We don't save/restore these across a trap, or with KPROBES_ON_FTRACE.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When called from xive_irq_startup(), the size of the cpumask can be
larger than nr_cpu_ids. This can result in a WARN_ON such as:
WARNING: CPU: 10 PID: 1 at ../arch/powerpc/sysdev/xive/common.c:476 xive_find_target_in_mask+0x110/0x2f0
...
NIP [c00000000008a310] xive_find_target_in_mask+0x110/0x2f0
LR [c00000000008a2e4] xive_find_target_in_mask+0xe4/0x2f0
Call Trace:
xive_find_target_in_mask+0x74/0x2f0 (unreliable)
xive_pick_irq_target.isra.1+0x200/0x230
xive_irq_startup+0x60/0x180
irq_startup+0x70/0xd0
__setup_irq+0x7bc/0x880
request_threaded_irq+0x14c/0x2c0
request_event_sources_irqs+0x100/0x180
__machine_initcall_pseries_init_ras_IRQ+0x104/0x134
do_one_initcall+0x68/0x1d0
kernel_init_freeable+0x290/0x374
kernel_init+0x24/0x170
ret_from_kernel_thread+0x5c/0x74
This happens because we're being called with our affinity mask set to
irq_default_affinity. That in turn was populated using
cpumask_setall(), which sets NR_CPUs worth of bits, not nr_cpu_ids
worth. Finally cpumask_weight() will return > nr_cpu_ids when passed a
mask which has > nr_cpu_ids bits set.
Fix it by limiting the value returned by cpumask_weight().
Signed-off-by: Cédric Le Goater <clg@kaod.org>
[mpe: Add change log details on actual cause]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On modern CPUs the CTRL register is read-only except bit 63 which is
the run latch control. This means it can be updated with a mtspr
rather than mfspr/mtspr.
To accomodate older CPUs (Cell at least), where there are other bits
in the register, we still do a read/modify/write on pre 2.06 CPUs.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Update change log to mention 2.06 workaround]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
HVI interrupts have always used 0x500, so remove the dead branch.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
POWER9 host external interrupts use the h_virt_irq_common handler, so
use that to replay them rather than using the hardware_interrupt_common
handler. Both call do_IRQ, but using the correct handler reduces
i-cache footprint.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This results in smaller code, and fewer branches. This relies on the
fact that both the 0xe80 and 0xa00 handlers call the same upper level
code, namely doorbell_exception().
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Mention we rely on the implementation of the 0xe80/0xa00 handlers]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Move the clearing of irq_happened bits into the condition where they
were found to be set. This reduces instruction count slightly, and
reduces stores into irq_happened.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Places in the kernel where r13 is not the PACA pointer must have
maskable interrupts disabled, so r13 does not have to be restored when
returning from a soft-masked interrupt. We should never have
interrupts soft disabled when we're in user space.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
MSR_EE is always enabled in SRR1 for masked interrupts, so we can use
xor to clear it.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Interrupts which do not require EE to be cleared can all be tested
with a single bitwise test.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
It's too big to be inline, there is no reason to keep it
that way.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[mpe: Rework to incorporate the comment changes via fixes branch]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Instead of comparing the whole CPU mask every time, let's
keep a counter of how many bits are set in the mask. Thus
testing for a local mm only requires testing if that counter
is 1 and the current CPU bit is set in the mask.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
We open-code testing for the mm being local to the current CPU
in a few places. Use our existing helper instead.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
It calls switch_mm() which already does the irq save/restore
these days.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Makes switch_mm_irqs_off() a bit more readable
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In __replay_interrupt() we take the address of a local label so we can
return to it later. However the assembler turns the local label into a
symbol with a name like ".L1^B42" - where "^B" is literally "\002".
This does not make for pleasant stack traces. Fix it by giving the
label a sensible name.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In preparation to stop storing the full node path in full_name, remove the
dependency on full_name from dlpar_attach_node(). Callers of
dlpar_attach_node() already have the parent device_node, so just pass the
parent node into dlpar_attach_node instead of the path. This avoids doing
a lookup of the parent node by the path.
Signed-off-by: Rob Herring <robh@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Now that we have a custom printf format specifier, convert users of
full_name to use %pOF instead. This is preparation to remove storing
of the full path string for each node.
Signed-off-by: Rob Herring <robh@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Anatolij Gustschin <agust@denx.de>
Cc: Scott Wood <oss@buserror.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: linuxppc-dev@lists.ozlabs.org
Reviewed-by: Tyrel Datwyler <tyreld@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently in the vio.c code we use a comparision against the parent
device node's full path to decide if the device is a PFO or VIO family
device.
Both the ibm,platform-facilities and vdevice nodes are defined by PAPR,
and must have a matching device_type. So instead of using the path we
can instead compare the device_type.
I've checked Qemu and kvmtool both do this correctly, and all the
PowerVM systems I have access to do also. So it seems to be safe.
This removes the dependency on full_name, which is being removed
upstream.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
There's a non-trivial dependency between some commits we want to put in
next and the KVM prefetch work around that went into fixes. So merge
fixes into next.
