linux/drivers/net/tokenring/3c359.c

1832 lines
58 KiB
C
Raw Normal View History

/*
* 3c359.c (c) 2000 Mike Phillips (mikep@linuxtr.net) All Rights Reserved
*
* Linux driver for 3Com 3c359 Tokenlink Velocity XL PCI NIC
*
* Base Driver Olympic:
* Written 1999 Peter De Schrijver & Mike Phillips
*
* This software may be used and distributed according to the terms
* of the GNU General Public License, incorporated herein by reference.
*
* 7/17/00 - Clean up, version number 0.9.0. Ready to release to the world.
*
* 2/16/01 - Port up to kernel 2.4.2 ready for submission into the kernel.
* 3/05/01 - Last clean up stuff before submission.
* 2/15/01 - Finally, update to new pci api.
*
* To Do:
*/
/*
* Technical Card Details
*
* All access to data is done with 16/8 bit transfers. The transfer
* method really sucks. You can only read or write one location at a time.
*
* Also, the microcode for the card must be uploaded if the card does not have
* the flashrom on board. This is a 28K bloat in the driver when compiled
* as a module.
*
* Rx is very simple, status into a ring of descriptors, dma data transfer,
* interrupts to tell us when a packet is received.
*
* Tx is a little more interesting. Similar scenario, descriptor and dma data
* transfers, but we don't have to interrupt the card to tell it another packet
* is ready for transmission, we are just doing simple memory writes, not io or mmio
* writes. The card can be set up to simply poll on the next
* descriptor pointer and when this value is non-zero will automatically download
* the next packet. The card then interrupts us when the packet is done.
*
*/
#define XL_DEBUG 0
#include <linux/jiffies.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/timer.h>
#include <linux/in.h>
#include <linux/ioport.h>
#include <linux/string.h>
#include <linux/proc_fs.h>
#include <linux/ptrace.h>
#include <linux/skbuff.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <linux/netdevice.h>
#include <linux/trdevice.h>
#include <linux/stddef.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/spinlock.h>
#include <linux/bitops.h>
#include <linux/firmware.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 09:04:11 +01:00
#include <linux/slab.h>
#include <net/checksum.h>
#include <asm/io.h>
#include "3c359.h"
static char version[] __devinitdata =
"3c359.c v1.2.0 2/17/01 - Mike Phillips (mikep@linuxtr.net)" ;
#define FW_NAME "3com/3C359.bin"
MODULE_AUTHOR("Mike Phillips <mikep@linuxtr.net>") ;
MODULE_DESCRIPTION("3Com 3C359 Velocity XL Token Ring Adapter Driver\n") ;
MODULE_FIRMWARE(FW_NAME);
/* Module parameters */
/* Ring Speed 0,4,16
* 0 = Autosense
* 4,16 = Selected speed only, no autosense
* This allows the card to be the first on the ring
* and become the active monitor.
*
* WARNING: Some hubs will allow you to insert
* at the wrong speed.
*
* The adapter will _not_ fail to open if there are no
* active monitors on the ring, it will simply open up in
* its last known ringspeed if no ringspeed is specified.
*/
static int ringspeed[XL_MAX_ADAPTERS] = {0,} ;
module_param_array(ringspeed, int, NULL, 0);
MODULE_PARM_DESC(ringspeed,"3c359: Ringspeed selection - 4,16 or 0") ;
/* Packet buffer size */
static int pkt_buf_sz[XL_MAX_ADAPTERS] = {0,} ;
module_param_array(pkt_buf_sz, int, NULL, 0) ;
MODULE_PARM_DESC(pkt_buf_sz,"3c359: Initial buffer size") ;
/* Message Level */
static int message_level[XL_MAX_ADAPTERS] = {0,} ;
module_param_array(message_level, int, NULL, 0) ;
MODULE_PARM_DESC(message_level, "3c359: Level of reported messages") ;
/*
* This is a real nasty way of doing this, but otherwise you
* will be stuck with 1555 lines of hex #'s in the code.
*/
static DEFINE_PCI_DEVICE_TABLE(xl_pci_tbl) =
{
{PCI_VENDOR_ID_3COM,PCI_DEVICE_ID_3COM_3C359, PCI_ANY_ID, PCI_ANY_ID, },
{ } /* terminate list */
};
MODULE_DEVICE_TABLE(pci,xl_pci_tbl) ;
static int xl_init(struct net_device *dev);
static int xl_open(struct net_device *dev);
static int xl_open_hw(struct net_device *dev) ;
static int xl_hw_reset(struct net_device *dev);
static netdev_tx_t xl_xmit(struct sk_buff *skb, struct net_device *dev);
static void xl_dn_comp(struct net_device *dev);
static int xl_close(struct net_device *dev);
static void xl_set_rx_mode(struct net_device *dev);
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 15:55:46 +02:00
static irqreturn_t xl_interrupt(int irq, void *dev_id);
static int xl_set_mac_address(struct net_device *dev, void *addr) ;
static void xl_arb_cmd(struct net_device *dev);
static void xl_asb_cmd(struct net_device *dev) ;
static void xl_srb_cmd(struct net_device *dev, int srb_cmd) ;
static void xl_wait_misr_flags(struct net_device *dev) ;
static int xl_change_mtu(struct net_device *dev, int mtu);
static void xl_srb_bh(struct net_device *dev) ;
static void xl_asb_bh(struct net_device *dev) ;
static void xl_reset(struct net_device *dev) ;
static void xl_freemem(struct net_device *dev) ;
/* EEProm Access Functions */
static u16 xl_ee_read(struct net_device *dev, int ee_addr) ;
static void xl_ee_write(struct net_device *dev, int ee_addr, u16 ee_value) ;
/* Debugging functions */
#if XL_DEBUG
static void print_tx_state(struct net_device *dev) ;
static void print_rx_state(struct net_device *dev) ;
static void print_tx_state(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
struct xl_tx_desc *txd ;
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
int i ;
printk("tx_ring_head: %d, tx_ring_tail: %d, free_ent: %d\n",xl_priv->tx_ring_head,
xl_priv->tx_ring_tail, xl_priv->free_ring_entries) ;
printk("Ring , Address , FSH , DnNextPtr, Buffer, Buffer_Len\n");
for (i = 0; i < 16; i++) {
txd = &(xl_priv->xl_tx_ring[i]) ;
printk("%d, %08lx, %08x, %08x, %08x, %08x\n", i, virt_to_bus(txd),
txd->framestartheader, txd->dnnextptr, txd->buffer, txd->buffer_length ) ;
}
printk("DNLISTPTR = %04x\n", readl(xl_mmio + MMIO_DNLISTPTR) );
printk("DmaCtl = %04x\n", readl(xl_mmio + MMIO_DMA_CTRL) );
printk("Queue status = %0x\n",netif_running(dev) ) ;
}
static void print_rx_state(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
struct xl_rx_desc *rxd ;
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
int i ;
printk("rx_ring_tail: %d\n", xl_priv->rx_ring_tail);
printk("Ring , Address , FrameState , UPNextPtr, FragAddr, Frag_Len\n");
for (i = 0; i < 16; i++) {
/* rxd = (struct xl_rx_desc *)xl_priv->rx_ring_dma_addr + (i * sizeof(struct xl_rx_desc)) ; */
rxd = &(xl_priv->xl_rx_ring[i]) ;
printk("%d, %08lx, %08x, %08x, %08x, %08x\n", i, virt_to_bus(rxd),
rxd->framestatus, rxd->upnextptr, rxd->upfragaddr, rxd->upfraglen ) ;
}
printk("UPLISTPTR = %04x\n", readl(xl_mmio + MMIO_UPLISTPTR));
printk("DmaCtl = %04x\n", readl(xl_mmio + MMIO_DMA_CTRL));
printk("Queue status = %0x\n",netif_running(dev));
}
#endif
/*
* Read values from the on-board EEProm. This looks very strange
* but you have to wait for the EEProm to get/set the value before
* passing/getting the next value from the nic. As with all requests
* on this nic it has to be done in two stages, a) tell the nic which
* memory address you want to access and b) pass/get the value from the nic.
* With the EEProm, you have to wait before and between access a) and b).
* As this is only read at initialization time and the wait period is very
* small we shouldn't have to worry about scheduling issues.
