sdk/emulator/qemu.git
8 years agomemory: Drop MemoryRegion.ram_addr
Fam Zheng [Tue, 1 Mar 2016 06:18:21 +0000 (14:18 +0800)]
memory: Drop MemoryRegion.ram_addr

All references to mr->ram_addr are replaced by
memory_region_get_ram_addr(mr) (except for a few assertions that are
replaced with mr->ram_block).

Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1456813104-25902-5-git-send-email-famz@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: Implement memory_region_get_ram_addr with mr->ram_block
Fam Zheng [Tue, 1 Mar 2016 06:18:20 +0000 (14:18 +0800)]
memory: Implement memory_region_get_ram_addr with mr->ram_block

Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1456813104-25902-4-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agomemory: Move assignment to ram_block to memory_region_init_*
Fam Zheng [Tue, 1 Mar 2016 06:18:19 +0000 (14:18 +0800)]
memory: Move assignment to ram_block to memory_region_init_*

We don't force "const" qualifiers with pointers in QEMU, but it's still
good to keep a clean function interface. Assigning to mr->ram_block is
in this sense ugly - one initializer mutating its owning object's state.

Move it to memory_region_init_*, where mr->ram_addr is assigned.

Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1456813104-25902-3-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoexec: Return RAMBlock pointer from allocating functions
Fam Zheng [Tue, 1 Mar 2016 06:18:18 +0000 (14:18 +0800)]
exec: Return RAMBlock pointer from allocating functions

Previously we return RAMBlock.offset; now return the pointer to the
whole structure.

ram_block_add returns void now, error is completely passed with errp.

Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <1456813104-25902-2-git-send-email-famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoi8257: fix Terminal Count status
Hervé Poussineau [Thu, 25 Feb 2016 12:45:32 +0000 (13:45 +0100)]
i8257: fix Terminal Count status

When a DMA transfer is done (ie all bytes have been transfered), the corresponding
Terminal Count bit must be set in the status register.
This bit is already cleared in i8257_read_cont and i8257_write_cont when required.

This fixes (at least) floppy transfer in IBM 40p firmware, which checks in DMA
controller if everything went fine.

Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
Message-Id: <1456404332-31556-1-git-send-email-hpoussin@reactos.org>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agolog: do not log if QEMU is daemonized but without -D
Paolo Bonzini [Mon, 29 Feb 2016 11:18:40 +0000 (12:18 +0100)]
log: do not log if QEMU is daemonized but without -D

Commit 96c33a4 ("log: Redirect stderr to logfile if deamonized",
2016-02-22) wanted to move stderr of a daemonized QEMU to the file
specified with -D.

However, if -D was not passed, the patch had the side effect of not
redirecting stderr to /dev/null.  This happened because qemu_logfile
was set to stderr rather than the expected value of NULL.  The fix
is simply in the "if" condition of do_qemu_set_log; the "if" for
closing the file is also changed to match.

Reported-by: Jan Tomko <jtomko@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2016-03-04' into staging
Peter Maydell [Sun, 6 Mar 2016 11:53:27 +0000 (11:53 +0000)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2016-03-04' into staging

QAPI patches for 2016-03-04

# gpg: Signature made Sat 05 Mar 2016 09:47:19 GMT using RSA key ID EB918653
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>"
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>"

* remotes/armbru/tags/pull-qapi-2016-03-04:
  qapi: Drop useless 'data' member of unions
  chardev: Drop useless ChardevDummy type
  qapi: Avoid use of 'data' member of QAPI unions
  ui: Shorten references into InputEvent
  util: Shorten references into SocketAddress
  chardev: Shorten references into ChardevBackend
  qapi: Update docs to match recent generator changes
  qapi-visit: Expose visit_type_FOO_members()
  qapi: Rename 'fields' to 'members' in generated C code
  qapi: Rename 'fields' to 'members' in generator
  qapi-dealloc: Reduce use outside of generated code
  qmp-shell: fix pretty printing of JSON responses

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoqapi: Drop useless 'data' member of unions
Eric Blake [Thu, 3 Mar 2016 16:16:52 +0000 (09:16 -0700)]
qapi: Drop useless 'data' member of unions

We started moving away from the use of the 'void *data' member
in the C union corresponding to a QAPI union back in commit
544a373; recent commits have gotten rid of other uses.  Now
that it is completely unused, we can remove the member itself
as well as the FIXME comment.  Update the testsuite to drop the
negative test union-clash-data.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1457021813-10704-11-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agochardev: Drop useless ChardevDummy type
Eric Blake [Fri, 4 Mar 2016 15:42:40 +0000 (08:42 -0700)]
chardev: Drop useless ChardevDummy type

Commit d0d7708b made ChardevDummy be an empty wrapper type around
ChardevCommon.  But there is no technical reason for this indirection,
so simplify the code by directly using the base type.

