sdk/emulator/qemu.git
10 years agotarget-arm: Add AArch32 FP VRINTA, VRINTN, VRINTP and VRINTM
Will Newton [Fri, 31 Jan 2014 14:47:33 +0000 (14:47 +0000)]
target-arm: Add AArch32 FP VRINTA, VRINTN, VRINTP and VRINTM

Add support for AArch32 ARMv8 FP VRINTA, VRINTN, VRINTP and VRINTM
instructions.

Signed-off-by: Will Newton <will.newton@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotarget-arm: Move arm_rmode_to_sf to a shared location.
Will Newton [Fri, 31 Jan 2014 14:47:33 +0000 (14:47 +0000)]
target-arm: Move arm_rmode_to_sf to a shared location.

This function will be needed for AArch32 ARMv8 support, so move it to
helper.c where it can be used by both targets. Also moves the code out
of line, but as it is quite a large function I don't believe this
should be a significant performance impact.

Signed-off-by: Will Newton <will.newton@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agodisplay: avoid multi-statement macro
Paolo Bonzini [Fri, 31 Jan 2014 14:47:33 +0000 (14:47 +0000)]
display: avoid multi-statement macro

For blizzard, pl110 and tc6393xb this is harmless, but for pxa2xx
Coverity noticed that it is used inside an "if" statement.
Fix it because it's the file with the highest number of defects
in the whole QEMU tree!  Use "do...while (0)", or just remove the
semicolon if there's a single statement in the macro.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoZYNQ: Implement board MIDR control for Zynq
Alistair Francis [Fri, 31 Jan 2014 14:47:33 +0000 (14:47 +0000)]
ZYNQ: Implement board MIDR control for Zynq

This patch uses the fact that the midr variable is now a property
This patch sets the midr variable to the boards custom midr

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: a3754b10d150af72e4688a993e484fa2b9b8fa21.1390176489.git.alistair.francis@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoARM: Convert MIDR to a property
Alistair Francis [Fri, 31 Jan 2014 14:47:32 +0000 (14:47 +0000)]
ARM: Convert MIDR to a property

Convert the MIDR register to a property. This allows boards to later set
a custom MIDR value. This has been done in such a way to maintain
compatibility with all existing CPUs and boards

Signed-off-by: Alistair Francis <alistair.francis@xilinx.com>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 878613f2f12d4162f12629522fd99de8df904856.1390176489.git.alistair.francis@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agohw/arm/boot: Don't set up ATAGS for autogenerated dtb booting
Peter Maydell [Fri, 31 Jan 2014 14:47:32 +0000 (14:47 +0000)]
hw/arm/boot: Don't set up ATAGS for autogenerated dtb booting

The code which decides whether to set up the ATAGS data structure on
reset was using the wrong conditional, which meant we were creating
an ATAGS structure when doing a device-tree boot if the dtb was
autogenerated by the board. This is harmless, but unnecessary, so
bring it in to line with user-provided-dtb boots.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Message-id: 1388326833-656-1-git-send-email-peter.maydell@linaro.org

10 years agotarget-arm: A64: Add SIMD scalar copy instructions
Peter Maydell [Fri, 31 Jan 2014 14:47:32 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD scalar copy instructions

Add support for the SIMD scalar copy instruction group (C3.6.7),
which consists of the single instruction DUP (element, scalar).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD modified immediate group
Alex Bennée [Fri, 31 Jan 2014 14:47:32 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD modified immediate group

This patch adds support for the AdvSIMD modified immediate group
(C3.6.6) with all its suboperations (movi, orr, fmov, mvni, bic).

Signed-off-by: Alexander Graf <agraf@suse.de>
[AJB: new decode struct, minor bug fixes, optimisation]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD copy operations
Alex Bennée [Fri, 31 Jan 2014 14:47:31 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD copy operations

This adds support for the all the AdvSIMD vector copy operations
(ARM ARM 3.6.5).

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD across-lanes instructions
Michael Matz [Fri, 31 Jan 2014 14:47:31 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD across-lanes instructions

Add support for the SIMD "across lanes" instruction group (C3.6.4).

Signed-off-by: Michael Matz <matz@suse.de>
[PMM: Updated to current codebase, added fp min/max ops,
 added unallocated encoding checks]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD ZIP/UZP/TRN
Michael Matz [Fri, 31 Jan 2014 14:47:31 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD ZIP/UZP/TRN

Add support for the SIMD ZIP/UZIP/TRN instruction group
(C3.6.3).

Signed-off-by: Michael Matz <matz@suse.de>
[PMM: use new do_vec_get/set etc functions and generally update to new
 codebase standards; refactor to pull per-element loop outside switch]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD TBL/TBLX
Michael Matz [Fri, 31 Jan 2014 14:47:31 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD TBL/TBLX

Add support for the SIMD TBL/TBLX instructions (group C3.6.2).

Signed-off-by: Michael Matz <matz@suse.de>
[PMM: rewritten to do more of the decode in translate-a64.c,
 and to do only one 64 bit pass at a time in the helper]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD EXT
Peter Maydell [Fri, 31 Jan 2014 14:47:30 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD EXT

Add support for the SIMD EXT instruction (the only one in its
group, C3.6.1).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add decode skeleton for SIMD data processing insns
Alex Bennée [Fri, 31 Jan 2014 14:47:30 +0000 (14:47 +0000)]
target-arm: A64: Add decode skeleton for SIMD data processing insns

Add decode skeleton and function placeholders for all the SIMD data
processing instructions. Due to the complexity of this part of the
table the normal extract and switch approach gets very messy very
quickly, so we use a simple data-driven pattern-and-mask approach.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD ld/st single
Peter Maydell [Fri, 31 Jan 2014 14:47:30 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD ld/st single

Implement the SIMD ld/st single structure instructions.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agotarget-arm: A64: Add SIMD ld/st multiple
Alex Bennée [Fri, 31 Jan 2014 14:47:30 +0000 (14:47 +0000)]
target-arm: A64: Add SIMD ld/st multiple

This adds support support for the SIMD load/store
multiple category of instructions.

This also brings in a couple of helper functions for manipulating
sections of the SIMD registers:

  * do_vec_get - fetch value from a slice of a vector register
  * do_vec_set - set a slice of a vector register

which use vec_reg_offset for consistent processing of offsets in an
endian aware manner. There are also additional helpers:

  * do_vec_ld - load value into SIMD
  * do_vec_st - store value from SIMD

which load or store a slice of a vector register to memory.
These don't zero extend like the fp variants.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
10 years agoMerge remote-tracking branch 'stefanha/tags/tracing-pull-request' into staging
Peter Maydell [Fri, 31 Jan 2014 11:13:08 +0000 (11:13 +0000)]
Merge remote-tracking branch 'stefanha/tags/tracing-pull-request' into staging

Tracing pull request

# gpg: Signature made Mon 27 Jan 2014 14:51:09 GMT using RSA key ID 81AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* stefanha/tags/tracing-pull-request:
  trace: fix simple trace "disable" keyword
  trace: add glib 2.32+ static GMutex support
  trace: [simple] Do not include "trace/simple.h" in generated tracer headers
  tracing: start trace processing thread in final child process

Message-id: 1390834386-23139-1-git-send-email-stefanha@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'mst/tags/for_anthony' into staging
Peter Maydell [Fri, 31 Jan 2014 00:23:27 +0000 (00:23 +0000)]
Merge remote-tracking branch 'mst/tags/for_anthony' into staging

acpi,pci,pc,virtio fixes and enhancements

This includes new unit-tests for acpi by Marcel,
hotplug for pci bridges by myself (piix only so far)
and cpu hotplug for q35.
And a bunch of fixes all over the place as usual.