*/
static u16 xl_ee_read(struct net_device *dev, int ee_addr)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
/* Wait for EEProm to not be busy */
writel(IO_WORD_READ | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while ( readw(xl_mmio + MMIO_MACDATA) & EEBUSY ) ;
/* Tell EEProm what we want to do and where */
writel(IO_WORD_WRITE | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(EEREAD + ee_addr, xl_mmio + MMIO_MACDATA) ;
/* Wait for EEProm to not be busy */
writel(IO_WORD_READ | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while ( readw(xl_mmio + MMIO_MACDATA) & EEBUSY ) ;
/* Tell EEProm what we want to do and where */
writel(IO_WORD_WRITE | EECONTROL , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(EEREAD + ee_addr, xl_mmio + MMIO_MACDATA) ;
/* Finally read the value from the EEProm */
writel(IO_WORD_READ | EEDATA , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
return readw(xl_mmio + MMIO_MACDATA) ;
}
/*
* Write values to the onboard eeprom. As with eeprom read you need to
* set which location to write, wait, value to write, wait, with the
* added twist of having to enable eeprom writes as well.
*/
static void xl_ee_write(struct net_device *dev, int ee_addr, u16 ee_value)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
/* Wait for EEProm to not be busy */
writel(IO_WORD_READ | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while ( readw(xl_mmio + MMIO_MACDATA) & EEBUSY ) ;
/* Enable write/erase */
writel(IO_WORD_WRITE | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(EE_ENABLE_WRITE, xl_mmio + MMIO_MACDATA) ;
/* Wait for EEProm to not be busy */
writel(IO_WORD_READ | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while ( readw(xl_mmio + MMIO_MACDATA) & EEBUSY ) ;
/* Put the value we want to write into EEDATA */
writel(IO_WORD_WRITE | EEDATA, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(ee_value, xl_mmio + MMIO_MACDATA) ;
/* Tell EEProm to write eevalue into ee_addr */
writel(IO_WORD_WRITE | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(EEWRITE + ee_addr, xl_mmio + MMIO_MACDATA) ;
/* Wait for EEProm to not be busy, to ensure write gets done */
writel(IO_WORD_READ | EECONTROL, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while ( readw(xl_mmio + MMIO_MACDATA) & EEBUSY ) ;
return ;
}
static const struct net_device_ops xl_netdev_ops = {
.ndo_open = xl_open,
.ndo_stop = xl_close,
.ndo_start_xmit = xl_xmit,
.ndo_change_mtu = xl_change_mtu,
.ndo_set_rx_mode = xl_set_rx_mode,
.ndo_set_mac_address = xl_set_mac_address,
};
static int __devinit xl_probe(struct pci_dev *pdev,
const struct pci_device_id *ent)
{
struct net_device *dev ;
struct xl_private *xl_priv ;
static int card_no = -1 ;
int i ;
card_no++ ;
if (pci_enable_device(pdev)) {
return -ENODEV ;
}
pci_set_master(pdev);
if ((i = pci_request_regions(pdev,"3c359"))) {
return i ;
}
/*
* Allowing init_trdev to allocate the private data will align
* xl_private on a 32 bytes boundary which we need for the rx/tx
* descriptors
*/
dev = alloc_trdev(sizeof(struct xl_private)) ;
if (!dev) {
pci_release_regions(pdev) ;
return -ENOMEM ;
}
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
xl_priv = netdev_priv(dev);
#if XL_DEBUG
printk("pci_device: %p, dev:%p, dev->priv: %p, ba[0]: %10x, ba[1]:%10x\n",
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
pdev, dev, netdev_priv(dev), (unsigned int)pdev->resource[0].start, (unsigned int)pdev->resource[1].start);
#endif
dev->irq=pdev->irq;
dev->base_addr=pci_resource_start(pdev,0) ;
xl_priv->xl_card_name = pci_name(pdev);
xl_priv->xl_mmio=ioremap(pci_resource_start(pdev,1), XL_IO_SPACE);
xl_priv->pdev = pdev ;
if ((pkt_buf_sz[card_no] < 100) || (pkt_buf_sz[card_no] > 18000) )
xl_priv->pkt_buf_sz = PKT_BUF_SZ ;
else
xl_priv->pkt_buf_sz = pkt_buf_sz[card_no] ;
dev->mtu = xl_priv->pkt_buf_sz - TR_HLEN ;
xl_priv->xl_ring_speed = ringspeed[card_no] ;
xl_priv->xl_message_level = message_level[card_no] ;
xl_priv->xl_functional_addr[0] = xl_priv->xl_functional_addr[1] = xl_priv->xl_functional_addr[2] = xl_priv->xl_functional_addr[3] = 0 ;
xl_priv->xl_copy_all_options = 0 ;
if((i = xl_init(dev))) {
iounmap(xl_priv->xl_mmio) ;
free_netdev(dev) ;
pci_release_regions(pdev) ;
return i ;
}
dev->netdev_ops = &xl_netdev_ops;
SET_NETDEV_DEV(dev, &pdev->dev);
pci_set_drvdata(pdev,dev) ;
if ((i = register_netdev(dev))) {
printk(KERN_ERR "3C359, register netdev failed\n") ;
pci_set_drvdata(pdev,NULL) ;
iounmap(xl_priv->xl_mmio) ;
free_netdev(dev) ;
pci_release_regions(pdev) ;
return i ;
}
printk(KERN_INFO "3C359: %s registered as: %s\n",xl_priv->xl_card_name,dev->name) ;
return 0;
}
static int xl_init_firmware(struct xl_private *xl_priv)
{
int err;
err = request_firmware(&xl_priv->fw, FW_NAME, &xl_priv->pdev->dev);
if (err) {
printk(KERN_ERR "Failed to load firmware \"%s\"\n", FW_NAME);
return err;
}
if (xl_priv->fw->size < 16) {
printk(KERN_ERR "Bogus length %zu in \"%s\"\n",
xl_priv->fw->size, FW_NAME);
release_firmware(xl_priv->fw);
err = -EINVAL;
}
return err;
}
static int __devinit xl_init(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
int err;
printk(KERN_INFO "%s\n", version);
printk(KERN_INFO "%s: I/O at %hx, MMIO at %p, using irq %d\n",
xl_priv->xl_card_name, (unsigned int)dev->base_addr ,xl_priv->xl_mmio, dev->irq);
spin_lock_init(&xl_priv->xl_lock) ;
err = xl_init_firmware(xl_priv);
if (err == 0)
err = xl_hw_reset(dev);
return err;
}
/*
* Hardware reset. This needs to be a separate entity as we need to reset the card
* when we change the EEProm settings.
*/
static int xl_hw_reset(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
unsigned long t ;
u16 i ;
u16 result_16 ;
u8 result_8 ;
u16 start ;
int j ;
if (xl_priv->fw == NULL)
return -EINVAL;
/*
* Reset the card. If the card has got the microcode on board, we have
* missed the initialization interrupt, so we must always do this.
*/
writew( GLOBAL_RESET, xl_mmio + MMIO_COMMAND ) ;
/*
* Must wait for cmdInProgress bit (12) to clear before continuing with
* card configuration.
*/
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 40 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL card not responding to global reset.\n", dev->name);
return -ENODEV;
}
}
/*
* Enable pmbar by setting bit in CPAttention
*/
writel( (IO_BYTE_READ | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
result_8 = readb(xl_mmio + MMIO_MACDATA) ;
result_8 = result_8 | CPA_PMBARVIS ;
writel( (IO_BYTE_WRITE | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(result_8, xl_mmio + MMIO_MACDATA) ;
/*
* Read cpHold bit in pmbar, if cleared we have got Flashrom on board.
* If not, we need to upload the microcode to the card
*/
writel( (IO_WORD_READ | PMBAR),xl_mmio + MMIO_MAC_ACCESS_CMD);
#if XL_DEBUG
printk(KERN_INFO "Read from PMBAR = %04x\n", readw(xl_mmio + MMIO_MACDATA));
#endif
if ( readw( (xl_mmio + MMIO_MACDATA)) & PMB_CPHOLD ) {
/* Set PmBar, privateMemoryBase bits (8:2) to 0 */
writel( (IO_WORD_READ | PMBAR),xl_mmio + MMIO_MAC_ACCESS_CMD);
result_16 = readw(xl_mmio + MMIO_MACDATA) ;
result_16 = result_16 & ~((0x7F) << 2) ;
writel( (IO_WORD_WRITE | PMBAR), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(result_16,xl_mmio + MMIO_MACDATA) ;
/* Set CPAttention, memWrEn bit */
writel( (IO_BYTE_READ | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
result_8 = readb(xl_mmio + MMIO_MACDATA) ;
result_8 = result_8 | CPA_MEMWREN ;
writel( (IO_BYTE_WRITE | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(result_8, xl_mmio + MMIO_MACDATA) ;
/*
* Now to write the microcode into the shared ram
* The microcode must finish at position 0xFFFF,
* so we must subtract to get the start position for the code
*
* Looks strange but ensures compiler only uses
* 16 bit unsigned int
*/
start = (0xFFFF - (xl_priv->fw->size) + 1) ;
printk(KERN_INFO "3C359: Uploading Microcode: ");
for (i = start, j = 0; j < xl_priv->fw->size; i++, j++) {
writel(MEM_BYTE_WRITE | 0XD0000 | i,
xl_mmio + MMIO_MAC_ACCESS_CMD);
writeb(xl_priv->fw->data[j], xl_mmio + MMIO_MACDATA);
if (j % 1024 == 0)
printk(".");
}
printk("\n") ;
for (i = 0; i < 16; i++) {
writel((MEM_BYTE_WRITE | 0xDFFF0) + i,
xl_mmio + MMIO_MAC_ACCESS_CMD);
writeb(xl_priv->fw->data[xl_priv->fw->size - 16 + i],
xl_mmio + MMIO_MACDATA);
}
/*
* Have to write the start address of the upload to FFF4, but
* the address must be >> 4. You do not want to know how long
* it took me to discover this.
