- Move the slow path fallbacks to their own assembly files
This makes them much easier to read and is needed for the next change.
- Add CFI annotations for unwinding (XXX need review)
- Remove constant case which can never happen with out of line spinlocks
- Use patchable LOCK prefixes
- Don't use lock sections anymore for inline code because they can't
be expressed by the unwinder (this adds one taken jump to the lock
fast path)
Cc: jbeulich@novell.com
Signed-off-by: Andi Kleen <ak@suse.de>
Use knowledge about EFLAGS layout (bits 22:63 are always 0) to distingush
EFLAGS word and kernel address in the spin lock stack frame.
Signed-off-by: Andi Kleen <ak@suse.de>
A few trivial spelling and grammar mistakes picked up in
"arch/x86_64/aperture.c", "arch/x86_64/crash.c" and
"arch/x86_64/apic.c". I think all are correct fixes but am ever aware
of my fallibility :o) This is my first patch submission so all
feedback is appreciated, esp. WRT CCing to Linus, Andi and
trivial@kernel.org, is this correct? And which is the most appropriate
kernel version to diff against? If any.
Should apply cleanly to 2.6.18-rc1
Signed-off-by: Adam Henley <adamazing@gmail.com>
Signed-off-by: Andi Kleen <ak@suse.de>
- adam
Hello,
Following my discussion with Andi. Here is a patch that introduces
two new TIF flags to simplify the context switch code in __switch_to().
The idea is to minimize the number of cache lines accessed in the common
case, i.e., when neither the debug registers nor the I/O bitmap are used.
This patch covers the x86-64 modifications. A patch for i386 follows.
Changelog:
- add TIF_DEBUG to track when debug registers are active
- add TIF_IO_BITMAP to track when I/O bitmap is used
- modify __switch_to() to use the new TIF flags
<signed-off-by>: eranian@hpl.hp.com
Signed-off-by: Andi Kleen <ak@suse.de>
For NUMA optimization and some other algorithms it is useful to have a fast
to get the current CPU and node numbers in user space.
x86-64 added a fast way to do this in a vsyscall. This adds a generic
syscall for other architectures to make it a generic portable facility.
I expect some of them will also implement it as a faster vsyscall.
The cache is an optimization for the x86-64 vsyscall optimization. Since
what the syscall returns is an approximation anyways and user space
often wants very fast results it can be cached for some time. The norma
methods to get this information in user space are relatively slow
The vsyscall is in a better position to manage the cache because it has direct
access to a fast time stamp (jiffies). For the generic syscall optimization
it doesn't help much, but enforce a valid argument to keep programs
portable
I only added an i386 syscall entry for now. Other architectures can follow
as needed.
AK: Also added some cleanups from Andrew Morton
Signed-off-by: Andi Kleen <ak@suse.de>
This patch adds a vgetcpu vsyscall, which depending on the CPU RDTSCP
capability uses either the RDTSCP or CPUID to obtain a CPU and node
numbers and pass them to the program.
AK: Lots of changes over Vojtech's original code:
Better prototype for vgetcpu()
It's better to pass the cpu / node numbers as separate arguments
to avoid mistakes when going from SMP to NUMA.
Also add a fast time stamp based cache using a user supplied
argument to speed things more up.
Use fast method from Chuck Ebbert to retrieve node/cpu from
GDT limit instead of CPUID
Made sure RDTSCP init is always executed after node is known.
Drop printk
Signed-off-by: Vojtech Pavlik <vojtech@suse.cz>
Signed-off-by: Andi Kleen <ak@suse.de>
This patch adds initalization of the RDTSCP auxilliary values to CPU numbers
to time.c. If RDTSCP is available, the MSRs are written with the respective
values. It can be later used to initalize per-cpu timekeeping variables.
AK: Some cleanups. Move externs into headers and fix CPU hotplug.
Signed-off-by: Vojtech Pavlik <vojtech@suse.cz>
Signed-off-by: Andi Kleen <ak@suse.de>
AK: This redoes the changes I temporarily reverted.
Intel now has support for Architectural Performance Monitoring Counters
( Refer to IA-32 Intel Architecture Software Developer's Manual
http://www.intel.com/design/pentium4/manuals/253669.htm ). This
feature is present starting from Intel Core Duo and Intel Core Solo processors.
What this means is, the performance monitoring counters and some performance
monitoring events are now defined in an architectural way (using cpuid).
And there will be no need to check for family/model etc for these architectural
events.