Also change the fallback assignment to assign u.null rather than
u.data, since a future patch will remove the data member of the C
struct generated for QAPI unions.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1457106160-23614-1-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqapi: Avoid use of 'data' member of QAPI unions
Eric Blake [Thu, 3 Mar 2016 16:16:50 +0000 (09:16 -0700)]
qapi: Avoid use of 'data' member of QAPI unions

QAPI code generators currently create a 'void *data' member as
part of the anonymous union embedded in the C struct corresponding
to a QAPI union.  However, directly assigning to this member of
the union feels a bit fishy, when we can assign to another member
of the struct instead.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1457021813-10704-9-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoui: Shorten references into InputEvent
Eric Blake [Thu, 3 Mar 2016 16:16:49 +0000 (09:16 -0700)]
ui: Shorten references into InputEvent

An upcoming patch will alter how simple unions, like InputEvent, are
laid out, which will impact all lines of the form 'evt->u.XXX'
(expanding it to the longer 'evt->u.XXX.data').  For better
legibility in that patch, and less need for line wrapping, it's better
to use a temporary variable to reduce the effect of a layout change to
just the variable initializations, rather than every reference within
an InputEvent.

There was one instance in hid.c:hid_pointer_event() where the code
was referring to evt->u.rel inside the case label where evt->u.abs
is the correct name; thankfully, both members of the union have the
same type, so it happened to work, but it is now cleaner.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457021813-10704-8-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoutil: Shorten references into SocketAddress
Eric Blake [Thu, 3 Mar 2016 16:16:48 +0000 (09:16 -0700)]
util: Shorten references into SocketAddress

An upcoming patch will alter how simple unions, like SocketAddress,
are laid out, which will impact all lines of the form 'addr->u.XXX'
(expanding it to the longer 'addr->u.XXX.data').  For better
legibility in that patch, and less need for line wrapping, it's better
to use a temporary variable to reduce the effect of a layout change to
just the variable initializations, rather than every reference within
a SocketAddress.  Also, take advantage of some C99 initialization where
it makes sense (simplifying g_new0() to g_new()).

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457021813-10704-7-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agochardev: Shorten references into ChardevBackend
Eric Blake [Thu, 3 Mar 2016 16:16:47 +0000 (09:16 -0700)]
chardev: Shorten references into ChardevBackend

An upcoming patch will alter how simple unions, like ChardevBackend,
are laid out, which will impact all lines of the form 'backend->u.XXX'
(expanding it to the longer 'backend->u.XXX.data').  For better
legibility in that patch, and less need for line wrapping, it's better
to use a temporary variable to reduce the effect of a layout change to
just the variable initializations, rather than every reference within
a ChardevBackend.  It doesn't hurt that this also makes the code more
consistent: some clients touched here already had a temporary variable
but weren't using it.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-By: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1457021813-10704-6-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqapi: Update docs to match recent generator changes
Eric Blake [Thu, 3 Mar 2016 16:16:46 +0000 (09:16 -0700)]
qapi: Update docs to match recent generator changes

Several commits have been changing the generator, but not updating
the docs to match:
- The implicit tag member is named "type", not "kind".  Screwed up in
commit 39a1815.
- Commit 9f08c8ec made list types lazy, and thereby dropped
UserDefOneList if nothing explicitly uses the list type.
- Commit 51e72bc1 switched the parameter order with 'name' occurring
earlier.
- Commit e65d89bf changed the layout of UserDefOneList.
- Prefer the term 'member' over 'field'.
- We now expose visit_type_FOO_members() for objects.
- etc.

Rework the examples to show slightly more output (we don't want to
show too much; that's what the testsuite is for), and regenerate the
output to match all recent changes.  Also, rearrange output to show
.h files before .c (understanding the interface first often makes
the implementation easier to follow).

Reported-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <1457021813-10704-5-git-send-email-eblake@redhat.com>

8 years agoqapi-visit: Expose visit_type_FOO_members()
Eric Blake [Thu, 3 Mar 2016 16:16:45 +0000 (09:16 -0700)]
qapi-visit: Expose visit_type_FOO_members()

Dan Berrange reported a case where he needs to work with a
QCryptoBlockOptions union type using the OptsVisitor, but only
visit one of the branches of that type (the discriminator is not
visited directly, but learned externally).  When things were
boxed, it was easy: just visit the variant directly, which took
care of both allocating the variant and visiting its members, then
store that pointer in the union type.  But now that things are
unboxed, we need a way to visit the members without allocation,
done by exposing visit_type_FOO_members() to the user.