I included the patch to fix memory alignment for q35
as well - even though it limits 32 bit guests to 3G (they
previously could address more memory with PAE).
To remove the limit, this will have to be fixed in seabios.

I also added self as virtio co-maintainer so I don't need
to troll the list for patches to review.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Sun 26 Jan 2014 11:12:09 GMT using RSA key ID D28D5469
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>"
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* mst/tags/for_anthony: (35 commits)
  MAINTAINERS: add self as virtio co-maintainer
  q35: document gigabyte_align
  q35: gigabyte alignment for ram
  acpi: Fix PCI hole handling on build_srat()
  pc: Save size of RAM below 4GB
  hw/pci: fix error flow in pci multifunction init
  acpi-test: update expected AML since recent changes
  pc: ACPI: update acpi-dsdt.hex.generated q35-acpi-dsdt.hex.generated
  pc: ACPI: unify source of CPU hotplug IO base/len
  pc: ACPI: expose PRST IO range via _CRS
  pc: Q35 DSDT: exclude CPU hotplug IO range from PCI bus resources
  pc: PIIX DSDT: exclude CPU/PCI hotplug & GPE0 IO range from PCI bus resources
  pc: set PRST base in DSDT depending on chipset
  acpi: ich9: add CPU hotplug handling to Q35 machine
  acpi: factor out common cpu hotplug code for PIIX4/Q35
  acpi-build: enable hotplug for PCI bridges
  piix4: add acpi pci hotplug support
  pcihp: generalization of piix4 acpi
  pci: add pci_for_each_bus_depth_first
  pc: make: fix dependencies: rebuild when included file is changed
  ...

Message-id: 1390735289-15563-1-git-send-email-mst@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'sstabellini/xen-170114' into staging
Peter Maydell [Fri, 31 Jan 2014 00:13:02 +0000 (00:13 +0000)]
Merge remote-tracking branch 'sstabellini/xen-170114' into staging

* sstabellini/xen-170114:
  xen_pt: Fix passthrough of device with ROM.
  xen_pt: Fix debug output.
  xenfb: map framebuffer read-only and handle unmap errors

Message-id: alpine.DEB.2.02.1401171537140.21510@kaball.uk.xensource.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'stefanha/tags/net-pull-request' into staging
Peter Maydell [Thu, 30 Jan 2014 22:25:39 +0000 (22:25 +0000)]
Merge remote-tracking branch 'stefanha/tags/net-pull-request' into staging

Net patches

# gpg: Signature made Mon 27 Jan 2014 14:45:35 GMT using RSA key ID 81AB73C8
# gpg: Can't check signature: public key not found

* stefanha/tags/net-pull-request:
  tap-linux: Get features once and use it many times
  Fix lan9118 buffer length handling
  Fix lan9118 TX "CMD A" handling
  net: Use g_strdup_printf instead of snprintf.

Message-id: 1390834129-19625-1-git-send-email-stefanha@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'rth/tcg-movbe' into staging
Peter Maydell [Thu, 30 Jan 2014 19:02:16 +0000 (19:02 +0000)]
Merge remote-tracking branch 'rth/tcg-movbe' into staging

* rth/tcg-movbe:
  tcg/i386: cleanup useless #ifdef
  tcg/i386: use movbe instruction in qemu_ldst routines
  tcg/i386: add support for three-byte opcodes
  tcg/i386: remove hardcoded P_REXW value
  disas/i386.c: disassemble movbe instruction

Message-id: 1390692772-15282-1-git-send-email-rth@twiddle.net
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoMerge remote-tracking branch 'mjt/tags/trivial-patches-2014-01-16' into staging
Peter Maydell [Thu, 30 Jan 2014 13:56:00 +0000 (13:56 +0000)]
Merge remote-tracking branch 'mjt/tags/trivial-patches-2014-01-16' into staging

trivial-patches for 2014-01-16

# gpg: Signature made Thu 16 Jan 2014 17:29:05 GMT using RSA key ID 74F0C838
# gpg: Good signature from "Michael Tokarev <mjt@tls.msk.ru>"
# gpg:                 aka "Michael Tokarev <mjt@corpit.ru>"
# gpg:                 aka "Michael Tokarev <mjt@debian.org>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6EE1 95D1 886E 8FFB 810D  4324 457C E0A0 8044 65C5
#      Subkey fingerprint: E190 8639 3B10 B51B AC2C  8B73 5253 C5AD 74F0 C838

Message-id: 1389893719-16336-1-git-send-email-mjt@msgid.tls.msk.ru
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agoTCG: Fix I64-on-32bit-host temporaries
Alexander Graf [Sun, 19 Jan 2014 15:53:31 +0000 (16:53 +0100)]
TCG: Fix I64-on-32bit-host temporaries

We have cache pools of temporaries that we can reuse later when they've
already been allocated before.

These cache pools differenciate between the target TCG variable type they
contain. So we have one pool for I32 and one pool for I64 variables.

On a 32bit system, we can't work with 64bit registers though. So instead we
spawn two I32 temporaries for every I64 temporary we create. All caching
works the same way as on a real 64-bit system though: We create a cache entry
in the 64bit array for the first i32 index.

However, when we free such a temporary we free it to the pool of its type
(which is always i32 on 32bit systems) rather than its base_type (which is
i64 or i32 depending on the variable). This means we put a temporary that
is of base_type == i64 into the i32 preallocated temporary pool.

Eventually, this results in failures like this on 32bit hosts:

  qemu-system-ppc64: tcg/tcg.c:515: tcg_temp_new_internal: Assertion `ts->base_type == type' failed.

This patch makes the free routine use the base_type instead for the free case,
so it's consistent with the temporary allocation. It fixes the above failure
for me.

Signed-off-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-id: 1390146811-59936-1-git-send-email-agraf@suse.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 years agotrace: fix simple trace "disable" keyword
Stefan Hajnoczi [Wed, 15 Jan 2014 03:10:30 +0000 (11:10 +0800)]
trace: fix simple trace "disable" keyword

The trace-events "disable" keyword turns an event into a nop at
compile-time.  This is important for high-frequency events that can
impact performance.

The "disable" keyword is currently broken in the simple trace backend.
This patch fixes the problem as follows:

Trace events are identified by their TraceEventID number.  When events
are disabled there are two options for assigning TraceEventID numbers:
1. Skip disabled events and don't assign them a number.
2. Assign numbers for all events regardless of the disabled keyword.

