All chips supported by the ADM1021 driver are also supported by the LM90
driver. Make that support official.

After this change, the adm1021 driver is only needed if the lm90 driver
is disabled. Also, the adm1021 driver misdetects a variety of chips as
MAX1617A, which is unwanted if any of those chips is in the system.
For this reason. make the adm1021 driver dependent on !SENSORS_LM90 to
show that it is not needed if the lm90 driver is enabled, and to avoid
misdetection if a chip supported by the lm90 driver is in the system.

Devicetree nodes are not added for the added chips since it is quite
unlikely that such old chips will ever be used in a devicetree based
system. They can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Both chips are quite similar to other chips of this series, so add
support for them to the lm90 driver. Also mention ON Semiconductor NCT210,
which is pin and register compatible to ADM1021A.

None of the chips support the secondary manufacturer and chip ID registers
at 0x3e and 0x3f, but return 0 when reading from those registers.
Use that information to improve the accuracy of chip detection code.

Devicetree nodes are not added for the added chips since it is quite
unlikely that such old chips will ever be used in a devicetree based
system. They can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

MAX1617 and LM84 are stripped-down versions of LM90, so they can easily
be supported by the LM90 driver. The most difficult part is chip detection,
since those old chips do not support manufacturer ID or chip ID registers.

The "alarms" attribute is enabled for both chips to match the functionality
of the adm1021 driver. Chip detection was improved and is less prone to
misdetection than the chip detection in the adm1021 driver.

Devicetree nodes are not added for the added chips since it is quite
unlikely that such old chips will ever be used in a devicetree based
system. They can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Introduce 16-bit register write function to simplify the code in
some places.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Simplify the code a bit by handling single-register read operations
in lm90_read16(). All we need to do is to skip the low-byte read
operation if the register address is 0.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

MAX6642 is a reduced version of LM90 with no low limits and no conversion
rate register. Its alert functionality is broken, similar to many other
chips supported by the lm90 driver.

After this change, the stand-alone max6642 driver is only needed if the
lm90 driver is disabled. Make it dependent on SENSORS_LM90=n to show that
it is not needed if the lm90 driver is enabled.

A devicetree node is not added for this chip since it is quite unlikely
that such an old chip will ever be used in a devicetree based system.
It can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

A flag indicating support for setting the conversion rate doesn't cost
much and will enable us to add support for MAX6642 to the lm90 driver.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

A flag indicating support for minimum temperature limits doesn't cost much
and will enable us to add support for MAX6642 to the lm90 driver.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

MAX6690 is all but identical to MAX6654. Revision 1 of its
datasheet lists the same chip ID as MAX6654, and a chip labeled
MAX6654 was found to have the chip ID listed as MAX6690 chip ID
in Revision 2 of its datasheet.

A devicetree node is not added for this chip since it is quite unlikely
that such an old chip will ever be used in a devicetree based system.
It can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

ADT7461A and NCT1008 support the undocumented manufacturer and chip ID
registers at 0x3e and 0x3f, and return 0x61 as chip ID. ADM1032 and
ADT7461 do not support those registers but return 0 when reading them.
Use this information to improve the accuracy of the chip detection code.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

ADT7481, ADT7482, and ADT7483 are similar to ADT7461, but support two
external temperature sensors, similar to MAX6695/6696. They support an
extended temperature range similar to ADT7461. Registers for the second
external channel can be accessed directly or by using the same method as
used by MAX6695/6696. For simplicity, the access method implemented for
MAX6695/6696 is used.

The chips support PEC (packet error checking). Set the PEC feature flag
and let the user decide if it should be enabled or not (it is by default
disabled).

Even though it is only documented for ADT7483, all three chips support a
secondary manufacturer ID register at 0x3e and a chip ID register at 0x3f.
Use the contents of those registers register for improved chip detection
accuracy. Add the same check to the ADT7461A detection code since this chip
also supports the same (undocumented) registers.

Devicetree nodes are not added for the added chips since it is quite
unlikely that such old chips will ever be used in a devicetree based
system. They can be added later if needed.
Reviewed-by: Slawomir Stepien <sst@poczta.fm>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Unlike MAX6646/MAX6647/MAX6649, MAX6648 and MAX6692 only support
a temperature range of 0..127 degrees C. Separate support for the
two sets of chips to be able to support maximum temperature ranges
correctly for all chips. Introduce new feature flag to indicate
temperature support up to 255 degrees C.

