For machines without the exrl instruction the BFP jit generates
code that uses an "br %r1" instruction located in the lowcore page.
Unfortunately there is a cut & paste error that puts an additional
"larl %r1,.+14" instruction in the code that clobbers the branch
target address in %r1. Remove the larl instruction.

Cc: <stable@vger.kernel.org> # v4.17+
Fixes: de5cb6eb ("s390: use expoline thunks in the BPF JIT")
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The memove, memset, memcpy, __memset16, __memset32 and __memset64
function have an additional indirect return branch in form of a
"bzr" instruction. These need to use expolines as well.

Cc: <stable@vger.kernel.org> # v4.17+
Fixes: 97489e06 ("s390/lib: use expoline for indirect branches")
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The numa_init_early initcall sets the node_to_cpumask_map[0] to the
full cpu_possible_mask. Unfortunately this early_initcall is too late,
the NUMA setup for numa=emu is done even earlier. The order of calls
is numa_setup() -> emu_update_cpu_topology(), then the early_initcalls(),
followed by sched_init_domains().

Starting with git commit 051f3ca0
"sched/topology: Introduce NUMA identity node sched domain"
the incorrect node_to_cpumask_map[0] really screws up the domain
setup and the kernel panics with the follow oops:

Cc: <stable@vger.kernel.org> # v4.15+
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Before the memory for the elfcorehdr is allocated the required size is
estimated with

       alloc_size = 0x1000 + get_cpu_cnt() * 0x4a0 +
               mem_chunk_cnt * sizeof(Elf64_Phdr);

Where 0x4a0 is used as size for the ELF notes to store the register
contend. This size is 8 bytes too small. Usually this does not immediately
cause a problem because the page reserved for overhead (Elf_Ehdr,
vmcoreinfo, etc.) is pretty generous. So usually there is enough spare
memory to counter the mis-calculated per cpu size. However, with growing
overhead and/or a huge cpu count the allocated size gets too small for the
elfcorehdr. Ultimately a BUG_ON is triggered causing the crash kernel to
panic.

Fix this by properly calculating the required size instead of relying on
magic numbers.

Fixes: a62bc073 ("s390/kdump: add support for vector extension")
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Processing the samples in the AUX-area by perf requires the computation
of respective time stamps.  The time stamps used by perf are based on
the monotonic clock.  To convert the TOD clock value contained in an
SDB to a monotonic clock value, the TOD clock base is required.  Hence,
also save the TOD clock base in the SDB.
Suggested-by: Thomas Richter <tmricht@linux.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Pull hlp_stage1 from Christian Borntraeger with the following changes:

KVM: s390: initial host large page support

- must be enabled via module parameter hpage=1
- cannot be used together with nested
- does support migration
- does support hugetlbfs
- no THP yet

General KVM huge page support on s390 has to be enabled via the
kvm.hpage module parameter. Either nested or hpage can be enabled, as
we currently do not support vSIE for huge backed guests. Once the vSIE
support is added we will either drop the parameter or enable it as
default.

For a guest the feature has to be enabled through the new
KVM_CAP_S390_HPAGE_1M capability and the hpage module
parameter. Enabling it means that cmm can't be enabled for the vm and
disables pfmf and storage key interpretation.

This is due to the fact that in some cases, in upcoming patches, we
have to split huge pages in the guest mapping to be able to set more
granular memory protection on 4k pages. These split pages have fake
page tables that are not visible to the Linux memory management which
subsequently will not manage its PGSTEs, while the SIE will. Disabling
these features lets us manage PGSTE data in a consistent matter and
solve that problem.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

Let's allow huge pmd linking when enabled through the
KVM_CAP_S390_HPAGE_1M capability. Also we can now restrict gmap
invalidation and notification to the cases where the capability has
been activated and save some cycles when that's not the case.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

Guests backed by huge pages could theoretically free unused pages via
the diagnose 10 instruction. We currently don't allow that, so we
don't have to refault it once it's needed again.
Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>

When doing skey emulation for huge guests, we now need to fault in
pmds, as we don't have PGSTES anymore to store them when we do not
have valid table entries.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>

Storage keys for guests with huge page mappings have to be managed in
hardware. There are no PGSTEs for PMDs that we could use to retain the
guests's logical view of the key.
Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

Similarly to the pte skey handling, where we set the storage key to
the default key for each newly mapped pte, we have to also do that for
huge pmds.