The simple trace backend and its binary file format uses approach #1.

The tracetool infrastructure has been using approach #2 for a while.

The result is that the numbers used in simple trace files do not
correspond with TraceEventIDs.  In trace/simple.c we assumed that they
are identical and therefore emitted bogus numbers.

This patch fixes the bug by using TraceEventID for trace_event_id()
while sticking to approach #1 for simple trace file numbers.  This
preserves simple trace file format compatibility.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agotrace: add glib 2.32+ static GMutex support
Stefan Hajnoczi [Thu, 12 Dec 2013 14:50:11 +0000 (15:50 +0100)]
trace: add glib 2.32+ static GMutex support

The GStaticMutex API was deprecated in glib 2.32.  We cannot switch over
to GMutex unconditionally since we would drop support for older glib
versions.  But the deprecated API warnings during build are annoying so
use static GMutex when possible.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agotrace: [simple] Do not include "trace/simple.h" in generated tracer headers
Lluís Vilanova [Tue, 14 Jan 2014 15:52:55 +0000 (16:52 +0100)]
trace: [simple] Do not include "trace/simple.h" in generated tracer headers

The header is not necessary, given that the simple backend does not define any
inlined tracing routines.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agotracing: start trace processing thread in final child process
Michael Mueller [Mon, 23 Sep 2013 14:36:54 +0000 (16:36 +0200)]
tracing: start trace processing thread in final child process

When running with trace backend e.g. "simple" the writer thread needs to be
implemented in the same process context as the trace points that will be
processed. Under libvirtd control, qemu gets first started in daemonized
mode to privide its capabilities. Creating the writer thread in the initial
process context then leads to a dead lock because the thread gets termined
together with the initial parent. (-daemonize)

Signed-off-by: Michael Mueller <mimu@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
[minor whitespace fixes]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agotap-linux: Get features once and use it many times
Kusanagi Kouichi [Sat, 18 Jan 2014 05:38:45 +0000 (14:38 +0900)]
tap-linux: Get features once and use it many times

Signed-off-by: Kusanagi Kouichi <slash@ac.auone-net.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agoFix lan9118 buffer length handling
Roy Franz [Wed, 8 Jan 2014 04:19:52 +0000 (20:19 -0800)]
Fix lan9118 buffer length handling

The 9118 ethernet controller supports transmission of multi-buffer packets
with arbitrary byte alignment of the start and end bytes.  All writes to
the packet fifo are 32 bits, so the controller discards bytes at the beginning
and end of each buffer based on the 'Data start offset' and 'Buffer size'
of the TX command 'A' format.

This patch uses the provided buffer length to limit the bytes transmitted.
Previously all the bytes of the last 32-bit word written to the TX fifo
were added to the internal transmit buffer structure resulting in more bytes
being transmitted than were submitted to the hardware in the command.  This
resulted in extra bytes being inserted into the middle of multi-buffer
packets when the non-final buffers had non-32bit aligned ending addresses.

Signed-off-by: Roy Franz <roy.franz@linaro.org>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agoFix lan9118 TX "CMD A" handling
Roy Franz [Wed, 8 Jan 2014 04:19:51 +0000 (20:19 -0800)]
Fix lan9118 TX "CMD A" handling

The 9118 ethernet controller supports transmission of multi-buffer packets
with arbitrary byte alignment of the start and end bytes.  All writes to
the packet fifo are 32 bits, so the controller discards bytes at the beginning
and end of each buffer based on the 'Data start offset' and 'Buffer size'
of the TX command 'A' format.

This patch changes the buffer size and offset internal state variables to be
updated on every "TX command A" write.  Previously they were only updated for
the first segment, which resulted incorrect behavior for packets with more
than one segment. Each segment of the packet has its own CMD A command, with
its own buffer size and start offset.

Also update extraction of fields from the CMD A word to use extract32().

Signed-off-by: Roy Franz <roy.franz@linaro.org>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agonet: Use g_strdup_printf instead of snprintf.
Hani Benhabiles [Thu, 9 Jan 2014 18:34:27 +0000 (19:34 +0100)]
net: Use g_strdup_printf instead of snprintf.

assign_name() in net/net.c is using snprintf + g_strdup to get the same
result as g_strdup_printf.

Signed-off-by: Hani Benhabiles <kroosec@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 years agoMAINTAINERS: add self as virtio co-maintainer
Michael S. Tsirkin [Sun, 26 Jan 2014 10:39:55 +0000 (12:39 +0200)]
MAINTAINERS: add self as virtio co-maintainer

This will help make sure I get Cc'd on patches.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoq35: document gigabyte_align
Michael S. Tsirkin [Mon, 16 Dec 2013 11:55:06 +0000 (13:55 +0200)]
q35: document gigabyte_align

Document the logic behind the below/above 4G split.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoq35: gigabyte alignment for ram
Gerd Hoffmann [Mon, 16 Dec 2013 09:11:28 +0000 (10:11 +0100)]
q35: gigabyte alignment for ram

Map 2G (q35) of memory below 4G, so the RAM pieces
are nicely aligned to gigabyte borders.

Keep old memory layout for (a) old machine types and (b) in case all
memory fits below 4G and thus we don't have to split RAM into pieces
in the first place.  The later makes sure this change doesn't take
away memory from 32bit guests.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi: Fix PCI hole handling on build_srat()
Eduardo Habkost [Thu, 9 Jan 2014 19:12:43 +0000 (17:12 -0200)]
acpi: Fix PCI hole handling on build_srat()

The original SeaBIOS code used the RamSize variable, that was used by
SeaBIOS for the size of RAM below 4GB, not for all RAM. When copied to
QEMU, the code was changed to use the full RAM size, and this broke the
build_srat() code that handles the PCI hole.

Change build_srat() to use ram_size_below_4g instead of ram_size, to
restore the original behavior from SeaBIOS.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: Save size of RAM below 4GB
Eduardo Habkost [Thu, 9 Jan 2014 19:12:42 +0000 (17:12 -0200)]
pc: Save size of RAM below 4GB

The ram_below_4g value will be useful in other places, such as the ACPI
table code, and other code that currently requires passing
below_4g_mem_size around in function arguments.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agohw/pci: fix error flow in pci multifunction init
Marcel Apfelbaum [Tue, 21 Jan 2014 16:37:51 +0000 (18:37 +0200)]
hw/pci: fix error flow in pci multifunction init

Scenario:
  - There is a non multifunction pci device A on 00:0X.0.
  - Hot-plug another multifunction pci device B at 00:0X.1.
  - The operation will fail of course.
  - Try to hot-plug the B device 2-3 more times, qemu will crash.

