The panic function will first print the panic message to the console,
then stop additional cpus with smp_send_stop and finally call the
function on the panic notifier list.
In case of an I/O based console the panic message will cause I/O to
be started and a function on the panic notifier list will wait for the
completion of the I/O. That does not work if an I/O completion interrupt
has already been delivered to a cpu that is then stopped by smp_send_stop.
To break this cyclic dependency add code to smp_send_stop that gives
the additional cpu the opportunity to complete outstanding interrupts.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Call generic IPC demultiplexer instead of having a nearly identical
s390 variant. Also make sure that native and compat handling now have
the same behaviour.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Move the program interruption code and the translation exception identifier
to the pt_regs structure as 'int_code' and 'int_parm_long' and make the
first level interrupt handler in entry[64].S store the two values. That
makes it possible to drop 'prot_addr' and 'trap_no' from the thread_struct
and to reduce the number of arguments to a lot of functions. Finally
un-inline do_trap. Overall this saves 5812 bytes in the .text section of
the 64 bit kernel.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Increase cpu topology change poll frequency if a change is anticipated.
Otherwise a user might be a bit confused to have to wait up to a minute
in order to see a change this should be visible immediatly.
However there is no guarantee that the change will happen during the
time frame the poll frequency is increased.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Another round of cleanup for entry[64].S, in particular the program check
handler looks more reasonable now. The code size for the 31 bit kernel
has been reduced by 616 byte and by 528 byte for the 64 bit version.
Even better the code is a bit faster as well.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

There is no reason for the cpu-measurement-facility host id constant to
reside in the lowcore where space is precious. Use an entry in the literal
pool in HANDLE_SIE_INTERCEPT and a stack slot in sie64a.
While we are at it replace the id -1 with 0 to indicate host execution.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Cleanup z10 topology handling. This adds some more code but hopefully
the result is more readable and easier to maintain.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove all ifdefs from topology code and also only compile it for the
CONFIG_SCHED_BOOK case. The new code selects SCHED_MC if SCHED_BOOK is
selected. SCHED_MC without SCHED_BOOK is not possible anymore.
Furthermore various sysfs attributes are not available anymore for the
!SCHED_BOOK case. In particular all attributes that correspond to
CPU polarization.
But since all real world kernels have SCHED_BOOK selected anyway this
doesn't matter too much.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Currently, when smp_switch_to_ipl_cpu() is done, the backchain in the dump
analysis tool crash looks like the following:

 #0 [1f746e70] __machine_kexec at 11dd92
 #1 [1f746eb8] smp_restart_cpu at 11820e
 #0 [00907eb0] cpu_idle at 10602e
 #1 [00907ef8] start_kernel at 979a08

It would be good to see the registers of the interrupted function.
To achieve this, the backchain on the new stack has to be set to zero.
This looks then like the following:

 #0 [1f746e70] __machine_kexec at 11dd8e
 #1 [1f746eb8] smp_restart_cpu at 11820a
 PSW:  0706000180000000 00000000005c6fe6 (vtime_stop_cpu+134)
 GPRS: 0000000000000000 00000000005c6fe6 0000000001ad0228 0000000001ad0248
       0000000000907f08 0000000001ad0b40 0000000000979344 0000000000000000
       00000000009c0000 00000000009c0010 00000000009ab024 0000000001ad0200
       0000000001ad0238 00000000005cc9d8 000000000010602e 0000000000907e68
 #0 [00907eb0] cpu_idle at 10602e
 #1 [00907ef8] start_kernel at 979a08

In addition to this, now also the correct PSW is stored in the pt_regs
structure that is located at the start of the panic stack.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The kernel address space of a 64 bit kernel currently uses a three level
page table and the vmemmap array has a fixed address and a fixed maximum
size. A three level page table is good enough for systems with less than
3.8TB of memory, for bigger systems four page table levels need to be
used. Each page table level costs a bit of performance, use 3 levels for
normal systems and 4 levels only for the really big systems.
To avoid bloating sparse.o too much set MAX_PHYSMEM_BITS to 46 for a
maximum of 64TB of memory.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

This patch makes the create_mem_hole() function more readable and
fixes some minor bugs (e.g. off-by-one problems).
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The current code in setup_boot_command_line() uses a heuristic to
detect an EBCDIC command line. It checks if any of the bytes in
the command line has bit one (0x80) set. In that case it is assumed
that we have an EBCDIC string and the complete command line is
converted.