With the PG_arch_1 flag we keep track if the area has already been
cleared of its skeys.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

When a guest starts using storage keys, we trap and set a default one
for its whole valid address space. With this patch we are now able to
do that for large pages.

To speed up the storage key insertion, we use
__storage_key_init_range, which in-turn will use sske_frame to set
multiple storage keys with one instruction. As it has been previously
used for debuging we have to get rid of the default key check and make
it quiescing.
Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
[replaced page_set_storage_key loop with __storage_key_init_range]
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

To do dirty loging with huge pages, we protect huge pmds in the
gmap. When they are written to, we unprotect them and mark them dirty.

We introduce the function gmap_test_and_clear_dirty_pmd which handles
dirty sync for huge pages.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Acked-by: David Hildenbrand <david@redhat.com>

If the host invalidates a pmd, we also have to invalidate the
corresponding gmap pmds, as well as flush them from the TLB. This is
necessary, as we don't share the pmd tables between host and guest as
we do with ptes.

The clearing part of these three new functions sets a guest pmd entry
to _SEGMENT_ENTRY_EMPTY, so the guest will fault on it and we will
re-link it.

Flushing the gmap is not necessary in the host's lazy local and csp
cases. Both purge the TLB completely.
Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Acked-by: David Hildenbrand <david@redhat.com>

Like for ptes, we also need invalidation notification for pmds, to
make sure the guest lowcore pages are always accessible and later
addition of shadowed pmds.

With PMDs we do not have PGSTEs or some other bits we could use in the
host PMD. Instead we pick one of the free bits in the gmap PMD. Every
time a host pmd will be invalidated, we will check if the respective
gmap PMD has the bit set and in that case fire up the notifier.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>

Let's allow pmds to be linked into gmap for the upcoming s390 KVM huge
page support.

Before this patch we copied the full userspace pmd entry. This is not
correct, as it contains SW defined bits that might be interpreted
differently in the GMAP context. Now we only copy over all hardware
relevant information leaving out the software bits.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

Currently we use the software PGSTE bits PGSTE_IN_BIT and
PGSTE_VSIE_BIT to notify before an invalidation occurs on a prefix
page or a VSIE page respectively. Both bits are pgste specific, but
are used when protecting a memory range.

Let's introduce abstract GMAP_NOTIFY_* bits that will be realized into
the respective bits when gmap DAT table entries are protected.
Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

This patch reworks the gmap_protect_range logic and extracts the pte
handling into an own function. Also we do now walk to the pmd and make
it accessible in the function for later use. This way we can add huge
page handling logic more easily.
Signed-off-by: Janosch Frank <frankja@linux.vnet.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Now that the early boot rework is upstream we can enable the gcc plugins
again. See git commit 72f108b308707f21499e0ac05bf7370360cf06d8
"s390: disable gcc plugins" for reference.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The s390 build currently fails with the latent entropy plugin:

arch/s390/kernel/als.o: In function `verify_facilities':
als.c:(.init.text+0x24): undefined reference to `latent_entropy'
als.c:(.init.text+0xae): undefined reference to `latent_entropy'
make[3]: *** [arch/s390/boot/compressed/vmlinux] Error 1
make[2]: *** [arch/s390/boot/compressed/vmlinux] Error 2
make[1]: *** [bzImage] Error 2

This will be fixed with the early boot rework from Vasily, which
is planned for the 4.19 merge window.