Reason: The error flow leaves the B's address space into global address spaces
list, but the device object is freed. Fixed that.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi-test: update expected AML since recent changes
Michael S. Tsirkin [Mon, 20 Jan 2014 12:00:12 +0000 (14:00 +0200)]
acpi-test: update expected AML since recent changes

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: ACPI: update acpi-dsdt.hex.generated q35-acpi-dsdt.hex.generated
Igor Mammedov [Thu, 9 Jan 2014 16:36:39 +0000 (17:36 +0100)]
pc: ACPI: update acpi-dsdt.hex.generated q35-acpi-dsdt.hex.generated

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: ACPI: unify source of CPU hotplug IO base/len
Igor Mammedov [Thu, 9 Jan 2014 16:36:38 +0000 (17:36 +0100)]
pc: ACPI: unify source of CPU hotplug IO base/len

use C headers defines as source of IO base/len for respective
values in ASL code.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: ACPI: expose PRST IO range via _CRS
Igor Mammedov [Thu, 9 Jan 2014 16:36:37 +0000 (17:36 +0100)]
pc: ACPI: expose PRST IO range via _CRS

.. so OSPM could notice resource conflict if there is any.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: Q35 DSDT: exclude CPU hotplug IO range from PCI bus resources
Igor Mammedov [Thu, 9 Jan 2014 16:36:36 +0000 (17:36 +0100)]
pc: Q35 DSDT: exclude CPU hotplug IO range from PCI bus resources

... for range defined at hw/acpi/ich9.c:ICH9_PROC_BASE

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: PIIX DSDT: exclude CPU/PCI hotplug & GPE0 IO range from PCI bus resources
Igor Mammedov [Thu, 9 Jan 2014 16:36:35 +0000 (17:36 +0100)]
pc: PIIX DSDT: exclude CPU/PCI hotplug & GPE0 IO range from PCI bus resources

.. so that they might not be used by PCI devices.

Note:
Resort to concatenating templates with preprocessor help,
because 1.0b spec isn't supporting ConcatenateResTemplate,
as result Windows XP fails to execute PCI0._CRS method if
ConcatenateResTemplate() is used.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: set PRST base in DSDT depending on chipset
Igor Mammedov [Thu, 9 Jan 2014 16:36:34 +0000 (17:36 +0100)]
pc: set PRST base in DSDT depending on chipset

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi: ich9: add CPU hotplug handling to Q35 machine
Igor Mammedov [Thu, 9 Jan 2014 16:36:32 +0000 (17:36 +0100)]
acpi: ich9: add CPU hotplug handling to Q35 machine

.. use IO port 0cd8-0xcf7 range for CPU present bitmap

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi: factor out common cpu hotplug code for PIIX4/Q35
Igor Mammedov [Thu, 9 Jan 2014 16:36:31 +0000 (17:36 +0100)]
acpi: factor out common cpu hotplug code for PIIX4/Q35

.. so it could be used for adding CPU hotplug to Q35 machine

Add an additional header with that will be shared between
C and ASL code: include/hw/acpi/cpu_hotplug_defs.h

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi-build: enable hotplug for PCI bridges
Michael S. Tsirkin [Mon, 14 Oct 2013 15:01:29 +0000 (18:01 +0300)]
acpi-build: enable hotplug for PCI bridges

This enables support for device hotplug behind
pci bridges. Bridge devices themselves need
to be pre-configured on qemu command line.

Design:
    - at machine init time, assign "bsel" property to bridges with
      hotplug support
    - dynamically (At ACPI table read) generate ACPI code to handle
      hotplug events for each bridge with "bsel" property

Note: ACPI doesn't support adding or removing bridges by hotplug.
We detect and prevent removal of bridges by hotplug,
unless they were added by hotplug previously
(and so, are not described by ACPI).

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopiix4: add acpi pci hotplug support
Michael S. Tsirkin [Mon, 14 Oct 2013 15:01:20 +0000 (18:01 +0300)]
piix4: add acpi pci hotplug support

Add support for acpi pci hotplug using the
new infrastructure.
PIIX4 legacy interface is maintained as is for
machine types 1.7 and older.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopcihp: generalization of piix4 acpi
Michael S. Tsirkin [Mon, 14 Oct 2013 15:01:11 +0000 (18:01 +0300)]
pcihp: generalization of piix4 acpi

Add ACPI based PCI hotplug library with bridge hotplug
support.
Design
   - each bus gets assigned "bsel" property.
   - ACPI code writes this number
     to a new BNUM register, then uses existing
     UP/DOWN registers to probe slot status;
     to eject, write number to BNUM register,
     then slot into existing EJ.

The interface is actually backwards-compatible with
existing PIIX4 ACPI (though not migration compatible).

This is split out from PIIX4 codebase so we can
reuse it for Q35 as well.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopci: add pci_for_each_bus_depth_first
Michael S. Tsirkin [Mon, 14 Oct 2013 15:01:07 +0000 (18:01 +0300)]
pci: add pci_for_each_bus_depth_first

Useful for ACPI hotplug.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agopc: make: fix dependencies: rebuild when included file is changed
Igor Mammedov [Thu, 9 Jan 2014 16:36:33 +0000 (17:36 +0100)]
pc: make: fix dependencies: rebuild when included file is changed

some *.dsl files include another *.dsl files but there weren't
any dependicies and when included file changed target table wasn't
rebuild. Fix this by using the same auto dependency generation
as for C files.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: do not fail on asl mismatch
Marcel Apfelbaum [Thu, 16 Jan 2014 15:50:48 +0000 (17:50 +0200)]
acpi unit-test: do not fail on asl mismatch

The asl comparison will break every time the ACPI
tables are updated. This may break the git bisect.
Instead of failing print a warning on stderr
including the retained asl files, so they can be
compared offline.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: resolved iasl crash
Marcel Apfelbaum [Thu, 16 Jan 2014 15:50:47 +0000 (17:50 +0200)]
acpi unit-test: resolved iasl crash

It seems that iasl has an issue when disassembles
some ACPI tables using the command line:
iasl -e DSDT -e SSDT -d HPET

Modified the iasl command line to "iasl -d HPET"
until the problem is solved. The command line
remained the same for DSDT and SSDT tables.

Reported-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: renamed ssdt_tables to tables
Marcel Apfelbaum [Thu, 16 Jan 2014 15:50:46 +0000 (17:50 +0200)]
acpi unit-test: renamed ssdt_tables to tables

Just a refactoring, ssdt_tables name was confusing as
it included other tables as well.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agotests: fix acpi to work on bigendian host
Alexey Kardashevskiy [Mon, 13 Jan 2014 07:33:53 +0000 (18:33 +1100)]
tests: fix acpi to work on bigendian host

Double endianness convertion make this test failing on POWERPC machine
running in big-endian.

