Debouncing failure is a good indicator of basic link problem.  Use
-EPIPE to indicate debouncing failure and make ata_eh_reset() invoke
sata_down_spd_limit() if the error occurs during reset.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

sata_down_spd_limit() first reads the current SPD from SStatus and
limit the speed to the lower one of one below the current limit or one
below the current SPD in SStatus.  SPD may not be accessible or valid
when SPD down is requested making sata_down_spd_limit() fail when it's
most needed.

This patch makes the current SPD cached after each successful reset
and forces GEN I speed (1.5Gbps) if neither of SStatus or the cached
value is valid, so sata_down_spd_limit() is now guaranteed to lower
the speed limit if lower speed is available.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

When an NCQ command fails, all commands in flight are aborted and the
offending one is reported using log page 10h.  Depending on controller
characteristics and LLD implementation, all commands may appear as
having a device error due to shared TF status making it hard to
determine what's actually going on.

This patch adds AC_ERR_NCQ, marks the command reported by log page 10h
with it and print extra "<F>" after the error report for the command
to help distinguishing the offending command.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Requiring LLDs to format multiple error description messages properly
doesn't work too well.  Help LLDs a bit by making ata_ehi_push_desc()
insert ", " on each invocation.  __ata_ehi_push_desc() is the raw
version without the automatic separator.

While at it, make ehi_desc interface proper functions instead of
macros.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Separate out ata_eh_handle_dev_fail() from ata_eh_recover().  This is
in preparation of ata_link and PMP support.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Implement _GTM/_STM support.  acpi_gtm is added to ata_port which
stores _GTM parameters over suspend/resume cycle.  A new hook
ata_acpi_on_suspend() is responsible for storing _GTM parameters
during suspend.  _STM is executed in ata_acpi_on_resume().  With this
change, invoking _GTF is safe on IDE hierarchy and acpi_sata check
before _GTF is removed.

ata_acpi_gtm() and ata_acpi_stm() implementation is taken from Alan
Cox's pata_acpi implementation.  ata_acpi_gtm() is fixed such that the
result parameter is not shifted by sizeof(union acpi_object).
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

This patch reimplements ACPI invocation such that, instead of
exporting ACPI details to the rest of libata, ACPI event handlers -
ata_acpi_on_resume() and ata_acpi_on_devcfg() - are used.  These two
functions are responsible for determining whether specific ACPI method
is used and when.

On resume, _GTF is scheduled by setting ATA_DFLAG_ACPI_PENDING device
flag.  This is done this way to avoid performing the action on wrong
device device (device swapping while suspended).

On every ata_dev_configure(), ata_acpi_on_devcfg() is called, which
performs _SDD and _GTF.  _GTF is performed only after resuming and, if
SATA, hardreset as the ACPI spec specifies.  As _GTF may contain
arbitrary commands, IDENTIFY page is re-read after _GTF taskfiles are
executed.

If one of ACPI methods fails, ata_acpi_on_devcfg() retries on the
first failure.  If it fails again on the second try, ACPI is disabled
on the device.  Note that successful configuration clears ACPI failed
status.

With all feature checks moved to the above two functions,
do_drive_set_taskfiles() is trivial and thus collapsed into
ata_acpi_exec_tfs(), which is now static and converted to return the
number of executed taskfiles to be used by ata_acpi_on_resume().  As
failures are handled properly, ata_acpi_push_id() now returns -errno
on errors instead of unconditional zero.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

When EH gives up after repeated exceptions, it doesn't't clear the
PENDING bit on exit which leaves PENDING bit set without EH actually
scheduled.  This makes ata_port_wait_eh() to wait forever makes rmmod
hang on such port.  Fix it by clearing the flag.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Removed unused variable did_followup_srst from ata_eh_reset().
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

prereset() is now allowed to set flag for unsupported reset method.
EH layer is responsible for selecting the fallback.  Remove non-sense
warning message.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Signed-off-by: NJeff Garzik <jeff@garzik.org>

