linux.git/arch/s390/kernel/setup.c, branch v6.9

s390/setup: fix virtual vs physical address confusion

2024-02-16T13:30:12+00:00

Fix virtual vs physical address confusion. This does not fix a bug
since virtual and physical address spaces are currently the same.

/proc/iomem should report the physical address ranges, so use __pa_symbol()
for resource registration, similar to other architectures.

Signed-off-by: Gerald Schaefer 
Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens

s390/fpu: remove ARCH_WANTS_DYNAMIC_TASK_STRUCT support

2023-12-11T13:33:06+00:00

s390 selects ARCH_WANTS_DYNAMIC_TASK_STRUCT in order to make the size of
the task structure dependent on the availability of the vector
facility. This doesn't make sense anymore because since many years all
machines provide the vector facility.

Therefore simplify the code a bit and remove s390 support for
ARCH_WANTS_DYNAMIC_TASK_STRUCT.

Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens 
Signed-off-by: Alexander Gordeev

s390/nmi: consistently enable machine checks in trap_init()

2023-12-11T13:33:05+00:00

The kernel starts with machine checks disabled (machine check mask bit in
the PSW is zero), and machine checks are enabled when trap_init() is
called. The rationale is that this allows to assume that the system is
initialized up to a certain point before the machine check handler may be
invoked.

However the implementation is incomplete: all new PSW masks in lowcore have
the machine check mask bit. This means that e.g. for any early program
check machine checks are enabled within the program check handler. This
contradicts the whole point of enabling machine checks at a single place.

Change this and initialize all new PSWs in lowcore so they have the machine
check mask bit not set. Set the bit in all masks in trap_init(). This way
machine check enabling is consistent.

Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens 
Signed-off-by: Alexander Gordeev

Merge tag 's390-6.7-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

2023-11-03T20:17:22+00:00

Pull s390 updates from Vasily Gorbik:

 - Get rid of private VM_FAULT flags

 - Add word-at-a-time implementation

 - Add DCACHE_WORD_ACCESS support

 - Cleanup control register handling

 - Disallow CPU hotplug of CPU 0 to simplify its handling complexity,
   following a similar restriction in x86

 - Optimize pai crypto map allocation

 - Update the list of crypto express EP11 coprocessor operation modes

 - Fixes and improvements for secure guests AP pass-through

 - Several fixes to address incorrect page marking for address
   translation with the "cmma no-dat" feature, preventing potential
   incorrect guest TLB flushes

 - Fix early IPI handling

 - Several virtual vs physical address confusion fixes

 - Various small fixes and improvements all over the code

* tag 's390-6.7-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (74 commits)
  s390/cio: replace deprecated strncpy with strscpy
  s390/sclp: replace deprecated strncpy with strtomem
  s390/cio: fix virtual vs physical address confusion
  s390/cio: export CMG value as decimal
  s390: delete the unused store_prefix() function
  s390/cmma: fix handling of swapper_pg_dir and invalid_pg_dir
  s390/cmma: fix detection of DAT pages
  s390/sclp: handle default case in sclp memory notifier
  s390/pai_crypto: remove per-cpu variable assignement in event initialization
  s390/pai: initialize event count once at initialization
  s390/pai_crypto: use PERF_ATTACH_TASK define for per task detection
  s390/mm: add missing arch_set_page_dat() call to gmap allocations
  s390/mm: add missing arch_set_page_dat() call to vmem_crst_alloc()
  s390/cmma: fix initial kernel address space page table walk
  s390/diag: add missing virt_to_phys() translation to diag224()
  s390/mm,fault: move VM_FAULT_ERROR handling to do_exception()
  s390/mm,fault: remove VM_FAULT_BADMAP and VM_FAULT_BADACCESS
  s390/mm,fault: remove VM_FAULT_SIGNAL
  s390/mm,fault: remove VM_FAULT_BADCONTEXT
  s390/mm,fault: simplify kfence fault handling
  ...

crash_core: change the prototype of function parse_crashkernel()

2023-10-04T17:41:58+00:00

Add two parameters 'low_size' and 'high' to function parse_crashkernel(),
later crashkernel=,high|low parsing will be added.  Make adjustments in
all call sites of parse_crashkernel() in arch.

Link: https://lkml.kernel.org/r/20230914033142.676708-3-bhe@redhat.com
Signed-off-by: Baoquan He 
Reviewed-by: Zhen Lei 
Cc: Catalin Marinas 
Cc: Chen Jiahao 
Signed-off-by: Andrew Morton

s390/setup: make use of system_ctl_load()

2023-09-19T11:26:57+00:00

Use system_ctl_load() instead of local_ctl_load() to reflect that
control register changes are supposed to be global.

Even though setup_cr() was ok, note that the usage of local_ctl_load()
would have been wrong, if it would have happened after the control
register save area was initialized: only local control register contents
would have been changed, but wouldn't be used for new CPUs.

With using system_ctl_load() the caller doesn't need to take care.

Signed-off-by: Heiko Carstens 
Reviewed-by: Alexander Gordeev 
Signed-off-by: Vasily Gorbik

s390/ctltreg: make initialization of control register save area explicit

2023-09-19T11:26:56+00:00

Commit e1b9c2749af0 ("s390/smp: ensure global control register contents
are in sync") made the control register save area contained within the
lowcore at absolute address zero a resource which is used when
initializing CPUs. However this is anything but obvious from the code.

Add an ctlreg_init_save_area() function in order to make this explicit.

Signed-off-by: Heiko Carstens 
Reviewed-by: Alexander Gordeev 
Signed-off-by: Vasily Gorbik

s390/ctlreg: add struct ctlreg

2023-09-19T11:26:56+00:00

Add struct ctlreg to enforce strict type checking / usage for control
register functions.

Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens 
Signed-off-by: Vasily Gorbik

s390/ctlreg: change parameters of __local_ctl_load() and __local_ctl_store()

2023-09-19T11:26:56+00:00

Change __local_ctl_load() and __local_ctl_store(), so that control
register parameters come first.

This way all control handling functions consistently have control
register(s) parameter first.

Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens 
Signed-off-by: Vasily Gorbik

s390/ctlreg: use local_ctl_load() and local_ctl_store() where possible

2023-09-19T11:26:56+00:00

Convert all single control register usages of __local_ctl_load() and
__local_ctl_store() to local_ctl_load() and local_ctl_store().

This also requires to change the type of some struct lowcore members
from __u64 to unsigned long.

Reviewed-by: Alexander Gordeev 
Signed-off-by: Heiko Carstens 
Signed-off-by: Vasily Gorbik