Instead of encoding shift in the table address, use an enumerated index value.
This allow us to do different things in the callback for pte and pmd.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

4K config use one full page at level 4 of the pagetable. Add support for single
fragment allocation in pagetable fragment code and and use that for 4K config.
This makes both 4k and 64k use the same code path. Later we will switch pmd to
use the page table fragment code. This is done only for 64bit platforms which
is using page table fragment support.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Now that we have removed 64K page size support, the RCU page table free can
be much simpler for nohash. Make a copy of the the rcu callback to pgalloc.h
header similar to nohash 32. We could possibly merge 32 and 64 bit there. But
that is for a later patch

We also move the book3s specific handler to pgtable_book3s64.c. This will be
updated in a later patch to handle split pmd ptlock.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

We have in Kconfig

config PPC_64K_PAGES
	bool "64k page size"
	depends on !PPC_FSL_BOOK3E && (44x || PPC_BOOK3S_64 || PPC_BOOK3E_64)
	select HAVE_ARCH_SOFT_DIRTY if PPC_BOOK3S_64

Only supported BOOK3E 64 bit platforms is FSL_BOOK3E. Remove the dead 64k page
support code from 64bit nohash.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

We rename the alloc and get_from_cache to indicate they operate on pte
fragments. In later patch we will add pmd fragment support.

No functional change in this patch.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

In later patch we switch pmd_lock from mm->page_table_lock to split pmd ptlock.
It avoid compilations issues, use pmd_lockptr helper.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Only code movement and avoid #ifdef.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

This brings in one commit that we may want to share with the kvm-ppc
tree, to avoid merge conflicts and get wider testing.

In the next set of patches, we will switch pmd allocator to use page fragments
and the locking will be updated to split pmd ptlock. We want to avoid using
fragments for partition-scoped table. Use slab cache similar to level 4 table
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Use new return type vm_fault_t for fault handler. For now, this is
just documenting that the function returns a VM_FAULT value rather
than an errno. Once all instances are converted, vm_fault_t will
become a distinct type. See commit 1c8f4220 ("mm: change return
type to vm_fault_t").

We are fixing a minor bug, that the error from vm_insert_pfn() was
being ignored and the effect of this is likely to be only felt in OOM
situations.
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Trivial fix to spelling mistake in debug messages of a structure
field name
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The exec_target binary could segfault calling _exit(2) because r13
is not set up properly (and libc looks at that when performing a
syscall). Call SYS_exit using syscall(2) which doesn't seem to
have this problem.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

At the moment we assume that IODA2 and newer PHBs can always do 4K/64K/16M
IOMMU pages, however this is not the case for POWER9 and now skiboot
advertises the supported sizes via the device so we use that instead
of hard coding the mask.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Currently memtrace doesn't build if NUMA=n:

  In function ‘memtrace_alloc_node’:
  arch/powerpc/platforms/powernv/memtrace.c:134:6:
  error: the address of ‘contig_page_data’ will always evaluate as ‘true’
    if (!NODE_DATA(nid) || !node_spanned_pages(nid))
        ^

This is because for NUMA=n NODE_DATA(nid) points to an always
allocated structure, contig_page_data.

But even in the NUMA=y case memtrace_alloc_node() is only called for
online nodes, and we should always have a NODE_DATA() allocated for an
online node. So remove the (hopefully) overly paranoid check, which
also means we can build when NUMA=n.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

In commit e6a6928c ("of/fdt: Convert FDT functions to use
libfdt") (Apr 2014), the generic flat device tree code dropped support
for flat device tree's older than version 0x10 (16).

We still have code in our CPU scanning to cope with flat device tree
versions earlier than 2, which can now never trigger, so drop it.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Add a test of the relative branch patching logic in the alternate
section feature fixup code. This tests that if we branch past the last
instruction of the alternate section, the branch is not patched.
That's because the assembler will have created a branch that already
points to the first instruction after the patched section, which is
correct and needs no further patching.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

We want this to remain the last test (because it's disabled by
default), so give it a non-numbered name so we don't have to renumber
it when adding new tests before it.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The code patching code has always been a bit confused about whether
it's best to use void *, unsigned int *, char *, etc. to point to
instructions. In fact in the feature fixups tests we use both unsigned
int[] and u8[] in different places.