*/
writel(MEM_WORD_WRITE | 0xDFFF4, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(start >> 4, xl_mmio + MMIO_MACDATA);
/* Clear the CPAttention, memWrEn Bit */
writel( (IO_BYTE_READ | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
result_8 = readb(xl_mmio + MMIO_MACDATA) ;
result_8 = result_8 & ~CPA_MEMWREN ;
writel( (IO_BYTE_WRITE | CPATTENTION), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(result_8, xl_mmio + MMIO_MACDATA) ;
/* Clear the cpHold bit in pmbar */
writel( (IO_WORD_READ | PMBAR),xl_mmio + MMIO_MAC_ACCESS_CMD);
result_16 = readw(xl_mmio + MMIO_MACDATA) ;
result_16 = result_16 & ~PMB_CPHOLD ;
writel( (IO_WORD_WRITE | PMBAR), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(result_16,xl_mmio + MMIO_MACDATA) ;
} /* If microcode upload required */
/*
* The card should now go though a self test procedure and get itself ready
* to be opened, we must wait for an srb response with the initialization
* information.
*/
#if XL_DEBUG
printk(KERN_INFO "%s: Microcode uploaded, must wait for the self test to complete\n", dev->name);
#endif
writew(SETINDENABLE | 0xFFF, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while ( !(readw(xl_mmio + MMIO_INTSTATUS_AUTO) & INTSTAT_SRB) ) {
schedule();
if (time_after(jiffies, t + 15 * HZ)) {
printk(KERN_ERR "3COM 3C359 Velocity XL card not responding.\n");
return -ENODEV;
}
}
/*
* Write the RxBufArea with D000, RxEarlyThresh, TxStartThresh,
* DnPriReqThresh, read the tech docs if you want to know what
* values they need to be.
*/
writel(MMIO_WORD_WRITE | RXBUFAREA, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(0xD000, xl_mmio + MMIO_MACDATA) ;
writel(MMIO_WORD_WRITE | RXEARLYTHRESH, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(0X0020, xl_mmio + MMIO_MACDATA) ;
writew( SETTXSTARTTHRESH | 0x40 , xl_mmio + MMIO_COMMAND) ;
writeb(0x04, xl_mmio + MMIO_DNBURSTTHRESH) ;
writeb(0x04, xl_mmio + DNPRIREQTHRESH) ;
/*
* Read WRBR to provide the location of the srb block, have to use byte reads not word reads.
* Tech docs have this wrong !!!!
*/
writel(MMIO_BYTE_READ | WRBR, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
xl_priv->srb = readb(xl_mmio + MMIO_MACDATA) << 8 ;
writel( (MMIO_BYTE_READ | WRBR) + 1, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
xl_priv->srb = xl_priv->srb | readb(xl_mmio + MMIO_MACDATA) ;
#if XL_DEBUG
writel(IO_WORD_READ | SWITCHSETTINGS, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if ( readw(xl_mmio + MMIO_MACDATA) & 2) {
printk(KERN_INFO "Default ring speed 4 mbps\n");
} else {
printk(KERN_INFO "Default ring speed 16 mbps\n");
}
printk(KERN_INFO "%s: xl_priv->srb = %04x\n",xl_priv->xl_card_name, xl_priv->srb);
#endif
return 0;
}
static int xl_open(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
u8 i ;
__le16 hwaddr[3] ; /* Should be u8[6] but we get word return values */
int open_err ;
u16 switchsettings, switchsettings_eeprom ;
if (request_irq(dev->irq, xl_interrupt, IRQF_SHARED , "3c359", dev))
return -EAGAIN;
/*
* Read the information from the EEPROM that we need.
*/
hwaddr[0] = cpu_to_le16(xl_ee_read(dev,0x10));
hwaddr[1] = cpu_to_le16(xl_ee_read(dev,0x11));
hwaddr[2] = cpu_to_le16(xl_ee_read(dev,0x12));
/* Ring speed */
switchsettings_eeprom = xl_ee_read(dev,0x08) ;
switchsettings = switchsettings_eeprom ;
if (xl_priv->xl_ring_speed != 0) {
if (xl_priv->xl_ring_speed == 4)
switchsettings = switchsettings | 0x02 ;
else
switchsettings = switchsettings & ~0x02 ;
}
/* Only write EEProm if there has been a change */
if (switchsettings != switchsettings_eeprom) {
xl_ee_write(dev,0x08,switchsettings) ;
/* Hardware reset after changing EEProm */
xl_hw_reset(dev) ;
}
memcpy(dev->dev_addr,hwaddr,dev->addr_len) ;
open_err = xl_open_hw(dev) ;
/*
* This really needs to be cleaned up with better error reporting.
*/
if (open_err != 0) { /* Something went wrong with the open command */
if (open_err & 0x07) { /* Wrong speed, retry at different speed */
printk(KERN_WARNING "%s: Open Error, retrying at different ringspeed\n", dev->name);
switchsettings = switchsettings ^ 2 ;
xl_ee_write(dev,0x08,switchsettings) ;
xl_hw_reset(dev) ;
open_err = xl_open_hw(dev) ;
if (open_err != 0) {
printk(KERN_WARNING "%s: Open error returned a second time, we're bombing out now\n", dev->name);
free_irq(dev->irq,dev) ;
return -ENODEV ;
}
} else {
printk(KERN_WARNING "%s: Open Error = %04x\n", dev->name, open_err) ;
free_irq(dev->irq,dev) ;
return -ENODEV ;
}
}
/*
* Now to set up the Rx and Tx buffer structures
*/
/* These MUST be on 8 byte boundaries */
xl_priv->xl_tx_ring = kzalloc((sizeof(struct xl_tx_desc) * XL_TX_RING_SIZE) + 7, GFP_DMA | GFP_KERNEL);
if (xl_priv->xl_tx_ring == NULL) {
free_irq(dev->irq,dev);
return -ENOMEM;
}
xl_priv->xl_rx_ring = kzalloc((sizeof(struct xl_rx_desc) * XL_RX_RING_SIZE) +7, GFP_DMA | GFP_KERNEL);
if (xl_priv->xl_rx_ring == NULL) {
free_irq(dev->irq,dev);
kfree(xl_priv->xl_tx_ring);
return -ENOMEM;
}
/* Setup Rx Ring */
for (i=0 ; i < XL_RX_RING_SIZE ; i++) {
struct sk_buff *skb ;
skb = dev_alloc_skb(xl_priv->pkt_buf_sz) ;
if (skb==NULL)
break ;
skb->dev = dev ;
xl_priv->xl_rx_ring[i].upfragaddr = cpu_to_le32(pci_map_single(xl_priv->pdev, skb->data,xl_priv->pkt_buf_sz, PCI_DMA_FROMDEVICE));
xl_priv->xl_rx_ring[i].upfraglen = cpu_to_le32(xl_priv->pkt_buf_sz) | RXUPLASTFRAG;
xl_priv->rx_ring_skb[i] = skb ;
}
if (i==0) {
printk(KERN_WARNING "%s: Not enough memory to allocate rx buffers. Adapter disabled\n",dev->name);
free_irq(dev->irq,dev) ;
kfree(xl_priv->xl_tx_ring);
kfree(xl_priv->xl_rx_ring);
return -EIO ;
}
xl_priv->rx_ring_no = i ;
xl_priv->rx_ring_tail = 0 ;
xl_priv->rx_ring_dma_addr = pci_map_single(xl_priv->pdev,xl_priv->xl_rx_ring, sizeof(struct xl_rx_desc) * XL_RX_RING_SIZE, PCI_DMA_TODEVICE) ;
for (i=0;i<(xl_priv->rx_ring_no-1);i++) {
xl_priv->xl_rx_ring[i].upnextptr = cpu_to_le32(xl_priv->rx_ring_dma_addr + (sizeof (struct xl_rx_desc) * (i+1)));
}
xl_priv->xl_rx_ring[i].upnextptr = 0 ;
writel(xl_priv->rx_ring_dma_addr, xl_mmio + MMIO_UPLISTPTR) ;
/* Setup Tx Ring */
xl_priv->tx_ring_dma_addr = pci_map_single(xl_priv->pdev,xl_priv->xl_tx_ring, sizeof(struct xl_tx_desc) * XL_TX_RING_SIZE,PCI_DMA_TODEVICE) ;
xl_priv->tx_ring_head = 1 ;
xl_priv->tx_ring_tail = 255 ; /* Special marker for first packet */
xl_priv->free_ring_entries = XL_TX_RING_SIZE ;
/*
* Setup the first dummy DPD entry for polling to start working.