On s390 there are cases where the boot loader provides a kernel
command line that is NULL terminated, but has random data after
the NULL termination. In that case, setup_boot_command_line()
might misinterpret an ASCII string for an EBCDIC string. A
subsequent string conversion can then damage the ASCII string.

This patch solves the problem by checking for NULL termination.
If no EBCDIC character has been found until the the NULL
termination has been found, we now assume that we have an ASCII
string.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Mask the extint_code parameter of the smp external interrupt handler
to get the interruption code. Otherwise emergency call interrupts
erroneously might be accounted as emergency signal interrupts.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

git commit 20b40a79 "signal race with restarting system calls"
added code to the poke_user/poke_user_compat to reset the system call
restart information in the thread-info if the PSW address is changed.
The purpose of that change has been to workaround old gdbs that do
not know about the REGSET_SYSTEM_CALL. It turned out that this is not
a good idea, it makes the behaviour of the debuggee dependent on the
order of specific ptrace call, e.g. the REGSET_SYSTEM_CALL register
set needs to be written last. And the workaround does not really fix
old gdbs, inferior calls on interrupted restarting system calls do not
work either way.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The last breaking event address is a read-only value, the regset misses the
.set function. If a PTRACE_SETREGSET is done for NT_S390_LAST_BREAK we
get an oops due to a branch to zero:

Kernel BUG at 0000000000000002 verbose debug info unavailable
illegal operation: 0001 #1 SMP
...
Call Trace:
(<0000000000158294> ptrace_regset+0x184/0x188)
 <00000000001595b6> ptrace_request+0x37a/0x4fc
 <0000000000109a78> arch_ptrace+0x108/0x1fc
 <00000000001590d6> SyS_ptrace+0xaa/0x12c
 <00000000005c7a42> sysc_noemu+0x16/0x1c
 <000003fffd5ec10c> 0x3fffd5ec10c
Last Breaking-Event-Address:
 <0000000000158242> ptrace_regset+0x132/0x188

Add a nop .set function to prevent the branch to zero.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>
Cc: stable@kernel.org

The TIF_SYSCALL bit needs to be cleared if the debugger changes the state
of the ptraced process in regard to the presence of a system call.
Otherwise the system call will be restarted although the debugger set up
an inferior call.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In order to have the same behavior for kdump based stand-alone dump
as for the kexec method, the is_kdump_kernel() check (only true for
the kexec method) has to be replaced by the OLDMEM_BASE check (true
for both methods).
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Make sure that all cpus in a book on a z10 appear as book siblings
and not as core siblings. This fixes some performance regressions that
appeared after the book scheduling domain got introduced.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In ESA mode STCKF is not defined even if the facility bit is enabled.
To prevent an illegal operation we must also check if we run a 64 bit kernel.
To make the check perform well add the STCKF bit to the machine flags.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

When the kernel is started in kdump mode, zfcpdump should not be
initialized because both dump methods can't be used at the same time.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

'readelf -n' on the s390 vmlinux file generates lots of warnings about
corrupt notes. The reason is that the 'NOTE' program header has incorrect
file and memory sizes. The problem is that the section following the
NOTES section do not switch to a different phdr and they get added to
the NOTE program section. Add a dummy entry to the linker script that
switches to the data phdr before the start of the RODATA section.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Currently it can happen that the pre-allocated ELF header contains a wrong
memory map which would result in errors when copying /proc/vmcore.
In order to still get a valid vmcore, we (temporarily) disable the error
checking in copy_oldmem_page(). This will then produce zero pages for those
memory regions.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

We use both the external call and emergency call IPIs to signal remote
cpus. Therefore it makes sense to account them differently withing
/proc/irqstats so we actually know what happened.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use __force to quiet sparse warnings about user address space.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix three sparse warnings in math-emu / sysinfo:

arch/s390/kernel/sysinfo.c:448:17: error: return expression in void function
arch/s390/kernel/sysinfo.c:445:25: warning: shift too big (32) for type unsigned int
arch/s390/kernel/sysinfo.c:445:25: warning: shift too big (32) for type unsigned int
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add prototypes and includes for functions used in different modules.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Make functions and data static to avoid sparse warnings.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove unnecessary code to avoid false positives from sparse, e.g.

