While trying to fix http://bugzilla.kernel.org/show_bug.cgi?id=8952
I looked at a few other drivers to figure out what drivers _should_
be doing for suspend/resume. I noticed typhoon driver is likely doing
more than it needs to. Patch below is untested since I don't have the HW.
Suspend/resume code across NIC drivers is fairly inconsistent.
And I couldn't find any documentation on what the canonical sequence
NICs need to do for suspend or resume. Is there any?
Barring contrary advice, I'm going model the tulip suspend/resume
fixes after tg3.c since a number of "modern" (< 5 years old) laptops
have that and I'm silly enough to assume it works.
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
This patch adds ibm_newemac PHY clock workaround for 440EP/440GR EMAC
attached to a PHY which doesn't generate RX clock if there is no link.
The code is based on the previous ibm_emac driver stuff. The 440EP/440GR
allows controlling each EMAC clock separately as opposed to global clock
selection for 440GX.
BenH: Made that #ifdef CONFIG_PPC_DCR_NATIVE for now as dcri_* stuff doesn't
exist for MMIO type DCRs like Cell. Some future rework & improvements of the
DCR infrastructure will make that cleaner but for now, this makes it work.
Signed-off-by: Valentine Barshak <vbarshak@ru.mvista.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The PowerPC 440GX Taishan board fails to reset EMAC3 (reset timeout
error) if there's no link. Because of that it fails to find PHY
chip. The older ibm_emac driver had a workaround for that: the
EMAC_CLK_INTERNAL/EMAC_CLK_EXTERNAL macros, which toggle the Ethernet
Clock Select bit in the SDR0_MFR register. This patch does the same for
"ibm,emac-440gx" compatible chips. The workaround forces clock on -all-
EMACs, so we select clock under global emac_phy_map_lock.
BenH: Made that #ifdef CONFIG_PPC_DCR_NATIVE for now as dcri_* stuff
doesn't exist for MMIO type DCRs like Cell. Some future rework &
improvements of the DCR infrastructure will make that cleaner but
for now, this makes it work.
Signed-off-by: Valentine Barshak <vbarshak@ru.mvista.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Convert ibm_newemac to use the of_device_is_available function when checking
for unused/unwired EMACs. We leave the current check for an "unused" property
to maintain backwards compatibility for older device trees. Newer device
trees should simply use the standard "status" property in the EMAC node.
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
This patch fixes several section mismatch warnings in the
ibm_newemac driver similar to:
WARNING: vmlinux.o(.devinit.text+0x3a04): Section mismatch in reference from the function emac_probe() to the function .devexit.text:tah_detach()
The function __devinit emac_probe() references
a function __devexit tah_detach().
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
On some 4xx PPC's (e.g. 460EX/GT), the rx channel number is a multiple
of 8 (e.g. 8 for EMAC1, 16 for EMAC2), but enabling in MAL_RXCASR needs
the divided by 8 value for the bitmask.
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
This fixes the jumbo frame support on EMAC V4 systems. Now the correct
bit is set depending on the EMAC version configured.
Tested on Kilauea (405EX) and Canyonlands (460EX).
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
To enable EEH support for pci-express network adapters, pcie/msi state
needs to be saved and restored for that adapter.
[after similar patches for ixgbe and e1000e from Wendy Xiong]
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Cc: Wendy Xiong <wendyx@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
To enable EEH support for pci-express network adapters, pcie/msi state
needs to be saved and restored for that adapter.
Tested this EEH patch with 2ports and 4ports pci-express e1000e
adapters.
Signed-off-by: Wendy Xiong <wendyx@us.ibm.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
To enable EEH support for pci-express network adapters, pcie/msi state
needs to be saved and restored for that adapter.
Tested this EEH patch with Intel 10G pci-express ixgbe adapter.
Signed-off-by: Wendy Xiong <wendyx@us.ibm.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The lower limit of 80 descriptors in the ring is only valid for
one older 8254x chipset. All e1000e devices can use as low as
64 descriptors.
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Several components to this complex fix. The es2lan cards occasionally
gave a "HW Error" especially when forcing speed. Some users also
reported that the BMC stole ARP packets.
The fixes include setting the proper SW_FW bits to tell the BMC
that we're active and not do any un-initialization at all, so the
setup routine is largely changed.
Signed-off-by: David Graham <david.graham@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The ethtool -c / -C interface can now be used to modify the
irq moderation algorithm. This change does not require an
adapter reset and can thus be used at all times. The adapter
only supports changing/reading rx-usecs which has special
values for 0, 1 and 3:
0 - no irq moderation whatsoever
1 - normal moderation favoring regular mixed traffic (default)
3 - best attempt at low latency possible at cost of CPU
For values between 10 and 10000 the rx-usecs defines "the minimum
time between successive irqs" in usec, unlike the module parameter.
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Several stats registers are completely unused and we just waste pci
bus time reading them. We also omit using the high 32 bits of the GORC/
GOTC counters. We can just read clear them and only read the low registers.