*/
xl_priv->xl_tx_ring[0].framestartheader = TXDPDEMPTY;
xl_priv->xl_tx_ring[0].buffer = 0 ;
xl_priv->xl_tx_ring[0].buffer_length = 0 ;
xl_priv->xl_tx_ring[0].dnnextptr = 0 ;
writel(xl_priv->tx_ring_dma_addr, xl_mmio + MMIO_DNLISTPTR) ;
writel(DNUNSTALL, xl_mmio + MMIO_COMMAND) ;
writel(UPUNSTALL, xl_mmio + MMIO_COMMAND) ;
writel(DNENABLE, xl_mmio + MMIO_COMMAND) ;
writeb(0x40, xl_mmio + MMIO_DNPOLL) ;
/*
* Enable interrupts on the card
*/
writel(SETINTENABLE | INT_MASK, xl_mmio + MMIO_COMMAND) ;
writel(SETINDENABLE | INT_MASK, xl_mmio + MMIO_COMMAND) ;
netif_start_queue(dev) ;
return 0;
}
static int xl_open_hw(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
u8 __iomem *xl_mmio = xl_priv->xl_mmio ;
u16 vsoff ;
char ver_str[33];
int open_err ;
int i ;
unsigned long t ;
/*
* Okay, let's build up the Open.NIC srb command
*
*/
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(OPEN_NIC, xl_mmio + MMIO_MACDATA) ;
/*
* Use this as a test byte, if it comes back with the same value, the command didn't work
*/
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb)+ 2, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0xff,xl_mmio + MMIO_MACDATA) ;
/* Open options */
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb) + 8, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0x00, xl_mmio + MMIO_MACDATA) ;
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb) + 9, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0x00, xl_mmio + MMIO_MACDATA) ;
/*
* Node address, be careful here, the docs say you can just put zeros here and it will use
* the hardware address, it doesn't, you must include the node address in the open command.
*/
if (xl_priv->xl_laa[0]) { /* If using a LAA address */
for (i=10;i<16;i++) {
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb) + i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(xl_priv->xl_laa[i-10],xl_mmio + MMIO_MACDATA) ;
}
memcpy(dev->dev_addr,xl_priv->xl_laa,dev->addr_len) ;
} else { /* Regular hardware address */
for (i=10;i<16;i++) {
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb) + i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(dev->dev_addr[i-10], xl_mmio + MMIO_MACDATA) ;
}
}
/* Default everything else to 0 */
for (i = 16; i < 34; i++) {
writel( (MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb) + i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0x00,xl_mmio + MMIO_MACDATA) ;
}
/*
* Set the csrb bit in the MISR register
*/
xl_wait_misr_flags(dev) ;
writel(MEM_BYTE_WRITE | MF_CSRB, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0xFF, xl_mmio + MMIO_MACDATA) ;
writel(MMIO_BYTE_WRITE | MISR_SET, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(MISR_CSRB , xl_mmio + MMIO_MACDATA) ;
/*
* Now wait for the command to run
*/
t=jiffies;
while (! (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_SRB)) {
schedule();
if (time_after(jiffies, t + 40 * HZ)) {
printk(KERN_ERR "3COM 3C359 Velocity XL card not responding.\n");
break ;
}
}
/*
* Let's interpret the open response
*/
writel( (MEM_BYTE_READ | 0xD0000 | xl_priv->srb)+2, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if (readb(xl_mmio + MMIO_MACDATA)!=0) {
open_err = readb(xl_mmio + MMIO_MACDATA) << 8 ;
writel( (MEM_BYTE_READ | 0xD0000 | xl_priv->srb) + 7, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
open_err |= readb(xl_mmio + MMIO_MACDATA) ;
return open_err ;
} else {
writel( (MEM_WORD_READ | 0xD0000 | xl_priv->srb) + 8, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
xl_priv->asb = swab16(readw(xl_mmio + MMIO_MACDATA)) ;
printk(KERN_INFO "%s: Adapter Opened Details: ",dev->name) ;
printk("ASB: %04x",xl_priv->asb ) ;
writel( (MEM_WORD_READ | 0xD0000 | xl_priv->srb) + 10, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
printk(", SRB: %04x",swab16(readw(xl_mmio + MMIO_MACDATA)) ) ;
writel( (MEM_WORD_READ | 0xD0000 | xl_priv->srb) + 12, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
xl_priv->arb = swab16(readw(xl_mmio + MMIO_MACDATA)) ;
printk(", ARB: %04x\n",xl_priv->arb );
writel( (MEM_WORD_READ | 0xD0000 | xl_priv->srb) + 14, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
vsoff = swab16(readw(xl_mmio + MMIO_MACDATA)) ;
/*
* Interesting, sending the individual characters directly to printk was causing klogd to use
* use 100% of processor time, so we build up the string and print that instead.
*/
for (i=0;i<0x20;i++) {
writel( (MEM_BYTE_READ | 0xD0000 | vsoff) + i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
ver_str[i] = readb(xl_mmio + MMIO_MACDATA) ;
}
ver_str[i] = '\0' ;
printk(KERN_INFO "%s: Microcode version String: %s\n",dev->name,ver_str);
}
/*
* Issue the AckInterrupt
*/
writew(ACK_INTERRUPT | SRBRACK | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
return 0 ;
}
/*
* There are two ways of implementing rx on the 359 NIC, either
* interrupt driven or polling. We are going to uses interrupts,
* it is the easier way of doing things.
*
* The Rx works with a ring of Rx descriptors. At initialise time the ring
* entries point to the next entry except for the last entry in the ring
* which points to 0. The card is programmed with the location of the first
* available descriptor and keeps reading the next_ptr until next_ptr is set
* to 0. Hopefully with a ring size of 16 the card will never get to read a next_ptr
* of 0. As the Rx interrupt is received we copy the frame up to the protocol layers
* and then point the end of the ring to our current position and point our current
* position to 0, therefore making the current position the last position on the ring.
* The last position on the ring therefore loops continually loops around the rx ring.
*
* rx_ring_tail is the position on the ring to process next. (Think of a snake, the head
* expands as the card adds new packets and we go around eating the tail processing the
* packets.)
*
* Undoubtably it could be streamlined and improved upon, but at the moment it works
* and the fast path through the routine is fine.
*
* adv_rx_ring could be inlined to increase performance, but its called a *lot* of times
* in xl_rx so would increase the size of the function significantly.
*/
static void adv_rx_ring(struct net_device *dev) /* Advance rx_ring, cut down on bloat in xl_rx */
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
int n = xl_priv->rx_ring_tail;
int prev_ring_loc;
prev_ring_loc = (n + XL_RX_RING_SIZE - 1) & (XL_RX_RING_SIZE - 1);
xl_priv->xl_rx_ring[prev_ring_loc].upnextptr = cpu_to_le32(xl_priv->rx_ring_dma_addr + (sizeof (struct xl_rx_desc) * n));
xl_priv->xl_rx_ring[n].framestatus = 0;
xl_priv->xl_rx_ring[n].upnextptr = 0;
xl_priv->rx_ring_tail++;
xl_priv->rx_ring_tail &= (XL_RX_RING_SIZE-1);
}
static void xl_rx(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
struct sk_buff *skb, *skb2 ;
int frame_length = 0, copy_len = 0 ;
int temp_ring_loc ;
/*
* Receive the next frame, loop around the ring until all frames
* have been received.
