If the RTNL is held when we invoke flush_scheduled_work() we could
deadlock. One such case is linkwatch, it is a work struct which tries
to grab the RTNL semaphore.
The most common case are net driver ->stop() methods. The
simplest conversion is to instead use cancel_{delayed_}work_sync()
explicitly on the various work struct the driver uses.
This is an OK transformation because these work structs are doing
things like resetting the chip, restarting link negotiation, and so
forth. And if we're bringing down the device, we're about to turn the
chip off and reset it anways. So if we cancel a pending work event,
that's fine here.
Some drivers were working around this deadlock by using a msleep()
polling loop of some sort, and those cases are converted to instead
use cancel_{delayed_}work_sync() as well.
Signed-off-by: David S. Miller <davem@davemloft.net>
Since 43cc71eed1, the platform modalias is
prefixed with "platform:". Add MODULE_ALIAS() to the hotpluggable network
platform drivers, to re-enable auto loading.
NOTE: didn't change drivers/net/fs_enet/fs_enet-main.c "old binding" support.
That looks problematic in the first place (it even uses the ancient "struct
device_driver" binding scheme for platform_bus!) and I suspect it will vanish
soonish when arch/powerpc rules the world. Also, drivers/net/ne.c would have
needed more thought to sort out.
[akpm@linux-foundation.org: fix sgiseeq.c]
[dbrownell@users.sourceforge.net: more drivers, registration fixes]
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Scott Wood <scottwood@freescale.com>
Cc: Vitaly Bordug <vitb@kernel.crashing.org>
Cc: Dale Farnsworth <dale@farnsworth.org>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Andrew Victor <andrew@sanpeople.com>
Cc: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
This patch introduces struct smc91x_platdata and modifies the driver so
bus width is checked during run time using SMC_nBIT() instead of
SMC_CAN_USE_nBIT.
V2 keeps static configuration lean using SMC_DYNAMIC_BUS_CONFIG.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Acked-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Pass a private data pointer to macros and functions. This makes it easy
to later on make run time decisions. This patch does not change any logic.
These changes should be optimized away during compilation.
V2 changes the macro argument name from "priv" to "lp".
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Acked-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
smc91x is shared between many different platforms. Each platform needs
to specify the interrupt type, and in some cases the irq type depends
on more than just the build configuration - it depends on runtime
checks.
Rather than throwing this code into the SMC_IRQ_FLAGS definition, provide
a way for these flags to be passed via the IRQ resource itself.
Note that IRQF_TRIGGER_* constants are intentionally defined to correspond
with the IORESOURCE_IRQ_* interrupt type flags, in much the same way that
the low bits of PCI iomem resources correspond with the BAR flag bits.
Also provide a way to configure smc91x to read the IRQ flags from the
resource. Once all platforms have been converted over (signified
by all definitions of SMC_IRQ_FLAGS being -1) SMC_IRQ_FLAGS should
be removed.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Nicolas Pitre <nico@cam.org>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
The PXA DMA support code for smc91x doesn't pass a struct device to
the dma_*map_single() functions, which leads to an oops in the dma
bounce code. We have a struct device which was used to probe the
SMC chip. Use it.
(This patch is slightly larger because it requires struct smc_local
to move into the header file.)
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
We now have struct net_device_stats embedded in struct net_device,
and the default ->get_stats() hook does the obvious thing for us.
Run through drivers/net/* and remove the driver-local storage of
statistics, and driver-local ->get_stats() hook where applicable.
This was just the low-hanging fruit in drivers/net; plenty more drivers
remain to be updated.
[ Resolved conflicts with napi_struct changes and fix sunqe build
regression... -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>
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>
This lets the network core have the ability to handle suspend/resume
issues, if it wants to.
Thanks to Frederik Deweerdt <frederik.deweerdt@gmail.com> for the arm
driver fixes.
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix up arch-specific work items where possible to use the new work_struct and
delayed_work structs.
Three places that enqueue bits of their stack and then return have been marked
with #error as this is not permitted.
Signed-Off-By: David Howells <dhowells@redhat.com>
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)
When booting using root-nfs, I'm seeing (independently) two lockdep dumps
in the smc91x driver. The patch below fixes both. Both dumps look like
real locking issues.