Mii-tool can also break es2lan if it executes a MII PHY register
ioctl while the device is in autonegotiation. Unfortunately it seems
that several applications and installations still perform this ioctl
call periodically and especially in this crucial startup time. We
can fool the ioctl by providing fail safe information that mimics
the "down" link state and only perform the dangerous PHY reads once
after link comes up to fill in the real values. As long as link
stays up the information will not change.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
- Removed receive buffer replenishment tasklet s2io_tasklet and instead
allocating the receive buffers in either the interrupt handler (no napi)
or the napi handler (napi enabled).
Signed-off-by: Surjit Reang <surjit.reang@neterion.com>
Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: Ramkrishna Vepa <ram.vepa@neterion.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The patch that changed mdio_bus to a string didn't conflict strongly enough
with the patch that added fixed PHY support to UCC. Gather it back into
the fold.
Fixes this error:
...
CC drivers/net/ucc_geth.o
'ucc_geth_probe':
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/net/ucc_geth.c:3935: error:
incompatible types in assignment
make[3]: *** [drivers/net/ucc_geth.o] Error 1
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
When count reaches 0 the postfix decrement still subtracts (to -1),
so bfin_reset_controller() returns as if the busy flag was cleared
while it was not.
Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
Acked-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Re-enable hotplug, now that the interrupt/error handling are mostly sane.
Also update the TODO list at the top.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Here it is again, minus the checkpatch.pl complaint:
Rework mv_err_intr() to leave the SError bits as-is,
so that libata-eh has a chance to see/use them.
We originally thought that clearing them here was necessary
before writing back to edma_err_cause (per the Marvell datasheets),
but we will end up reseting the chip regardless in those cases.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Continue fixing the interrupt handling logic.
Get rid of mv_intr_pio(), by using ata_sff_host_intr() for PIO..
Add a mv_unexpected_intr() catch-all for "impossible" scenarios,
where we get an interrupt that shouldn't have happened
(never seen in testing, but just in case..).
Rearrange the logic so that we always process completed
response queue entries before looking for other events,
This avoids having to re-issue commands that had already succeeded.
As part of this, we split out some duplicated functionality
into a new function, mv_get_active_qc().
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Tidy up host controller interrupt handling, by moving the weirdo
bit shifting from mv_interrupt() to mv_host_intr().
This lets us take advantage of the MV_PORT_TO_SHIFT_AND_HARDPORT() macro
from an earlier patch to greatly simplify the port numbering logic.
Also, defer reading the hc_irq_cause (one per hc) until it is
actually proven to be needed. This may save a microsecond or
so per interrupt, on average (a later patchset will further reduce
unnecessary register reads throughout the driver).
Apart from that, we still leave the actual IRQ handling logic alone.
Subsequent patches in this series will address that.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Try and simplify handling of the request/response queues.
Maintain the cached copies of queue indexes in a fully-masked state,
rather than having each use of them have to do the masking.
Split off handling of a single crpb response into a separate function,
to reduce complexity in the main mv_process_crpb_entries() routine.
Ignore the rarely-valid error bits from the crpb status field,
as we already handle that information in mv_err_intr().
For now, preserve the rest of the original logic.
A later patch will deal with fixing that separately.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Introduce the MV_PORT_TO_SHIFT_AND_HARDPORT() macro,
to centralize/simplify various scattered bits of logic
for calculating bit shifts and the like.
Some of the places that do this get it wrong, too,
so consolidating the algorithm at one place will help
keep the code correct.
For now, we use the new macro in mv_eh_{freeze,thaw}.
A subsequent patch will re-use this in the interrupt handlers
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Ignore *all* interrupt coalescing bits on all controllers,
not just some of each.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
More cosmetic cleanups prior to the interrupt/error handling logic changes.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The drive is directly soldered to the controller, so there is no cable at
all. Remove the 40-wire assumption so the drive can operate at max speed.
Before patch:
$ dd if=/dev/sda of=/dev/null bs=2M count=64 iflag=direct
134217728 bytes (134 MB) copied, 5.29612 s, 25.3 MB/s
After patch:
$ dd if=/dev/sda of=/dev/null bs=2M count=64 iflag=direct
134217728 bytes (134 MB) copied, 3.94955 s, 34.0 MB/s
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Fix mis-reporting of NCQ errors by ensuring that result_tf->flags
is properly initialized in libata-eh. This allows ata_gen_ata_sense()
to report the failed block number correctly to SCSI after a media error
during NCQ.
This patch may also be a candidate for backporting to earlier kernels.
Without this fix, SCSI will fail I/O on the entire request rather
than just the bad sector. That can be bad for a request that was
merged from many independent read reads from different tasks.
Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
These #if's are unneeded since they:
- did anyway not handle the CONFIG_ACPI_DOCK_MODULE case correctly and
- this is already handled in include/acpi/acpi_drivers.h and
- it's now correctly handled in kconfig.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Acked-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
sata_nv hardreset can't classify but was left out while unifying
follow-up SRST request mechanism[1]. This caused detection failures
on those controllers. Fix it.