There is no guarantee that the various isync's involved with
the context switch will order the update of the CPU mask with
the first TLB entry for the new context being loaded by the HW.
Be safe here and add a memory barrier to order any subsequent
load/store which may bring entries into the TLB.
The corresponding barrier on the other side already exists as
pte updates use pte_xchg() which uses __cmpxchg_u64 which has
a sync after the atomic operation.
Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Add comments in the code]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
We use mm cpumask for serializing against lockless page table walk.
Anybody who is doing a lockless page table walk is expected to disable
irq and only cpus in mm cpumask is expected do the lockless walk. This
ensure that a THP split can send IPI to only cpus in the mm cpumask,
to make sure there are no parallel lockless page table walk.
Add the CAPI fault handling cpu to the mm cpumask so that we can do
the lockless page table walk while inserting hash page table entries.
Reviewed-by: Frederic Barrat <fbarrat@linux.vnet.ibm.com>
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 made sure that lockless walk of linux page table is mostly
limitted to current task(current->mm->pgdir) we can update the THP
update sequence to only send IPI to CPUs on which this task has run.
This helps in reducing the IPI overload on systems with large number
of CPUs.
WRT kvm even though kvm is walking page table with vpc->arch.pgdir,
it is done only on secondary CPUs and in that case we have primary CPU
added to task's mm cpumask. Sending an IPI to primary will force the
secondary to do a vm exit and hence this mm cpumask usage is safe
here.
WRT CAPI, we still end up walking linux page table with capi context
MM. For now the pte lookup serialization sends an IPI to all CPUs in
CPI is in use. We can further improve this by adding the CAPI
interrupt handling CPU to task mm cpumask. That will be done in a
later patch.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Bring in the commit to rename find_linux_pte_or_hugepte() which touches
arch and KVM code, and might need to be merged with the kvmppc tree to
avoid conflicts.
Add newer helpers to make the function usage simpler. It is always
recommended to use find_current_mm_pte() for walking the page table.
If we cannot use find_current_mm_pte(), it should be documented why
the said usage of __find_linux_pte() is safe against a parallel THP
split.
For now we have KVM code using __find_linux_pte(). This is because kvm
code ends up calling __find_linux_pte() in real mode with MSR_EE=0 but
with PACA soft_enabled = 1. We may want to fix that later and make
sure we keep the MSR_EE and PACA soft_enabled in sync. When we do that
we can switch kvm to use find_linux_pte().
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Use the newly introduced memset32() to pre-fill BPF page(s) with trap
instructions.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Based on Matthew Wilcox's patches for other architectures.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
nest_imc_refc is a reference count struct, used to track number of
active perf sessions using the nest units.
Currently the code accesses nest_imc_refc using node_id, which is
incorrect, the array is indexed by node number. Meaning in the case of
sparse node ids we index off the end of the array.
Fix it to use get_nest_pmu_ref() which uses the existing per-cpu
variable local_nest_imc_refc.
Fixes: 885dcd709b ('powerpc/perf: Add nest IMC PMU support')
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
[mpe: Tweak change log]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Declare bin_attribute structures as const as they are only passed as an
argument to the function sysfs_create_bin_file. This argument is of
type const, so declare the structure as const.
Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
__giveup_vsx/save_vsx are completely equivalent to testing MSR_FP
and MSR_VEC and calling the corresponding giveup/save function so
just remove the spurious VSX cases. Also add WARN_ONs checking that
we never have VSX enabled without the two other.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
__giveup_fpu() already does it and we cannot have MSR_VSX set
without having MSR_FP also set.
This also adds a warning to check we indeed do
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
__giveup_vsx() already calls those two functions.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Commit a7be6e5a7f ("mm: drop useless local parameters of
__register_one_node()") removes the last user of parent_node().
The parent_node() macro in POWERPC platform is unnecessary.
Remove it for cleanup.
Reported-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Dou Liyang <douly.fnst@cn.fujitsu.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
VSX uses a combination of the old vector registers, the old FP
registers and new "second halves" of the FP registers.
Thus when we need to see the VSX state in the thread struct
(flush_vsx_to_thread()) or when we'll use the VSX in the kernel
(enable_kernel_vsx()) we need to ensure they are all flushed into
the thread struct if either of them is individually enabled.
Unfortunately we only tested if the whole VSX was enabled, not if they
were individually enabled.
Fixes: 72cd7b44bc ("powerpc: Uncomment and make enable_kernel_vsx() routine available")
Cc: stable@vger.kernel.org # v4.3+
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Now that we have GIGANTIC_PAGE enabled on powerpc, use this for 16G hugepages
with hash translation mode. Depending on the total system memory we have, we may
be able to allocate 16G hugepages runtime. This also remove the hugetlb setup
difference between hash/radix translation mode.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
With commit aa888a7497 ("hugetlb: support larger than MAX_ORDER") we added
support for allocating gigantic hugepages via kernel command line. Switch
ppc64 arch specific code to use that.
W.r.t FSL support, we now limit our allocation range using BOOTMEM_ALLOC_ACCESSIBLE.