Unfortunately the tests that use unsigned int[] calculate the size of
the code blocks using subtraction of those unsigned int pointers, and
then pass the result to memcmp(). This means we're only comparing 1/4
of the bytes we need to, because we need to multiply by
sizeof(unsigned int) to get the number of *bytes*.

The result is that the tests do all the patching and then only compare
some of the resulting code, so patching bugs that only effect that
last 3/4 of the code could slip through undetected. It turns out that
hasn't been happening, although one test had a bad expected case (see
previous commit).

Fix it for now by multiplying the size by 4 in the affected functions.

Fixes: 362e7701 ("powerpc: Add self-tests of the feature fixup code")
Epic-brown-paper-bag-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The expected case for this test was wrong, the source of the alternate
code sequence is:

  FTR_SECTION_ELSE
  2:	or	2,2,2
  	PPC_LCMPI	r3,1
  	beq	3f
  	blt	2b
  	b	3f
  	b	1b
  ALT_FTR_SECTION_END(0, 1)
  3:	or	1,1,1
  	or	2,2,2
  4:	or	3,3,3

So when it's patched the '3' label should still be on the 'or 1,1,1',
and the 4 label is irrelevant and can be removed.

Fixes: 362e7701 ("powerpc: Add self-tests of the feature fixup code")
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

If the systbl_chk.sh checks fail we print a message, but with no
indication that it's an error. That makes it hard to find in build
logs with eg. grep.

So prefix any output with "Error:".
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

it had always been pointless - compat_sys_select() sign-extends
the first argument just fine on its own.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
[mpe: Use COMPAT_SPU_NEW() to keep systbl_chk.sh happy]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Currently the select system call is wired up with the SYSX_SPU()
macro. The SYSX_SPU() is not handled by systbl_chk.c, which means the
syscall number for select is not checked.

That hides the fact that the syscall number for select is actually
__NR__newselect not __NR_select.

In a following patch we'd like to drop ppc32_select() which means
select will become a regular COMPAT_SYS_SPU() syscall. But
COMPAT_SYS_SPU() can't deal with the fact that the syscall number is
actually __NR__newselect. We also can't just redefine __NR_select
because that's still used for the old select call.

So add a new COMPAT_NEW_SPU() that does the same thing as
COMPAT_SYS_SPU() except it encodes that we're using the new number.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
[mpe: Update sys_ni.c for s/ppc_rtas/sys_rtas/]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
[mpe: Fix sys_debug_setcontext() prototype to return long]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The "Power Architecture 64-Bit ELF V2 ABI" says in section 2.3.2.3:

[...] There are several rules that must be adhered to in order to ensure
reliable and consistent call chain backtracing:

* Before a function calls any other function, it shall establish its
  own stack frame, whose size shall be a multiple of 16 bytes.

 – In instances where a function’s prologue creates a stack frame, the
   back-chain word of the stack frame shall be updated atomically with
   the value of the stack pointer (r1) when a back chain is implemented.
   (This must be supported as default by all ELF V2 ABI-compliant
   environments.)
[...]
 – The function shall save the link register that contains its return
   address in the LR save doubleword of its caller’s stack frame before
   calling another function.

To me this sounds like the equivalent of HAVE_RELIABLE_STACKTRACE.
This patch may be unneccessarily limited to ppc64le, but OTOH the only
user of this flag so far is livepatching, which is only implemented on
PPCs with 64-LE, a.k.a. ELF ABI v2.

Feel free to add other ppc variants, but so far only ppc64le got tested.

This change also implements save_stack_trace_tsk_reliable() for ppc64le
that checks for the above conditions, where possible.
Signed-off-by: Torsten Duwe <duwe@suse.de>
Signed-off-by: Nicolai Stange <nstange@suse.de>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Provide timebase and timebase of last heartbeat in watchdog lockup
messages. Also provide a stack trace of when a CPU becomes un-stuck,
which can be useful -- it could be where irqs are re-enabled, so it
may be the end of the critical section which is responsible for the
latency which is useful information.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The watchdog heartbeat timestamp is updated when the local heartbeat
timer fires (or touch_nmi_watchdog() is called).

This is an interesting data point, so don't overwrite it when the
soft-NMI interrupt detects a hard lockup. That code came from a pre-
merge version to prevent hard lockup messages flood, but that's taken
care of with the stuck CPU logic now, so there is no reason to
update the heartbeat timestamp here.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

This is not the case for the moment, but future releases of pHyp might
need to introduce some synchronisation routines under the hood which
would make the XIVE hcalls longer to complete.