Below is the patch to use this performance counters in nmi watchdog driver.
Patch handles both i386 and x86-64 kernels.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Andi Kleen <ak@suse.de>
After a crash we should wait for NMI IPI event and not for external NMI or
NMI watchdog tick.
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
When a unknown NMI happened the panic would claim a NMI watchdog timeout.
Also it would check the variable set by nmi_watchdog=panic and panic then.
Fix up the panic message to be generic
Unconditionally panic on unknown NMI when panic on unknown nmi is enabled.
Noticed by Jan Beulich
Cc: jbeulich@novell.com
Signed-off-by: Andi Kleen <ak@suse.de>
Making NMI suspend/resume work with SMP. We use CPU hotplug to offline
APs in SMP suspend/resume. Only BSP executes sysdev's .suspend/.resume
method. APs should follow CPU hotplug code path.
And:
+From: Don Zickus <dzickus@redhat.com>
Makes the start/stop paths of nmi watchdog more robust to handle the
suspend/resume cases more gracefully.
AK: I merged the two patches together
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Clean up some of the output messages on the nmi error paths to make more
sense when they are displayed. This is mainly a cosmetic fix and
shouldn't impact any normal code path.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
To quote Alan Cox:
The default Linux behaviour on an NMI of either memory or unknown is to
continue operation. For many environments such as scientific computing
it is preferable that the box is taken out and the error dealt with than
an uncorrected parity/ECC error get propogated.
A small number of systems do generate NMI's for bizarre random reasons
such as power management so the default is unchanged. In other respects
the new proc/sys entry works like the existing panic controls already in
that directory.
This is separate to the edac support - EDAC allows supported chipsets to
handle ECC errors well, this change allows unsupported cases to at least
panic rather than cause problems further down the line.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Adds a new /proc/sys/kernel/nmi_watchdog call that will enable/disable the
nmi watchdog.
By entering a non-zero value here, a user can enable the nmi watchdog to
monitor the online cpus in the system. By entering a zero value here, a
user can disable the nmi watchdog and free up a performance counter which
could then be utilized by the oprofile subsystem, otherwise oprofile may be
short a counter when in use.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Adds a new /proc/sys/kernel/nmi call that will enable/disable the nmi
watchdog.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Removes the un/set_nmi_callback and reserve/release_lapic_nmi functions as
they are no longer needed. The various subsystems are modified to register
with the die_notifier instead.
Also includes compile fixes by Andrew Morton.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
We need TIF_RESTORE_SIGMASK in order to support ppoll() and pselect()
system calls. This patch originally came from Andi, and was based
heavily on David Howells' implementation of same on i386. I fixed a typo
which was causing do_signal() to use the wrong signal mask.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Andi Kleen <ak@suse.de>
This patch cleans up the NMI interrupt path. Instead of being gated by if
the 'nmi callback' is set, the interrupt handler now calls everyone who is
registered on the die_chain and additionally checks the nmi watchdog,
reseting it if enabled. This allows more subsystems to hook into the NMI if
they need to (without being block by set_nmi_callback).
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
This patch includes the changes to make the nmi watchdog on x86_64 SMP
aware. A bunch of code was moved around to make it simpler to read. In
addition, it is now possible to determine if a particular NMI was the result
of the watchdog or not. This feature allows the kernel to filter out
unknown NMIs easier.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Adds basic infrastructure to allow subsystems to reserve performance
counters on the x86 chips. Only UP kernels are supported in this patch to
make reviewing easier. The SMP portion makes a lot more changes.
Think of this as a locking mechanism where each bit represents a different
counter. In addition, each subsystem should also reserve an appropriate
event selection register that will correspond to the performance counter it
will be using (this is mainly neccessary for the Pentium 4 chips as they
break the 1:1 relationship to performance counters).
This will help prevent subsystems like oprofile from interfering with the
nmi watchdog.
Signed-off-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Now that the tfm is passed directly to setkey instead of the ctx, we no
longer need to pass the &tfm->crt_flags pointer.
This patch also gets rid of a few unnecessary checks on the key length
for ciphers as the cipher layer guarantees that the key length is within
the bounds specified by the algorithm.
Rather than testing dia_setkey every time, this patch does it only once
during crypto_alloc_tfm. The redundant check from crypto_digest_setkey
is also removed.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This reverts commits 11012d419c and
40dd2d20f2, which allowed us to use the
MMIO accesses for PCI config cycles even without the area being marked
reserved in the e820 memory tables.