This fixes the test to success on big-endian host.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: hook to rebuild expected aml files
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:25 +0000 (16:54 +0200)]
acpi unit-test: hook to rebuild expected aml files

When running the test with TEST_ACPI_REBUILD_AML=y environment
variable, the test will rebuild and validate the expected aml
files.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: added script to rebuild the expected aml files
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:24 +0000 (16:54 +0200)]
acpi unit-test: added script to rebuild the expected aml files

Acpi unit-test will fail every time the acpi tables change.
This script rebuild the expected aml files, so the test
will pass. It also validates the modifications.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: extract iasl executable from configuration
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:23 +0000 (16:54 +0200)]
acpi unit-test: extract iasl executable from configuration

The test checked if iasl is installed by running "iasl"
and checking the error output.
It is better to use the iasl executable as appears
in configuration.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoconfigure: add CONFIG_IASL to config-host.h
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:22 +0000 (16:54 +0200)]
configure: add CONFIG_IASL to config-host.h

Acpi unit-tests will extract iasl executable
from CONFIG_IASL define.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: compare DSDT and SSDT tables against expected values
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:21 +0000 (16:54 +0200)]
acpi unit-test: compare DSDT and SSDT tables against expected values

This test will run only if iasl is installed on the host machine.
The test plan:
 1. Dumps the ACPI tables as AML on the disk.
 2. Runs iasl to disassembly the tables into ASL files.
 3. Runs iasl to disassembly the offline AML files into ASL files.
 4. Compares the ASL files.

The test runs for both default machine and q35.
In case the test fails, it can be easily tweaked to
show the differences between the ASL files and
understand the issue.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoconfigure: added acpi unit-test files
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:20 +0000 (16:54 +0200)]
configure: added acpi unit-test files

Ensure configure will set-up links for the files
if the build is created in other directory.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoacpi unit-test: add test files
Marcel Apfelbaum [Thu, 26 Dec 2013 14:54:19 +0000 (16:54 +0200)]
acpi unit-test: add test files

Added unit-test's expected aml files to be compared
with the actual ACPI tables.

Signed-off-by: Marcel Apfelbaum <marcel.a@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agovirtio: Fix return value for dummy function vhost_net_virtqueue_pending
Stefan Weil [Sun, 22 Dec 2013 14:51:22 +0000 (15:51 +0100)]
virtio: Fix return value for dummy function vhost_net_virtqueue_pending

cgcc complains that -ENOSYS is not a good value for 'bool'.

A dummy virtio will never have pending queue entries, so let us return
false.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoACPI: Fix AppleSMC _STA size
Gabriel L. Somlo [Mon, 13 Jan 2014 20:27:13 +0000 (15:27 -0500)]
ACPI: Fix AppleSMC _STA size

Minimize the storage used for AppleSMC's _STA (8bit), relying on ASL
to implicitly convert it to the officially specified 32bit value.

Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoAdd DSDT node for AppleSMC
Gabriel L. Somlo [Sun, 22 Dec 2013 15:34:56 +0000 (10:34 -0500)]
Add DSDT node for AppleSMC

AppleSMC (-device isa-applesmc) is required to boot OS X guests.
OS X expects a SMC node to be present in the ACPI DSDT. This patch
adds a SMC node to the DSDT, and dynamically patches the return value
of SMC._STA to either 0x0B if the chip is present, or otherwise to 0x00,
before booting the guest.

Signed-off-by: Gabriel Somlo <somlo@cmu.edu>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agoPython-lang gdb script to extract x86_64 guest vmcore from qemu coredump
Laszlo Ersek [Tue, 17 Dec 2013 00:37:06 +0000 (01:37 +0100)]
Python-lang gdb script to extract x86_64 guest vmcore from qemu coredump

When qemu dies unexpectedly, for example in response to an explicit
abort() call, or (more importantly) when an external signal is delivered
to it that results in a coredump, sometimes it is useful to extract the
guest vmcore from the qemu process' memory image. The guest vmcore might
help understand an emulation problem in qemu, or help debug the guest.

This script reimplements (and cuts many features of) the
qmp_dump_guest_memory() command in gdb/Python,

  https://sourceware.org/gdb/current/onlinedocs/gdb/Python-API.html

working off the saved memory image of the qemu process. The docstring in
the patch (serving as gdb help text) describes the limitations relative to
the QMP command.

Dependencies of qmp_dump_guest_memory() have been reimplemented as needed.
I sought to follow the general structure, sticking to original function
names where possible. However, keeping it simple prevailed in some places.

The patch has been tested with a 4 VCPU, 768 MB, RHEL-6.4
(2.6.32-358.el6.x86_64) guest:

- The script printed

> guest RAM blocks:
> target_start     target_end       host_addr        message count
> ---------------- ---------------- ---------------- ------- -----
0000000000000000 00000000000a0000 00007f95d0000000 added       1
00000000000a0000 00000000000b0000 00007f960ac00000 added       2
00000000000c0000 00000000000ca000 00007f95d00c0000 added       3
00000000000ca000 00000000000cd000 00007f95d00ca000 joined      3
00000000000cd000 00000000000d0000 00007f95d00cd000 joined      3
00000000000d0000 00000000000f0000 00007f95d00d0000 joined      3
00000000000f0000 0000000000100000 00007f95d00f0000 joined      3
0000000000100000 0000000030000000 00007f95d0100000 joined      3
00000000fc000000 00000000fc800000 00007f960ac00000 added       4
00000000fffe0000 0000000100000000 00007f9618800000 added       5
> dumping range at 00007f95d0000000 for length 00000000000a0000
> dumping range at 00007f960ac00000 for length 0000000000010000
> dumping range at 00007f95d00c0000 for length 000000002ff40000
> dumping range at 00007f960ac00000 for length 0000000000800000
> dumping range at 00007f9618800000 for length 0000000000020000

- The vmcore was checked with "readelf", comparing the results against a
  vmcore written by qmp_dump_guest_memory():

> --- theirs      2013-09-12 17:38:59.797289404 +0200
> +++ mine        2013-09-12 17:39:03.820289404 +0200
> @@ -27,16 +27,16 @@
>    Type           Offset             VirtAddr           PhysAddr
>                   FileSiz            MemSiz              Flags  Align
>    NOTE           0x0000000000000190 0x0000000000000000 0x0000000000000000
> -                 0x0000000000000ca0 0x0000000000000ca0         0
> -  LOAD           0x0000000000000e30 0x0000000000000000 0x0000000000000000
> +                 0x000000000000001c 0x000000000000001c         0
> +  LOAD           0x00000000000001ac 0x0000000000000000 0x0000000000000000
>                   0x00000000000a0000 0x00000000000a0000         0
> -  LOAD           0x00000000000a0e30 0x0000000000000000 0x00000000000a0000
> +  LOAD           0x00000000000a01ac 0x0000000000000000 0x00000000000a0000
>                   0x0000000000010000 0x0000000000010000         0
> -  LOAD           0x00000000000b0e30 0x0000000000000000 0x00000000000c0000
> +  LOAD           0x00000000000b01ac 0x0000000000000000 0x00000000000c0000
>                   0x000000002ff40000 0x000000002ff40000         0
> -  LOAD           0x000000002fff0e30 0x0000000000000000 0x00000000fc000000
> +  LOAD           0x000000002fff01ac 0x0000000000000000 0x00000000fc000000
>                   0x0000000000800000 0x0000000000800000         0
> -  LOAD           0x00000000307f0e30 0x0000000000000000 0x00000000fffe0000
> +  LOAD           0x00000000307f01ac 0x0000000000000000 0x00000000fffe0000
>                   0x0000000000020000 0x0000000000020000         0
>
>  There is no dynamic section in this file.
> @@ -47,13 +47,6 @@
>
>  No version information found in this file.
>
> -Notes at offset 0x00000190 with length 0x00000ca0:
> +Notes at offset 0x00000190 with length 0x0000001c:
>    Owner                Data size       Description
> -  CORE         0x00000150      NT_PRSTATUS (prstatus structure)
> -  CORE         0x00000150      NT_PRSTATUS (prstatus structure)
> -  CORE         0x00000150      NT_PRSTATUS (prstatus structure)
> -  CORE         0x00000150      NT_PRSTATUS (prstatus structure)
> -  QEMU         0x000001b0      Unknown note type: (0x00000000)
> -  QEMU         0x000001b0      Unknown note type: (0x00000000)
> -  QEMU         0x000001b0      Unknown note type: (0x00000000)
> -  QEMU         0x000001b0      Unknown note type: (0x00000000)
> +  NONE         0x00000005      Unknown note type: (0x00000000)