Before the patch, we had quite a bit of code associated with
object_members_seen to make sure that a declaration of the helper
was in scope before any use of the function.  But now that the
helper is public and declared in the header, the .c file no
longer needs to worry about topological sorting (the helper is
always in scope), which leads to some nice cleanups.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457021813-10704-4-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqapi: Rename 'fields' to 'members' in generated C code
Eric Blake [Thu, 3 Mar 2016 16:16:44 +0000 (09:16 -0700)]
qapi: Rename 'fields' to 'members' in generated C code

C types and JSON objects don't have fields, but members.  We
shouldn't gratuitously invent terminology.  This patch is a
strict renaming of static genarated functions, plus the naming
of the dummy filler member for empty structs, before the next
patch exposes some of that naming to the rest of the code base.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457021813-10704-3-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqapi: Rename 'fields' to 'members' in generator
Eric Blake [Thu, 3 Mar 2016 16:16:43 +0000 (09:16 -0700)]
qapi: Rename 'fields' to 'members' in generator

C types and JSON objects don't have fields, but members.  We
shouldn't gratuitously invent terminology.  This patch is a
strict renaming of generator code internals (including testsuite
comments), before later patches rename C interfaces.

No change to generated code with this patch.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1457021813-10704-2-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqapi-dealloc: Reduce use outside of generated code
Eric Blake [Tue, 23 Feb 2016 21:14:33 +0000 (14:14 -0700)]
qapi-dealloc: Reduce use outside of generated code

No need to roll our own use of the dealloc visitors when we can
just directly use the qapi_free_FOO() functions that do what we
want in one line.

In net.c, inline net_visit() into its remaining lone caller.

After this patch, test-visitor-serialization.c is the only
non-generated file that needs to use a dealloc visitor, because
it is testing low level aspects of the visitor interface.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1456262075-3311-2-git-send-email-eblake@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoqmp-shell: fix pretty printing of JSON responses
Daniel P. Berrange [Tue, 23 Feb 2016 10:51:46 +0000 (10:51 +0000)]
qmp-shell: fix pretty printing of JSON responses

Pretty printing of JSON responses is important to be able to understand
large responses from query commands in particular. Unfortunately this
was broken during the addition of the verbose flag in

  commit 1ceca07e48ead0dd2e41576c81d40e6a91cafefd
  Author: John Snow <jsnow@redhat.com>
  Date:   Wed Apr 29 15:14:04 2015 -0400

    scripts: qmp-shell: Add verbose flag

This is because that change turned the python data structure into a
formatted JSON string before the pretty print was given it. So we're
just pretty printing a string, which is a no-op.

The original pretty printer would output python objects.

(QEMU) query-chardev
{   u'return': [   {   u'filename': u'vc',
                       u'frontend-open': False,
                       u'label': u'parallel0'},
                   {   u'filename': u'vc',
                       u'frontend-open': True,
                       u'label': u'serial0'},
                   {   u'filename': u'unix:/tmp/qemp,server',
                       u'frontend-open': True,
                       u'label': u'compat_monitor0'}]}

This fixes the problem by switching to outputting pretty formatted JSON
text instead. This has the added benefit that the pretty printed output
is now valid JSON text. Due to the way the verbose flag was handled, the
pretty printing now applies to the command sent, as well as its response:

(QEMU) query-chardev
{
    "execute": "query-chardev",
    "arguments": {}
}
{
    "return": [
        {
            "frontend-open": false,
            "label": "parallel0",
            "filename": "vc"
        },
        {
            "frontend-open": true,
            "label": "serial0",
            "filename": "vc"
        },
        {
            "frontend-open": true,
            "label": "compat_monitor0",
            "filename": "unix:/tmp/qmp,server"
        }
    ]
}

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1456224706-1591-1-git-send-email-berrange@redhat.com>
Tested-by: Kashyap Chamarthy <kchamart@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
[Bonus fix: multiple -p now work]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20160304' into...
Peter Maydell [Fri, 4 Mar 2016 11:46:32 +0000 (11:46 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20160304' into staging

target-arm queue:
 * Correct handling of writes to CPSR from gdbstub in user mode
 * virt: lift maximum RAM limit to 255GB
 * sdhci: implement reset
 * virt: if booting in Secure mode, provide secure-only RAM, make first
   flash device secure-only, and assume the EL3 boot rom will handle PSCI
 * bcm2835: use explicit endianness accessors rather than ldl/stl_phys
 * support big-endian in system mode for ARM
 * implement SETEND instruction
 * arm_gic: implement the GICv2 GICC_DIR register
 * fix SRS bug: only trap from S-EL1 to EL3 if specified mode is Mon

# gpg: Signature made Fri 04 Mar 2016 11:38:53 GMT using RSA key ID 14360CDE
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>"
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>"
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>"