Since the chips are almost identical except for the supported temperature
range, automatic chip detection is limited. Effectively this means that
MAX6648 may be mis-detected as MAX6649 when auto-detected, but there is
nothing we can do about that.

Devicetree nodes are not added for the added chips since it is quite
unlikely that such old chips will ever be used in a devicetree based
system. They can be added later if needed.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

We don't want to support the obsolete 'alarms' attribute for new
chips supported by this driver. Add flag to indicate 'alarms' attribute
support and use it for existing chips. This flag will not be set for
additional chips supported by this driver in the future.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

When support for G781 was added, chips with ID 0x01 were found at I2C
addresses 0x4c and 0x4d. The G781 datasheet (version 1.3 from October 2003)
says that the device ID for G781-1 is 0x03, not 0x01. Also, the datasheet
states that the chip at I2C address is G781 and the chip at I2C address
0x4d is G781-1.

A G781-1 at I2C address 0x4d was now found to have a chip ID of 0x03
as suggested by the datasheet. Accept both 0x01 and 0x03 chip IDs at both
addresses to ensure that all variants of G781 are detected properly.

While at it, improve chip detection accuracy by reading two additional
registers and ensuring that only expected bits are set in those registers.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

The NCT1008 datasheet, Revision 3, states that its chip revision is
0x57. This matches the ADT7461A chip revision, and NCT1008 is therefore
detected as ADT7461A. In revision 6 of the datasheet, the chip revision
register is no longer documented. Multiple samples of NCT1008 were found
to report a chip revision of 0x54. As it turns out, one of the patches
submitted to add NCT1008 support to the lm90 driver already included a
check for chip revision 0x54. Unfortunately, that patch never made it into
the kernel. Remedy the situation and explicitly detect chips with revision
0x54 as NCT1008.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

The detect function is getting larger and larger and difficult to
understand or review. Split it into per-manufacturer detect functions
to improve readability.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Since temperature conversion functions are now unified, there is no need
to keep "the chip supports a configurable extended temperature range" and
"the chip has extended temperature range enabled" flags separate.
Use a single flag instead to reflect both.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

While most LM90 compatible chips support a temperature sensor resolution
of 11 bit, this is not the case for all chips. ADT7461 only supports a
resolution of 10 bit, and TMP451/TMP461 support a resolution of 12 bit.

Add support for various temperature sensor resolutions. To do this,
model all temperature sensors as 16 bit sensors, and use unified
temperature conversion functions which take the sensor resolution
as parameter.

While enhancing functionality, this has the positive side effect of
reducing code size by about 5%.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

When reading 16-bit volatile registers, the code uses a trick to
determine if a temperature is consistent: It reads the high part
of the register twice. If the values are the same, the code assumes
that the reading is consistent. If the value differs, the code
re-reads the second register as well and assumes that it now has
correct values.

This is only necessary for volatile registers. Add a parameter to
lm90_read16() to indicate if the register is volatile to avoid the
extra overhead for non-volatile registers.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

ADT7461 and TMP451 temperature sensors support extended temperature ranges.
If standard temperature range is selected, the temperature range is
unsigned and limited to 0 .. 127 degrees C. For TMP461, the standard
temperature range is -128000 ... 127000 degrees C. Distinguish between
the two chips by introducing a feature flag indicating if the standard
temperature range is signed or unsigned. Use the same flag for MAX6646/
MAX6647 as well since those chips also support unsigned temperatures.

Note that while the datasheet for ADT7461 suggests that the default
temperature range is unsigned, tests with a real chip suggest that this
is not the case: If the temperature offset is set to a value << 0,
the temperature register does report negative values.

Tests with real chips show that MAX6680/MAX6681 and SA56004 report
temperatures of 128 degrees C and higher as negative temperatures.
Add respective comments to the code.