- The vmcore was checked with "crash" too, again comparing the results
  against a vmcore written by qmp_dump_guest_memory():

> --- guest.vmcore.log2   2013-09-12 17:52:27.074289201 +0200
> +++ example.dump.log2   2013-09-12 17:52:15.904289203 +0200
> @@ -22,11 +22,11 @@
>  This GDB was configured as "x86_64-unknown-linux-gnu"...
>
>       KERNEL: /usr/lib/debug/lib/modules/2.6.32-358.el6.x86_64/vmlinux
> -    DUMPFILE: /home/lacos/tmp/guest.vmcore
> +    DUMPFILE: /home/lacos/tmp/example.dump
>          CPUS: 4
> -        DATE: Thu Sep 12 17:16:11 2013
> -      UPTIME: 00:01:09
> -LOAD AVERAGE: 0.07, 0.03, 0.00
> +        DATE: Thu Sep 12 17:17:41 2013
> +      UPTIME: 00:00:38
> +LOAD AVERAGE: 0.18, 0.05, 0.01
>         TASKS: 130
>      NODENAME: localhost.localdomain
>       RELEASE: 2.6.32-358.el6.x86_64
> @@ -38,12 +38,12 @@
>       COMMAND: "swapper"
>          TASK: ffffffff81a8d020  (1 of 4)  [THREAD_INFO: ffffffff81a00000]
>           CPU: 0
> -       STATE: TASK_RUNNING (PANIC)
> +       STATE: TASK_RUNNING (ACTIVE)
> +     WARNING: panic task not found
>
>  crash> bt
>  PID: 0      TASK: ffffffff81a8d020  CPU: 0   COMMAND: "swapper"
> - #0 [ffffffff81a01ed0] default_idle at ffffffff8101495d
> - #1 [ffffffff81a01ef0] cpu_idle at ffffffff81009fc6
> + #0 [ffffffff81a01ef0] cpu_idle at ffffffff81009fc6
>  crash> task ffffffff81a8d020
>  PID: 0      TASK: ffffffff81a8d020  CPU: 0   COMMAND: "swapper"
>  struct task_struct {
> @@ -75,7 +75,7 @@
>        prev = 0xffffffff81a8d080
>      },
>      on_rq = 0,
> -    exec_start = 8618466836,
> +    exec_start = 7469214014,
>      sum_exec_runtime = 0,
>      vruntime = 0,
>      prev_sum_exec_runtime = 0,
> @@ -149,7 +149,7 @@
>    },
>    tasks = {
>      next = 0xffff88002d621948,
> -    prev = 0xffff880029618f28
> +    prev = 0xffff880023b74488
>    },
>    pushable_tasks = {
>      prio = 140,
> @@ -165,7 +165,7 @@
>      }
>    },
>    mm = 0x0,
> -  active_mm = 0xffff88002929b780,
> +  active_mm = 0xffff8800297eb980,
>    exit_state = 0,
>    exit_code = 0,
>    exit_signal = 0,
> @@ -177,7 +177,7 @@
>    sched_reset_on_fork = 0,
>    pid = 0,
>    tgid = 0,
> -  stack_canary = 2483693585637059287,
> +  stack_canary = 7266362296181431986,
>    real_parent = 0xffffffff81a8d020,
>    parent = 0xffffffff81a8d020,
>    children = {
> @@ -224,14 +224,14 @@
>    set_child_tid = 0x0,
>    clear_child_tid = 0x0,
>    utime = 0,
> -  stime = 3,
> +  stime = 2,
>    utimescaled = 0,
> -  stimescaled = 3,
> +  stimescaled = 2,
>    gtime = 0,
>    prev_utime = 0,
>    prev_stime = 0,
>    nvcsw = 0,
> -  nivcsw = 1000,
> +  nivcsw = 1764,
>    start_time = {
>      tv_sec = 0,
>      tv_nsec = 0

- <name_dropping>I asked for Dave Anderson's help with verifying the
  extracted vmcore, and his comments make me think I should post
  this.</name_dropping>

Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
10 years agotcg/i386: cleanup useless #ifdef
Aurelien Jarno [Wed, 6 Nov 2013 18:56:58 +0000 (19:56 +0100)]
tcg/i386: cleanup useless #ifdef

TCG_TARGET_HAS_movcond_i32 is always defined to 1 in tcg-target.h, so
remove the corresponding #ifdef #endif sequence, left from a previous
refactoring.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>
10 years agotcg/i386: use movbe instruction in qemu_ldst routines
Aurelien Jarno [Wed, 6 Nov 2013 18:51:21 +0000 (19:51 +0100)]
tcg/i386: use movbe instruction in qemu_ldst routines

The movbe instruction has been added on some Intel Atom CPUs and on
recent Intel Haswell CPUs. It allows to load/store a value and at the
same time bswap it.

This patch detects the avaibility of this instruction and when available
use it in the qemu load/store routines in replacement of load/store +
bswap. Note that for 16-bit unsigned loads, movbe + movzw is basically the
same as movzw + bswap, so the patch doesn't touch this case.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
[RTH: Reduced the number of conditionals using "movop".]
Signed-off-by: Richard Henderson <rth@twiddle.net>
10 years agotcg/i386: add support for three-byte opcodes
Aurelien Jarno [Wed, 6 Nov 2013 18:49:08 +0000 (19:49 +0100)]
tcg/i386: add support for three-byte opcodes

Add support for three-byte opcodes, starting with the 0x0f 0x38 prefix.
Use P_EXT38 as the new constant, and shift all other constants so that
P_EXT and P_EXT38 have neighbouring values.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
[RTH: Changed the name from P_EXT2 to P_EXT38.]
Signed-off-by: Richard Henderson <rth@twiddle.net>
10 years agotcg/i386: remove hardcoded P_REXW value
Aurelien Jarno [Wed, 6 Nov 2013 17:32:23 +0000 (18:32 +0100)]
tcg/i386: remove hardcoded P_REXW value

P_REXW is defined has a constant at the beginning of i386/tcg-target.c,
but the corresponding bit is later used in a harcoded way, which defeat
the purpose of a constant.