* remotes/pmaydell/tags/pull-target-arm-20160304: (30 commits)
  target-arm: Only trap SRS from S-EL1 if specified mode is MON
  hw/intc/arm_gic.c: Implement GICv2 GICC_DIR
  arm: boot: Support big-endian elfs
  loader: Add data swap option to load-elf
  loader: load_elf(): Add doc comment
  loader: add API to load elf header
  target-arm: implement BE32 mode in system emulation
  target-arm: implement setend
  target-arm: introduce tbflag for endianness
  target-arm: a64: Add endianness support
  target-arm: introduce disas flag for endianness
  target-arm: pass DisasContext to gen_aa32_ld*/st*
  target-arm: implement SCTLR.EE
  linux-user: arm: handle CPSR.E correctly in strex emulation
  linux-user: arm: set CPSR.E/SCTLR.E0E correctly for BE mode
  arm: cpu: handle BE32 user-mode as BE
  target-arm: cpu: Move cpu_is_big_endian to header
  target-arm: implement SCTLR.B, drop bswap_code
  linux-user: arm: pass env to get_user_code_*
  linux-user: arm: fix coding style for some linux-user signal functions
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: Only trap SRS from S-EL1 if specified mode is MON
Ralf-Philipp Weinmann [Fri, 4 Mar 2016 11:30:22 +0000 (11:30 +0000)]
target-arm: Only trap SRS from S-EL1 if specified mode is MON

Commit cbc0326b6fb9 caused SRS instructions executed from Secure
EL1 to trap to EL3 even if the specified mode was not monitor mode.

According to the ARMv8 Architecture reference manual [F6.1.203], ALL
of the following conditions need to be met for SRS to trap to EL3:
* It is executed at Secure PL1.
* The specified mode is monitor mode.
* EL3 is using AArch64.

Correct the condition governing the trap to EL3 to check the
specified mode.

Signed-off-by: Ralf-Philipp Weinmann <ralf+devel@comsecuris.com>
Message-id: 20160222224251.GA11654@beta.comsecuris.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: tweaked comment text to read 'specified mode'; edited
 commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/intc/arm_gic.c: Implement GICv2 GICC_DIR
Peter Maydell [Fri, 4 Mar 2016 11:30:22 +0000 (11:30 +0000)]
hw/intc/arm_gic.c: Implement GICv2 GICC_DIR

The GICv2 introduces a new CPU interface register GICC_DIR, which
allows an OS to split the "priority drop" and "deactivate interrupt"
parts of interrupt completion. Implement this register.
(Note that the register is at offset 0x1000 in the CPU interface,
which means it is on a different 4K page from all the other registers.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Sergey Fedorov <serge.fdrv@gmail.com>
Message-id: 1456854176-7813-1-git-send-email-peter.maydell@linaro.org

8 years agoarm: boot: Support big-endian elfs
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
arm: boot: Support big-endian elfs

Support ARM big-endian ELF files in system-mode emulation. When loading
an elf, determine the endianness mode expected by the elf, and set the
relevant CPU state accordingly.

With this, big-endian modes are now fully supported via system-mode LE,
so there is no need to restrict the elf loading to the TARGET
endianness so the ifdeffery on TARGET_WORDS_BIGENDIAN goes away.

Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: fix typo in comments]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoloader: Add data swap option to load-elf
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
loader: Add data swap option to load-elf

Some CPUs are of an opposite data-endianness to other components in the
system. Sometimes elfs have the data sections layed out with this CPU
data-endianness accounting for when loaded via the CPU, so byte swaps
(relative to other system components) will occur.

The leading example, is ARM's BE32 mode, which is is basically LE with
address manipulation on half-word and byte accesses to access the
hw/byte reversed address. This means that word data is invariant
across LE and BE32. This also means that instructions are still LE.
The expectation is that the elf will be loaded via the CPU in this
endianness scheme, which means the data in the elf is reversed at
compile time.

As QEMU loads via the system memory directly, rather than the CPU, we
need a mechanism to reverse elf data endianness to implement this
possibility.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoloader: load_elf(): Add doc comment
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
loader: load_elf(): Add doc comment

Document the usage of load_elf() for clarity on current features.

Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoloader: add API to load elf header
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
loader: add API to load elf header

Add an API to load an elf header header from a file. Populates a
buffer with the header contents, as well as a boolean for whether the
elf is 64b or not. Both arguments are optional.

Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: Fix typo in comment]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: implement BE32 mode in system emulation
Paolo Bonzini [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
target-arm: implement BE32 mode in system emulation

System emulation only has a little-endian target; BE32 mode
is implemented by adjusting the low bits of the address
for every byte and halfword load and store.  64-bit accesses
flip the low and high words.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[PC changes:
  * rebased against master (Jan 2016)
]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: implement setend
Paolo Bonzini [Fri, 4 Mar 2016 11:30:21 +0000 (11:30 +0000)]
target-arm: implement setend

Since this is not a high-performance path, just use a helper to
flip the E bit and force a lookup in the hash table since the
flags have changed.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: introduce tbflag for endianness
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:20 +0000 (11:30 +0000)]
target-arm: introduce tbflag for endianness

Introduce a tbflags for endianness, set based upon the CPUs current
endianness. This in turn propagates through to the disas endianness
flag.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: a64: Add endianness support
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:20 +0000 (11:30 +0000)]
target-arm: a64: Add endianness support

Set the dc->mo_endianness flag for AA64 and use it in all ldst ops.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: introduce disas flag for endianness
Paolo Bonzini [Fri, 4 Mar 2016 11:30:20 +0000 (11:30 +0000)]
target-arm: introduce disas flag for endianness

Introduce a disas flag for setting the CPU data endianness. This allows
control of the endianness from the CPU state rather than hard-coding it
to TARGET_WORDS_BIGENDIAN.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[ PC changes:
  * Split off as new patch from original:
        "target-arm: introduce tbflag for CPSR.E"
  * Wrote commit message from scratch
]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: pass DisasContext to gen_aa32_ld*/st*
Paolo Bonzini [Fri, 4 Mar 2016 11:30:20 +0000 (11:30 +0000)]
target-arm: pass DisasContext to gen_aa32_ld*/st*

We'll need the DisasContext in the next patch to retrieve the
desired endianness, so pass it as a whole to gen_aa32_ld*/st*.

Unfortunately we cannot let those functions call get_mem_index,
because of user-mode load/store instructions.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[ PC changes:
 * Fix long lines
]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: implement SCTLR.EE
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:20 +0000 (11:30 +0000)]
target-arm: implement SCTLR.EE

Implement SCTLR.EE bit which controls data endianess for exceptions
and page table translations. SCTLR.EE is mirrored to the CPSR.E bit
on exception entry.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: arm: handle CPSR.E correctly in strex emulation
Paolo Bonzini [Fri, 4 Mar 2016 11:30:19 +0000 (11:30 +0000)]
linux-user: arm: handle CPSR.E correctly in strex emulation

Now that CPSR.E is set correctly, prepare for when setend will be able
to change it; bswap data in and out of strex manually by comparing
SCTLR.B, CPSR.E and TARGET_WORDS_BIGENDIAN (we do not have the luxury
of using TCGMemOps).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[ PC changes:
  * Moved SCTLR/CPSR logic to arm_cpu_data_is_big_endian
]
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: arm: set CPSR.E/SCTLR.E0E correctly for BE mode
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:19 +0000 (11:30 +0000)]
linux-user: arm: set CPSR.E/SCTLR.E0E correctly for BE mode

If doing big-endian linux-user mode, set both the CPSR.E and SCTLR.E0E
bits. This sets big-endian mode for data accesses.

Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoarm: cpu: handle BE32 user-mode as BE
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:19 +0000 (11:30 +0000)]
arm: cpu: handle BE32 user-mode as BE

endian with address manipulations on subword accesses (to give the
illusion of BE). But user-mode cannot tell the difference and is
already implemented as straight BE. So handle the difference in the
endianess query, where USER mode is BE and system is not.

Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: cpu: Move cpu_is_big_endian to header
Peter Crosthwaite [Fri, 4 Mar 2016 11:30:19 +0000 (11:30 +0000)]
target-arm: cpu: Move cpu_is_big_endian to header

There is a CPU data endianness test that is used to drive the
virtio_big_endian test.

Move this up to the header so it can be more generally used for endian
tests. The KVM specific cpu_syncronize_state call is left behind in the
virtio specific function.

Rename it arm_cpu-data_is_big_endian() to more accurately capture that
this is for data accesses only.

Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Signed-off-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotarget-arm: implement SCTLR.B, drop bswap_code
Paolo Bonzini [Fri, 4 Mar 2016 11:30:19 +0000 (11:30 +0000)]
target-arm: implement SCTLR.B, drop bswap_code

bswap_code is a CPU property of sorts ("is the iside endianness the
opposite way round to TARGET_WORDS_BIGENDIAN?") but it is not the
actual CPU state involved here which is SCTLR.B (set for BE32
binaries, clear for BE8).

