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)
this patch fix leak of memory allocated to intr if allocation of
sc->urb_int fails.
Found by the Coverity checker.
Signed-off-by: Duncan Sands <baldrick@free.fr>
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
- increase ack timeout for slow system (geode 233MHz where HZ=100)
- reset the cmv ack flag when rebooting
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
- improve debug trace in order to make easy to solve user problems.
- indent some code
- increase version number
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
this patch correct a possible bug with cmv_name being static. If there
is 2 modems and the driver is scheduled when filling cmv_name this could
result with garbage in cmv_name. We allocate cmv_name on the stack but
with a small size in order to avoid that.
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch adds the support for isochronous pipe.
A new module parameter is added to select iso mode. It is set to iso by
default because bulk mode doesn't work well at high speed rate (>3 Mbps
for upload).
We use UDSL_IGNORE_EILSEQ flags because ADI firmware doesn't reply to
ISO IN when it has nothing to send [1].
[1]
from cypress datasheet :
The ISOSEND0 Bit (bit 7 in the USBPAIR Register) is used when the EZ-USB
FX chip receives an isochronous IN token while the IN FIFO is empty. If
ISOSEND0=0 (the default value), the USB core does not respond to the IN
token. If ISOSEND0=1, the USB core sends a zero-length data packet in
response to the IN token. The action to take depends on the overall
system design. The ISOSEND0 Bit applies to all of the isochronous IN
endpoints, IN-8 through IN-15.
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is the usbatm part of the Arjan, Jes and Ingo
mass semaphore to mutex conversion, reworked to apply on top
of the patches I just sent to you. This time, with correct
attribution and signed-off lines.
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Duncan Sands <baldrick@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
While the usbatm core has had some support for using isoc urbs
for some time, there was no way for users to turn it on. While
use of isoc transfer should still be considered experimental, it
now works well enough to let users turn it on. Minidrivers signal
to the core that they want to use isoc transfer by setting the new
UDSL_USE_ISOC flag. The speedtch minidriver gets a new module
parameter enable_isoc (defaults to false), plus some logic that
checks for the existence of an isoc receive endpoint (not all
speedtouch modems have one).
Signed-off-by: Duncan Sands <baldrick@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Have minidrivers and the core signal special requirements
using a flags field in struct usbatm_data. For the moment
this is only used to replace the need_heavy_init bind
parameter, but there'll be new flags in later patches.
Signed-off-by: Duncan Sands <baldrick@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
More care on loading firmware, take into account fw->size can't be zero.
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
A driver for USB ADSL modems based on the ADI eagle chipset using the
usb_atm infrastructure.
The managing part was taken from bsd ueagle driver, other parts were
written from scratch.
The driver uses the in-kernel firmware loader :
- to load a first usb firmware when the modem is in pre-firmware state
- to load the dsp firmware that are swapped in host memory.
- to load CMV (configuration and management variables) when the modem
boot. (We can't use options or sysfs for this as there many possible
values. See
https://mail.gna.org/public/eagleusb-dev/2005-04/msg00031.html for a
description of some)
- to load fpga code for 930 chipset.
The device had 4 endpoints :
* 2 for data (use by usbatm). The incoming
endpoint could be iso or bulk. The modem seems buggy and produce lot's
of atm errors when using it in bulk mode for speed > 3Mbps, so iso
endpoint is need for speed > 3Mbps. At the moment iso endpoint need a
patched usbatm library and for this reason is not included in this patch.
* One bulk endpoint for uploading dsp firmware
* One irq endpoint that notices the driver
- if we need to upload a page of the dsp firmware
- an ack for read or write CMV and the value (for the read case).
If order to make the driver cleaner, we design synchronous
(read|write)_cmv :
-send a synchronous control message to the modem
-wait for an ack or a timeout
-return the value if needed.
In order to run these synchronous usb messages we need a kernel thread.
The driver has been tested with sagem fast 800 modems with different
eagle chipset revision and with ADI 930 since April 2005.
Signed-off-by: Matthieu CASTET <castet.matthieu@free.fr>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