Fix that by using a conditional expression operator instead of a shift.
On x86 this actually makes the code slightly smaller as GCC does in
practice (opc >> 8) & 8 instead of (opc & 0x800) >> 8 so the constants
are smaller to load.

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>
10 years agodisas/i386.c: disassemble movbe instruction
Aurelien Jarno [Wed, 6 Nov 2013 17:31:43 +0000 (18:31 +0100)]
disas/i386.c: disassemble movbe instruction

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>
10 years agoMerge remote-tracking branch 'qemu-kvm/uq/master' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:52:44 +0000 (15:52 -0800)]
Merge remote-tracking branch 'qemu-kvm/uq/master' into staging

* qemu-kvm/uq/master:
  kvm: always update the MPX model specific register
  KVM: fix addr type for KVM_IOEVENTFD
  KVM: Retry KVM_CREATE_VM on EINTR
  mempath prefault: fix off-by-one error
  kvm: x86: Separately write feature control MSR on reset
  roms: Flush icache when writing roms to guest memory
  target-i386: clear guest TSC on reset
  target-i386: do not special case TSC writeback
  target-i386: Intel MPX

Conflicts:
exec.c

aliguori: fix trivial merge conflict in exec.c

Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'otubo/seccomp' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:52:16 +0000 (15:52 -0800)]
Merge remote-tracking branch 'otubo/seccomp' into staging

* otubo/seccomp:
  seccomp: add some basic shared memory syscalls to the whitelist
  seccomp: add mkdir() and fchmod() to the whitelist

Message-id: 1390231004-18392-1-git-send-email-otubo@linux.vnet.ibm.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'sweil/tags/for_anthony' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:52:08 +0000 (15:52 -0800)]
Merge remote-tracking branch 'sweil/tags/for_anthony' into staging

Initial patch for QEMU GTK support on Windows

# gpg: Signature made Mon 20 Jan 2014 11:37:58 AM PST using RSA key ID FAD62069
# gpg: Can't check signature: public key not found

* sweil/tags/for_anthony:
  gtk: Support keyboard translation for hosts running Windows

Message-id: 1390246909-18757-1-git-send-email-sw@weilnetz.de
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'kraxel/tags/pull-audio-2' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:51:38 +0000 (15:51 -0800)]
Merge remote-tracking branch 'kraxel/tags/pull-audio-2' into staging

hda-codec: disable streams on reset

# gpg: Signature made Tue 21 Jan 2014 02:17:12 AM PST using RSA key ID D3E87138
# gpg: Can't check signature: public key not found

* kraxel/tags/pull-audio-2:
  hda-codec: disable streams on reset

Message-id: 1390299589-5082-1-git-send-email-kraxel@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'kraxel/tags/pull-usb-2' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:51:23 +0000 (15:51 -0800)]
Merge remote-tracking branch 'kraxel/tags/pull-usb-2' into staging

usb core+hid: add support for microsoft os descriptors

# gpg: Signature made Tue 21 Jan 2014 02:21:29 AM PST using RSA key ID D3E87138
# gpg: Can't check signature: public key not found

* kraxel/tags/pull-usb-2:
  usb-hid: add microsoft os descriptor support
  usb: add support for microsoft os descriptors

Message-id: 1390299772-5368-1-git-send-email-kraxel@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'bonzini/scsi-next' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:50:14 +0000 (15:50 -0800)]
Merge remote-tracking branch 'bonzini/scsi-next' into staging

* bonzini/scsi-next:
  scsi: Support TEST UNIT READY in the dummy LUN0
  block: add .bdrv_reopen_prepare() stub for iscsi
  virtio-scsi: Prevent assertion on missed events
  virtio-scsi: Cleanup of I/Os that never started
  scsi: Assign cancel_io vector for scsi_disk_emulate_ops

Conflicts:
block/iscsi.c

aliguori: resolve trivial merge conflict in block/iscsi.c

Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoMerge remote-tracking branch 'kwolf/tags/for-anthony' into staging
Anthony Liguori [Fri, 24 Jan 2014 23:43:30 +0000 (15:43 -0800)]
Merge remote-tracking branch 'kwolf/tags/for-anthony' into staging

Block patches

# gpg: Signature made Fri 24 Jan 2014 08:40:53 AM PST using RSA key ID C88F2FD6
# gpg: Can't check signature: public key not found

* kwolf/tags/for-anthony: (93 commits)
  block: Switch bdrv_io_limits_intercept() to byte granularity
  qemu-iotests: Test pwritev RMW logic
  qemu-io: New command 'sleep'
  blkdebug: Make required alignment configurable
  iscsi: Set bs->request_alignment
  block: Make bdrv_pwrite() a bdrv_prwv_co() wrapper
  block: Make bdrv_pread() a bdrv_prwv_co() wrapper
  block: Change coroutine wrapper to byte granularity
  block: Assert serialisation assumptions in pwritev
  block: Align requests in bdrv_co_do_pwritev()
  block: Allow wait_serialising_requests() at any point
  block: Make overlap range for serialisation dynamic
  block: Generalise and optimise COR serialisation
  block: Make zero-after-EOF work with larger alignment
  block: Allow waiting for overlapping requests between begin/end
  block: Switch BdrvTrackedRequest to byte granularity
  block: Introduce bdrv_co_do_pwritev()
  block: write: Handle COR dependency after I/O throttling
  block: Introduce bdrv_aligned_pwritev()
  block: Introduce bdrv_co_do_preadv()
  ...

Message-id: 1390584136-24703-1-git-send-email-kwolf@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>
10 years agoblock: Switch bdrv_io_limits_intercept() to byte granularity
Kevin Wolf [Thu, 16 Jan 2014 12:29:10 +0000 (13:29 +0100)]
block: Switch bdrv_io_limits_intercept() to byte granularity

Request sizes used to be rounded down to the next sector boundary,
allowing to bypass the I/O limit. Now all requests are accounted for
with their exact byte size.

Reported-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoqemu-iotests: Test pwritev RMW logic
Kevin Wolf [Tue, 14 Jan 2014 14:37:03 +0000 (15:37 +0100)]
qemu-iotests: Test pwritev RMW logic

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoqemu-io: New command 'sleep'
Kevin Wolf [Wed, 15 Jan 2014 14:39:10 +0000 (15:39 +0100)]
qemu-io: New command 'sleep'

There is no easy way to check that a request correctly waits for a
different request. With a sleep command we can at least approximate it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
10 years agoblkdebug: Make required alignment configurable
Kevin Wolf [Tue, 14 Jan 2014 12:44:35 +0000 (13:44 +0100)]
blkdebug: Make required alignment configurable

The new 'align' option of blkdebug can be used in order to emulate
backends with a required 4k alignment on hosts which only really require
512 byte alignment.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
10 years agoiscsi: Set bs->request_alignment
Paolo Bonzini [Tue, 29 Nov 2011 11:41:35 +0000 (12:41 +0100)]
iscsi: Set bs->request_alignment

The iSCSI backend already gets the block size from the READ CAPACITY
command it sends.  Save it so that the generic block layer gets it
too.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoblock: Make bdrv_pwrite() a bdrv_prwv_co() wrapper
Kevin Wolf [Thu, 5 Dec 2013 11:34:02 +0000 (12:34 +0100)]
block: Make bdrv_pwrite() a bdrv_prwv_co() wrapper

Instead of implementing the alignment adjustment here, use the now
existing functionality of bdrv_co_do_pwritev().