arch/s390/kernel/compat_signal.c:221:61: warning: invalid access past the end of 'set32' (8 8)
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

On sie_fault we need to switch back to user ASCE. Otherwise we get
interresting effects when exiting to "userspace" while the guest
space is still active.
Signed-off-by: NCarsten Otte <cotte@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use a sigp sense running to decide which signal processor order to use
for an ipi. If the target cpu is running use external call, if the target
cpu is not running use emergency signal.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add support for CHSC I/O interrupt statistics in /proc/interrupts.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The user space program can change its addressing mode between the
24-bit, 31-bit and the 64-bit mode if the kernel is 64 bit. Currently
the kernel always forces the standard amode on signal delivery and
signal return and on ptrace: 64-bit for a 64-bit process, 31-bit for
a compat process and 31-bit kernels. Change the signal and ptrace code
to allow the full range of addressing modes. Signal handlers are
run in the standard addressing mode for the process.

One caveat is that even an 31-bit compat process can switch to the
64-bit mode. The next signal will switch back into the 31-bit mode
and there is no room in the 31-bit compat signal frame to store the
information that the program came from the 64-bit mode.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Split out addressing mode bits from PSW_BASE_BITS, rename PSW_BASE_BITS
to PSW_MASK_BASE, get rid of psw_user32_bits, remove unused function
enabled_wait(), introduce PSW_MASK_USER, and drop PSW_MASK_MERGE macros.
Change psw_kernel_bits / psw_user_bits to contain only the bits that
are always set in the respective mode.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add an explicit TIF_SYSCALL bit that indicates if a task is inside
a system call. The svc_code in the pt_regs structure is now only
valid if TIF_SYSCALL is set. With this definition TIF_RESTART_SVC
can be replaced with TIF_SYSCALL. Overall do_signal is a bit more
readable and it saves a few lines of code.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

An instruction with an address right below the adress limit for the
current addressing mode will wrap. The instruction restart logic in
the protection fault handler and the signal code need to follow the
wrapping rules to find the correct instruction address.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

For a ERESTARTNOHAND/ERESTARTSYS/ERESTARTNOINTR restarting system call
do_signal will prepare the restart of the system call with a rewind of
the PSW before calling get_signal_to_deliver (where the debugger might
take control). For A ERESTART_RESTARTBLOCK restarting system call
do_signal will set -EINTR as return code.
There are two issues with this approach:
1) strace never sees ERESTARTNOHAND, ERESTARTSYS, ERESTARTNOINTR or
   ERESTART_RESTARTBLOCK as the rewinding already took place or the
   return code has been changed to -EINTR
2) if get_signal_to_deliver does not return with a signal to deliver
   the restart via the repeat of the svc instruction is left in place.
   This opens a race if another signal is made pending before the
   system call instruction can be reexecuted. The original system call
   will be restarted even if the second signal would have ended the
   system call with -EINTR.

These two issues can be solved by dropping the early rewind of the
system call before get_signal_to_deliver has been called and by using
the TIF_RESTART_SVC magic to do the restart if no signal has to be
delivered. The only situation where the system call restart via the
repeat of the svc instruction is appropriate is when a SA_RESTART
signal is delivered to user space.

Unfortunately this breaks inferior calls by the debugger again. The
system call number and the length of the system call instruction is
lost over the inferior call and user space will see ERESTARTNOHAND/
ERESTARTSYS/ERESTARTNOINTR/ERESTART_RESTARTBLOCK. To correct this a
new ptrace interface is added to save/restore the system call number
and system call instruction length.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove the save_area_64 field from the 0xe00 - 0xf00 area in the lowcore.
Use a free slot in the save_area array instead.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

This patch implements the crash_map_pages() function for s390.
KEXEC_CRASH_MEM_ALIGN is set to HPAGE_SIZE, in order to support
kernel mappings that use large pages. We also use HPAGE_SIZE alignment
for CONFIG_HUGETLB_PAGE=n in order to have the same 1 MiB alignment on
all s390 systems.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>