Also use clamp_val() and DIV_ROUND_CLOSEST where appropriate in
calculations.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Experiments show that ADT7461A and NCT1008 support PEC, even though it is
not documented. Enable support for it in the driver. Since ADT7461 only
supports partial PEC, this means that the configuration for ADT7461A
needs to be separated from ADT7461.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Revision 0 of the ADT7461 datasheet suggests that the chip supports PEC
(packet error checking). This information is gone in later versions of the
datasheet. Experiments show that PEC support on ADT7461 is similar to PEC
support in ADM1032, ie it is only supported for read operations. Add
support for it to the driver.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

PEC (packet error checking) support for ADM1032 is currently only enabled
if the chip was auto-detected, but not if a chip is instantiated
explicitly. Always enable PEC support by introducing a chip feature flag
indicating partial PEC support. Also, for consistency, disable PEC support
by default to match existing functionality if the chip was not auto-
detected.

At the same time, introduce generic support for PEC with a separate feature
flag. This will be used when support for chips with full PEC functionality
is added.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

The register write address either matches the read address, or it is the
read address plus 6. Simplify the code and resolve the write address
programmatically instead of having two defines for each register.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Handling bit shifts necessary to extract status bits during compile time
reduces code and data size by almost 5% when building for x86_64.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Use BIT macro instead of shift operation to improve readability.
No functional change.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Reorder chip enumeration in alphabetical order to make it easier to
see which chips are supported, and to clarify where to add support
new chip types. No functional change.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Reorder include files in alphabetical order to reduce the chance of
duplicates and to make it clear where new include files should be
added. Drop the unnecessary include of linux/sysfs.h. Include
linux/device.h instead because that is what is actually used.

No functional change.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Many chips supported by this driver clear status registers after it
is read and update it in the next measurement cycle. Normally this falls
under the radar because all registers are only read once per measurement
cycle. However, there is an exception: Status registers are always read
during interrupt and laert handling. This can result in invalid status
reports if userspace reads an alarm attribute immediately afterwards.

Rework alarm/status handling by keeping a shadow register with 'current'
alarms, and by ensuring that the register is either only updated once per
measurement cycle or not cleared.

A second problem is related to alert handling: Alert handling is disabled
for chips with broken alert after an alert was reported, but only
re-enabled if attributes are read by the user. This means that alert
conditions may appear and disappear unnoticed. Remedy the situation
by introducing a worker to periodically read the status register(s) while
alert handling is disabled, and re-enable alerts after the alert condition
clears.

Yet another problem is that sysfs and udev events are currently only
reported to userspace if an alarm is raised, but not if an alarm condition
clears. Use the new worker to detect that situation and also generate
sysfs and udev events in that case.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

So far the driver only generated sysfs and udev events for minimum and
maximum alarms. Also generate events for critical and emergency alarms.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Enabling and disabling PEC for PMBus devices is currently only supported
with a debugfs attribute, which requires debugfs to be enabled and is
thus less than perfect. Take the lm90 driver as example and add a 'pec'
attribute to the I2C device if both the I2C adapter and the PMBus device
support it. Remove the now obsolete 'pec' attribute from debugfs.

Cc: Andrew Jeffery <andrew@aj.id.au>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>

Looking at the conditional lock acquire functions in the kernel due to
the new sparse support (see commit 4a557a5d "sparse: introduce
conditional lock acquire function attribute"), it became obvious that
the lockref code has a couple of them, but they don't match the usual
naming convention for the other ones, and their return value logic is
also reversed.

In the other very similar places, the naming pattern is '*_and_lock()'
(eg 'atomic_put_and_lock()' and 'refcount_dec_and_lock()'), and the
function returns true when the lock is taken.

The lockref code is superficially very similar to the refcount code,
only with the special "atomic wrt the embedded lock" semantics.  But
instead of the '*_and_lock()' naming it uses '*_or_lock()'.

And instead of returning true in case it took the lock, it returns true
if it *didn't* take the lock.

Now, arguably the reflock code is quite logical: it really is a "either
decrement _or_ lock" kind of situation - and the return value is about
whether the operation succeeded without any special care needed.

So despite the similarities, the differences do make some sense, and
maybe it's not worth trying to unify the different conditional locking
primitives in this area.

But while looking at this all, it did become obvious that the
'lockref_get_or_lock()' function hasn't actually had any users for
almost a decade.

The only user it ever had was the shortlived 'd_rcu_to_refcount()'
function, and it got removed and replaced with 'lockref_get_not_dead()'
back in 2013 in commits 0d98439e ("vfs: use lockred 'dead' flag to
mark unrecoverably dead dentries") and e5c832d5 ("vfs: fix dentry
RCU to refcounting possibly sleeping dput()")

In fact, that single use was removed less than a week after the whole
function was introduced in commit b3abd802 ("lockref: add
'lockref_get_or_lock() helper") so this function has been around for a
decade, but only had a user for six days.