*/
while (xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].framestatus & (RXUPDCOMPLETE | RXUPDFULL) ) { /* Descriptor to process */
if (xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].framestatus & RXUPDFULL ) { /* UpdFull, Multiple Descriptors used for the frame */
/*
* This is a pain, you need to go through all the descriptors until the last one
* for this frame to find the framelength
*/
temp_ring_loc = xl_priv->rx_ring_tail ;
while (xl_priv->xl_rx_ring[temp_ring_loc].framestatus & RXUPDFULL ) {
temp_ring_loc++ ;
temp_ring_loc &= (XL_RX_RING_SIZE-1) ;
}
frame_length = le32_to_cpu(xl_priv->xl_rx_ring[temp_ring_loc].framestatus) & 0x7FFF;
skb = dev_alloc_skb(frame_length) ;
if (skb==NULL) { /* No memory for frame, still need to roll forward the rx ring */
printk(KERN_WARNING "%s: dev_alloc_skb failed - multi buffer !\n", dev->name) ;
while (xl_priv->rx_ring_tail != temp_ring_loc)
adv_rx_ring(dev) ;
adv_rx_ring(dev) ; /* One more time just for luck :) */
dev->stats.rx_dropped++ ;
writel(ACK_INTERRUPT | UPCOMPACK | LATCH_ACK , xl_mmio + MMIO_COMMAND) ;
return ;
}
while (xl_priv->rx_ring_tail != temp_ring_loc) {
copy_len = le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfraglen) & 0x7FFF;
frame_length -= copy_len ;
pci_dma_sync_single_for_cpu(xl_priv->pdev,le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr),xl_priv->pkt_buf_sz,PCI_DMA_FROMDEVICE);
skb_copy_from_linear_data(xl_priv->rx_ring_skb[xl_priv->rx_ring_tail],
skb_put(skb, copy_len),
copy_len);
pci_dma_sync_single_for_device(xl_priv->pdev,le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr),xl_priv->pkt_buf_sz,PCI_DMA_FROMDEVICE);
adv_rx_ring(dev) ;
}
/* Now we have found the last fragment */
pci_dma_sync_single_for_cpu(xl_priv->pdev,le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr),xl_priv->pkt_buf_sz,PCI_DMA_FROMDEVICE);
skb_copy_from_linear_data(xl_priv->rx_ring_skb[xl_priv->rx_ring_tail],
skb_put(skb,copy_len), frame_length);
/* memcpy(skb_put(skb,frame_length), bus_to_virt(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr), frame_length) ; */
pci_dma_sync_single_for_device(xl_priv->pdev,le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr),xl_priv->pkt_buf_sz,PCI_DMA_FROMDEVICE);
adv_rx_ring(dev) ;
skb->protocol = tr_type_trans(skb,dev) ;
netif_rx(skb) ;
} else { /* Single Descriptor Used, simply swap buffers over, fast path */
frame_length = le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].framestatus) & 0x7FFF;
skb = dev_alloc_skb(xl_priv->pkt_buf_sz) ;
if (skb==NULL) { /* Still need to fix the rx ring */
printk(KERN_WARNING "%s: dev_alloc_skb failed in rx, single buffer\n",dev->name);
adv_rx_ring(dev) ;
dev->stats.rx_dropped++ ;
writel(ACK_INTERRUPT | UPCOMPACK | LATCH_ACK , xl_mmio + MMIO_COMMAND) ;
return ;
}
skb2 = xl_priv->rx_ring_skb[xl_priv->rx_ring_tail] ;
pci_unmap_single(xl_priv->pdev, le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr), xl_priv->pkt_buf_sz,PCI_DMA_FROMDEVICE) ;
skb_put(skb2, frame_length) ;
skb2->protocol = tr_type_trans(skb2,dev) ;
xl_priv->rx_ring_skb[xl_priv->rx_ring_tail] = skb ;
xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr = cpu_to_le32(pci_map_single(xl_priv->pdev,skb->data,xl_priv->pkt_buf_sz, PCI_DMA_FROMDEVICE));
xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfraglen = cpu_to_le32(xl_priv->pkt_buf_sz) | RXUPLASTFRAG;
adv_rx_ring(dev) ;
dev->stats.rx_packets++ ;
dev->stats.rx_bytes += frame_length ;
netif_rx(skb2) ;
} /* if multiple buffers */
} /* while packet to do */
/* Clear the updComplete interrupt */
writel(ACK_INTERRUPT | UPCOMPACK | LATCH_ACK , xl_mmio + MMIO_COMMAND) ;
return ;
}
/*
* This is ruthless, it doesn't care what state the card is in it will
* completely reset the adapter.
*/
static void xl_reset(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
unsigned long t;
writew( GLOBAL_RESET, xl_mmio + MMIO_COMMAND ) ;
/*
* Must wait for cmdInProgress bit (12) to clear before continuing with
* card configuration.
*/
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
if (time_after(jiffies, t + 40 * HZ)) {
printk(KERN_ERR "3COM 3C359 Velocity XL card not responding.\n");
break ;
}
}
}
static void xl_freemem(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
int i ;
for (i=0;i<XL_RX_RING_SIZE;i++) {
dev_kfree_skb_irq(xl_priv->rx_ring_skb[xl_priv->rx_ring_tail]) ;
pci_unmap_single(xl_priv->pdev,le32_to_cpu(xl_priv->xl_rx_ring[xl_priv->rx_ring_tail].upfragaddr),xl_priv->pkt_buf_sz, PCI_DMA_FROMDEVICE);
xl_priv->rx_ring_tail++ ;
xl_priv->rx_ring_tail &= XL_RX_RING_SIZE-1;
}
/* unmap ring */
pci_unmap_single(xl_priv->pdev,xl_priv->rx_ring_dma_addr, sizeof(struct xl_rx_desc) * XL_RX_RING_SIZE, PCI_DMA_FROMDEVICE) ;
pci_unmap_single(xl_priv->pdev,xl_priv->tx_ring_dma_addr, sizeof(struct xl_tx_desc) * XL_TX_RING_SIZE, PCI_DMA_TODEVICE) ;
kfree(xl_priv->xl_rx_ring) ;
kfree(xl_priv->xl_tx_ring) ;
return ;
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 15:55:46 +02:00
static irqreturn_t xl_interrupt(int irq, void *dev_id)
{
struct net_device *dev = (struct net_device *)dev_id;
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv =netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
u16 intstatus, macstatus ;
intstatus = readw(xl_mmio + MMIO_INTSTATUS) ;
if (!(intstatus & 1)) /* We didn't generate the interrupt */
return IRQ_NONE;
spin_lock(&xl_priv->xl_lock) ;
/*
* Process the interrupt
*/
/*
* Something fishy going on here, we shouldn't get 0001 ints, not fatal though.
*/
if (intstatus == 0x0001) {
writel(ACK_INTERRUPT | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
printk(KERN_INFO "%s: 00001 int received\n",dev->name);
} else {
if (intstatus & (HOSTERRINT | SRBRINT | ARBCINT | UPCOMPINT | DNCOMPINT | HARDERRINT | (1<<8) | TXUNDERRUN | ASBFINT)) {
/*
* Host Error.
* It may be possible to recover from this, but usually it means something
* is seriously fubar, so we just close the adapter.
*/
if (intstatus & HOSTERRINT) {
printk(KERN_WARNING "%s: Host Error, performing global reset, intstatus = %04x\n",dev->name,intstatus);
writew( GLOBAL_RESET, xl_mmio + MMIO_COMMAND ) ;
printk(KERN_WARNING "%s: Resetting hardware:\n", dev->name);
netif_stop_queue(dev) ;
xl_freemem(dev) ;
free_irq(dev->irq,dev);
xl_reset(dev) ;
writel(ACK_INTERRUPT | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
spin_unlock(&xl_priv->xl_lock) ;
return IRQ_HANDLED;
} /* Host Error */
if (intstatus & SRBRINT ) { /* Srbc interrupt */
writel(ACK_INTERRUPT | SRBRACK | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
if (xl_priv->srb_queued)
xl_srb_bh(dev) ;
} /* SRBR Interrupt */
if (intstatus & TXUNDERRUN) { /* Issue DnReset command */
writel(DNRESET, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) { /* Wait for command to run */
/* !!! FIX-ME !!!!