Nico - please review and ack if you think the patch is correct.
Dump 1:
Sending DHCP requests .
=================================
[ INFO: inconsistent lock state ]
Signed-off-by: Jeff Garzik <jeff@garzik.org>
All accessor's different methods are now selected with C code and unused
ones statically optimized away at compile time instead of being selected
with #if's and #ifdef's. This has many advantages such as allowing the
compiler to validate the syntax of the whole code, making it cleaner and
easier to understand, and ultimately allowing people to define
configuration symbols in terms of variables if they really want to
dynamically support multiple bus configurations at the same time (with
the unavoidable performance cost).
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
platform_get_irq*() now returns on -ENXIO when the resource cannot be
found. Ensure all users of platform_get_irq*() handle this error
appropriately.
Signed-off-by: David Vrabel <dvrabel@arcom.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Some ARM platforms have the ability to program the interrupt controller to
detect various interrupt edges and/or levels. For some platforms, this is
critical to setup correctly, particularly those which the setting is dependent
on the device.
Currently, ARM drivers do (eg) the following:
err = request_irq(irq, ...);
set_irq_type(irq, IRQT_RISING);
However, if the interrupt has previously been programmed to be level sensitive
(for whatever reason) then this will cause an interrupt storm.
Hence, if we combine set_irq_type() with request_irq(), we can then safely set
the type prior to unmasking the interrupt. The unfortunate problem is that in
order to support this, these flags need to be visible outside of the ARM
architecture - drivers such as smc91x need these flags and they're
cross-architecture.
Finally, the SA_TRIGGER_* flag passed to request_irq() should reflect the
property that the device would like. The IRQ controller code should do its
best to select the most appropriate supported mode.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The smc91x driver relies upon register bank 2 being selected whenever
the interrupt handler is called. This isn't always so, especially if
we have a link change event during PHY configuration.
This results in register bank 0 being selected when the interrupt
handler is called, causing the wrong registers to be read for the
IRQ mask and status. In turn, this causes us to spin with a
permanently asserted IRQ.
The patch ensures that smc_phy_configure always exits with register
bank 2 selected.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Not only SMC_ACK_INT(IM_TX_EMPTY_INT) in in smc_hardware_send_pkt)
appears to be unnecessary (tested with an SMC91C94 and SMC91C111), but
it seems to trigger spurious interrupts on some machines as well.
Removed.
While at it, let's log any remaining spurious interrupts if any (and
clean usage of the max IRQ loop count value).
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
This allows us to eliminate the casts in the drivers, and eventually
remove the use of the device_driver function pointer methods for
platform device drivers.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Convert everyone who uses platform_bus_type to include
linux/platform_device.h.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
If the interface is not used right away after being probed it wastes
power needlessly. Noted by Holger Schurig.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
In PM v1, all devices were called at SUSPEND_DISABLE level. Then
all devices were called at SUSPEND_SAVE_STATE level, and finally
SUSPEND_POWER_DOWN level. However, with PM v2, to maintain
compatibility for platform devices, I arranged for the PM v2
suspend/resume callbacks to call the old PM v1 suspend/resume
callbacks three times with each level in order so that existing
drivers continued to work.
Since this is obsolete infrastructure which is no longer necessary,
we can remove it. Here's an (untested) patch to do exactly that.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For boards that invert the SMC91x IRQ line (maybe an FPGA inverts it),
the set_irq_type() call can't assume IRQT_RISING. These particular
boards currently use OMAP-specific calls to change the trigger type,
but the boards break when set_irq_type() stops being a NOP.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
The race causes a kernel oops when smc_hardware_send_pkt() tries to
dereference pending_tx_skb which would have been freed from one of the
driver reset paths just after the tx_task tasklet has been scheduled.
This race is possible on SMP but was uncovered by the kernel RT work.
Signed-off-by: Nicolas Pitre <nico@cam.org>
... and remove duplicate status defines.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Index: linux-2.6/drivers/net/smc91x.c
===================================================================
The PAGE_SIZE mask is indeed confusing. Use the exact mask for
this context which has nothing to do with memory pages at all.
Also cast to int since the value to compare with is an int.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
This fixes remaining u32s in drivers/ net.
Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