After certain errors, some devices report complete garbage on
IDENTIFY.  This can cause ata_dev_read_id() to fail with -EINVAL
resulting in immediate disabling of the device.  Give the device one
last chance after -EINVAL to allow recovery from such situations.  As
-EINVAL is triggered very rarely, this shouldn't cause any noticeable
affect on more common error paths.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: Harald Dunkel <harald.dunkel@t-online.de>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata enables SCSI host during ATA host activation which happens
after IRQ handler is registered and IRQ is enabled.  All ATA ports are
in frozen state when IRQ is enabled but frozen ports may raise limited
number of IRQs after being frozen - IOW, ->freeze() is not responsible
for clearing pending IRQs.  During normal operation, the IRQ handler
is responsible for clearing spurious IRQs on frozen ports and it
usually doesn't require any extra code.

Unfortunately, during host initialization, the IRQ handler can end up
scheduling EH for a port whose SCSI host isn't initialized yet.  This
results in OOPS in the SCSI midlayer.  This is relatively short window
and scheduling EH for probing is the first thing libata does after
initialization, so ignoring EH scheduling until initialization is
complete solves the problem nicely.

This problem was spotted by Berck E. Nash in the following thread.

  http://thread.gmane.org/gmane.linux.kernel/519412Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: Berck E. Nash <flyboy@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Reimplement suspend/resume support using sdev->manage_start_stop.

* Device suspend/resume is now SCSI layer's responsibility and the
  code is simplified a lot.

* DPM is dropped.  This also simplifies code a lot.  Suspend/resume
  status is port-wide now.

* ata_scsi_device_suspend/resume() and ata_dev_ready() removed.

* Resume now has to wait for disk to spin up before proceeding.  I
  couldn't find easy way out as libata is in EH waiting for the
  disk to be ready and sd is waiting for EH to complete to issue
  START_STOP.

* sdev->manage_start_stop is set to 1 in ata_scsi_slave_config().
  This fixes spindown on shutdown and suspend-to-disk.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata previously depended upon waits in prereset to get resets after
hotplug right for both spin up and device ready wait.  This was
necessary both for reliablity and speed as reset was likely to fail if
initiated too early and each try usually took more than 30secs to
fail.  Previous patches fixed the reliability part by fixing status
and SCR handling in resets.  This patch remedies the speed part by
improving reset sequencing.

Prereset waiting timeout is adjusted to 10s because spinup wait is
replaced by reset sequencing and !BSY wait is not as important as
before.  During boot or module loading where the drive is already
fully spun up, !BSY wait succeeds immediately, so 10s should be enough
in most cases.  It matters after hotplugging or other error
conditions, but in those cases, !BSY wait in prereset simply can't be
relied upon due to the varied and weird behaviors ATA controllers and
devices show.

Reset is now driven by ata_eh_reset_timeouts[] table which contains
timeouts for each reset try.  The first reset can be softreset but the
following ones are always hardreset if available.  Each timeout
defines deadline for the reset try.  If a reset try fails, reset is
retried with the next timeout till the end of the timeout table is
reached.  If a reset try fails before the timeout with error, libata
waits till the deadline of the failed try before retrying.

IOW, the timeout table defines timetable of reset tries such that the
n'th try always begins at least after the sum of all previous timeouts
has passed.  The current timetable defines 4 tries and takes around 1
minute.

@0	: First try.  This should succeed most of the time during boot.
@10	: 10s is enough to spin up most consumer harddrives.  Give it
	  another shot.
@20	: 20s should spin up > 99% of working drives.  This has 30s
	  timeout for retarded devices needing long idleness post reset.
@55	: Final try with 5s timeout just in case.

The above timetable is trade off between not annoying the device too
much with frequent resets and taking reasonable amount of time in most
cases.  Some controllers may do better with shorter timeouts while
others may fare better with longer but we just can't rely upon LLD
writers to test each controller with wide variety of devices using
various scenarios.  We need default behavior which reasonably fits
most cases.

I've tested the above timetable on a dozen SATA controllers and a few
PATA controllers with about a dozen different drives from all major
vendors and 4 different ODDs from three different vendors for both
boot and hotplug (if available) cases.