Let's just put this mis-designed and unused function out of its misery.

We can think about the naming and semantic oddities of the remaining
'lockref_put_or_lock()' later, but at least that function has users.

And while the naming is different and the return value doesn't match,
that function matches the whole '{atomic,refcount}_dec_and_test()'
pattern much better (ie the magic happens when the count goes down to
zero, not when it is incremented from zero).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

The kernel tends to try to avoid conditional locking semantics because
it makes it harder to think about and statically check locking rules,
but we do have a few fundamental locking primitives that take locks
conditionally - most obviously the 'trylock' functions.

That has always been a problem for 'sparse' checking for locking
imbalance, and we've had a special '__cond_lock()' macro that we've used
to let sparse know how the locking works:

    # define __cond_lock(x,c)        ((c) ? ({ __acquire(x); 1; }) : 0)

so that you can then use this to tell sparse that (for example) the
spinlock trylock macro ends up acquiring the lock when it succeeds, but
not when it fails:

    #define raw_spin_trylock(lock)  __cond_lock(lock, _raw_spin_trylock(lock))

and then sparse can follow along the locking rules when you have code like

        if (!spin_trylock(&dentry->d_lock))
                return LRU_SKIP;
	.. sparse sees that the lock is held here..
        spin_unlock(&dentry->d_lock);

and sparse ends up happy about the lock contexts.

However, this '__cond_lock()' use does result in very ugly header files,
and requires you to basically wrap the real function with that macro
that uses '__cond_lock'.  Which has made PeterZ NAK things that try to
fix sparse warnings over the years [1].

To solve this, there is now a very experimental patch to sparse that
basically does the exact same thing as '__cond_lock()' did, but using a
function attribute instead.  That seems to make PeterZ happy [2].

Note that this does not replace existing use of '__cond_lock()', but
only exposes the new proposed attribute and uses it for the previously
unannotated 'refcount_dec_and_lock()' family of functions.

For existing sparse installations, this will make no difference (a
negative output context was ignored), but if you have the experimental
sparse patch it will make sparse now understand code that uses those
functions, the same way '__cond_lock()' makes sparse understand the very
similar 'atomic_dec_and_lock()' uses that have the old '__cond_lock()'
annotations.

Note that in some cases this will silence existing context imbalance
warnings.  But in other cases it may end up exposing new sparse warnings
for code that sparse just didn't see the locking for at all before.

This is a trial, in other words.  I'd expect that if it ends up being
successful, and new sparse releases end up having this new attribute,
we'll migrate the old-style '__cond_lock()' users to use the new-style
'__cond_acquires' function attribute.

The actual experimental sparse patch was posted in [3].

Link: https://lore.kernel.org/all/20130930134434.GC12926@twins.programming.kicks-ass.net/ [1]
Link: https://lore.kernel.org/all/Yr60tWxN4P568x3W@worktop.programming.kicks-ass.net/ [2]
Link: https://lore.kernel.org/all/CAHk-=wjZfO9hGqJ2_hGQG3U_XzSh9_XaXze=HgPdvJbgrvASfA@mail.gmail.com/ [3]
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Alexander Aring <aahringo@redhat.com>
Cc: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull xfs fixes from Darrick Wong:
 "This fixes some stalling problems and corrects the last of the
  problems (I hope) observed during testing of the new atomic xattr
  update feature.

   - Fix statfs blocking on background inode gc workers

   - Fix some broken inode lock assertion code

   - Fix xattr leaf buffer leaks when cancelling a deferred xattr update
     operation

   - Clean up xattr recovery to make it easier to understand.

   - Fix xattr leaf block verifiers tripping over empty blocks.

   - Remove complicated and error prone xattr leaf block bholding mess.

   - Fix a bug where an rt extent crossing EOF was treated as "posteof"
     blocks and cleaned unnecessarily.