Those changes were needed for EFI-environment Intel macs, but broke some
newer Intel 965 boards, so for now it's better to revert to our old
2.6.17 behaviour and at least avoid introducing any new breakage.
Andi Kleen has a set of patches that work with both EFI and the broken
Intel 965 boards, which will be applied once they get wider testing.
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Edgar Hucek <hostmaster@ed-soft.at>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Take default arch/*/kernel/audit.c to lib/, have those with special
needs (== biarch) define AUDIT_ARCH in their Kconfig.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
It's possible to get an invalid page fault in kernel mode when we try to
write out segments from vsyscall32 when dumping core for a 32bit process if
the vsyscall32 DSO is not mapped in its address space (which can happen if,
for example, ulimit -v 100 is run).
Signed-off-by: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The values in init_tss.ist[] can change when an IST event occurs. Save
the original IST values for checking stack addresses when debugging or
doing stack traces.
Signed-off-by: Keith Owens <kaos@ocs.com.au>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
As a replacement for the earlier removal of the e820 MCFG check
we blacklist the Intel SDV with the original BIOS bug that
motivated that check. On those machines don't use MMCONFIG.
This also adds a new pci=mmconf parameter to override the blacklist.
Cc: Greg KH <gregkh@suse.de>
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Noticed by Jan Beulich.
When the kernel was moved from 1MB to 2MB in 2.6.17 the kernel reservation
code wasn't adjusted and it still reserved starting with 1MB. This means 1MB always
were lost.
This patch fixes this by reserving only starting with _text.
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The unwinder fallback logic still had potential for falling through to
the legacy stack trace code without printing an indication (at once
serving as a separator) of this.
Further, the stack pointer retrieval for the fallback should be as
restrictive as possible (in order to avoid having the legacy stack
tracer try to access invalid memory). The patch tightens that, but
this could certainly be further improved.
Also making the call_trace command line option now conditional upon
CONFIG_STACK_UNWIND (as it's meaningless otherwise).
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
One open question: Should these added pushes perhaps be made
conditional upon CONFIG_STACK_UNWIND or CONFIG_UNWIND_INFO?
[AK: Not needed -- these are all very slow paths]
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The check for the MCFG table being reserved in the e820 map was originally
added to detect a broken BIOS in a preproduction Intel SDV. However it also
breaks the Apple x86 Macs, which can't supply this properly, but need
a working MCFG. With this patch they wouldn't use the MCFG and not work.
After some discussion I think it's best to remove the heuristic again.
It also failed on some other boxes (although it didn't cause much
problems there because old style port access for PCI config space
still works as fallback), but the preproduction SDVs can just use
pci=nommcfg. Supporting production machines properly is more
important.
Edgar Hucek did all the debugging work.
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Edgar Hucek <hostmaster@ed-soft.at>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Previously the message was "Fatal exception: panic_on_oops", as introduced
in a recent patch whith removed a somewhat dangerous call to ssleep() in
the panic_on_oops path. However, Paul Mackerras suggested that this was
somewhat confusing, leadind people to believe that it was panic_on_oops
that was the root cause of the fatal exception. On his suggestion, this
patch changes the message to simply "Fatal exception". A suitable oops
message should already have been displayed.
Signed-off-by: Simon Horman <horms@verge.net.au>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If CONFIG_IOMMU_DEBUG is set force_iommu defaults to 1. In the case
where no HW IOMMU is present in the machine and we end up using nommu,
leaving force_iommu set to 1 causes dma_alloc_coherent to do the wrong
thing. Therefore, if we end up using nommu, make sure force_iommu is
0.
Signed-off-by: Muli Ben-Yehuda <muli@il.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Re-add backlink for old style unwinder to stack switching. Add proper
stack frame and CFI annotations to call_softirq
This prevents a oops when backtracing with fallback through the
interrupt stack top.
Suggested by Jan Beulich and Herbert Xu wanted it in 2.6.18.
Cc: jbeulich@novell.com
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The latest toolchains can produce a new ELF section in DSOs and
dynamically-linked executables. The new section ".gnu.hash" replaces
".hash", and allows for more efficient runtime symbol lookups by the
dynamic linker. The new ld option --hash-style={sysv|gnu|both} controls
whether to produce the old ".hash", the new ".gnu.hash", or both. In some
new systems such as Fedora Core 6, gcc by default passes --hash-style=gnu
to the linker, so that a standard invocation of "gcc -shared" results in
producing a DSO with only ".gnu.hash". The new ".gnu.hash" sections need
to be dealt with the same way as ".hash" sections in all respects; only the
dynamic linker cares about their contents. To work with older dynamic
linkers (i.e. preexisting releases of glibc), a binary must have the old
".hash" section. The --hash-style=both option produces binaries that a new
dynamic linker can use more efficiently, but an old dynamic linker can
still handle.