Must put a timeout check here ! */
/* Empty Loop */
}
printk(KERN_WARNING "%s: TX Underrun received\n",dev->name);
writel(ACK_INTERRUPT | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
} /* TxUnderRun */
if (intstatus & ARBCINT ) { /* Arbc interrupt */
xl_arb_cmd(dev) ;
} /* Arbc */
if (intstatus & ASBFINT) {
if (xl_priv->asb_queued == 1) {
xl_asb_cmd(dev) ;
} else if (xl_priv->asb_queued == 2) {
xl_asb_bh(dev) ;
} else {
writel(ACK_INTERRUPT | LATCH_ACK | ASBFACK, xl_mmio + MMIO_COMMAND) ;
}
} /* Asbf */
if (intstatus & UPCOMPINT ) /* UpComplete */
xl_rx(dev) ;
if (intstatus & DNCOMPINT ) /* DnComplete */
xl_dn_comp(dev) ;
if (intstatus & HARDERRINT ) { /* Hardware error */
writel(MMIO_WORD_READ | MACSTATUS, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
macstatus = readw(xl_mmio + MMIO_MACDATA) ;
printk(KERN_WARNING "%s: MacStatusError, details: ", dev->name);
if (macstatus & (1<<14))
printk(KERN_WARNING "tchk error: Unrecoverable error\n");
if (macstatus & (1<<3))
printk(KERN_WARNING "eint error: Internal watchdog timer expired\n");
if (macstatus & (1<<2))
printk(KERN_WARNING "aint error: Host tried to perform invalid operation\n");
printk(KERN_WARNING "Instatus = %02x, macstatus = %02x\n",intstatus,macstatus) ;
printk(KERN_WARNING "%s: Resetting hardware:\n", dev->name);
netif_stop_queue(dev) ;
xl_freemem(dev) ;
free_irq(dev->irq,dev);
unregister_netdev(dev) ;
free_netdev(dev) ;
xl_reset(dev) ;
writel(ACK_INTERRUPT | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
spin_unlock(&xl_priv->xl_lock) ;
return IRQ_HANDLED;
}
} else {
printk(KERN_WARNING "%s: Received Unknown interrupt : %04x\n", dev->name, intstatus);
writel(ACK_INTERRUPT | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
}
}
/* Turn interrupts back on */
writel( SETINDENABLE | INT_MASK, xl_mmio + MMIO_COMMAND) ;
writel( SETINTENABLE | INT_MASK, xl_mmio + MMIO_COMMAND) ;
spin_unlock(&xl_priv->xl_lock) ;
return IRQ_HANDLED;
}
/*
* Tx - Polling configuration
*/
static netdev_tx_t xl_xmit(struct sk_buff *skb, struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
struct xl_tx_desc *txd ;
int tx_head, tx_tail, tx_prev ;
unsigned long flags ;
spin_lock_irqsave(&xl_priv->xl_lock,flags) ;
netif_stop_queue(dev) ;
if (xl_priv->free_ring_entries > 1 ) {
/*
* Set up the descriptor for the packet
*/
tx_head = xl_priv->tx_ring_head ;
tx_tail = xl_priv->tx_ring_tail ;
txd = &(xl_priv->xl_tx_ring[tx_head]) ;
txd->dnnextptr = 0 ;
txd->framestartheader = cpu_to_le32(skb->len) | TXDNINDICATE;
txd->buffer = cpu_to_le32(pci_map_single(xl_priv->pdev, skb->data, skb->len, PCI_DMA_TODEVICE));
txd->buffer_length = cpu_to_le32(skb->len) | TXDNFRAGLAST;
xl_priv->tx_ring_skb[tx_head] = skb ;
dev->stats.tx_packets++ ;
dev->stats.tx_bytes += skb->len ;
/*
* Set the nextptr of the previous descriptor equal to this descriptor, add XL_TX_RING_SIZE -1
* to ensure no negative numbers in unsigned locations.
*/
tx_prev = (xl_priv->tx_ring_head + XL_TX_RING_SIZE - 1) & (XL_TX_RING_SIZE - 1) ;
xl_priv->tx_ring_head++ ;
xl_priv->tx_ring_head &= (XL_TX_RING_SIZE - 1) ;
xl_priv->free_ring_entries-- ;
xl_priv->xl_tx_ring[tx_prev].dnnextptr = cpu_to_le32(xl_priv->tx_ring_dma_addr + (sizeof (struct xl_tx_desc) * tx_head));
/* Sneaky, by doing a read on DnListPtr we can force the card to poll on the DnNextPtr */
/* readl(xl_mmio + MMIO_DNLISTPTR) ; */
netif_wake_queue(dev) ;
spin_unlock_irqrestore(&xl_priv->xl_lock,flags) ;
return NETDEV_TX_OK;
} else {
spin_unlock_irqrestore(&xl_priv->xl_lock,flags) ;
return NETDEV_TX_BUSY;
}
}
/*
* The NIC has told us that a packet has been downloaded onto the card, we must
* find out which packet it has done, clear the skb and information for the packet
* then advance around the ring for all transmitted packets
*/
static void xl_dn_comp(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
struct xl_tx_desc *txd ;
if (xl_priv->tx_ring_tail == 255) {/* First time */
xl_priv->xl_tx_ring[0].framestartheader = 0 ;
xl_priv->xl_tx_ring[0].dnnextptr = 0 ;
xl_priv->tx_ring_tail = 1 ;
}
while (xl_priv->xl_tx_ring[xl_priv->tx_ring_tail].framestartheader & TXDNCOMPLETE ) {
txd = &(xl_priv->xl_tx_ring[xl_priv->tx_ring_tail]) ;
pci_unmap_single(xl_priv->pdev, le32_to_cpu(txd->buffer), xl_priv->tx_ring_skb[xl_priv->tx_ring_tail]->len, PCI_DMA_TODEVICE);
txd->framestartheader = 0 ;
txd->buffer = cpu_to_le32(0xdeadbeef);
txd->buffer_length = 0 ;
dev_kfree_skb_irq(xl_priv->tx_ring_skb[xl_priv->tx_ring_tail]) ;
xl_priv->tx_ring_tail++ ;
xl_priv->tx_ring_tail &= (XL_TX_RING_SIZE - 1) ;
xl_priv->free_ring_entries++ ;
}
netif_wake_queue(dev) ;
writel(ACK_INTERRUPT | DNCOMPACK | LATCH_ACK , xl_mmio + MMIO_COMMAND) ;
}
/*
* Close the adapter properly.
* This srb reply cannot be handled from interrupt context as we have
* to free the interrupt from the driver.
*/
static int xl_close(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
unsigned long t ;
netif_stop_queue(dev) ;
/*
* Close the adapter, need to stall the rx and tx queues.
*/
writew(DNSTALL, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-DNSTALL not responding.\n", dev->name);
break ;
}
}
writew(DNDISABLE, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-DNDISABLE not responding.\n", dev->name);
break ;
}
}
writew(UPSTALL, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-UPSTALL not responding.\n", dev->name);
break ;
}
}
/* Turn off interrupts, we will still get the indication though
* so we can trap it
*/
writel(SETINTENABLE, xl_mmio + MMIO_COMMAND) ;
xl_srb_cmd(dev,CLOSE_NIC) ;
t=jiffies;
while (!(readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_SRB)) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-CLOSENIC not responding.\n", dev->name);
break ;
}
}
/* Read the srb response from the adapter */
writel(MEM_BYTE_READ | 0xd0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD);
if (readb(xl_mmio + MMIO_MACDATA) != CLOSE_NIC) {
printk(KERN_INFO "%s: CLOSE_NIC did not get a CLOSE_NIC response\n",dev->name);
} else {
writel((MEM_BYTE_READ | 0xd0000 | xl_priv->srb) +2, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if (readb(xl_mmio + MMIO_MACDATA)==0) {
printk(KERN_INFO "%s: Adapter has been closed\n",dev->name);
writew(ACK_INTERRUPT | SRBRACK | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
xl_freemem(dev) ;
free_irq(dev->irq,dev) ;
} else {
printk(KERN_INFO "%s: Close nic command returned error code %02x\n",dev->name, readb(xl_mmio + MMIO_MACDATA)) ;
}
}
/* Reset the upload and download logic */
writew(UPRESET, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-UPRESET not responding.\n", dev->name);
break ;
}
}
writew(DNRESET, xl_mmio + MMIO_COMMAND) ;
t=jiffies;
while (readw(xl_mmio + MMIO_INTSTATUS) & INTSTAT_CMD_IN_PROGRESS) {
schedule();
if (time_after(jiffies, t + 10 * HZ)) {
printk(KERN_ERR "%s: 3COM 3C359 Velocity XL-DNRESET not responding.\n", dev->name);
break ;
}
}
xl_hw_reset(dev) ;
return 0 ;
}
static void xl_set_rx_mode(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
struct netdev_hw_addr *ha;
unsigned char dev_mc_address[4] ;
u16 options ;
if (dev->flags & IFF_PROMISC)
options = 0x0004 ;
else
options = 0x0000 ;
if (options ^ xl_priv->xl_copy_all_options) { /* Changed, must send command */
xl_priv->xl_copy_all_options = options ;
xl_srb_cmd(dev, SET_RECEIVE_MODE) ;
return ;
}
dev_mc_address[0] = dev_mc_address[1] = dev_mc_address[2] = dev_mc_address[3] = 0 ;
netdev_for_each_mc_addr(ha, dev) {
dev_mc_address[0] |= ha->addr[2];
dev_mc_address[1] |= ha->addr[3];
dev_mc_address[2] |= ha->addr[4];
dev_mc_address[3] |= ha->addr[5];
}
if (memcmp(xl_priv->xl_functional_addr,dev_mc_address,4) != 0) { /* Options have changed, run the command */
memcpy(xl_priv->xl_functional_addr, dev_mc_address,4) ;
xl_srb_cmd(dev, SET_FUNC_ADDRESS) ;
}
return ;
}
/*
* We issued an srb command and now we must read
* the response from the completed command.