Reported and bisected by Roland Dreier, Petr Vandrovec and Marc
Dionne. Thanks guys.
[1] 305d2a1ab1
Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: Roland Dreier <rdreier@cisco.com>
Cc: Petr Vandrovec <vandrove@vc.cvut.cz>
Cc: Marc Dionne <marc.c.dionne@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Some chips need AHCI_EN set more than once to actually set it. Try a
few times before giving up and spitting out WARN_ON().
Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: Peer Chen <pchen@nvidia.com>
Cc: Volker Armin Hemmann <volker.armin.hemmann@tu-clausthal.de>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
WARN_ON()'s in ata_hsm_move() was too liberal and got triggerred when
it shouldn't (e.g. hotplug events at the right moment). As the HSM
only deals with device errors and state machine violations, make it
check only against them.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: Mark Lord <liml@rtr.ca>
Cc: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Since 43cc71eed1, the platform modalias
is prefixed with "platform:". Add MODULE_ALIAS() to the hotpluggable
ATA and IDE platform drivers, to re-enable auto loading.
NOTE: both ata/pata_platform.c and ide/legacy/ide_platform.c claim
to provide "the" platform_pata driver, and there's no build-time
mutual exclusion mechanism. This means that configs which enable
both drivers will make some trouble when hotplugging...
[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: Tejun Heo <htejun@gmail.com>
Acked-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
For WUSB devices, usb_dev.devnum is a device index and not the real
device address (which is managed by wusbcore). Therefore, only set
devnum once (in choose_address()) and never change it.
Signed-off-by: David Vrabel <david.vrabel@csr.com>
Cc: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We need to be able to call ep0_reinit() [renamed to usb_ep0_reinit()]
from the WUSB security code. The reason is that when we authenticate
the device, it's address changes (from having bit 7 set to having it
cleared). Thus, we need to signal the USB stack to reinitialize EP0,
so the status with the previous address kept at the HCD layer is
cleared and properly reinitialized.
Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
A WUSB device gets his address during the connection phase; later on,
during the authenthication phase (driven from user space) we assign
the final address. So we need to skip in hub_port_init() the actual
setting of the address for WUSB devices.
Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Modify choose_address() so it knows about our special scheme of
addressing WUSB devices (1:1 w/ port number).
Signed-off-by: Inaky Perez-Gonzalez <inaky@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
- If a termios change fails due to lack of memory we should copy the
old settings back over as the device has not changed
- Note various locking problems
- kl5kusb105 had various remaining tty flag handling problems
- Make safe_serial use tty_insert_flip_string not open coded loops
- set termios speed properly in usb_serial
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1079) cleans up the way URB_* flags are exported in
usbfs.
The URB_NO_INTERRUPT flag is now exported (this is the
only behavioral change).
USBDEVFS_URB_* macros are added for URB_NO_FSBR,
URB_ZERO_PACKET, and URB_NO_INTERRUPT, making explicit the
fact that the kernel accepts them.
The flag matching takes into account that the URB_* values
may change as the kernel evolves, whereas the USBDEVFS_URB_*
values must remain fixed since they are a user API.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1072) fixes some recently-introduced compile problems
that show up in ehci-hcd when CONFIG_PM is turned off.
PORT_WAKE_BITS needs to be defined always.
ehci_port_power() is called during initialization by all the
EHCI variants other than the PCI version, in which it is
"defined but not used". So add a call to it.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Simplify processing of completed qtds, and correct handling of short
reads, by removing two state variables:
- "qtd_status" wasn't needed. The current URB's status is either
OK (-EINPROGRESS) or some fault status. Once a fault appears,
the queue halts and any later QTDs are immediately removed, so
no temporary status is needed. (Or for typical short reads,
it's not treated as a fault, so no queue halt is needed.)
- "do_status" was erroneous. Because of how the queue is set up,
short control reads can (and should!) be treated like full size
reads, and cleaned up the usual way. The status stage will be
executed transparently, and usbcore handles the choice of whether
to report this status as unexected.
The "do_status" problem caused a rather perplexing timing-dependent
problem with usbtest case 10. Sometimes it would make the controller
skip a dozen transactions while (wrongly) trying to clean up after a
short transfer. Fortunately, removing a dcache contention issue made
this become trivial to reproduce (on one test rig), so enough clues
finally presented themselves ... I think this has been around for a
very long time, but was worsened by recent urb->status changes.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix bogus assignment of "unsigned char *" to "char *": preserve
unsignedness. These values are used directly as descriptor lengths
when iterating through the buffer, so this *could* cause oddness
that potentially includes oopsing. (IMO not likely, except as
part of a malicious device...)
Fix the bogus warning in CDC ACM which highlighted this problem
(by showing a negative descriptor type). It uses the undesirable
legacy err() for something that's not even an error; switch to
use dev_dbg, and show descriptor types in hex notation to match
the convention for such codes.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Acked-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Francois Romieu