In commit b2d4383480 ("sparc64: Make
PAGE_OFFSET variable."), the MAX_PHYS_ADDRESS_BITS value was increased
(to 47).
This constant reference to '41UL' was missed.
Signed-off-by: David S. Miller <davem@davemloft.net>
Make get_user_insn() able to cope with huge PMDs.
Next, make do_fault_siginfo() more robust when get_user_insn() can't
actually fetch the instruction. In particular, use the MMU announced
fault address when that happens, instead of calling
compute_effective_address() and computing garbage.
Signed-off-by: David S. Miller <davem@davemloft.net>
If we have a 32-bit task we must chop off the top 32-bits of the
64-bit value just as the cpu would.
Signed-off-by: David S. Miller <davem@davemloft.net>
The large PMD path needs to check _PAGE_VALID not _PAGE_PRESENT, to
decide if it needs to bail and return 0.
pmd_large() should therefore just check _PAGE_PMD_HUGE.
Calls to gup_huge_pmd() are guarded with a check of pmd_large(), so we
just need to add a valid bit check.
Signed-off-by: David S. Miller <davem@davemloft.net>
On sparc64 "present" and "valid" are seperate PTE bits, this allows us to
naturally distinguish between the user explicitly asking for PROT_NONE
with mprotect() and other situations.
However we weren't handling this properly in the huge PMD paths.
First of all, the page table walker in the TSB miss path only checks
for _PAGE_PMD_HUGE. So the generic pmdp_invalidate() would clear
_PAGE_PRESENT but the TLB miss paths would still load it into the TLB
as a valid huge PMD.
Fix this by clearing the valid bit in pmdp_invalidate(), and also
checking the valid bit in USER_PGTABLE_CHECK_PMD_HUGE using "brgez"
since _PAGE_VALID is bit 63 in both the sun4u and sun4v pte layouts.
Signed-off-by: David S. Miller <davem@davemloft.net>
This code was mistakenly using the exec bit from the PMD in all
cases, even when the PMD isn't a huge PMD.
If it's not a huge PMD, test the exec bit in the individual ptes down
in tlb_batch_pmd_scan().
Signed-off-by: David S. Miller <davem@davemloft.net>
Bring this code in line with the perf based generic NMI watchdog
in kernel/watchdog.c (which we should convert over to at some
point).
In particular, don't do anything super fancy when the watchdog
triggers, and specifically don't do a do_exit() which only makes
things worse.
Either panic(), or WARN(). The latter of which will do all of
the actions such as give us a stack backtrace.
Signed-off-by: David S. Miller <davem@davemloft.net>
One more place where we must not be able
to be preempted or to be interrupted in RT.
Always actually disable interrupts during
synchronization cycle.
Signed-off-by: Kirill Tkhai <tkhai@yandex.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull parisc fixes from Helge Deller:
"Drop the architecture-specifc value for_STK_LIM_MAX to fix stack
related problems with GNU make"
* 'parisc-3.15-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
parisc: Use generic uapi/asm/resource.h file
parisc: remove _STK_LIM_MAX override
There are only a couple of architectures that override _STK_LIM_MAX to
a non-infinity value. This changes the stack allocation semantics in
subtle ways. For example, GNU make changes its stack allocation to the
hard maximum defined by _STK_LIM_MAX. As a results, threads executed
by processes running under make are allocated a stack size of
_STK_LIM_MAX rather than a sensible default value. This causes various
thread stress tests to fail when they can't muster more than about 50
threads.
The attached change implements the default behavior used by the
majority of architectures.
Signed-off-by: John David Anglin <dave.anglin@bell.net>
Reviewed-by: Carlos O'Donell <carlos@systemhalted.org>
Cc: stable@vger.kernel.org # 3.14
Signed-off-by: Helge Deller <deller@gmx.de>
Commit 93ea02bb84 ("arch: Clean up asm/barrier.h implementations")
wired generic barrier.h for hexagon, but failed to delete the existing
file.
Cc: Richard Kuo <rkuo@codeaurora.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Compile-tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull perf fixes from Ingo Molnar:
"Mostly tooling fixes, plus an Intel RAPL PMU driver fix"
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf tests x86: Fix stack map lookup in dwarf unwind test
perf x86: Fix perf to use non-executable stack, again
perf tools: Remove extra '/' character in events file path
perf machine: Search for modules in %s/lib/modules/%s
perf tests: Add static build make test
perf tools: Fix bfd dependency libraries detection
perf tools: Use LDFLAGS instead of ALL_LDFLAGS
perf/x86: Fix RAPL rdmsrl_safe() usage
tools lib traceevent: Fix memory leak in pretty_print()
tools lib traceevent: Fix backward compatibility macros for pevent filter enums
perf tools: Disable libdw unwind for all but x86 arch
perf tests x86: Fix memory leak in sample_ustack()
There was a very small race window where resume to kernel mode from a
Exception Path (or pure kernel mode which is true for most of ARC
exceptions anyways), was not disabling interrupts in restore_regs,
clobbering the exception regs
Anton found the culprit call flow (after many sleepless nights)
| 1. we got a Trap from user land
| 2. started to service it.
