Valdis Kletnieks bisected a boot failure back to this recent commit:
360cb4d155 ("x86/mm/cpa: In populate_pgd(), don't set the PGD entry until it's populated")
I broke the case where a PUD table got allocated -- populate_pud()
would wander off a pgd_none entry and get lost. I'm not sure how
this survived my testing.
Fix the original issue in a much simpler way. The problem
was that, if we allocated a PUD table, failed to populate it, and
freed it, another CPU could potentially keep using the PGD entry we
installed (either by copying it via vmalloc_fault or by speculatively
caching it). There's a straightforward fix: simply leave the
top-level entry in place if this happens. This can't waste any
significant amount of memory -- there are at most 256 entries like
this systemwide and, as a practical matter, if we hit this failure
path repeatedly, we're likely to reuse the same page anyway.
For context, this is a reversion with this hunk added in:
if (ret < 0) {
+ /*
+ * Leave the PUD page in place in case some other CPU or thread
+ * already found it, but remove any useless entries we just
+ * added to it.
+ */
- unmap_pgd_range(cpa->pgd, addr,
+ unmap_pud_range(pgd_entry, addr,
addr + (cpa->numpages << PAGE_SHIFT));
return ret;
}
This effectively open-codes what the now-deleted unmap_pgd_range()
function used to do except that unmap_pgd_range() used to try to
free the page as well.
Reported-by: Valdis Kletnieks <Valdis.Kletnieks@vt.edu>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Luis R. Rodriguez <mcgrof@suse.com>
Cc: Mike Krinkin <krinkin.m.u@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Toshi Kani <toshi.kani@hp.com>
Link: http://lkml.kernel.org/r/21cbc2822aa18aa812c0215f4231dbf5f65afa7f.1469249789.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kvm_setup_default_irq_routing and kvm_setup_empty_irq_routing are
not used by generic code. So let's move the declarations in x86 irq.h
header instead of kvm_host.h.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Suggested-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
I don't think it is really possible to have a system where CPUID
enumerates support for XSAVE but that it does not have FP/SSE
(they are "legacy" features and always present).
But, I did manage to hit this case in qemu when I enabled its
somewhat shaky XSAVE support. The bummer is that the FPU is set
up before we parse the command-line or have *any* console support
including earlyprintk. That turned what should have been an easy
thing to debug in to a bit more of an odyssey.
So a BUG() here is worthless. All it does it guarantee that
if/when we hit this case we have an empty console. So, remove
the BUG() and try to limp along by disabling XSAVE and trying to
continue. Add a comment on why we are doing this, and also add
a common "out_disable" path for leaving fpu__init_system_xstate().
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave@sr71.net>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20160720194551.63BB2B58@viggo.jf.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Add support for Intel's AVX-512 instructions to the instruction decoder.
AVX-512 instructions are documented in Intel Architecture Instruction
Set Extensions Programming Reference (February 2016).
AVX-512 instructions are identified by a EVEX prefix which, for the
purpose of instruction decoding, can be treated as though it were a
4-byte VEX prefix.
Existing instructions which can now accept an EVEX prefix need not be
further annotated in the op code map (x86-opcode-map.txt). In the case
of new instructions, the op code map is updated accordingly.
Also add associated Mask Instructions that are used to manipulate mask
registers used in AVX-512 instructions.
The 'perf tools' instruction decoder is updated in a subsequent patch.
And a representative set of instructions is added to the perf tools new
instructions test in a subsequent patch.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: X86 ML <x86@kernel.org>
Link: http://lkml.kernel.org/r/1469003437-32706-3-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
So the reserve_ebda_region() code has accumulated a number of
problems over the years that make it really difficult to read
and understand:
- The calculation of 'lowmem' and 'ebda_addr' is an unnecessarily
interleaved mess of first lowmem, then ebda_addr, then lowmem tweaks...
- 'lowmem' here means 'super low mem' - i.e. 16-bit addressable memory. In other
parts of the x86 code 'lowmem' means 32-bit addressable memory... This makes it
super confusing to read.
- It does not help at all that we have various memory range markers, half of which
are 'start of range', half of which are 'end of range' - but this crucial
property is not obvious in the naming at all ... gave me a headache trying to
understand all this.
- Also, the 'ebda_addr' name sucks: it highlights that it's an address (which is
obvious, all values here are addresses!), while it does not highlight that it's
the _start_ of the EBDA region ...
- 'BIOS_LOWMEM_KILOBYTES' says a lot of things, except that this is the only value
that is a pointer to a value, not a memory range address!
- The function name itself is a misnomer: it says 'reserve_ebda_region()' while
its main purpose is to reserve all the firmware ROM typically between 640K and
1MB, while the 'EBDA' part is only a small part of that ...