*/
static void xl_srb_bh(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
u8 srb_cmd, ret_code ;
int i ;
writel(MEM_BYTE_READ | 0xd0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
srb_cmd = readb(xl_mmio + MMIO_MACDATA) ;
writel((MEM_BYTE_READ | 0xd0000 | xl_priv->srb) +2, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
ret_code = readb(xl_mmio + MMIO_MACDATA) ;
/* Ret_code is standard across all commands */
switch (ret_code) {
case 1:
printk(KERN_INFO "%s: Command: %d - Invalid Command code\n",dev->name,srb_cmd) ;
break ;
case 4:
printk(KERN_INFO "%s: Command: %d - Adapter is closed, must be open for this command\n",dev->name,srb_cmd);
break ;
case 6:
printk(KERN_INFO "%s: Command: %d - Options Invalid for command\n",dev->name,srb_cmd);
break ;
case 0: /* Successful command execution */
switch (srb_cmd) {
case READ_LOG: /* Returns 14 bytes of data from the NIC */
if(xl_priv->xl_message_level)
printk(KERN_INFO "%s: READ.LOG 14 bytes of data ",dev->name) ;
/*
* We still have to read the log even if message_level = 0 and we don't want
* to see it
*/
for (i=0;i<14;i++) {
writel(MEM_BYTE_READ | 0xd0000 | xl_priv->srb | i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if(xl_priv->xl_message_level)
printk("%02x:",readb(xl_mmio + MMIO_MACDATA)) ;
}
printk("\n") ;
break ;
case SET_FUNC_ADDRESS:
if(xl_priv->xl_message_level)
printk(KERN_INFO "%s: Functional Address Set\n",dev->name);
break ;
case CLOSE_NIC:
if(xl_priv->xl_message_level)
printk(KERN_INFO "%s: Received CLOSE_NIC interrupt in interrupt handler\n",dev->name);
break ;
case SET_MULTICAST_MODE:
if(xl_priv->xl_message_level)
printk(KERN_INFO "%s: Multicast options successfully changed\n",dev->name) ;
break ;
case SET_RECEIVE_MODE:
if(xl_priv->xl_message_level) {
if (xl_priv->xl_copy_all_options == 0x0004)
printk(KERN_INFO "%s: Entering promiscuous mode\n", dev->name);
else
printk(KERN_INFO "%s: Entering normal receive mode\n",dev->name);
}
break ;
} /* switch */
break ;
} /* switch */
return ;
}
static int xl_set_mac_address (struct net_device *dev, void *addr)
{
struct sockaddr *saddr = addr ;
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
if (netif_running(dev)) {
printk(KERN_WARNING "%s: Cannot set mac/laa address while card is open\n", dev->name) ;
return -EIO ;
}
memcpy(xl_priv->xl_laa, saddr->sa_data,dev->addr_len) ;
if (xl_priv->xl_message_level) {
printk(KERN_INFO "%s: MAC/LAA Set to = %x.%x.%x.%x.%x.%x\n",dev->name, xl_priv->xl_laa[0],
xl_priv->xl_laa[1], xl_priv->xl_laa[2],
xl_priv->xl_laa[3], xl_priv->xl_laa[4],
xl_priv->xl_laa[5]);
}
return 0 ;
}
static void xl_arb_cmd(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
u8 arb_cmd ;
u16 lan_status, lan_status_diff ;
writel( ( MEM_BYTE_READ | 0xD0000 | xl_priv->arb), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
arb_cmd = readb(xl_mmio + MMIO_MACDATA) ;
if (arb_cmd == RING_STATUS_CHANGE) { /* Ring.Status.Change */
writel( ( (MEM_WORD_READ | 0xD0000 | xl_priv->arb) + 6), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
printk(KERN_INFO "%s: Ring Status Change: New Status = %04x\n", dev->name, swab16(readw(xl_mmio + MMIO_MACDATA) )) ;
lan_status = swab16(readw(xl_mmio + MMIO_MACDATA));
/* Acknowledge interrupt, this tells nic we are done with the arb */
writel(ACK_INTERRUPT | ARBCACK | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
lan_status_diff = xl_priv->xl_lan_status ^ lan_status ;
if (lan_status_diff & (LSC_LWF | LSC_ARW | LSC_FPE | LSC_RR) ) {
if (lan_status_diff & LSC_LWF)
printk(KERN_WARNING "%s: Short circuit detected on the lobe\n",dev->name);
if (lan_status_diff & LSC_ARW)
printk(KERN_WARNING "%s: Auto removal error\n",dev->name);
if (lan_status_diff & LSC_FPE)
printk(KERN_WARNING "%s: FDX Protocol Error\n",dev->name);
if (lan_status_diff & LSC_RR)
printk(KERN_WARNING "%s: Force remove MAC frame received\n",dev->name);
/* Adapter has been closed by the hardware */
netif_stop_queue(dev);
xl_freemem(dev) ;
free_irq(dev->irq,dev);
printk(KERN_WARNING "%s: Adapter has been closed\n", dev->name);
} /* If serious error */
if (xl_priv->xl_message_level) {
if (lan_status_diff & LSC_SIG_LOSS)
printk(KERN_WARNING "%s: No receive signal detected\n", dev->name);
if (lan_status_diff & LSC_HARD_ERR)
printk(KERN_INFO "%s: Beaconing\n",dev->name);
if (lan_status_diff & LSC_SOFT_ERR)
printk(KERN_WARNING "%s: Adapter transmitted Soft Error Report Mac Frame\n",dev->name);
if (lan_status_diff & LSC_TRAN_BCN)
printk(KERN_INFO "%s: We are transmitting the beacon, aaah\n",dev->name);
if (lan_status_diff & LSC_SS)
printk(KERN_INFO "%s: Single Station on the ring\n", dev->name);
if (lan_status_diff & LSC_RING_REC)
printk(KERN_INFO "%s: Ring recovery ongoing\n",dev->name);
if (lan_status_diff & LSC_FDX_MODE)
printk(KERN_INFO "%s: Operating in FDX mode\n",dev->name);
}
if (lan_status_diff & LSC_CO) {
if (xl_priv->xl_message_level)
printk(KERN_INFO "%s: Counter Overflow\n", dev->name);
/* Issue READ.LOG command */
xl_srb_cmd(dev, READ_LOG) ;
}
/* There is no command in the tech docs to issue the read_sr_counters */
if (lan_status_diff & LSC_SR_CO) {
if (xl_priv->xl_message_level)
printk(KERN_INFO "%s: Source routing counters overflow\n", dev->name);
}
xl_priv->xl_lan_status = lan_status ;
} /* Lan.change.status */
else if ( arb_cmd == RECEIVE_DATA) { /* Received.Data */
#if XL_DEBUG
printk(KERN_INFO "Received.Data\n");
#endif
writel( ((MEM_WORD_READ | 0xD0000 | xl_priv->arb) + 6), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
xl_priv->mac_buffer = swab16(readw(xl_mmio + MMIO_MACDATA)) ;
/* Now we are going to be really basic here and not do anything
* with the data at all. The tech docs do not give me enough
* information to calculate the buffers properly so we're
* just going to tell the nic that we've dealt with the frame
* anyway.
