ipath_driver.c:1260: warning: format '%Lx' expects type 'long long unsigned int', but argument 6 has type 'long unsigned int'
    ipath_driver.c:1459: warning: format '%Lx' expects type 'long long unsigned int', but argument 4 has type 'u64'
    ipath_intr.c:358: warning: format '%Lx' expects type 'long long unsigned int', but argument 3 has type 'u64'
    ipath_intr.c:358: warning: format '%Lu' expects type 'long long unsigned int', but argument 6 has type 'u64'
    ipath_intr.c:1119: warning: format '%Lx' expects type 'long long unsigned int', but argument 5 has type 'u64'
    ipath_intr.c:1119: warning: format '%Lx' expects type 'long long unsigned int', but argument 3 has type 'u64'
    ipath_intr.c:1123: warning: format '%Lx' expects type 'long long unsigned int', but argument 3 has type 'u64'
    ipath_intr.c:1130: warning: format '%Lx' expects type 'long long unsigned int', but argument 4 has type 'u64'
    ipath_iba7220.c:1032: warning: format '%llx' expects type 'long long unsigned int', but argument 4 has type 'u64'
    ipath_iba7220.c:1045: warning: format '%llX' expects type 'long long unsigned int', but argument 3 has type 'u64'
    ipath_iba7220.c:2506: warning: format '%Lu' expects type 'long long unsigned int', but argument 4 has type 'u64'
Signed-off-by: NAlexander Beregalov <a.beregalov@gmail.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Now that we always use PIO for vl15 on 7220, we could get stuck forever
if we happened to run out of PIO buffers from the verbs code, because
the setup code wouldn't run; the interrupt was also ignored if SDMA was
supported.  We also have to reduce the pio update threshold if we have
fewer kernel buffers than the existing threshold.

Clean up the initialization a bit to get ordering safer and more
sensible, and use the existing ipath_chg_kernavail call to do init,
rather than doing it separately.

Drop unnecessary clearing of pio buffer on pio parity error.

Drop incorrect updating of pioavailshadow when exitting freeze mode
(software state may not match chip state if buffer has been allocated
and not yet written).

If we couldn't get a kernel buffer for a while, make sure we are
in sync with hardware, mainly to handle the exitting freeze case.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The semantics of cancel_sends changed, but the code using it was missed.
Don't leave sends and pioavail updates disabled, and add a comment as to
why the force update is needed.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

This patch adds the initialization calls into the new 7220 HCA files,
changes the Makefile to compile and link the new files, and code to
handle send DMA.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The patch adds a number of minor changes to support newer HCAs:
 - New send buffer control bits
 - New error condition bits
 - Locking and initialization changes
 - More send buffers
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The hardware-based recovery doesn't need any intervention, and in a few
cases we can get a bit confused about state and skip steps such as
turning off the link state LED when we consider recovery to be "down".
So ignore this transition, and either we recover in hardware, or we
transition to down, and will handle it then.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Newer HCAs have a HW option to write a sequence number to each receive
queue entry and avoid a separate DMA of the tail register to memory.
This patch adds support for these changes.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

This patch makes some white space changes and minor non-functional
changes to more closely match the code in OFED-1.3.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

A fixed partitioning of send buffers is determined at driver load time
for user processes and kernel use.  Since send buffers are a scarce
resource, it makes sense to allow the kernel to use the buffers if they
are not in use by a user process.

Also, eliminate code duplication for ipath_force_pio_avail_update().
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The link can be put in LINKDOWN_DISABLE state either locally or via a
MAD.  However, the link-recovery code will take it out of that state as
a side-effect of attempts to clear SerDes/XGXS issues.

We add a flag to indicate "link is down on purpose, leave it alone."
Signed-off-by: NMichael Albaugh <michael.albaugh@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Workaround a QLE7140 problem that in rare cases causes flow control
problems after link recovery by forcing a link retrain after recovery.
A module parameter is provided to control the behavior in case it causes
problems.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

There's a conflict between our need to quiesce PSM-based applications
to avoid HoL blocking when the IB link goes down and the apps' desire
to remain running so that their quiescence timout mechanism can keep
running.

The compromise is to STOP the processes for a fixed period of time and
then alternate between CONT and STOP until the link is again active.

If there are poor interactions with subnet manager configuration at a
given site, the interval can be adjusted via a module paramter.
Signed-off-by: NJohn Gregor <john.gregor@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The error recovery code for updating the driver's cached status information
for which send buffers are busy or free wasn't updated for IBA7220.
It should be similar to the initialization code in enable_chip().
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Fix byte order of value assigned to pioavailshadow.  This bug was
detected by sparse endianness warnings.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Recent sparse versions and kernel cleanups knock down the false positive
rate of the ipath driver code to a point where having it be sparse clean
is worthwhile. Here we fixup the sparse warnings.  Some of these warnings
(and the impetus to run sparse again) are due to work by Roland Dreier.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Different chips have different width interrupt status registers, so add
a flag and accessor function to decide which width register read to use.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The 6110 had a bug that caused some registers to be swapped; it was
fixed for the 7220 (and didn't affect the 6120 because it had fewer
registers).  This adds a flag and related code to handle that, and
includes some minor cleanups in the same area.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>

This patch moves some arrays that were defined per-device to be
variables defined in the per context data structure, thus avoiding extra
kzalloc() calls.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

In preparation for upcoming chips that have different values for
INFINIPATH_R_PORTENABLE_SHIFT, INFINIPATH_R_INTRAVAIL_SHIFT,
INFINIPATH_R_TAILUPD_SHIFT, and portcfg_shift, remove the shared
#defines and use device-specific variables instead.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

kreceive is now portdata * instead of devdata * and other kreceive
related cleanups....
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

An internal code review found the comment here lacking -- update it with 
more specifics of how and why the rmb() is there.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Code review pointed out that the locking around uses of ipath_sendctrl 
and kr_sendctrl were, in several places, incorrect and/or inconsistent.
Signed-off-by: NJohn Gregor <john.gregor@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

At one point in time there was code to allow a user process to
wait for a send buffer if none were available. This feature was
never used and most of the code was removed. This removes
some missed unused code.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Improve interrupt handler cache footprint by noinline'ing error
functions that are rarely called.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Inside a function declared as

    void foo(char bar[512])

the value of sizeof bar is the size of a pointer, not 512.  So avoid
constructions like this by passing the size explicitly.