| 3. While doing some stuff on user-land memory (I think it is padzero()),
| we got a DataTlbMiss
| 4. On return from it we are taking "resume_kernel_mode" path
| 5. NEED_RESHED is not set, so we go to "return from exception" path in
| restore regs.
| 6. there seems to be IRQ happening
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: <stable@vger.kernel.org> #3.10, 3.12, 3.13, 3.14
Cc: Anton Kolesov <Anton.Kolesov@synopsys.com>
Cc: Francois Bedard <Francois.Bedard@synopsys.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This branch contains a pair of important bug fixes for the DT code:
- Fix some incorrect binding property names before they enter common usage
- Fix bug where some platform devices will be unable to get their
interrupt number when they depend on an interrupt controller that is
not available at device creation time. This is a problem causing
mainline to fail on a number of ARM platforms.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=B2JT
-----END PGP SIGNATURE-----
Merge tag 'dt-for-linus' of git://git.secretlab.ca/git/linux
Pull devicetree bug fixes from Grant Likely:
"These are some important bug fixes that need to get into v3.15.
This branch contains a pair of important bug fixes for the DT code:
- Fix some incorrect binding property names before they enter common
usage
- Fix bug where some platform devices will be unable to get their
interrupt number when they depend on an interrupt controller that
is not available at device creation time. This is a problem
causing mainline to fail on a number of ARM platforms"
* tag 'dt-for-linus' of git://git.secretlab.ca/git/linux:
of/irq: do irq resolution in platform_get_irq
of: selftest: add deferred probe interrupt test
dt: Fix binding typos in clock-names and interrupt-names
Pull powerpc fixes from Ben Herrenschmidt:
"Here is a bunch of post-merge window fixes that have been accumulating
in patchwork while I was on vacation or buried under other stuff last
week.
We have the now usual batch of LE fixes from Anton (sadly some new
stuff that went into this merge window had endian issues, we'll try to
make sure we do better next time)
Some fixes and cleanups to the new 24x7 performance monitoring stuff
(mostly typos and cleaning up printk's)
A series of fixes for an issue with our runlatch bit, which wasn't set
properly for offlined threads/cores and under KVM, causing potentially
some counters to misbehave along with possible power management
issues.
A fix for kexec nasty race where the new kernel wouldn't "see" the
secondary processors having reached back into firmware in time.
And finally a few other misc (and pretty simple) bug fixes"
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (33 commits)
powerpc/4xx: Fix section mismatch in ppc4xx_pci.c
ppc/kvm: Clear the runlatch bit of a vcpu before napping
ppc/kvm: Set the runlatch bit of a CPU just before starting guest
ppc/powernv: Set the runlatch bits correctly for offline cpus
powerpc/pseries: Protect remove_memory() with device hotplug lock
powerpc: Fix error return in rtas_flash module init
powerpc: Bump BOOT_COMMAND_LINE_SIZE to 2048
powerpc: Bump COMMAND_LINE_SIZE to 2048
powerpc: Rename duplicate COMMAND_LINE_SIZE define
powerpc/perf/hv-24x7: Catalog version number is be64, not be32
powerpc/perf/hv-24x7: Remove [static 4096], sparse chokes on it
powerpc/perf/hv-24x7: Use (unsigned long) not (u32) values when calling plpar_hcall_norets()
powerpc/perf/hv-gpci: Make device attr static
powerpc/perf/hv_gpci: Probe failures use pr_debug(), and padding reduced
powerpc/perf/hv_24x7: Probe errors changed to pr_debug(), padding fixed
powerpc/mm: Fix tlbie to add AVAL fields for 64K pages
powerpc/powernv: Fix little endian issues in OPAL dump code
powerpc/powernv: Create OPAL sglist helper functions and fix endian issues
powerpc/powernv: Fix little endian issues in OPAL error log code
powerpc/powernv: Fix little endian issues with opal_do_notifier calls
...