   - Fix a UAF when log shutdown races with unmount"

* tag 'xfs-5.19-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: prevent a UAF when log IO errors race with unmount
  xfs: dont treat rt extents beyond EOF as eofblocks to be cleared
  xfs: don't hold xattr leaf buffers across transaction rolls
  xfs: empty xattr leaf header blocks are not corruption
  xfs: clean up the end of xfs_attri_item_recover
  xfs: always free xattri_leaf_bp when cancelling a deferred op
  xfs: use invalidate_lock to check the state of mmap_lock
  xfs: factor out the common lock flags assert
  xfs: introduce xfs_inodegc_push()
  xfs: bound maximum wait time for inodegc work

Pull nfsd fixes from Chuck Lever:
 "Notable regression fixes:

   - Fix NFSD crash during NFSv4.2 READ_PLUS operation

   - Fix incorrect status code returned by COMMIT operation"

* tag 'nfsd-5.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
  SUNRPC: Fix READ_PLUS crasher
  NFSD: restore EINVAL error translation in nfsd_commit()

Pull parisc architecture fixes from Helge Deller:
 "Two important fixes for bugs in code which was added in 5.18:

   - Fix userspace signal failures on 32-bit kernel due to a bug in vDSO

   - Fix 32-bit load-word unalignment exception handler which returned
     wrong values"

* tag 'for-5.19/parisc-4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Fix vDSO signal breakage on 32-bit kernel
  parisc/unaligned: Fix emulate_ldw() breakage

Addition of vDSO support for parisc in kernel v5.18 suddenly broke glibc
signal testcases on a 32-bit kernel.

The trampoline code (sigtramp.S) which is mapped into userspace includes
an offset to the context data on the stack, which is used by gdb and
glibc to get access to registers.

In a 32-bit kernel we used by mistake the offset into the compat context
(which is valid on a 64-bit kernel only) instead of the offset into the
"native" 32-bit context.
Reported-by: John David Anglin <dave.anglin@bell.net>
Tested-by: John David Anglin <dave.anglin@bell.net>
Fixes: 	df24e178 ("parisc: Add vDSO support")
CC: stable@vger.kernel.org # 5.18
Signed-off-by: Helge Deller <deller@gmx.de>

Merge tag 'perf-tools-fixes-for-v5.19-2022-07-02' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - BPF program info linear (BPIL) data is accessed assuming 64-bit
   alignment resulting in undefined behavior as the data is just byte
   aligned. Fix it, Found using -fsanitize=undefined.

 - Fix 'perf offcpu' build on old kernels wrt task_struct's
   state/__state field.

 - Fix perf_event_attr.sample_type setting on the 'offcpu-time' event
   synthesized by the 'perf offcpu' tool.

 - Don't bail out when synthesizing PERF_RECORD_ events for pre-existing
   threads when one goes away while parsing its procfs entries.

 - Don't sort the task scan result from /proc, its not needed and
   introduces bugs when the main thread isn't the first one to be
   processed.

 - Fix uninitialized 'offset' variable on aarch64 in the unwind code.

 - Sync KVM headers with the kernel sources.

* tag 'perf-tools-fixes-for-v5.19-2022-07-02' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  perf synthetic-events: Ignore dead threads during event synthesis
  perf synthetic-events: Don't sort the task scan result from /proc
  perf unwind: Fix unitialized 'offset' variable on aarch64
  tools headers UAPI: Sync linux/kvm.h with the kernel sources
  perf bpf: 8 byte align bpil data
  tools kvm headers arm64: Update KVM headers from the kernel sources
  perf offcpu: Accept allowed sample types only
  perf offcpu: Fix build failure on old kernels

Pull powerpc fixes from Michael Ellerman:

 - Fix BPF uapi confusion about the correct type of bpf_user_pt_regs_t.

 - Fix virt_addr_valid() when memory is hotplugged above the boot-time
   high_memory value.

 - Fix a bug in 64-bit Book3E map_kernel_page() which would incorrectly
   allocate a PMD page at PUD level.

 - Fix a couple of minor issues found since we enabled KASAN for 64-bit
   Book3S.

Thanks to Aneesh Kumar K.V, Cédric Le Goater, Christophe Leroy, Kefeng
Wang, Liam Howlett, Nathan Lynch, and Naveen N. Rao.

* tag 'powerpc-5.19-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/memhotplug: Add add_pages override for PPC
  powerpc/bpf: Fix use of user_pt_regs in uapi
  powerpc/prom_init: Fix kernel config grep
  powerpc/book3e: Fix PUD allocation size in map_kernel_page()
  powerpc/xive/spapr: correct bitmap allocation size