Boot probing is not affected unless the device is broken in which
cases new code gives up on the port after a minute rather than five or
nine minutes.  When hotplugging, most devices get detected on the
first or second try.  Multi-platter drives with long spin up time
which sometimes took > 40 secs with the original code, now usually
comes up during the second try and at least right after the third try
@20.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Add @deadline to prereset and reset methods and make them honor it.
ata_wait_ready() which directly takes @deadline is implemented to be
used as the wait function.  This patch is in preparation for EH timing
improvements.

* ata_wait_ready() never does busy sleep.  It's only used from EH and
  no wait in EH is that urgent.  This function also prints 'be
  patient' message automatically after 5 secs of waiting if more than
  3 secs is remaining till deadline.

* ata_bus_post_reset() now fails with error code if any of its wait
  fails.  This is important because earlier reset tries will have
  shorter timeout than the spec requires.  If a device fails to
  respond before the short timeout, reset should be retried with
  longer timeout rather than silently ignoring the device.

  There are three behavior differences.

  1. Timeout is applied to both devices at once, not separately.  This
     is more consistent with what the spec says.

  2. When a device passes devchk but fails to become ready before
     deadline.  Previouly, post_reset would just succeed and let
     device classification remove the device.  New code fails the
     reset thus causing reset retry.  After a few times, EH will give
     up disabling the port.

  3. When slave device passes devchk but fails to become accessible
     (TF-wise) after reset.  Original code disables dev1 after 30s
     timeout and continues as if the device doesn't exist, while the
     patched code fails reset.  When this happens, new code fails
     reset on whole port rather than proceeding with only the primary
     device.

  If the failing device is suffering transient problems, new code
  retries reset which is a better behavior.  If the failing device is
  actually broken, the net effect is identical to it, but not to the
  other device sharing the channel.  In the previous code, reset would
  have succeeded after 30s thus detecting the working one.  In the new
  code, reset fails and whole port gets disabled.  IMO, it's a
  pathological case anyway (broken device sharing bus with working
  one) and doesn't really matter.

* ata_bus_softreset() is changed to return error code from
  ata_bus_post_reset().  It used to return 0 unconditionally.

* Spin up waiting is to be removed and not converted to honor
  deadline.

* To be on the safe side, deadline is set to 40s for the time being.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Separate ATA_EHI_DID_RESET into ATA_EHI_DID_SOFTRESET and
ATA_EHI_DID_HARDRESET.  ATA_EHI_DID_RESET is redefined as OR of the
two flags.  This patch doesn't introduce any behavior change.  This
will be used later to determine whether _SDD is necessary or not.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

->cable_detect() used to be called on by the old ata_bus_probe() path.
Add invocation to ata_eh_revalidate_and_attach() right after IDENTIFYs
are done.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

->post_internal_cmd is simplified EH for internal commands.  Its
primary mission is to stop the controller such that no rogue memory
access or other activities occur after the internal command is
released.  It may provide error diagnostics by setting qc->err_mask
but this hasn't been a requirement.

To ignore SETXFER failure for CFA devices, libata needs to know
whether a command was failed by the device or for any other reason.
ie. internal command needs to get AC_ERR_DEV right.

This patch makes the following changes to AC_ERR_DEV handling and
->post_internal_cmd semantics to accomodate this need and simplify
callback implementation.

1. As long as the correct bits in the result TF registers are set,
   there is no need to set AC_ERR_DEV explicitly.  libata EH core
   takes care of that for both normal and internal commands.

2. The only requirement for ->post_internal_cmd() is to put the
   controller into quiescent state.  It needs not to set any err_mask.

3. ata_exec_internal_sg() performs minimal error analysis such that
   AC_ERR_DEV is automatically set as long as result_tf is filled
   correctly.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

There was a rare report where SB600 reported SERR_INTERNAL and SRST
couldn't get it out of the failure mode.  Hardreset on SERR_INTERNAL.
As the problem is intermittent, whether this fixes the problem or not
hasn't been verified yet, but hardresetting the channel on internal
error is a good idea anyway.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

patch 2/4:
  Clear tf before doing request sense.