This patch fixes this section mismatch:
WARNING: vmlinux.o(.text+0x1efc4): Section mismatch in reference from
the function apm821xx_pciex_init_port_hw() to the function
.init.text:ppc4xx_pciex_wait_on_sdr.isra.9()
The function apm821xx_pciex_init_port_hw() references the function
__init ppc4xx_pciex_wait_on_sdr.isra.9(). This is often because
apm821xx_pciex_init_port_hw lacks a __init annotation or the
annotation of ppc4xx_pciex_wait_on_sdr.isra.9 is wrong.
apm821xx_pciex_init_port_hw is only referenced by a struct in
__initdata, so it should be safe to add __init to
apm821xx_pciex_init_port_hw.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When the guest cedes the vcpu or the vcpu has no guest to
run it naps. Clear the runlatch bit of the vcpu before
napping to indicate an idle cpu.
Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The secondary threads in the core are kept offline before launching guests
in kvm on powerpc: "371fefd6f2dc4666:KVM: PPC: Allow book3s_hv guests to use
SMT processor modes."
Hence their runlatch bits are cleared. When the secondary threads are called
in to start a guest, their runlatch bits need to be set to indicate that they
are busy. The primary thread has its runlatch bit set though, but there is no
harm in setting this bit once again. Hence set the runlatch bit for all
threads before they start guest.
Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Up until now we have been setting the runlatch bits for a busy CPU and
clearing it when a CPU enters idle state. The runlatch bit has thus
been consistent with the utilization of a CPU as long as the CPU is online.
However when a CPU is hotplugged out the runlatch bit is not cleared. It
needs to be cleared to indicate an unused CPU. Hence this patch has the
runlatch bit cleared for an offline CPU just before entering an idle state
and sets it immediately after it exits the idle state.
Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
While testing memory hot-remove, I found following dead lock:
Process #1141 is drmgr, trying to remove some memory, i.e. memory499.
It holds the memory_hotplug_mutex, and blocks when trying to remove file
"online" under dir memory499, in kernfs_drain(), at
wait_event(root->deactivate_waitq,
atomic_read(&kn->active) == KN_DEACTIVATED_BIAS);
Process #1120 is trying to online memory499 by
echo 1 > memory499/online
In .kernfs_fop_write, it uses kernfs_get_active() to increase
&kn->active, thus blocking process #1141. While itself is blocked later
when trying to acquire memory_hotplug_mutex, which is held by process
The backtrace of both processes are shown below:
[<c000000001b18600>] 0xc000000001b18600
[<c000000000015044>] .__switch_to+0x144/0x200
[<c000000000263ca4>] .online_pages+0x74/0x7b0
[<c00000000055b40c>] .memory_subsys_online+0x9c/0x150
[<c00000000053cbe8>] .device_online+0xb8/0x120
[<c00000000053cd04>] .online_store+0xb4/0xc0
[<c000000000538ce4>] .dev_attr_store+0x64/0xa0
[<c00000000030f4ec>] .sysfs_kf_write+0x7c/0xb0
[<c00000000030e574>] .kernfs_fop_write+0x154/0x1e0
[<c000000000268450>] .vfs_write+0xe0/0x260
[<c000000000269144>] .SyS_write+0x64/0x110
[<c000000000009ffc>] syscall_exit+0x0/0x7c
[<c000000001b18600>] 0xc000000001b18600
[<c000000000015044>] .__switch_to+0x144/0x200
[<c00000000030be14>] .__kernfs_remove+0x204/0x300
[<c00000000030d428>] .kernfs_remove_by_name_ns+0x68/0xf0
[<c00000000030fb38>] .sysfs_remove_file_ns+0x38/0x60
[<c000000000539354>] .device_remove_attrs+0x54/0xc0
[<c000000000539fd8>] .device_del+0x158/0x250
[<c00000000053a104>] .device_unregister+0x34/0xa0
[<c00000000055bc14>] .unregister_memory_section+0x164/0x170
[<c00000000024ee18>] .__remove_pages+0x108/0x4c0
[<c00000000004b590>] .arch_remove_memory+0x60/0xc0
[<c00000000026446c>] .remove_memory+0x8c/0xe0
[<c00000000007f9f4>] .pseries_remove_memblock+0xd4/0x160
[<c00000000007fcfc>] .pseries_memory_notifier+0x27c/0x290
[<c0000000008ae6cc>] .notifier_call_chain+0x8c/0x100
[<c0000000000d858c>] .__blocking_notifier_call_chain+0x6c/0xe0
[<c00000000071ddec>] .of_property_notify+0x7c/0xc0
[<c00000000071ed3c>] .of_update_property+0x3c/0x1b0
[<c0000000000756cc>] .ofdt_write+0x3dc/0x740
[<c0000000002f60fc>] .proc_reg_write+0xac/0x110
[<c000000000268450>] .vfs_write+0xe0/0x260
[<c000000000269144>] .SyS_write+0x64/0x110
[<c000000000009ffc>] syscall_exit+0x0/0x7c
This patch uses lock_device_hotplug() to protect remove_memory() called
in pseries_remove_memblock(), which is also stated before function
remove_memory():
* NOTE: The caller must call lock_device_hotplug() to serialize hotplug
* and online/offline operations before this call, as required by
* try_offline_node().
