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)
The UHCI controller in my laptop takes longer to turn off the
Resume-Detect bit than the 4 us allowed by uhci-hcd. Presumably other
computers will have the same problem.
This patch (as752) increases the maximum delay to 10 us, which should be
plenty, and uses polling to avoid penalizing systems which can turn the
bit off more quickly.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as706) removes the private hc_inaccessible flag from
uhci-hcd. It's not needed because it conveys exactly the same
information as the generic HCD_FLAG_HW_ACCESSIBLE bit.
In its place goes a new flag recording whether the controller is dead.
The new code allows a complete device reset to resurrect a dead
controller (although usbcore doesn't yet implement such a facility).
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as683) re-implements Full-Speed Bandwidth Reclamation (FSBR)
properly. It keeps track of which endpoint queues have advanced, and
when none have advanced for a sufficiently long time, FSBR is turned
off. The next TD on each of the non-moving queues is modified to
generate an interrupt on completion, so that FSBR can be re-enabled as
soon as the hardware starts to make some progress.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Someone recently posted a bug report where it turned out that uhci-hcd
was disagreeing with the UHCI controller over whether or not a port was
suspended: The driver thought it wasn't and the hardware thought it was.
This patch (as665) fixes the problem and simplifies the driver by
removing the internal state-tracking completely. Now the driver just
asks the hardware whether a port is suspended.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as657) increases the port-reset completion delay in uhci-hcd
for HP's embedded controllers. Unlike other UHCI controllers, the HP
chips can take as long as 250 us to carry out the processing associated
with finishing a port reset.
This fixes Novell bug #148761.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as558) removes from the UHCI driver a kernel timer used for
checking Full Speed Bandwidth Reclamation (FSBR). The checking can be
done during normal root-hub polling; it doesn't need a separate timer.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch, which has as478b as a prerequisite, enables the uhci-hcd
driver to take advantage of root-hub IRQs rather than polling during the
time it is suspended. (Unfortunately the hardware doesn't support
port-change interrupts while the controller is running.) It also turns
off the driver's private timer while the controller is suspended, as it
isn't needed then. The combined elimination of polling interrupts and
timer interrupts ought to be enough to allow some systems to save a
noticeable amount of power while they are otherwise idle.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch implements (finally!) separate suspend and resume routines
for the root hub and the controller in the UHCI driver. It also
changes the sequence used to reset the controller during initial
probing, so as to preserve the existing state during a Resume-From-Disk.
(This new sequence is what should be used in the PCI Quirks code for
early USB handoffs, incidentally.) Lastly it adds a notion of the
controller being "inaccessible" while in a PCI low-power state, when
normal I/O operations shouldn't be allowed.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch starts making some serious changes to the UHCI driver.
There's a set of private states for the root hub, and the internal
routines for suspending and resuming work completely differently, with
transitions based on the new states. Now the driver distinguishes
between a privately auto-stopped state and a publicly suspended state,
and it will properly suspend controllers with broken resume-detect
interrupts instead of resetting them.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch makes a few small improvements in the UHCI driver. Some
code is moved between different source files and a more useful pointer
is passed to a callback routine.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
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!
David Howells