*/
/* Acknowledge interrupt, this tells nic we are done with the arb */
writel(ACK_INTERRUPT | ARBCACK | LATCH_ACK, xl_mmio + MMIO_COMMAND) ;
/* Is the ASB free ? */
xl_priv->asb_queued = 0 ;
writel( ((MEM_BYTE_READ | 0xD0000 | xl_priv->asb) + 2), xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if (readb(xl_mmio + MMIO_MACDATA) != 0xff) {
xl_priv->asb_queued = 1 ;
xl_wait_misr_flags(dev) ;
writel(MEM_BYTE_WRITE | MF_ASBFR, xl_mmio + MMIO_MAC_ACCESS_CMD);
writeb(0xff, xl_mmio + MMIO_MACDATA) ;
writel(MMIO_BYTE_WRITE | MISR_SET, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(MISR_ASBFR, xl_mmio + MMIO_MACDATA) ;
return ;
/* Drop out and wait for the bottom half to be run */
}
xl_asb_cmd(dev) ;
} else {
printk(KERN_WARNING "%s: Received unknown arb (xl_priv) command: %02x\n",dev->name,arb_cmd);
}
/* Acknowledge the arb interrupt */
writel(ACK_INTERRUPT | ARBCACK | LATCH_ACK , xl_mmio + MMIO_COMMAND) ;
return ;
}
/*
* There is only one asb command, but we can get called from different
* places.
*/
static void xl_asb_cmd(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
if (xl_priv->asb_queued == 1)
writel(ACK_INTERRUPT | LATCH_ACK | ASBFACK, xl_mmio + MMIO_COMMAND) ;
writel(MEM_BYTE_WRITE | 0xd0000 | xl_priv->asb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0x81, xl_mmio + MMIO_MACDATA) ;
writel(MEM_WORD_WRITE | 0xd0000 | xl_priv->asb | 6, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(swab16(xl_priv->mac_buffer), xl_mmio + MMIO_MACDATA) ;
xl_wait_misr_flags(dev) ;
writel(MEM_BYTE_WRITE | MF_RASB, xl_mmio + MMIO_MAC_ACCESS_CMD);
writeb(0xff, xl_mmio + MMIO_MACDATA) ;
writel(MMIO_BYTE_WRITE | MISR_SET, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(MISR_RASB, xl_mmio + MMIO_MACDATA) ;
xl_priv->asb_queued = 2 ;
return ;
}
/*
* This will only get called if there was an error
* from the asb cmd.
*/
static void xl_asb_bh(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
u8 ret_code ;
writel(MMIO_BYTE_READ | 0xd0000 | xl_priv->asb | 2, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
ret_code = readb(xl_mmio + MMIO_MACDATA) ;
switch (ret_code) {
case 0x01:
printk(KERN_INFO "%s: ASB Command, unrecognized command code\n",dev->name);
break ;
case 0x26:
printk(KERN_INFO "%s: ASB Command, unexpected receive buffer\n", dev->name);
break ;
case 0x40:
printk(KERN_INFO "%s: ASB Command, Invalid Station ID\n", dev->name);
break ;
}
xl_priv->asb_queued = 0 ;
writel(ACK_INTERRUPT | LATCH_ACK | ASBFACK, xl_mmio + MMIO_COMMAND) ;
return ;
}
/*
* Issue srb commands to the nic
*/
static void xl_srb_cmd(struct net_device *dev, int srb_cmd)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
switch (srb_cmd) {
case READ_LOG:
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(READ_LOG, xl_mmio + MMIO_MACDATA) ;
break;
case CLOSE_NIC:
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(CLOSE_NIC, xl_mmio + MMIO_MACDATA) ;
break ;
case SET_RECEIVE_MODE:
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(SET_RECEIVE_MODE, xl_mmio + MMIO_MACDATA) ;
writel(MEM_WORD_WRITE | 0xD0000 | xl_priv->srb | 4, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writew(xl_priv->xl_copy_all_options, xl_mmio + MMIO_MACDATA) ;
break ;
case SET_FUNC_ADDRESS:
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(SET_FUNC_ADDRESS, xl_mmio + MMIO_MACDATA) ;
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb | 6 , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(xl_priv->xl_functional_addr[0], xl_mmio + MMIO_MACDATA) ;
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb | 7 , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(xl_priv->xl_functional_addr[1], xl_mmio + MMIO_MACDATA) ;
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb | 8 , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(xl_priv->xl_functional_addr[2], xl_mmio + MMIO_MACDATA) ;
writel(MEM_BYTE_WRITE | 0xD0000 | xl_priv->srb | 9 , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(xl_priv->xl_functional_addr[3], xl_mmio + MMIO_MACDATA) ;
break ;
} /* switch */
xl_wait_misr_flags(dev) ;
/* Write 0xff to the CSRB flag */
writel(MEM_BYTE_WRITE | MF_CSRB , xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0xFF, xl_mmio + MMIO_MACDATA) ;
/* Set csrb bit in MISR register to process command */
writel(MMIO_BYTE_WRITE | MISR_SET, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(MISR_CSRB, xl_mmio + MMIO_MACDATA) ;
xl_priv->srb_queued = 1 ;
return ;
}
/*
* This is nasty, to use the MISR command you have to wait for 6 memory locations
* to be zero. This is the way the driver does on other OS'es so we should be ok with
* the empty loop.
*/
static void xl_wait_misr_flags(struct net_device *dev)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u8 __iomem * xl_mmio = xl_priv->xl_mmio ;
int i ;
writel(MMIO_BYTE_READ | MISR_RW, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
if (readb(xl_mmio + MMIO_MACDATA) != 0) { /* Misr not clear */
for (i=0; i<6; i++) {
writel(MEM_BYTE_READ | 0xDFFE0 | i, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
while (readb(xl_mmio + MMIO_MACDATA) != 0) {
; /* Empty Loop */
}
}
}
writel(MMIO_BYTE_WRITE | MISR_AND, xl_mmio + MMIO_MAC_ACCESS_CMD) ;
writeb(0x80, xl_mmio + MMIO_MACDATA) ;
return ;
}
/*
* Change mtu size, this should work the same as olympic
*/
static int xl_change_mtu(struct net_device *dev, int mtu)
{
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv = netdev_priv(dev);
u16 max_mtu ;
if (xl_priv->xl_ring_speed == 4)
max_mtu = 4500 ;
else
max_mtu = 18000 ;
if (mtu > max_mtu)
return -EINVAL ;
if (mtu < 100)
return -EINVAL ;
dev->mtu = mtu ;
xl_priv->pkt_buf_sz = mtu + TR_HLEN ;
return 0 ;
}
static void __devexit xl_remove_one (struct pci_dev *pdev)
{
struct net_device *dev = pci_get_drvdata(pdev);
dev->priv to netdev_priv(dev), drivers/net/tokenring/ Replacing accesses to dev->priv to netdev_priv(dev). The replacment is safe when netdev_priv is used to access a private structure that is right next to the net_device structure in memory. Cf http://groups.google.com/group/comp.os.linux.development.system/browse_thread/thread/de19321bcd94dbb8/0d74a4adcd6177bd This is the case when the net_device structure was allocated with a call to alloc_netdev or one of its derivative. Here is an excerpt of the semantic patch that performs the transformation @ rule1 @ type T; struct net_device *dev; @@ dev = ( alloc_netdev | alloc_etherdev | alloc_trdev ) (sizeof(T), ...) @ rule1bis @ struct net_device *dev; expression E; @@ dev->priv = E @ rule2 depends on rule1 && !rule1bis @ struct net_device *dev; type rule1.T; @@ - (T*) dev->priv + netdev_priv(dev) PS: I have performed the same transformation on the whole kernel and it affects around 70 files, most of them in drivers/net/. Should I split my patch for each subnet directories ? (wireless/, wan/, etc) Thanks to Thomas Surrel for helping me refining my semantic patch. Signed-off-by: Yoann Padioleau <padator@wanadoo.fr> 3c359.c | 58 +++++++++++++++++++++++++++++----------------------------- ibmtr.c | 38 +++++++++++++++++++------------------- lanstreamer.c | 32 ++++++++++++++++---------------- madgemc.c | 4 ++-- olympic.c | 36 ++++++++++++++++++------------------ tmspci.c | 4 ++-- 6 files changed, 86 insertions(+), 86 deletions(-) Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-07-23 15:18:21 +02:00
struct xl_private *xl_priv=netdev_priv(dev);
release_firmware(xl_priv->fw);
unregister_netdev(dev);
iounmap(xl_priv->xl_mmio) ;
pci_release_regions(pdev) ;
pci_set_drvdata(pdev,NULL) ;
free_netdev(dev);
return ;
}
static struct pci_driver xl_3c359_driver = {
.name = "3c359",
.id_table = xl_pci_tbl,
.probe = xl_probe,
.remove = __devexit_p(xl_remove_one),
};
module_pci_driver(xl_3c359_driver);
MODULE_LICENSE("GPL") ;