*/
void __ref remove_memory(int nid, u64 start, u64 size)
With this lock held, the other process(#1120 above) trying to online the
memory block will retry the system call when calling
lock_device_hotplug_sysfs(), and finally find No such device error.
Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
module_init should return 0 or a negative errno.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Bump the boot wrapper BOOT_COMMAND_LINE_SIZE to match the
kernel.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
I've had a report that the current limit is too small for
an automated network based installer. Bump it.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We have two definitions of COMMAND_LINE_SIZE, one for the kernel
and one for the boot wrapper. I assume this is so the boot
wrapper can be self sufficient and not rely on kernel headers.
Having two defines with the same name is confusing, I just
updated the wrong one when trying to bump it.
Make the boot wrapper define unique by calling it
BOOT_COMMAND_LINE_SIZE.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The catalog version number was changed from a be32 (with proceeding
32bits of padding) to a be64, update the code to treat it as a be64
Signed-off-by: Cody P Schafer <cody@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
fixup for "powerpc/perf: Add support for the hv gpci (get performance
counter info) interface".
Makes the "not enabled" message less awful (and hidden unless
debugging).
Signed-off-by: Cody P Schafer <cody@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
fixup for "powerpc/perf: Add support for the hv 24x7 interface"
Makes the "not enabled" message less awful (and hides it in most cases).
Signed-off-by: Cody P Schafer <cody@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The if condition check was based on a draft ISA doc. Remove the same.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We have two copies of code that creates an OPAL sg list. Consolidate
these into a common set of helpers and fix the endian issues.
The flash interface embedded a version number in the num_entries
field, whereas the dump interface did did not. Since versioning
wasn't added to the flash interface and it is impossible to add
this in a backwards compatible way, just remove it.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Fix little endian issues with the OPAL error log code.
Signed-off-by: Anton Blanchard <anton@samba.org>
Reviewed-by: Stewart Smith <stewart@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The bitmap in opal_poll_events and opal_handle_interrupt is
big endian, so we need to byteswap it on little endian builds.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We had some duplication of the internal OPAL functions.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Using size_t in our APIs is asking for trouble, especially
when some OPAL calls use size_t pointers.
Signed-off-by: Anton Blanchard <anton@samba.org>
Reviewed-by: Stewart Smith <stewart@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
On PowerNV platform, we are holding an unnecessary refcount on a pci_dev, which
leads to the pci_dev is not destroyed when hotplugging a pci device.
This patch release the unnecessary refcount.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
With this patch I was able to update firmware on an LE kernel.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We have a subtle race when sending CPUs back to OPAL on kexec.
We mark them as "in real mode" right before we send them down. Once
we've booted the new kernel, it might try to call opal_reinit_cpus()
to change endianness, and that requires all CPUs to be spinning inside
OPAL.
However there is no synchronization here and we've observed cases
where the returning CPUs hadn't established their new state inside
OPAL before opal_reinit_cpus() is called, causing it to fail.
The proper fix is to actually wait for them to go down all the way
from the kexec'ing kernel.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The size of the sysparam sysfs files is determined from the device tree
at boot. However the buffer is hard coded to 64 bytes. If we encounter a
parameter that is larger than 64, or miss-parse the device tree, the
buffer will overflow when reading or writing to the parameter.
Check it at discovery time, and if the parameter is too large, do not
create a sysfs entry for it.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The sysparam code currently uses the userspace supplied number of
bytes when memcpy()ing in to a local 64-byte buffer.
Limit the maximum number of bytes by the size of the buffer.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The OPAL calls are returning int64_t values, which the sysparam code
stores in an int, and the sysfs callback returns ssize_t. Make code a
easier to read by consistently using ssize_t.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When a sysparam query in OPAL returned a negative value (error code),
sysfs would spew out a decent chunk of memory; almost 64K more than
expected. This was traced to a sign/unsigned mix up in the OPAL sysparam
sysfs code at sys_param_show.
The return value of sys_param_show is a ssize_t, calculated using
return ret ? ret : attr->param_size;
Alan Modra explains:
"attr->param_size" is an unsigned int, "ret" an int, so the overall
expression has type unsigned int. Result is that ret is cast to
unsigned int before being cast to ssize_t.
Instead of using the ternary operator, set ret to the param_size if an
error is not detected. The same bug exists in the sysfs write callback;
this patch fixes it in the same way.
A note on debugging this next time: on my system gcc will warn about
this if compiled with -Wsign-compare, which is not enabled by -Wall,
only -Wextra.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