- Likewise, the paravirt quirk flag name 'ebda_search' is misleading as well: this
too should be about whether to reserve firmware areas in the paravirt case.
- In fact thinking about this as 'end of RAM' is confusing: what this function
*really* wants to reserve is firmware data and code areas! Once the thinking is
inverted from a mixed 'ram' and 'reserved firmware area' notion to a pure
'reserved area' notion everything becomes a lot clearer.
To improve all this rewrite the whole code (without changing the logic):
- Firstly invert the naming from 'lowmem end' to 'BIOS reserved area start'
and propagate this concept through all the variable names and constants.
BIOS_RAM_SIZE_KB_PTR // was: BIOS_LOWMEM_KILOBYTES
BIOS_START_MIN // was: INSANE_CUTOFF
ebda_start // was: ebda_addr
bios_start // was: lowmem
BIOS_START_MAX // was: LOWMEM_CAP
- Then clean up the name of the function itself by renaming it
to reserve_bios_regions() and renaming the ::ebda_search paravirt
flag to ::reserve_bios_regions.
- Fix up all the comments (fix typos), harmonize and simplify their
formulation and remove comments that become unnecessary due to
the much better naming all around.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
In commit:
eb008eb6f8 ("x86: Audit and remove any remaining unnecessary uses of module.h")
... we looked for instances of module.h that were not supporting anything
more than exported symbols.
To facilitate the exchange of module.h to the much smaller export.h
we occasionally remove tags like MODULE_AUTHOR() etc. which in the case
of built in files, are no-ops and hence that is fine, assuming the
info is already in the comments at the top of the file..
However the error here is that I overlooked that this file was used
not as a driver, but as a library of functions, and hence has no
explicit modular linkage functions or similar, making it _appear_
non-modular. We can see that in retrospect with:
arch/x86/crypto/Makefile:obj-$(CONFIG_CRYPTO_GLUE_HELPER_X86) += glue_helper.o
crypto/Kconfig:config CRYPTO_GLUE_HELPER_X86
crypto/Kconfig: tristate
Since we removed what was an active MODULE_LICENSE(), the module failed
to load and then automated testing showed the missing glue helpers as:
glue_helper: Unknown symbol blkcipher_walk_done (err 0)
glue_helper: Unknown symbol blkcipher_walk_virt (err 0)
glue_helper: Unknown symbol kernel_fpu_end (err 0)
glue_helper: Unknown symbol kernel_fpu_begin (err 0)
glue_helper: Unknown symbol blkcipher_walk_virt_block (err 0)
So we do a partial revert of that change to just this one file, and
watch for similar MODULE_LICENSE() only cases in future audits.
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-crypto@vger.kernel.org
Cc: lkp@01.org
Fixes: eb008eb6f8 ("x86: Audit and remove any remaining unnecessary uses of module.h")
Link: http://lkml.kernel.org/r/20160719144243.GK21225@windriver.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
There are very few files that need add an -I$(obj) gcc for the preprocessor
or the assembler. For C files, we add always these for both the objtree and
srctree, but for the other ones we require the Makefile to add them, and
Kbuild then adds it for both trees.
As a preparation for changing the meaning of the -I$(obj) directive to
only refer to the srctree, this changes the two instances in arch/x86 to use
an explictit $(objtree) prefix where needed, otherwise we won't find the
headers any more, as reported by the kbuild 0day builder.
arch/x86/realmode/rm/realmode.lds.S:75:20: fatal error: pasyms.h: No such file or directory
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Michal Marek <mmarek@suse.com>
With PML enabled, guest will shut down if a PML full VMEXIT occurs during
event delivery. According to Intel SDM 27.2.3, PML full VMEXIT can occur when
event is being delivered through IDT, so KVM should not exit to user space
with error. Instead, it should let EXIT_REASON_PML_FULL go through and the
event will be re-injected on the next VMENTRY.
Signed-off-by: Lei Cao <lei.cao@stratus.com>
Cc: stable@vger.kernel.org
Fixes: 843e433057 ("KVM: VMX: Add PML support in VMX")
[Shortened the summary and Cc'd stable.]
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
This reverts commit 9770404a00.
The reverted patch is not needed as only userspace uses RDTSCP and
MSR_TSC_AUX is in host_save_user_msrs[] and therefore properly saved in
svm_vcpu_load() and restored in svm_vcpu_put() before every switch to
userspace.
The reverted patch did not allow the kernel to use RDTSCP in the future,
because of missed trashing in svm_set_msr() and 64-bit ifdef.
This reverts commit 2b23c3a6e3.