Also reduce the size of the buffer to 128 bytes (512 was overly generous).
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The link state event calls were being generated when the SM told the SMA
to change link states. This works for IB_EVENT_PORT_ACTIVE but not if
the link goes down and stays down. The fix is to generate event calls
from the interrupt handler when the HW link state changes.
Signed-off-by: NRalph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The General Purpose I/O pins can be configured to cause interrupts. At
the end of the interrupt code dealing with all known causes, a message
is output if any bits remain un-handled. Since this is a "can't happen"
scenario, it should only be triggered by bugs elsewhere. It is harmless,
and potentially beneficial, to limit the damage by masking any such
unexpected interrupts.

This patch adds disabling of interrupts from any pins that should
not have been allowed to interrupt, in addition to emitting a message.
Signed-off-by: NMichael Albaugh <Michael.Albaugh@Qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

ipath_poll() suffered from a couple subtle bugs.  Under the right
conditions we could leave recv interrupts enabled on an ipath user
context on close, thereby taking potentially unwanted interrupts on the
next open -- this is fixed by unconditionally turning off recv
interrupts on close.  Also, we now use counters rather than set/clear
bits which allows us to make sure we catch all interrupts at the cost of
changing the semantics slightly (it's now give me all events since the
last time I called poll() rather than give me all events since I called
_this_ poll routine).  We also added some memory barriers which may help
ensure we get all notifications in a timely manner.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

On iba6110 rev4, support for three more IB counters were added.  The
LocalLinkIntegrityError counter, the ExcessiveBufferOverrunErrors
counter and support for error counting of flow control packets on an
invalid VL.  These counters trigger GPIO interrupts and the sw keeps
track of the counts.  Since we also use GPIO interrupts to signal packet
reception, we need to turn off the fast interrupts, or we risk losing a
GPIO interrupt.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

On some system hardware, we are seeing moderately common cases of the
chip errormask register being overwritten due to a chip bug in iba6120
that is triggered by a vendor-specific PCIe broadcast message.  This
patch merely checks periodically, and corrects it if needed (the
overwrite can cause us to not get error and hardware error
interrupts).  Also, make dd->ipath_errormask the one, true canonical
source for kr_errormask, and remove references to ipath_ignorederrs as
it is currently unused.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NJohn Gregor <john.gregor@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

There was confused use of INFINIPATH_S_PIOBUFAVAILUPD (value) and
IPATH_S_PIOBUFAVAILUPD (bit position).  Also, some callers of
ipath_cancel_sends() need kr_sendctrl restored, and some want to do it
later.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

The fastrcvint code's purpose was to avoid reading the interrupt
status if kernel packets were in the receive queue (to reduce
overhead).  Because intstatus was not read, we could miss the error
interrupt bit indicating freeze mode, since it only delivers a single
interrupt, even if still pending after intclear is written.

This patch removes that unsafe optimization.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Make some functions that are only used in a single .c file static.  In
addition to being a cleanup, this shrinks the generated code.  On x86_64:

add/remove: 1/3 grow/shrink: 2/1 up/down: 4777/-4956 (-179)
function                                     old     new   delta
handle_errors                                  -    3994   +3994
__verbs_timer                                 42     710    +668
ipath_do_ruc_send                           2131    2246    +115
ipath_no_bufs_available                      136       -    -136
ipath_disarm_senderrbufs                     639       -    -639
ipath_ib_timer                               658       -    -658
ipath_intr                                  5878    2355   -3523
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

All too often, interrupts do not get enabled for our card due to BIOS
misconfiguration and other issues.  This patch checks for that
condition on startup and warns the user.  This patch is based on work
(check LID availability) by Robert Walsh.
Signed-off-by: NArthur Jones <arthur.jones@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

We are more careful to be sure that we don't lose information about
changes that occurred while we were in freeze mode, when the chip will
not notify us, and try to avoid false error interrupts while doing
cleanup.  Put all of this logic in a new function ipath_clear_freeze().
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Now that it's June, it's about time to update
the copyright notices of files that have changed.
Signed-off-by: NJohn Gregor <john.gregor@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

Fix ipath_poll and enhance it so we can poll for urgent packets or
regular packets and receive notifications of when a header queue
overflows.
Signed-off-by: NRobert Walsh <robert.walsh@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

This centralizes the use of the abort functionality, removes the
unneeded buffer cancel (abort does the same thing), sets up to ignore
launch errors after abort, same as cancel.  We need abort on exit from
freeze mode to avoid having buffers stuck in the busy state, if a user
process happened to complete the send while we were in freeze mode
doing the recovery.
Signed-off-by: NDave Olson <dave.olson@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>

We currently track various errors, now we enhance that capability by
logging some of them to EEPROM.  We also now log a cumulative "active"
time defined by traffic though the InfiniPath HCA beyond the normal SM
traffic.
Signed-off-by: NMichael Albaugh <michael.albaugh@qlogic.com>
Signed-off-by: NRoland Dreier <rolandd@cisco.com>