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoblock: Make bdrv_pread() a bdrv_prwv_co() wrapper
Kevin Wolf [Thu, 5 Dec 2013 11:29:59 +0000 (12:29 +0100)]
block: Make bdrv_pread() a bdrv_prwv_co() wrapper

Instead of implementing the alignment adjustment here, use the now
existing functionality of bdrv_co_do_preadv().

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoblock: Change coroutine wrapper to byte granularity
Kevin Wolf [Thu, 5 Dec 2013 11:09:38 +0000 (12:09 +0100)]
block: Change coroutine wrapper to byte granularity

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoblock: Assert serialisation assumptions in pwritev
Kevin Wolf [Tue, 14 Jan 2014 10:41:35 +0000 (11:41 +0100)]
block: Assert serialisation assumptions in pwritev

If a request calls wait_serialising_requests() and actually has to wait
in this function (i.e. a coroutine yield), other requests can run and
previously read data (like the head or tail buffer) could become
outdated. In this case, we would have to restart from the beginning to
read in the updated data.

However, we're lucky and don't actually need to do that: A request can
only wait in the first call of wait_serialising_requests() because we
mark it as serialising before that call, so any later requests would
wait. So as we don't wait in practice, we don't have to reload the data.

This is an important assumption that may not be broken or data
corruption will happen. Document it with some assertions.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoblock: Align requests in bdrv_co_do_pwritev()
Kevin Wolf [Tue, 3 Dec 2013 15:34:41 +0000 (16:34 +0100)]
block: Align requests in bdrv_co_do_pwritev()

This patch changes bdrv_co_do_pwritev() to actually be what its name
promises. If requests aren't properly aligned, it performs a RMW.

Requests touching the same block are serialised against the RMW request.
Further optimisation of this is possible by differentiating types of
requests (concurrent reads should actually be okay here).

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Allow wait_serialising_requests() at any point
Kevin Wolf [Fri, 13 Dec 2013 12:04:35 +0000 (13:04 +0100)]
block: Allow wait_serialising_requests() at any point

We can only have a single wait_serialising_requests() call per request
because otherwise we can run into deadlocks where requests are waiting
for each other. The same is true when wait_serialising_requests() is not
at the very beginning of a request, so that other requests can be issued
between the start of the tracking and wait_serialising_requests().

Fix this by changing wait_serialising_requests() to ignore requests that
are already (directly or indirectly) waiting for the calling request.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Make overlap range for serialisation dynamic
Kevin Wolf [Wed, 4 Dec 2013 16:08:50 +0000 (17:08 +0100)]
block: Make overlap range for serialisation dynamic

Copy on Read wants to serialise with all requests touching the same
cluster, so wait_serialising_requests() rounded to cluster boundaries.
Other users like alignment RMW will have different requirements, though
(requests touching the same sector), so make it dynamic.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Generalise and optimise COR serialisation
Kevin Wolf [Wed, 4 Dec 2013 15:43:44 +0000 (16:43 +0100)]
block: Generalise and optimise COR serialisation

Change the API so that specific requests can be marked serialising. Only
these requests are checked for overlaps then.

This means that during a Copy on Read operation, not all requests
overlapping other requests are serialised any more, but only those that
actually overlap with the specific COR request.

Also remove COR from function and variable names because this
functionality can be useful in other contexts.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Make zero-after-EOF work with larger alignment
Kevin Wolf [Wed, 4 Dec 2013 11:13:10 +0000 (12:13 +0100)]
block: Make zero-after-EOF work with larger alignment

Odd file sizes could make bdrv_aligned_preadv() shorten the request in
non-aligned ways. Fix it by rounding to the required alignment instead
of 512 bytes.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Allow waiting for overlapping requests between begin/end
Kevin Wolf [Tue, 3 Dec 2013 13:55:55 +0000 (14:55 +0100)]
block: Allow waiting for overlapping requests between begin/end

Previously, it was not possible to use wait_for_overlapping_requests()
between tracked_request_begin()/end() because it would wait for itself.

Ignore the current request in the overlap check and run more of the
bdrv_co_do_preadv/pwritev code with a BdrvTrackedRequest present.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Switch BdrvTrackedRequest to byte granularity
Kevin Wolf [Tue, 3 Dec 2013 14:31:25 +0000 (15:31 +0100)]
block: Switch BdrvTrackedRequest to byte granularity

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Introduce bdrv_co_do_pwritev()
Kevin Wolf [Tue, 3 Dec 2013 13:40:18 +0000 (14:40 +0100)]
block: Introduce bdrv_co_do_pwritev()

This is going to become the bdrv_co_do_preadv() equivalent for writes.
In this patch, however, just a function taking byte offsets is created,
it doesn't align anything yet.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: write: Handle COR dependency after I/O throttling
Kevin Wolf [Tue, 3 Dec 2013 13:30:44 +0000 (14:30 +0100)]
block: write: Handle COR dependency after I/O throttling

First waiting for all COR requests to complete and calling the
throttling function afterwards means that the request could be delayed
and we still need to wait for the COR request even if it was issued only
after the throttled write request.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Introduce bdrv_aligned_pwritev()
Kevin Wolf [Tue, 3 Dec 2013 13:02:23 +0000 (14:02 +0100)]
block: Introduce bdrv_aligned_pwritev()

This separates the part of bdrv_co_do_writev() that needs to happen
before the request is modified to match the backend alignment, and a
part that needs to be executed afterwards and passes the request to the
BlockDriver.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Introduce bdrv_co_do_preadv()
Kevin Wolf [Mon, 2 Dec 2013 15:09:46 +0000 (16:09 +0100)]
block: Introduce bdrv_co_do_preadv()

Similar to bdrv_pread(), which aligns byte-aligned request to 512 byte
sectors, bdrv_co_do_preadv() takes a byte-aligned request and aligns it
to the alignment specified in bs->request_alignment.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
10 years agoblock: Introduce bdrv_aligned_preadv()
Kevin Wolf [Mon, 2 Dec 2013 14:07:48 +0000 (15:07 +0100)]
block: Introduce bdrv_aligned_preadv()

This separates the part of bdrv_co_do_readv() that needs to happen
before the request is modified to match the backend alignment, and a
part that needs to be executed afterwards and passes the request to the
BlockDriver.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
10 years agoraw: Probe required direct I/O alignment
Paolo Bonzini [Tue, 29 Nov 2011 11:42:20 +0000 (12:42 +0100)]
raw: Probe required direct I/O alignment

Add a bs->request_alignment field that contains the required
offset/length alignment for I/O requests and fill it in the raw block
drivers. Use ioctls if possible, else see what alignment it takes for
O_DIRECT to succeed.

While at it, also expose the memory alignment requirements, which may be
(and in practice are) different from the disk alignment requirements.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>