2b23c3a6e3 ("KVM: SVM: do not set MSR_TSC_AUX on 32-bit builds") is a
build fix for 9770404a00 and reverting them separately would only
break more bisections.
Cc: stable@vger.kernel.org
This patch adds support for non-linear data on raw records. It
extends raw records to have one or multiple fragments that will
be written linearly into the ring slot, where each fragment can
optionally have a custom callback handler to walk and extract
complex, possibly non-linear data.
If a callback handler is provided for a fragment, then the new
__output_custom() will be used instead of __output_copy() for
the perf_output_sample() part. perf_prepare_sample() does all
the size calculation only once, so perf_output_sample() doesn't
need to redo the same work anymore, meaning real_size and padding
will be cached in the raw record. The raw record becomes 32 bytes
in size without holes; to not increase it further and to avoid
doing unnecessary recalculations in fast-path, we can reuse
next pointer of the last fragment, idea here is borrowed from
ZERO_OR_NULL_PTR(), which should keep the perf_output_sample()
path for PERF_SAMPLE_RAW minimal.
This facility is needed for BPF's event output helper as a first
user that will, in a follow-up, add an additional perf_raw_frag
to its perf_raw_record in order to be able to more efficiently
dump skb context after a linear head meta data related to it.
skbs can be non-linear and thus need a custom output function to
dump buffers. Currently, the skb data needs to be copied twice;
with the help of __output_custom() this work only needs to be
done once. Future users could be things like XDP/BPF programs
that work on different context though and would thus also have
a different callback function.
The few users of raw records are adapted to initialize their frag
data from the raw record itself, no change in behavior for them.
The code is based upon a PoC diff provided by Peter Zijlstra [1].
[1] http://thread.gmane.org/gmane.linux.network/421294
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
On Intel hardware, native_play_dead() uses mwait_play_dead() by
default and only falls back to the other methods if that fails.
That also happens during resume from hibernation, when the restore
(boot) kernel runs disable_nonboot_cpus() to take all of the CPUs
except for the boot one offline.
However, that is problematic, because the address passed to
__monitor() in mwait_play_dead() is likely to be written to in the
last phase of hibernate image restoration and that causes the "dead"
CPU to start executing instructions again. Unfortunately, the page
containing the address in that CPU's instruction pointer may not be
valid any more at that point.
First, that page may have been overwritten with image kernel memory
contents already, so the instructions the CPU attempts to execute may
simply be invalid. Second, the page tables previously used by that
CPU may have been overwritten by image kernel memory contents, so the
address in its instruction pointer is impossible to resolve then.
A report from Varun Koyyalagunta and investigation carried out by
Chen Yu show that the latter sometimes happens in practice.
To prevent it from happening, temporarily change the smp_ops.play_dead
pointer during resume from hibernation so that it points to a special
"play dead" routine which uses hlt_play_dead() and avoids the
inadvertent "revivals" of "dead" CPUs this way.
A slightly unpleasant consequence of this change is that if the
system is hibernated with one or more CPUs offline, it will generally
draw more power after resume than it did before hibernation, because
the physical state entered by CPUs via hlt_play_dead() is higher-power
than the mwait_play_dead() one in the majority of cases. It is
possible to work around this, but it is unclear how much of a problem
that's going to be in practice, so the workaround will be implemented
later if it turns out to be necessary.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=106371
Reported-by: Varun Koyyalagunta <cpudebug@centtech.com>
Original-by: Chen Yu <yu.c.chen@intel.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Dell Optiplex 7450 AIO works with BOOT_ACPI; however, the quirk for
"OptiPlex 745" changes its boot method to BOOT_BIOS and causes 7450 AIO
hangs when rebooting; as a result, 7450 AIO is appended to overwrite
BOOT_BIOS by BOOT_ACPI in order not to break the original 745 series
Signed-off-by: Alex Hung <alex.hung@canonical.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
When freeing the nested resources of a vcpu, there is an assumption that
the vcpu's vmcs01 is the current VMCS on the CPU that executes
nested_release_vmcs12(). If this assumption is violated, the vcpu's
vmcs01 may be made active on multiple CPUs at the same time, in
violation of Intel's specification. Moreover, since the vcpu's vmcs01 is
not VMCLEARed on every CPU on which it is active, it can linger in a
CPU's VMCS cache after it has been freed and potentially
repurposed. Subsequent eviction from the CPU's VMCS cache on a capacity
miss can result in memory corruption.
It is not sufficient for vmx_free_vcpu() to call vmx_load_vmcs01(). If
the vcpu in question was last loaded on a different CPU, it must be
migrated to the current CPU before calling vmx_load_vmcs01().
Signed-off-by: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