As this was done for H_INT_RESET, let's wrap the other hcalls in a
loop catching the H_LONG_BUSY_* codes.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The hcall H_INT_RESET should be called to make sure XIVE is fully
reseted.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The hcall H_INT_RESET can take some time to complete and in such cases
it returns H_LONG_BUSY_* codes requiring the machine to sleep for a
while before retrying.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The kexec_state KEXEC_STATE_IRQS_OFF barrier is reached by all
secondary CPUs before the kexec_cpu_down() operation is called on
secondaries. This can raise conflicts and provoque errors in the XIVE
hcalls when XIVE is shutdown with H_INT_RESET on the primary CPU.

To synchronize the kexec_cpu_down() operations and make sure the
secondaries have completed their task before the primary starts doing
the same, let's move the primary kexec_cpu_down() after the
KEXEC_STATE_REAL_MODE barrier.

This change of the ending sequence of kexec is mostly useful on the
pseries platform but it impacts also the powernv, ps3 and 85xx
platforms. powernv can be easily tested and fixed but some caution is
required for the other two.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

For consideration:

* Add NVDIMM support - Enables greater testing, mambo device.
* Add IPv6 support built in + additional modules - Because it's 2018 maan.
* Add DEFERRED_STRUCT_PAGE_INIT - Let's see what breaks.
* Add PPC_MEMTRACE - Small powernv debugfs driver for getting hardware traces.
* Add MEMORY_FAILURE - Machine check exceptions can now drive memory failure.
* Turn on FANOTIFY - This is the current filesystem notification feature.
* Turn on SCOM_DEBUGFS - Handy for hardware/firmware debugging, security risk?
* Turn on async SCSI scanning - Let's see what breaks.
* Add MLX5 driver as a module - Popular demand.
* Add CRYPTO_CRCT10DIF_VPMSUM - POWER8 T10DIF acceleration.

* Make a bunch of USB hid drivers modules.
* Make SCSI SG, SR, and FC modules - FC is huge.
* Make video drivers except AST GPU modules - Also huge.
* Make PCI serial driver a module - Uncommon.
* Make more things modules, NFS FS, RAM disk, netconsole, MS-DOS fs.

* Get rid of /dev/port - Not used.
* Remove PPS and PTP subsystms - Unusual.
* Remove legacy BSD ttys - Long dead.
* Remove IDE - Deprecated and replaced with ATA.
* Remove WIRELESS - Until we get POWER9 laptops.
* Remove RAW - Long deprecated in favour of direct IO.
* Remove floppy, parport, and PS2 input devices - not supported.
* Remove virtio drivers, ballooning - We're host only.
* Remove PPP - Sorry Paulus.

This results in a significantly smaller vmlinux:

    text      data       bss        dec   filename
13143383   5277944   1317856   19739183   vanilla
12263281   4852074   1341720   18457075   patched
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The B43 driver only needs CONFIG_SSB to support the WLAN card found in
the Wii. Configure it accordingly, and disable BCMA bus support to save
a bit of space.
Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

This allows access to the SD card and the BCM4318 Wifi module.
Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Now that there's a GPIO driver for the Wii, let's enable the following
drivers:

- the GPIO driver itself
- gpio-keys
- gpio-poweroff
- gpio-leds and a few LED triggers
Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The Wii doesn't have built-in Ethernet and USB Ethernet adapters are in
a different menu. Disable CONFIG_ETHERNET to save some space in support
code for Ethernet drivers.

Note that this patch doesn't disable any Ethernet drivers, because they
are not enabled by default.
Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

The hcall_exit() tracepoint has retval defined as unsigned long. That
leads to humours results like:

  bash-3686  [009] d..2   854.134094: hcall_entry: opcode=24
  bash-3686  [009] d..2   854.134095: hcall_exit: opcode=24 retval=18446744073709551609

It's normal for some hcalls to return negative values, displaying them
as unsigned isn't very helpful. So change it to signed.

  bash-3711  [001] d..2   471.691008: hcall_entry: opcode=24
  bash-3711  [001] d..2   471.691008: hcall_exit: opcode=24 retval=-7

Which can be more easily compared to H_NOT_FOUND in hvcall.h
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Acked-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Tested-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>