Replace bswap_code with SCTLR.B, and pass that to arm_ld*_code.
The next patches will make data fetches honor both SCTLR.B and
CPSR.E appropriately.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[PC changes:
 * rebased on master (Jan 2016)
 * s/TARGET_USER_ONLY/CONFIG_USER_ONLY
 * Use bswap_code() for disas_set_info() instead of raw sctlr_b
]
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: arm: pass env to get_user_code_*
Paolo Bonzini [Fri, 4 Mar 2016 11:30:18 +0000 (11:30 +0000)]
linux-user: arm: pass env to get_user_code_*

This matches the idiom used by get_user_data_* later in the series,
and will help when bswap_code will be replaced by SCTLR.B.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agolinux-user: arm: fix coding style for some linux-user signal functions
Paolo Bonzini [Fri, 4 Mar 2016 11:30:18 +0000 (11:30 +0000)]
linux-user: arm: fix coding style for some linux-user signal functions

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agobcm2835_mbox/property: replace ldl_phys/stl_phys with endian-specific accesses
Andrew Baumann [Fri, 4 Mar 2016 11:30:18 +0000 (11:30 +0000)]
bcm2835_mbox/property: replace ldl_phys/stl_phys with endian-specific accesses

PMM pointed out that ldl_phys and stl_phys are dependent on the CPU's
endianness, whereas device model code should be independent of
it. This changes the relevant Raspberry Pi devices to explicitly call
the little-endian variants.

Signed-off-by: Andrew Baumann <Andrew.Baumann@microsoft.com>
Message-id: 1456880233-22568-1-git-send-email-Andrew.Baumann@microsoft.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agohw/arm/virt: Assume EL3 boot rom will handle PSCI if one is provided
Peter Maydell [Fri, 4 Mar 2016 11:30:18 +0000 (11:30 +0000)]
hw/arm/virt: Assume EL3 boot rom will handle PSCI if one is provided

If the user passes us an EL3 boot rom, then it is going to want to
implement the PSCI interface itself. In this case, disable QEMU's
internal PSCI implementation so it does not get in the way, and
instead start all CPUs in an SMP configuration at once (the boot
rom will catch them all and pen up the secondaries until needed).
The boot rom code is also responsible for editing the device tree
to include any necessary information about its own PSCI implementation
before eventually passing it to a NonSecure guest.

(This "start all CPUs at once" approach is what both ARM Trusted
Firmware and UEFI expect, since it is what the ARM Foundation Model
does; the other approach would be to provide some emulated hardware
for "start the secondaries" but this is simplest.)

This is a compatibility break, but I don't believe that anybody
was using a secure boot ROM with an SMP configuration. Such a setup
would be somewhat broken since there was nothing preventing nonsecure
guest code from calling the QEMU PSCI function to start up a secondary
core in a way that completely bypassed the secure world.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-id: 1456853976-7592-1-git-send-email-peter.maydell@linaro.org

8 years agohw/arm/virt: Make first flash device Secure-only if booting secure
Peter Maydell [Fri, 4 Mar 2016 11:30:18 +0000 (11:30 +0000)]
hw/arm/virt: Make first flash device Secure-only if booting secure

If the virt board is started with the 'secure' property set to
request a Secure setup, then make the first flash device be
visible only to the Secure world.

This is a breaking change, but I don't expect it to be noticed
by anybody, because running TZ-aware guests isn't common and
those guests are generally going to be booting from the flash
and implicitly expecting their Non-secure guests to not touch it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1455288361-30117-5-git-send-email-peter.maydell@linaro.org

8 years agohw/arm/virt: Load bios image to MemoryRegion, not physaddr
Peter Maydell [Fri, 4 Mar 2016 11:30:17 +0000 (11:30 +0000)]
hw/arm/virt: Load bios image to MemoryRegion, not physaddr

If we're loading a BIOS image into the first flash device,
load it into the flash's memory region specifically, not
into the physical address where the flash resides. This will
make a difference when the flash might be in the Secure
address space rather than the Nonsecure one.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1455288361-30117-4-git-send-email-peter.maydell@linaro.org

8 years agoloader: Add load_image_mr() to load ROM image to a MemoryRegion
Peter Maydell [Fri, 4 Mar 2016 11:30:17 +0000 (11:30 +0000)]
loader: Add load_image_mr() to load ROM image to a MemoryRegion

Add a new function load_image_mr(), which behaves like
load_image_targphys() except that it loads the ROM image to
a specified MemoryRegion rather than to a specified physical
address. This is useful when a ROM blob needs to be loaded
to a particular flash or ROM device but the address of that
device in the machine's address space is not known. (For
instance, ROMs in devices, or ROMs which might exist in
a different address space to the system address space.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1455288361-30117-3-git-send-email-peter.maydell@linaro.org
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
8 years agohw/arm/virt: Provide a secure-only RAM if booting in Secure mode
Peter Maydell [Fri, 4 Mar 2016 11:30:17 +0000 (11:30 +0000)]
hw/arm/virt: Provide a secure-only RAM if booting in Secure mode

If we're booting in Secure mode, provide a secure-only RAM
(just 16MB) so that secure firmware has somewhere to run
from that won't be accessible to the Non-secure guest.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1455288361-30117-2-git-send-email-peter.maydell@linaro.org