This fixes the AOpen 56X/AKH timeout problem.
(http://bugzilla.kernel.org/show_bug.cgi?id=8244)
Signed-off-by: NAlbert Lee <albertcc@tw.ibm.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

For drive side cable detection to work correctly, drives need to be
identified backwards such that the slave device releases PDIAG- before
the mater drive tries to detect cable type.  ata_bus_probe() was fixed
by commit f31f0cc2 but the new EH path
wasn't fixed.  This patch makes new EH path do IDENTIFY backwards.

ata_dev_configure() for new devices are still performed master first.
This is to keep the detection messages in forward order.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

->prereset() returns -ENOENT to tell libata that the port is empty and
reset sequencing should be stopped.  This is not an error condition.
Update ata_eh_reset() such that it sets device classes to ATA_DEV_NONE
and return success in on -ENOENT.  This makes spurious error message
go away.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Conditionalize all PM related stuff in libata core layer using
CONFIG_PM.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

ata_port has two different id fields - id and port_no.  id is
system-wide 1-based unique id for the port while port_no is 0-based
host-wide port number.  The former is primarily used to identify the
ATA port to the user in printk messages while the latter is used in
various places in libata core and LLDs to index the port inside the
host.

The two fields feel quite similar and sometimes ap->id is used in
place of ap->port_no, which is very difficult to spot.  This patch
renames ap->id to ap->print_id to reduce the possibility of such bugs.

Some printk messages are adjusted such that id string (ata%u[.%u])
isn't printed twice and/or to use ata_*_printk() instead of hardcoded
id format.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The current EH speed down code is more of a proof that the EH
framework is capable of adjusting transfer speed in response to error.
This patch puts some intelligence into EH speed down sequence.  The
rules are..

* If there have been more than three timeout, HSM violation or
  unclassified DEV errors for known supported commands during last 10
  mins, NCQ is turned off.

* If there have been more than three timeout or HSM violation for known
  supported command, transfer mode is slowed down.  If DMA is active,
  it is first slowered by one grade (e.g. UDMA133->100).  If that
  doesn't help, it's slowered to 40c limit (UDMA33).  If PIO is
  active, it's slowered by one grade first.  If that doesn't help,
  PIO0 is forced.  Note that this rule does not change transfer mode.
  DMA is never degraded into PIO by this rule.

* If there have been more than ten ATA bus, timeout, HSM violation or
  unclassified device errors for known supported commands && speeding
  down DMA mode didn't help, the device is forced into PIO mode.  Note
  that this rule is considered only for PATA devices and is pretty
  difficult to trigger.

One error can only trigger one rule at a time.  After a rule is
triggered, error history is cleared such that the next speed down
happens only after some number of errors are accumulated.  This makes
sense because now speed down is done in bigger stride.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

* Move forcing device to PIO0 on device disable into
  ata_dev_disable().  This makes both old and new EHs act the same
  way.

* Speed down only PIO mode on probe failure.  All commands used during
  probing are PIO commands.  There's no point in speeding down DMA.

* Retry at least once after -ENODEV.  Some devices report garbled
  IDENTIFY data after certain events.  This shouldn't cause device
  detach and re-attach.

* Rearrange EH failure path for simplicity.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Make ata_down_xfermask_limit() accept @sel instead of @force_pio0.
@sel selects how the xfermask limit will be adjusted.  The following
selectors are defined.

* ATA_DNXFER_PIO	: only speed down PIO
* ATA_DNXFER_DMA	: only speed down DMA, don't cause transfer mode change
* ATA_DNXFER_40C	: apply 40c cable limit
* ATA_DNXFER_FORCE_PIO	: force PIO
* ATA_DNXFER_FORCE_PIO0	: force PIO0 (same as original with @force_pio0 == 1)
* ATA_DNXFER_ANY	: same as original with @force_pio0 == 0

Currently, only ANY and FORCE_PIO0 are used to maintain the original
behavior.  Other selectors will be used later to improve EH speed down
sequence.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata used two separate sets of variables to record request size and
current offset for ATA and ATAPI.  This is confusing and fragile.
This patch replaces qc->nsect/cursect with qc->nbytes/curbytes and
kills them.  Also, ata_pio_sector() is updated to use bytes for
qc->cursg_ofs instead of sectors.  The field used to be used in bytes
for ATAPI and in sectors for ATA.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

ata_eh_suspend() was returning 0 regardless of failure.  This bug has
potential to lose data on suspend.  Fix it.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata EH ignores port-wide actions in per-dev action mask.  However,
device resume requests EH_SOFTRESET using per-dev action mask.  Under
certain circumstances, this results in not resetting frozen port after
resuming which causes failure of all commands.

