When we add the generic napi_disable_pending() breakout
logic to net_rx_action() it means that napi_disable()
can cause NAPI poll interrupt events to be disabled.
And this is exactly what we want. If a napi_disable()
is pending, and we are looping in the ->poll(), we want
->poll() event interrupts to stay disabled and we want
to complete the NAPI poll ASAP.
When ->poll() break out during device down was being handled on a
per-driver basis, often these drivers would turn interrupts back on
when '!netif_running()' was detected.
And this would just cause a reschedule of the NAPI ->poll() in the
interrupt handler before the napi_disable() could get in there and
grab the NAPI_STATE_SCHED bit.
The vast majority of drivers don't care if napi_disable() might have
the side effect of disabling NAPI ->poll() event interrupts. In all
such cases, when a napi_disable() is performed, the driver just
disabled interrupts or is about to.
However there were three exceptions to this in PCNET32, R8169, and
SKY2. To fix those cases, at the subsequent napi_enable() points, I
added code to ensure that the ->poll() interrupt events are enabled in
the hardware.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Don Fry <pcnet32@verizon.net>
Remove the statistics from the private structure.
Use the net_device_stats in netn_device structure.
Following Jeff Garzik's massive cleanup Sep 01.
pcnet32 was not "low-hanging fruit".
Tested x86_64.
Signed-off-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Recent changes to the driver for the new napi API broke the reception
of packets when in non-napi mode. The initialization of napi.weight
was removed for the non-napi case leaving the value zero.
Tested NAPI and non-NAPI on x86_64.
Signed-off-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
These have been superceded by the new ->get_sset_count() hook.
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the operations
get-tx-csum
get-sg
get-tso
get-ufo
the default ethtool_op_xxx behavior is fine for all drivers, so we
permit op==NULL to imply the default behavior.
This provides a more uniform behavior across all drivers, eliminating
ethtool(8) "ioctl not supported" errors on older drivers that had
not been updated for the latest sub-ioctls.
The ethtool_op_xxx() functions are left exported, in case anyone
wishes to call them directly from a driver-private implementation --
a not-uncommon case. Should an ethtool_op_xxx() helper remain unused
for a while, except by net/core/ethtool.c, we can un-export it at a
later date.
[ Resolved conflicts with set/get value ethtool patch... -DaveM ]
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's been a useless no-op for long enough in 2.6 so I figured it's time to
remove it. The number of people that could object because they're
maintaining unified 2.4 and 2.6 drivers is probably rather small.
[ Handled drivers added by netdev tree and some missed IRDA cases... -DaveM ]
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add suspend and resume capability to the driver.
Tested both to ram and to disk on x86_64 platform.
Signed-off-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Several devices have multiple independant RX queues per net
device, and some have a single interrupt doorbell for several
queues.
In either case, it's easier to support layouts like that if the
structure representing the poll is independant from the net
device itself.
The signature of the ->poll() call back goes from:
int foo_poll(struct net_device *dev, int *budget)
to
int foo_poll(struct napi_struct *napi, int budget)
The caller is returned the number of RX packets processed (or
the number of "NAPI credits" consumed if you want to get
abstract). The callee no longer messes around bumping
dev->quota, *budget, etc. because that is all handled in the
caller upon return.
The napi_struct is to be embedded in the device driver private data
structures.
Furthermore, it is the driver's responsibility to disable all NAPI
instances in it's ->stop() device close handler. Since the
napi_struct is privatized into the driver's private data structures,
only the driver knows how to get at all of the napi_struct instances
it may have per-device.
With lots of help and suggestions from Rusty Russell, Roland Dreier,
Michael Chan, Jeff Garzik, and Jamal Hadi Salim.
Bug fixes from Thomas Graf, Roland Dreier, Peter Zijlstra,
Joseph Fannin, Scott Wood, Hans J. Koch, and Michael Chan.