The new section runs afoul of the custom linker scripts used to build vDSO
images for the kernel. On ia64, the failure mode for this is a boot-time
panic because the vDSO's PT_IA_64_UNWIND segment winds up ill-formed.
This patch addresses the problem in two ways.
First, it mentions ".gnu.hash" in all the linker scripts alongside ".hash".
This produces correct vDSO images with --hash-style=sysv (or old tools),
with --hash-style=gnu, or with --hash-style=both.
Second, it passes the --hash-style=sysv option when building the vDSO
images, so that ".gnu.hash" is not actually produced. This is the most
conservative choice for compatibility with any old userland. There is some
concern that some ancient glibc builds (though not any known old production
system) might choke on --hash-style=both binaries. The optimizations
provided by the new style of hash section do not really matter for a DSO
with a tiny number of symbols, as the vDSO has. If someone wants to use
=gnu or =both for their vDSO builds and worry less about that
compatibility, just change the option and the linker script changes will
make any choice work fine.
Signed-off-by: Roland McGrath <roland@redhat.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Use hotplug version of register_cpu_notifier in late init functions.
Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch is part of an effort to unify the panic_on_oops behaviour across
all architectures that implement it.
It was pointed out to me by Andi Kleen that if an oops has occured in
interrupt context, then calling sleep() in the oops path will only cause a
panic, and that it would be really better for it not to be in the path at
all.
This patch removes the ssleep() call and reworks the console message
accordinly. I have a slght concern that the resulting console message is
too long, feedback welcome.
For powerpc it also unifies the 32bit and 64bit behaviour.
Fror x86_64, this patch only updates the console message, as ssleep() is
already not present.
Signed-off-by: Horms <horms@verge.net.au>
Acked-by: Paul Mackerras <paulus@samba.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
One of my original comments in machine_kexec was unclear
and this should fix it.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Andi Kleen <ak@muc.de>
Acked-by: Horms <horms@verge.net.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
It was broken before. But having it is important as possible hardware
bug workaround.
And previously there was no way to force swiotlb if there is another IOMMU.
Side effect is that iommu=force won't force swiotlb anymore even if there
isn't another IOMMU.
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
As Travis Betak points out it accesses the wrong northbridge subfunction
now. Switch back to the old code.
Cc: "Travis Betak" <betak@mpdtxmail.amd.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Calgary hits a NULL pointer dereference when booting in a multi-chassis
NUMA system. See Redhat bugzilla number 198498, found by Konrad
Rzeszutek (konradr@redhat.com).
There are many issues that had to be resolved to fix this problem.
Firstly when I originally wrote the code to handle NUMA systems, I
had a large misunderstanding that was not corrected until now. That was
that I thought the "number of nodes online" referred to number of
physical systems connected. So that if NUMA was disabled, there
would only be 1 node and it would only show that node's PCI bus.
In reality if NUMA is disabled, the system displays all of the
connected chassis as one node but is only ignorant of the delays
in accessing main memory. Therefore, references to num_online_nodes()
and MAX_NUMNODES are incorrect and need to be set to the maximum
number of nodes that can be accessed (which are 8). I created a
variable, MAX_NUM_CHASSIS, and set it to 8 to fix this.
Secondly, when walking the PCI in detect_calgary, the code only
checked the first "slot" when looking to see if a device is present.
This will work for most cases, but unfortunately it isn't always the
case. In the NUMA MXE drawers, there are USB devices present on the
3rd slot (with slot 1 being empty). So, to work around this, all
slots (up to 8) are scanned to see if there are any devices present.
Lastly, the bus is being enumerated on large systems in a different
way the we originally thought. This throws the ugly logic we had
out the window. To more elegantly handle this, I reorganized the
kva array to be sparse (which removed the need to have any bus number
to kva slot logic in tce.c) and created a secondary space array to
contain the bus number to phb mapping.
With these changes Calgary boots on an x460 with 4 nodes with and
without NUMA enabled.
Signed-off-by: Jon Mason <jdmason@us.ibm.com>
Signed-off-by: Muli Ben-Yehuda <muli@il.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