This patch allows port-wide actions to be requested in per-dev action
mask.  Before EH recovery starts, port-wide actions will be collected.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata switched to IRQ-driven IDENTIFY when IRQ-driven PIO was
introduced.  This has caused a lot of problems including device
misdetection and phantom device.

ATA_FLAG_DETECT_POLLING was added recently to selectively use polling
IDENTIFY on problemetic drivers but many controllers and devices are
affected by this problem and trying to adding ATA_FLAG_DETECT_POLLING
for each such case is diffcult and not very rewarding.

This patch makes libata always use polling IDENTIFY.  This is
consistent with libata's original behavior and drivers/ide's behavior.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

If EH command is issued to a frozen port, it fails with AC_ERR_SYSTEM.
libata used to request sense even when the port is frozen needlessly
adding AC_ERR_SYSTEM to err_mask.  Don't do it.
Signed-off-by: NTejun Heo <htejun@gmail.com>

Print cdb[0] in failed qc report.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Improve failed qc reporting.  The original message didn't include the
actual command nor full error status and it was necessary to
temporarily patch the code to find out exactly which command is
causing problem.  This patch makes EH report full command and result
TFs along with data direction and length.  This change will make bug
reports more useful.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

On some controllers (ICHs in piix mode), there is *NO* reliable way to
determine device presence other than issuing IDENTIFY and see how the
transaction proceeds by watching the TF status register.

libata acted this way before irq-pio and phantom devices caused very
little problem but now that IDENTIFY is performed using IRQ drive PIO,
such phantom devices now result in multiple 30sec timeouts during
boot.

This patch implements ATA_FLAG_DETECT_POLLING.  If a LLD sets this
flag, libata core issues the initial IDENTIFY in polling mode and if
the initial data transfer fails w/ HSM violation, the port is
considered to be empty thus replicating the old libata and IDE
behavior.
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Make ata_dev_read_id() take @flags instead of @post_reset.  Currently
there is only one flag defined - ATA_READID_POSTRESET, which is
equivalent to @post_reset.  This is preparation for polling presence
detection.
Signed-off-by: NJeff Garzik <jeff@garzik.org>

libata EH used to perform ata_set_mode() iff the EH session performed
reset as indicated by ATA_EHI_DID_RESET.  This is incorrect because
->dev_config() called by revalidation is allowed to modify transfer
mode which ata_set_mode() should take care of.  This patch implements
the following two flags.

* ATA_EHI_SETMODE: set during EH to schedule ata_set_mode().  Both new
  device attachment and revalidation set this flag.

* ATA_EHI_POST_SETMODE: set while the device is revalidated after
  ata_set_mode().  Post-setmode revalidation is different from initial
  configuaration and EH revalidation in that ->dev_config() is not
  allowed tune transfer mode.  LLD can use this flag to determine
  whether it's allowed to tune transfer mode.  Note that POST_SETMODE
  ->dev_config() is guaranteed to be preceded by non-POST_SETMODE
  ->dev_config().
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Implement ehi flag ATA_EHI_PRINTINFO.  This flag is set when device
configuration needs to print out device info.  This used to be handled
by @print_info argument to ata_dev_configure() but LLDs also need to
know about it in ->dev_config() callback.

This patch replaces @print_info w/ ATA_EHI_PRINTINFO and make sata_sil
print workaround messages only on the initial configuration.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Separate delayable work items from non-delayable work items be splitting them
into a separate structure (delayed_work), which incorporates a work_struct and
the timer_list removed from work_struct.

The work_struct struct is huge, and this limits it's usefulness.  On a 64-bit
architecture it's nearly 100 bytes in size.  This reduces that by half for the
non-delayable type of event.
Signed-Off-By: NDavid Howells <dhowells@redhat.com>