[ Ported to current tree and all drivers converted. Integrated
Stephen's follow-on kerneldoc additions, and restored poll_list
handling to the old style to fix mutual exclusion issues. -DaveM ]
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
All drivers implement ethtool get_perm_addr the same way -- by calling
the generic function. So we can inline the generic function into the
caller and avoid going through the drivers.
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: David S. Miller <davem@davemloft.net>
It hasn't "summed" anything in over 7 years, and it's
just a straight mempcy ala skb_copy_to_linear_data()
so just get rid of it.
Signed-off-by: David S. Miller <davem@davemloft.net>
use netdev_priv() instead of dev->priv
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The patch below moves the init_block out of the private struct and
only allocates init block with pci_alloc_consistent.
This has two effects:
1. Performance increase for non cache coherent machines, because the
CPU only data in the private struct are now cached
2. locks are working now for platforms, which need to have locks
in cached memory
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Acked-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
One less thing for drivers writers to worry about.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The PCnet32 driver always passed the the size of the largest possible packet
to the pci_dma_sync_single_for_cpu and pci_dma_sync_single_for_device.
This results in a fairly large "colateral damage" in the caches and makes
the flush operation itself much slower. On a system with a 40MHz CPU this
patch increases network bandwidth by about 12%.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Acked-by: Don Fry <pcnet32@verizon.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
- Eliminate check for irq handler 'dev_id==NULL' where the
condition never occurs.
- Eliminate needless casts to/from void*
Signed-off-by: Jeff Garzik <jeff@garzik.org>
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)
Implement NAPI changes to pcnet32 driver. Compile default is off.
Listed as experimental.
Len and Don both worked on a NAPI implementation and have both tested
these changes.
An e1000 blasting short packets to the pcnet32 will lockup Don's system
until the receive storm stops. Without NAPI Len's system watchdog would
expire causing the system to reboot. With NAPI the system will stay
operational.
Tested ia32 and ppc64. Tested '970A, '971, '972, '973, '975, '976, and
'978.
The Kconfig changes came from Len. Don is to blame for all the others.
Signed-off-by: Len Sorensen <lsorense@csclub.uwaterloo.ca>
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Breaking the receive frame processing into two routines for greater clarity.
Tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Move the receive routine and create the transmit routine.
Tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Change some magic numbers to clearer names. A few whitespace changes.
Tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Delete unnecessary save/restore of rap in interrupt handler and statistics.
tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
A change I made for 2.6.17 and another for 2.6.18 do not work on older
pcnet32 chips which I do not have access to. If the chip is a 79C970 or
79C965, do not try and suspend or check the link status.
I have tested with a 79C970A, 79C971, 79C972, 79C973, 79C975, 79C976,
and 79C978.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
From: Michal Piotrowski <michal.k.k.piotrowski@gmail.com>
Signed-off-by: Michal Piotrowski <michal.k.k.piotrowski@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
More cleanup to pcnet32_loopback_test to release receive buffers if
device is not up. Created common routine to free rx buffers.
Tested ia32 and ppc64
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Suspend the chip if possible rather than stop and discard all tx and rx
frames, when changing the mcast list or entering/leaving promiscuous
mode. Created common pcnet32_suspend routine.
Tested ia32 and ppc64
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Fix pcnet32_set_ringparam to handle memory allocation errors without
leaving the adapter in an inoperative state and null pointers waiting to
be dereferenced.
Tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
On 2006-03-08 Eric Sesterhenn wrote:
converts drivers/net to kzalloc usage.
Don Fry modified it to use netif_msg_drv. Tested ia32 and ppc64.
Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Jon Mason wrote on Thu, 12 Jan 2006 17:07:49 -0600:
This patch adds the PCI_DEVICE macro to the pcnet32 driver.
This has been tested on my opteron with my "trident" adapter.
Don Fry modified it slightly and tested on ia32 and ppc64.
Signed-off-by: Jon Mason <jdmason@us.ibm.com>
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Symbols such as PCI_USES_IO, PCI_ADDR0, etc. originated from Donald
Becker's net driver template, but have been long unused. Remove.
In a few drivers, this allows the further eliminate of the pci_flags (or
just plain flags) member in the template driver probe structure.