For 4.18 the simplest solution is to disable the gcc plugins and
reenable them after the early boot rework is upstream.
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
(cherry picked from commit 2fba3573)

Use new return type vm_fault_t for fault handler vdso_fault.
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
__field macro for the necessary members.
Keep the __fieldstruct macro to extract the complete
structure when dumps are analysed.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
__field macro for the necessary members.
Keep the __fieldstruct macro to extract the complete
structure when dumps are analysed.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
__field macro for the necessary members.
Keep the __fieldstruct macro to extract the complete
structure when dumps are analysed.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
the __field macro for the missing members.
Keep the __fieldstruct macro to extract the complete
structure when dumps are analysed.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
__field macro for the members adapter_IO, isc and type
of struct tpi_info.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Tools like 'perf stat' parse the trace point format files defined
in /sys/kernel/debug/tracing/events/s390/.../format to handle
the print fmt: statement. The kernel provides a library in
directory linux/tools/lib/traceevent/* for this reason.

This library can not handle structures or unions defined in
the TRACE_EVENT/TP_STRUCT__entry macros with __field_struct macro.
There is no possibility to extract a structure member
(which might be a bit field) since there is no packing
information nor bit field offset by parsing the printf fmt line.

Therefore rewrite the TRACE_EVENT macro and add the
__field macro for the necessary fields.
Keep the __fieldstruct macro to extract the complete
structure when dumps are analysed.

Note that the same information is displayed, this is no
interface change.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

PTR_RET is deprecated, use PTR_ERR_OR_ZERO instead.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

PTR_RET is deprecated, use PTR_ERR_OR_ZERO instead.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

PTR_RET is deprecated, use PTR_ERR_OR_ZERO instead.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

PTR_RET is deprecated, use PTR_ERR_OR_ZERO instead.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The kbd_ioctl uses two user controlled indexes for KDGKBENT/KDSKBENT.
Use array_index_nospec to prevent any out of bounds speculation.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Detect and report the etoken facility. With spectre_v2=auto or
CONFIG_EXPOLINE_AUTO=y automatically disable expolines and use
the full branch prediction mode for the kernel.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Remove attribute packed where possible failing this add proper alignment
information to fix warnings like the one below:

drivers/s390/cio/chsc.c: In function 'chsc_siosl':
drivers/s390/cio/chsc.c:1287:2: warning: alignment 1 of 'struct <anonymous>' is less than 4 [-Wpacked-not-aligned]
  } __attribute__ ((packed)) *siosl_area;

Note: this patch should be a nop since non of these structs use auto
storage but allocated pages. However there are changes to the generated
code because of additional padding at the end of some of the structs due
to alignment when memset(foo, 0, sizeof(*foo)) is used.
Signed-off-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Both css_evaluate_new_subchannel and cio_validate_subchannel used
stsch and css_sch_is_valid to check for a valid device.

Reduce stsch calls during subchannel evaluation by re-using schib
data. Also the type/devno valid information is only checked once.
Signed-off-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

In css_alloc_subchannel we allocate the subchannel and do a
validation of the subchannel (to decide if we should look for
devices via this subchannel). On a typical LPAR we find lots
of subchannels to be invalid (because there is no device
attached or the device is blacklisted) leading to lots of
useless kmalloc and kfree calls.

This patch changes the order to only allocate the subchannels
that have been found valid.
Signed-off-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

The css bus code uses 2 initcalls: channel_subsystem_init to
initialize internal data and channel_subsystem_init_sync to
start scanning for devices and wait for it to finish.

The start scanning for devices part is moved to the first
initcall such that more work happens in parallel.
Signed-off-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

Improve locking in chp_new to make sure that we don't register
the same chpid twice. Chpid registration was synchronized via
the machine check handler thread but we also have codepaths to
look for new chpids triggered independent of that thread (during
IPL or resume from hibernate).
Signed-off-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

When the oom killer kills a userspace process in the page fault handler
while in guest context, the fault handler fails to release the mm_sem
if the FAULT_FLAG_RETRY_NOWAIT option is set. This leads to a deadlock
when tearing down the mm when the process terminates. This bug can only
happen when pfault is enabled, so only KVM clients are affected.

The problem arises in the rare cases in which handle_mm_fault does not
release the mm_sem. This patch fixes the issue by manually releasing
the mm_sem when needed.

Fixes: 24eb3a82 ("KVM: s390: Add FAULT_FLAG_RETRY_NOWAIT for guest fault")
Cc: <stable@vger.kernel.org> # 3.15+
Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>