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>
Fix 4032 chip undocumented "feature" where bit-8 is set
if the inbound completion is for a VLAN.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
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>
The code for checksum is more complex than needed when dealing with VLAN's;
the higher layers already pass down the location of the IP header.
Compile tested only, no hardware available.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Reading the code for ql_hw_csum_setup(), it is obvious that
this driver is broken for IPV6. The driver sets the NETIF_F_HW_SUM
flag, but the code for checksum setup only deals with IPV4.
Compile tested only, no hardware available.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
On Fri, Mar 30, 2007 at 01:05:59AM -0700, Andrew Morton wrote:
>...
> Changes since 2.6.21-rc5-mm2:
>...
> git-netdev-all.patch
>...
> git trees
>...
This patch makes the needlessly global PHY_DEVICES[] static.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This PHY support patch was written by Benjamin Li.
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
To clearly state the intent of copying from linear sk_buffs, _offset being a
overly long variant but interesting for the sake of saving some bytes.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
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>
Fixed rx checksum bits. Turn on TCP processing for rx checksum.
Fixed max frame length register write. It wasn't getting set
in multi-port system. Set rx buffer queue length properly
for jumbo frames.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Update the rx queue pointer when exiting NAPI poll rather than
at the end of each iteration. Remove unnecessary PCI flushes
that occurred after every write. Now write all regs and
flush once.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This was removed in a previous patch to increase performance, but
caused a transmit error for the 4032 chip.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This missing line caused transmit errors on the Qlogic 4032 chip.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
To reproduce this panic consistently, we run an intensive network
application like 'netperf'. After waiting for a couple of seconds,
you will see a stack trace and a kernel panic where we are calling
pci_unmap_single() in ql_poll().
Changes:
1) Check the flags on the Response MAC IO Control block to check for
errors
2) Ensure that if we are on the 4022 we only use one segment
3) Before, we were reading the memory mapped producer index register
everytime we iterated in the loop when clearing the queue. We should
only be iterating to a known point, not as the producer index
is being updated.
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
To Reproduce the Problem:
To reproduce this panic consistently, we run an intensive network
application like 'netperf' and then switch to a different console.
After waiting for a couple of seconds, you will see a tx reset has occured.
Reason:
We enable interrupts even if we were not running.
Solution:
Now we will enable interrupts only after we are ready to give up the poll
routine.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
pci_map_single() could fail. We need to properly check the return
code from pci_map_single(). If we can not properly map this address,
then we should cleanup and return the proper return code.
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The scatter/gather lists were not being build correctly. When
large frames spanned several buffers the chip would panic.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This change removes use of constants for rx buffer queue size
and instead calculates the queue length based on what he MTU
is set to.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
This network device driver shares the same hardware as the qla4xxx
iSCSI driver. Changing the MTU via the device interface will
cause qla4xxx to crash as there is no way to make notification.
Users wishing to change the MTU must do so using an iSCSI
utility such as Qlogic SanSurfer. This forces the user to
unload/reload this network device driver after the MTU
value has been changed in flash.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
1) Fix deadlock issue when in QL_RESET_ACTIVE state and traversing
through the Link State Machine
2) Fix deadlock issue when ethtool would call ql_get_settings()
3) Fix deadlock issue when adaptor is ifup'ed but adaptor fails to initialize
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
We do not need to zero out the 64 byte MAC request I/O control block.
By zeroing out the control block and setting it to proper fields is
redundant work. This is because in the qla3xxx_send() function we will
already set the proper fields in this structure. The unused fields are
not looked at by the hardware and do not need to be zeroed out.
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
The return code was not properly set when when allocating memory or mapping
memory failed. Depending on the stack, the return code would sometimes
return 0, which indicates everything was ok, when in fact there was an error.
This would cause trouble when the module was removed. Now, we will pass
back the proper return code when an error occurs during the PCI probe.
Signed-off-by: Benjamin Li <benjamin.li@qlogic.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Qlogic 4032 chip is an incremental change from the 4022.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
When accessing the 93LC86 serial prom the clock high and low times must be at least 250ns each. We have seen on some systems where the access times were much lower casing bit errors.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Driver TX locking was removed some time ago, but the flag was overlooked.
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
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)
ql3xxx_probe() does ioremap and stores result in ->mem_map_registers.
On failure exit it does iounmap() of the same thing.
OTOH, ql3xxx_remove() does iounmap() of ->mmap_virt_base which is
(a) never assigned and
(b) never used other than in that iounmap() call.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This is a complementary network driver for our ISP4XXX parts.
There is a concurrent effort underway to get the iSCSI driver (qla4xxx)
integrated upstream as well.
I have been through several iterations with the linux-netdev list and have had
much response from Stephen Hemminger.
- Built and tested using kernel 2.6.17-rc4.
- The chip supports two ethernet and two iSCSI functions.
- The functions ql_sem_lock, ql_sem_spinlock, ql_sem_unlock, and
ql_wait_for_drvr_lock are used to protect resources that are shared across
the network and iSCSI functions. This protection is mostly during chip
initialization and resets, but also include link management.
- The PHY/MII are not exported through ethtool due to the fact that the
iSCSI function will control the common link at least 50% of the time.
This driver has been through several iterations on the netdev list and we feel
this driver is ready for inclusion in the upstream kernel.
It has been built and tested on x86 and PPC64 platforms.
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