Most of this logic is simply open-coded in most drivers, since it never
changes.
Made a few other cleanups while I was in there, too:
* constify, __devinitdata several PCI ID tables
* replace table terminating entries such as "{0,}," and "{NULL},"
with a more-clean "{ }".
Signed-off-by: Jeff Garzik <jeff@garzik.org>
During a code scan for another change I discovered that this call to
pcnet32_free_ring must be removed. If the open fails due to a lack of
memory all the ring structures are removed via the call to free_ring
and a subsequent call to open will dereference a null pointer in
pcnet32_init_ring.
Please apply to 2.6.17.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The comments concerning how the pcnet32 ethernet device driver selects
the MAC addr to use are incorrect. A recent patch (in the last 3 months)
changed how the code worked, but did not change the comments.
Side comment: the new behaviour is good; I've got a pcnet32 card which
powers up with garbage in the CSR's, and a good MAC addr in the PROM.
Signed-off-by: Linas Vepstas <linas@linas.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
- undo some Lindent damage by indenting member names
- remove history at top of .c file, this is stored in the kernel
repo changelog (in greater detail, even).
Boards with multiple PHYs were not being handled properly by the pcnet32
driver. This patch by Thomas Bogendoerfer with changes by me will allow
Allied Telesyn 2700FTX and 2701FTX boards to use either the copper or
the fiber interfaces. It has been tested on ia32 and ppc64 hardware.
Philippe Seewer also tested and improved the patch.
ethtool for pcnet32 already supports multiple phys.
See also bugzilla bug 4219.
Please apply to 2.6.16
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The CSR contains garbage after a coldboot on RS/6000.
One some systems (like my 44p 270) the MAC address is all FF,
on others (like my B50) it is ff:ff:ff:fd:ff:6b.
It can eventually be fixed by loading pcnet32, set the interface
into the UP state, rmmod pcnet32 and load it again. But this worked
only on the 270.
Only netbooting after a cold start provides the correct MAC address
via prom and CSR. This makes it very unreliable.
I dont know why the MAC is stored in two different places. Remove
the special case for powerpc, which was added in early 2.4 development.
Signed-off-by: Olaf Hering <olh@suse.de>
drivers/net/pcnet32.c | 5 -----
1 files changed, 5 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Some boards using the 79c976 pcnet32 chip will hang the system if the
ethtool --register-dump is performed with the device operational. The
request to read bcr30 is retried by the PCI device infinitely without
returning data, hanging the system.
Tested ia32 and ppc64.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
This patch is a better fix for Allied Telesyn 2700/2701 FX boards than
the change made in early January this year. It allows the user to
select the speed/duplex via module_param, but if no selection is made,
forces the speed to 100 FD. It fixes both Bugzilla bugs 2669 and 4551.
Tested ia32 and ppc64 by myself, and by the originator of bug 2669.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Display the name eth%d or pci_name() of device which fails to allocate
memory. When changing ring size via ethtool, it also releases the
lock before returning on error. Added comment that the caller of
pcnet32_alloc_ring must call pcnet32_free_ring on error, to avoid leak.
Tested ia32 by forcing allocation errors.
Signed-off-by: Don Fry <brazilnut@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Don Fry reminded me that the pcnet32_loopback_test() asssumes the ring size
is no less than 4. The minimum ring size was changed to 4 in
pcnet32_set_ringparam() to allow the loopback test to work unchanged.
- Set minimum ring size to 4 to allow loopback test to work unchanged
- Moved variable init_block to first field in struct pcnet32_private
Signed-off-by: Hubert WS Lin <wslin@tw.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
This patch implements the set_ringparam(), one of the ethtool operations,
which allows changing tx/rx ring sizes via ethtool.
- Changed memery allocation of tx/rx ring from static to dynamic
- Implemented set_ringparam()
- Tested on i386 and ppc64
Signed-off-by: Hubert WS Lin <wslin@tw.ibm.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Cc: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
David S. Miller