8 years agosdhci: Implement DeviceClass reset
Peter Maydell [Fri, 4 Mar 2016 11:30:17 +0000 (11:30 +0000)]
sdhci: Implement DeviceClass reset

The sdhci device was missing a DeviceClass reset method;
implement it. Poweron reset looks the same as reset commanded
by the guest via the device registers, apart from modelling of
the rpi 'pending insert interrupt on powerup' quirk.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Andrew Baumann <Andrew.Baumann@microsoft.com>
Message-id: 1456493044-10025-3-git-send-email-peter.maydell@linaro.org

8 years agosd.c: Handle NULL block backend in sd_get_inserted()
Peter Maydell [Fri, 4 Mar 2016 11:30:17 +0000 (11:30 +0000)]
sd.c: Handle NULL block backend in sd_get_inserted()

The sd.c SD card emulation code can be in a state where the
SDState BlockBackend pointer is NULL; this is treated as
"card not present". Add a missing check to sd_get_inserted()
so that we don't segfault in this situation.

(This could be provoked by the guest writing to the SDHCI
register to do a reset on a xilinx-zynq-a9 board; it will
also happen at startup when sdhci implements its DeviceClass
reset method.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
Message-id: 1456493044-10025-2-git-send-email-peter.maydell@linaro.org

8 years agovirt: Lift the maximum RAM limit from 30GB to 255GB
Peter Maydell [Fri, 4 Mar 2016 11:30:16 +0000 (11:30 +0000)]
virt: Lift the maximum RAM limit from 30GB to 255GB

The virt board restricts guests to only 30GB of RAM. This is a
hangover from the vexpress-a15 board, and there's no inherent reason
for it. 30GB is smaller than you might reasonably want to provision
a VM for on a beefy server machine. Raise the limit to 255GB.

We choose 255GB because the available space we currently have
below the 1TB boundary is up to the 512GB mark, but we don't
want to paint ourselves into a corner by assigning it all to
RAM. So we make half of it available for RAM, with the 256GB..512GB
range available for future non-RAM expansion purposes.

If we need to provide more RAM to VMs in the future then we need to:
 * allocate a second bank of RAM starting at 2TB and working up
 * fix the DT and ACPI table generation code in QEMU to correctly
   report two split lumps of RAM to the guest
 * fix KVM in the host kernel to allow guests with >40 bit address spaces

The last of these is obviously the trickiest, but it seems
reasonable to assume that anybody configuring a VM with a quarter
of a terabyte of RAM will be doing it on a host with more than a
terabyte of physical address space.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Tested-by: Wei Huang <wei@redhat.com>
Message-id: 1456402182-11651-1-git-send-email-peter.maydell@linaro.org

8 years agotarget-arm: Correct handling of writes to CPSR mode bits from gdb in usermode
Peter Maydell [Fri, 4 Mar 2016 11:30:16 +0000 (11:30 +0000)]
target-arm: Correct handling of writes to CPSR mode bits from gdb in usermode

In helper.c the expression
  (env->uncached_cpsr & CPSR_M) != CPSR_USER
is always true; the right hand side was supposed to be ARM_CPU_MODE_USR
(an error in commit cb01d391).

Since the incorrect expression was always true, this just meant that
commit cb01d391 had no effect.

However simply changing the RHS here would reveal a logic error: if
the mode is USR we wish to completely ignore the attempt to set the
mode bits, which means that we must clear the CPSR_M bits from mask
to avoid the uncached_cpsr bits being updated at the end of the
function.

Move the condition into the correct place in the code, fix its RHS
constant, and add a comment about the fact that we must be doing a
gdbstub write if we're in user mode.

Fixes: https://bugs.launchpad.net/qemu/+bug/1550503
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Sergey Fedorov <serge.fdrv@gmail.com>
Message-id: 1456764438-30015-1-git-send-email-peter.maydell@linaro.org

8 years agoMerge remote-tracking branch 'remotes/amit-virtio-rng/tags/rng-for-2.6-1' into staging
Peter Maydell [Thu, 3 Mar 2016 13:13:35 +0000 (13:13 +0000)]
Merge remote-tracking branch 'remotes/amit-virtio-rng/tags/rng-for-2.6-1' into staging

rng:
- implement a request queue for rng-random so multiple guest requests
  don't result in vq buffers getting forgotten
- remove unused request cancellation code
- a VM with multiple vq buffers, when migrated, could get in a situation
  where not all buffers are handed back to the guest.  This is now
  fixed.

