Pull more s390 updates from Heiko Carstens:
 "This is mainly to decouple udelay() and arch_cpu_idle() and simplify
  both of them.

  Summary:

   - Always initialize kernel stack backchain when entering the kernel,
     so that unwinding works properly.

   - Fix stack unwinder test case to avoid rare interrupt stack
     corruption.

   - Simplify udelay() and just let it busy loop instead of implementing
     a complex logic.

   - arch_cpu_idle() cleanup.

   - Some other minor improvements"

* tag 's390-5.11-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/zcrypt: convert comma to semicolon
  s390/idle: allow arch_cpu_idle() to be kprobed
  s390/idle: remove raw_local_irq_save()/restore() from arch_cpu_idle()
  s390/idle: merge enabled_wait() and arch_cpu_idle()
  s390/delay: remove udelay_simple()
  s390/irq: select HAVE_IRQ_EXIT_ON_IRQ_STACK
  s390/delay: simplify udelay
  s390/test_unwind: use timer instead of udelay
  s390/test_unwind: fix CALL_ON_STACK tests
  s390: make calls to TRACE_IRQS_OFF/TRACE_IRQS_ON balanced
  s390: always clear kernel stack backchain before calling functions

Pull TIF_NOTIFY_SIGNAL updates from Jens Axboe:
 "This sits on top of of the core entry/exit and x86 entry branch from
  the tip tree, which contains the generic and x86 parts of this work.

  Here we convert the rest of the archs to support TIF_NOTIFY_SIGNAL.

  With that done, we can get rid of JOBCTL_TASK_WORK from task_work and
  signal.c, and also remove a deadlock work-around in io_uring around
  knowing that signal based task_work waking is invoked with the sighand
  wait queue head lock.

  The motivation for this work is to decouple signal notify based
  task_work, of which io_uring is a heavy user of, from sighand. The
  sighand lock becomes a huge contention point, particularly for
  threaded workloads where it's shared between threads. Even outside of
  threaded applications it's slower than it needs to be.

  Roman Gershman <romger@amazon.com> reported that his networked
  workload dropped from 1.6M QPS at 80% CPU to 1.0M QPS at 100% CPU
  after io_uring was changed to use TIF_NOTIFY_SIGNAL. The time was all
  spent hammering on the sighand lock, showing 57% of the CPU time there
  [1].

  There are further cleanups possible on top of this. One example is
  TIF_PATCH_PENDING, where a patch already exists to use
  TIF_NOTIFY_SIGNAL instead. Hopefully this will also lead to more
  consolidation, but the work stands on its own as well"

[1] https://github.com/axboe/liburing/issues/215

* tag 'tif-task_work.arch-2020-12-14' of git://git.kernel.dk/linux-block: (28 commits)
  io_uring: remove 'twa_signal_ok' deadlock work-around
  kernel: remove checking for TIF_NOTIFY_SIGNAL
  signal: kill JOBCTL_TASK_WORK
  io_uring: JOBCTL_TASK_WORK is no longer used by task_work
  task_work: remove legacy TWA_SIGNAL path
  sparc: add support for TIF_NOTIFY_SIGNAL
  riscv: add support for TIF_NOTIFY_SIGNAL
  nds32: add support for TIF_NOTIFY_SIGNAL
  ia64: add support for TIF_NOTIFY_SIGNAL
  h8300: add support for TIF_NOTIFY_SIGNAL
  c6x: add support for TIF_NOTIFY_SIGNAL
  alpha: add support for TIF_NOTIFY_SIGNAL
  xtensa: add support for TIF_NOTIFY_SIGNAL
  arm: add support for TIF_NOTIFY_SIGNAL
  microblaze: add support for TIF_NOTIFY_SIGNAL
  hexagon: add support for TIF_NOTIFY_SIGNAL
  csky: add support for TIF_NOTIFY_SIGNAL
  openrisc: add support for TIF_NOTIFY_SIGNAL
  sh: add support for TIF_NOTIFY_SIGNAL
  um: add support for TIF_NOTIFY_SIGNAL
  ...

udelay is implemented by using quite subtle details to make it
possible to load an idle psw and waiting for an interrupt even in irq
context or when interrupts are disabled. Also handling (or better: no
handling) of softirqs is taken into account.

All this is done to optimize for something which should in normal
circumstances never happen: calling udelay to busy wait. Therefore get
rid of the whole complexity and just busy loop like other
architectures are doing it also.

It could have been possible to use diag 0x44 instead of cpu_relax() in
the busy loop, however we have seen too many bad things happen with
diag 0x44 that it seems to be better to simply busy loop.

Also note that with this new implementation kernel preemption does
work when within the udelay loop. This did not work before.

To get a feeling what the former code optimizes for: IPL'ing a kernel
with 'defconfig' and afterwards compiling a kernel ends with a total
of zero udelay calls.

Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

In case of udelay CIF_IGNORE_IRQ is set. This leads to an unbalanced
call of TRACE_IRQS_OFF and TRACE_IRQS_ON. That is: from lockdep's
point of view TRACE_IRQS_ON is called one time too often.

This doesn't fix any real bug, just makes the calls balanced.

Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

Clear the kernel stack backchain before potentially calling the
lockdep trace_hardirqs_off/on functions. Without this walking the
kernel backchain, e.g. during a panic, might stop too early.

Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

Pull s390 updates from Heiko Carstens:

 - Add support for the hugetlb_cma command line option to allocate
   gigantic hugepages using CMA

 - Add arch_get_random_long() support.

 - Add ap bus userspace notifications.

 - Increase default size of vmalloc area to 512GB and otherwise let it
   increase dynamically by the size of physical memory. This should fix
   all occurrences where the vmalloc area was not large enough.

 - Completely get rid of set_fs() (aka select SET_FS) and rework address
   space handling while doing that; making address space handling much
   more simple.

 - Reimplement getcpu vdso syscall in C.

 - Add support for extended SCLP responses (> 4k). This allows e.g. to
   handle also potential large system configurations.

 - Simplify KASAN by removing 3-level page table support and only
   supporting 4-levels from now on.

 - Improve debug-ability of the kernel decompressor code, which now
   prints also stack traces and symbols in case of problems to the
   console.

 - Remove more power management leftovers.

 - Other various fixes and improvements all over the place.

* tag 's390-5.11-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (62 commits)
  s390/mm: add support to allocate gigantic hugepages using CMA
  s390/crypto: add arch_get_random_long() support
  s390/smp: perform initial CPU reset also for SMT siblings
  s390/mm: use invalid asce for user space when switching to init_mm
  s390/idle: fix accounting with machine checks
  s390/idle: add missing mt_cycles calculation
  s390/boot: add build-id to decompressor
  s390/kexec_file: fix diag308 subcode when loading crash kernel
  s390/cio: fix use-after-free in ccw_device_destroy_console
  s390/cio: remove pm support from ccw bus driver
  s390/cio: remove pm support from css-bus driver
  s390/cio: remove pm support from IO subchannel drivers
  s390/cio: remove pm support from chsc subchannel driver
  s390/vmur: remove unused pm related functions
  s390/tape: remove unsupported PM functions
  s390/cio: remove pm support from eadm-sch drivers
  s390: remove pm support from console drivers
  s390/dasd: remove unused pm related functions
  s390/zfcp: remove pm support from zfcp driver
  s390/ap: let bus_register() add the AP bus sysfs attributes
  ...

When a machine check interrupt is triggered during idle, the code
is using the async timer/clock for idle time calculation. It should use
the machine check enter timer/clock which is passed to the macro.

Fixes: 0b0ed657fe00 ("s390: remove critical section cleanup from entry.S")
Cc: <stable@vger.kernel.org> # 5.8
Reviewed-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

During removal of the critical section cleanup the calculation
of mt_cycles during idle was removed. This causes invalid
accounting on systems with SMT enabled.

Fixes: 0b0ed657fe00 ("s390: remove critical section cleanup from entry.S")
Cc: <stable@vger.kernel.org> # 5.8
Reviewed-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

With commit 58c644ba512c ("sched/idle: Fix arch_cpu_idle() vs
tracing") common code calls arch_cpu_idle() with a lockdep state that
tells irqs are on.

This doesn't work very well for s390: psw_idle() will enable interrupts
to wait for an interrupt. As soon as an interrupt occurs the interrupt
handler will verify if the old context was psw_idle(). If that is the
case the interrupt enablement bits in the old program status word will
be cleared.

A subsequent test in both the external as well as the io interrupt
handler checks if in the old context interrupts were enabled. Due to
the above patching of the old program status word it is assumed the
old context had interrupts disabled, and therefore a call to
TRACE_IRQS_OFF (aka trace_hardirqs_off_caller) is skipped. Which in
turn makes lockdep incorrectly "think" that interrupts are enabled
within the interrupt handler.

Fix this by unconditionally calling TRACE_IRQS_OFF when entering
interrupt handlers. Also call unconditionally TRACE_IRQS_ON when
leaving interrupts handlers.

This leaves the special psw_idle() case, which now returns with
interrupts disabled, but has an "irqs on" lockdep state. So callers of
psw_idle() must adjust the state on their own, if required. This is
currently only __udelay_disabled().

Fixes: 58c644ba512c ("sched/idle: Fix arch_cpu_idle() vs tracing")
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

Verify on exit to user space that always
- the primary ASCE (cr1) is set to kernel ASCE
- the secondary ASCE (cr7) is set to user ASCE

If this is not the case: panic since something went terribly wrong.

Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>