}
static void sky2_phy_reset(struct sky2_hw *hw, unsigned port)
@@ -3428,6 +3425,10 @@ static int sky2_suspend(struct pci_dev *
{
struct sky2_hw *hw = pci_get_drvdata(pdev);
int i;
+ pci_power_t pstate = pci_choose_state(pdev, state);
+
+ if (!(pstate == PCI_D3hot || pstate == PCI_D3cold))
+ return -EINVAL;
for (i = 0; i < 2; i++) {
struct net_device *dev = hw->dev[i];
@@ -3442,7 +3443,8 @@ static int sky2_suspend(struct pci_dev *
}
pci_save_state(pdev);
- return sky2_set_power_state(hw, pci_choose_state(pdev, state));
+ sky2_set_power_state(hw, pstate);
+ return 0;
}
static int sky2_resume(struct pci_dev *pdev)
@@ -3452,9 +3454,7 @@ static int sky2_resume(struct pci_dev *p
pci_restore_state(pdev);
pci_enable_wake(pdev, PCI_D0, 0);
- err = sky2_set_power_state(hw, PCI_D0);
- if (err)
- goto out;
+ sky2_set_power_state(hw, PCI_D0);
err = sky2_reset(hw);
if (err)
--
^ permalink raw reply
* [RFT 0/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
There were a several problems buried in suspend/resume. The real
failure was caused by the idle timer not being stopped/restarted.
But several other races, and cleanups were needed.
Since I don't have a machine that will suspend successfully with
that hardware, I can't test it.
--
^ permalink raw reply
* [RFT 5/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
In-Reply-To: <20060613081726.695812000@localhost.localdomain>
[-- Attachment #1: sky2-resume.patch --]
[-- Type: text/plain, Size: 1685 bytes --]
Subject: [RFT 5/5] sky2: stop/start hardware idle timer on suspend/resume
The resume bug was cause not by an early interrupt but because
the idle timeout was not being stopped on suspend. Also disable hardware
IRQ's on suspend. Will need to revisit this with hotplug?
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -2161,6 +2161,13 @@ static void sky2_descriptor_error(struct
/* If idle then force a fake soft NAPI poll once a second
* to work around cases where sharing an edge triggered interrupt.
*/
+static inline void sky2_idle_start(struct sky2_hw *hw)
+{
+ if (idle_timeout > 0)
+ mod_timer(&hw->idle_timer,
+ jiffies + msecs_to_jiffies(idle_timeout));
+}
+
static void sky2_idle(unsigned long arg)
{
struct sky2_hw *hw = (struct sky2_hw *) arg;
@@ -3350,9 +3357,7 @@ static int __devinit sky2_probe(struct p
sky2_write32(hw, B0_IMSK, Y2_IS_BASE);
setup_timer(&hw->idle_timer, sky2_idle, (unsigned long) hw);
- if (idle_timeout > 0)
- mod_timer(&hw->idle_timer,
- jiffies + msecs_to_jiffies(idle_timeout));
+ sky2_idle_start(hw);
pci_set_drvdata(pdev, hw);
@@ -3430,6 +3435,8 @@ static int sky2_suspend(struct pci_dev *
if (!(pstate == PCI_D3hot || pstate == PCI_D3cold))
return -EINVAL;
+ del_timer_sync(&hw->idle_timer);
+
for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
@@ -3472,10 +3479,12 @@ static int sky2_resume(struct pci_dev *p
printk(KERN_ERR PFX "%s: could not up: %d\n",
dev->name, err);
dev_close(dev);
- break;
+ goto out;
}
}
}
+
+ sky2_idle_start(hw);
out:
return err;
}
--
^ permalink raw reply
* [RFT 4/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
In-Reply-To: <20060613081726.695812000@localhost.localdomain>
[-- Attachment #1: sky2-suspend-nohwirq.patch --]
[-- Type: text/plain, Size: 725 bytes --]
Subject: [RFT 4/5] sky2: save/restore base hardware irq during suspend/resume
The hardware should be fully shut off during suspend, and the base
irq mask restored during resume.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -3442,6 +3442,7 @@ static int sky2_suspend(struct pci_dev *
}
}
+ sky2_write32(hw, B0_IMSK, 0);
pci_save_state(pdev);
sky2_set_power_state(hw, pstate);
return 0;
@@ -3460,6 +3461,8 @@ static int sky2_resume(struct pci_dev *p
if (err)
goto out;
+ sky2_write32(hw, B0_IMSK, Y2_IS_BASE);
+
for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
if (dev && netif_running(dev)) {
--
^ permalink raw reply
* [RFT 2/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
In-Reply-To: <20060613081726.695812000@localhost.localdomain>
[-- Attachment #1: sky2-ports.patch --]
[-- Type: text/plain, Size: 761 bytes --]
Subject: [RFT 2/5] sky2: don't hard code number of ports
It is cleaner, to not loop over both ports if only one exists.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -3430,7 +3430,7 @@ static int sky2_suspend(struct pci_dev *
if (!(pstate == PCI_D3hot || pstate == PCI_D3cold))
return -EINVAL;
- for (i = 0; i < 2; i++) {
+ for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
if (dev) {
@@ -3460,7 +3460,7 @@ static int sky2_resume(struct pci_dev *p
if (err)
goto out;
- for (i = 0; i < 2; i++) {
+ for (i = 0; i < hw->ports; i++) {
struct net_device *dev = hw->dev[i];
if (dev && netif_running(dev)) {
netif_device_attach(dev);
--
^ permalink raw reply
* [RFT 3/5] sky2: suspend/resume patchlets
From: Stephen Hemminger @ 2006-06-13 8:17 UTC (permalink / raw)
To: Jeff Garzik; +Cc: linus, netdev
In-Reply-To: <20060613081726.695812000@localhost.localdomain>
[-- Attachment #1: sky2-poll-complete.patch --]
[-- Type: text/plain, Size: 700 bytes --]
Subject: [RFT 3/5] sky2: fix hotplug detect during poll
Linus, doesn't understand NAPI.
If the poll routine detects no hardware available, it needs to dequeue
it self from the network poll list.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
--- test.orig/drivers/net/sky2.c
+++ test/drivers/net/sky2.c
@@ -2181,7 +2181,7 @@ static int sky2_poll(struct net_device *
u32 status = sky2_read32(hw, B0_Y2_SP_EISR);
if (!~status)
- return 0;
+ goto out;
if (status & Y2_IS_HW_ERR)
sky2_hw_intr(hw);
@@ -2219,7 +2219,7 @@ static int sky2_poll(struct net_device *
if (sky2_more_work(hw))
return 1;
-
+out:
netif_rx_complete(dev0);
sky2_read32(hw, B0_Y2_SP_LISR);
--
^ permalink raw reply
* Re: Problem authenticating using WPA with bcm43xx-softmac
From: Johannes Berg @ 2006-06-13 8:40 UTC (permalink / raw)
To: Larry Finger; +Cc: netdev
In-Reply-To: <448CBF43.6000200@lwfinger.net>
[-- Attachment #1: Type: text/plain, Size: 215 bytes --]
> I expect that softmac should be listening to the driver as to whether this capability is available;
> however, I'm now up and running once again.
Yeah, I think a patch was floating around too!
johannes
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 793 bytes --]
^ 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