We use the kernel command line to do reservation of hugetlb pages on powernv
platforms. On pseries hash mmu mode the supported gigantic huge page size is
16GB and that can only be allocated with hypervisor assist. For pseries the
command line option doesn't do the allocation. Instead pseries does gigantic
hugepage allocation based on hypervisor hint that is specified via
"ibm,expected#pages" property of the memory node.
Cc: Scott Wood <oss@buserror.net>
Cc: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
gup_hugepte() checks if pages are present and readable, and
when 'write' is set, also checks if the pages are writable.
Initially this was done by checking if _PAGE_PRESENT and
_PAGE_READ were set. In addition, _PAGE_WRITE was verified for write
accesses.
The problem is that we have to handle the three following cases:
1/ The target defines __PAGE_READ and __PAGE_WRITE
2/ The target defines __PAGE_RW
3/ The target defines __PAGE_RO
In case 1/, this is obvious
In case 2/, __PAGE_READ is defined as 0 and __PAGE_WRITE as __PAGE_RW
so it works as well.
But in case 3, __PAGE_RW is defined as 0, which means __PAGE_WRITE is 0
and then the test returns true (page writable) in all cases.
A first correction was attempted in commit 6b8cb66a6a ("powerpc: Fix
usage of _PAGE_RO in hugepage"), but that fix is wrong:
instead of checking that the page is writable when write is requested,
it checks that the page is NOT writable when write is NOT requested.
This patch adds a new pte_read() helper to check whether a page is
readable or not. This avoids handling all possible cases in
gup_hugepte().
Then gup_hugepte() is modified to use pte_present(), pte_read()
and pte_write() instead of the raw flags.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
__set_fixmap() uses __fix_to_virt() then does the boundary checks
by it self. Instead, we can use fix_to_virt() which does the
verification at build time. For this, we need to use it inline
so that GCC can see the real value of idx at buildtime.
In the meantime, we remove the 'fixmaps' variable.
This variable is set but has never been used from the beginning
(commit 2c419bdeca ("[POWERPC] Port fixmap from x86 and use
for kmap_atomic"))
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
get_pteptr() and __mapin_ram_chunk() are only used locally,
so define them static
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This patch implements STRICT_KERNEL_RWX on PPC32.
As for CONFIG_DEBUG_PAGEALLOC, it deactivates BAT and LTLB mappings
in order to allow page protection setup at the level of each page.
As BAT/LTLB mappings are deactivated, there might be a performance
impact.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
As seen below, allthough the init sections have been freed, the
associated memory area is still marked as executable in the
page tables.
~ dmesg
[ 5.860093] Freeing unused kernel memory: 592K (c0570000 - c0604000)
~ cat /sys/kernel/debug/kernel_page_tables
---[ Start of kernel VM ]---
0xc0000000-0xc0497fff 4704K rw X present dirty accessed shared
0xc0498000-0xc056ffff 864K rw present dirty accessed shared
0xc0570000-0xc059ffff 192K rw X present dirty accessed shared
0xc05a0000-0xc7ffffff 125312K rw present dirty accessed shared
---[ vmalloc() Area ]---
This patch fixes that.
The implementation is done by reusing the change_page_attr()
function implemented for CONFIG_DEBUG_PAGEALLOC
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
__change_page_attr() uses flush_tlb_page().
flush_tlb_page() uses tlbie instruction, which also invalidates
pinned TLBs, which is not what we expect.
This patch modifies the implementation to use flush_tlb_kernel_range()
instead. This will make use of tlbia which will preserve pinned TLBs.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This reduces the DTLB miss handler hot path (user address path)
by one instruction by preserving r10.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
setup_initial_memory_limit() is only called during init.
mmu_patch_cmp_limit() is only called from 8xx_mmu.c
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Pinning TLBs bypasses STRICT_KERNEL_RWX or DEBUG_PAGEALLOC protections
so it should only be allowed when those are not selected
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
As stated in a comment in head_8xx.S, today we "Always pin the first
8 MB ITLB to prevent ITLB misses while mucking around with SRR0/SRR1
in asm".
This issue has just been cleared by the preceding patch, therefore
we can make this pinning optional (on by default) and independent
of DATA pinning.
This patch also makes pinning of IMMR independent of pinning of DATA.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
By default, the 8xx pins an ITLB on the first 8M of memory in order
to avoid any ITLB miss on kernel code.
However, with some debug functions like DEBUG_PAGEALLOC and
DEBUG_RODATA, pinning TLBs is contradictory.
In order to avoid any ITLB miss in a critical section without pinning
TLBs, we have to ensure that there is no page boundary crossed between
the setup of a new value in SRR0/SRR1 and the associated RFI.
The functions modifying srr0/srr1 are all located in setup_32.S.
They are spread over almost 4kbytes.
The patch forces a 12 bits (4kbytes) alignment for those
functions. This garanties that the functions remain in a
single 4k page.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The macro to check if an address is a kernel address or not is
not used anymore in DTLBmiss handler. It is used in ITLB miss handler
and in DTLB error handler. DTLB error handler is not a hot path, it
doesn't need such optimisation.
In order to simplify a following patch which will rework ITLB miss
handler, we remove the macros and reintroduce them inside the handler.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Naveen N. Rao