# gpg: Signature made Thu 03 Mar 2016 12:18:54 GMT using RSA key ID 854083B6
# gpg: Good signature from "Amit Shah <amit@amitshah.net>"
# gpg:                 aka "Amit Shah <amit@kernel.org>"
# gpg:                 aka "Amit Shah <amitshah@gmx.net>"

* remotes/amit-virtio-rng/tags/rng-for-2.6-1:
  virtio-rng: ask for more data if queue is not fully drained
  rng: add request queue support to rng-random
  rng: move request queue cleanup from RngEgd to RngBackend
  rng: move request queue from RngEgd to RngBackend
  rng: remove the unused request cancellation code
  MAINTAINERS: Add an entry for the include/sysemu/rng*.h files

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agovirtio-rng: ask for more data if queue is not fully drained
Ladi Prosek [Thu, 3 Mar 2016 09:48:34 +0000 (10:48 +0100)]
virtio-rng: ask for more data if queue is not fully drained

This commit effectively reverts:

  commit 4621c1768ef5d12171cca2aa1473595ecb9f1c9e
  Author: Amit Shah <amit.shah@redhat.com>
  Date:   Wed Nov 21 11:21:19 2012 +0530

  virtio-rng: remove extra request for entropy

but instead of calling virtio_rng_process unconditionally, it
first checks to see if the queue is empty as a little bit of
optimization.

Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456998514-19271-1-git-send-email-lprosek@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agorng: add request queue support to rng-random
Ladi Prosek [Thu, 3 Mar 2016 08:37:18 +0000 (09:37 +0100)]
rng: add request queue support to rng-random

Requests are now created in the RngBackend parent class and the
code path is shared by both rng-egd and rng-random.

This commit fixes the rng-random implementation which processed
only one request at a time and simply discarded all but the most
recent one. In the guest this manifested as delayed completion
of reads from virtio-rng, i.e. a read was completed only after
another read was issued.

By switching rng-random to use the same request queue as rng-egd,
the unsafe stack-based allocation of the entropy buffer is
eliminated and replaced with g_malloc.

Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456994238-9585-5-git-send-email-lprosek@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agorng: move request queue cleanup from RngEgd to RngBackend
Ladi Prosek [Thu, 3 Mar 2016 08:37:17 +0000 (09:37 +0100)]
rng: move request queue cleanup from RngEgd to RngBackend

RngBackend is now in charge of cleaning up the linked list on
instance finalization. It also exposes a function to finalize
individual RngRequest instances, called by its child classes.

Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456994238-9585-4-git-send-email-lprosek@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agorng: move request queue from RngEgd to RngBackend
Ladi Prosek [Thu, 3 Mar 2016 08:37:16 +0000 (09:37 +0100)]
rng: move request queue from RngEgd to RngBackend

The 'requests' field now lives in the RngBackend parent class.
There are no functional changes in this commit.

Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456994238-9585-3-git-send-email-lprosek@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agorng: remove the unused request cancellation code
Ladi Prosek [Thu, 3 Mar 2016 08:37:15 +0000 (09:37 +0100)]
rng: remove the unused request cancellation code

rng_backend_cancel_requests had no callers and none of the code
deleted in this commit ever ran.

Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456994238-9585-2-git-send-email-lprosek@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoMAINTAINERS: Add an entry for the include/sysemu/rng*.h files
Thomas Huth [Thu, 25 Feb 2016 12:44:20 +0000 (13:44 +0100)]
MAINTAINERS: Add an entry for the include/sysemu/rng*.h files

These headers are used by the virtio-rng and rng backends code,
so they should be listed in the same section in MAINTAINERS, too.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Message-Id: <1456404260-26928-1-git-send-email-thuth@redhat.com>
Signed-off-by: Amit Shah <amit.shah@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/stefanha/tags/tracing-pull-request' into staging
Peter Maydell [Tue, 1 Mar 2016 15:54:03 +0000 (15:54 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/tracing-pull-request' into staging

# gpg: Signature made Tue 01 Mar 2016 15:48:04 GMT using RSA key ID 81AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>"
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>"

* remotes/stefanha/tags/tracing-pull-request:
  trace: Add a proper API to manage auto-generated events from the 'tcg' property
  trace: Add 'vcpu' event property to trace guest vCPU
  typedefs: Add CPUState
  trace: Add helper function to cast event arguments
  tcg: Move definition of type TCGv
  tcg: Add type for vCPU pointers
  trace: Remove unnecessary intermediate event copies
  trace: Extend API to manage event arguments
  vl: fix tracing initialization
  trace: use addresses instead of offsets in memory tracepoints
  trace: split subpage MMIOs into their own trace events.
  trace: docs: "simple" backend does support strings
  trace: drop trailing empty strings

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agotrace: Add a proper API to manage auto-generated events from the 'tcg' property
Lluís Vilanova [Thu, 25 Feb 2016 13:06:30 +0000 (14:06 +0100)]
trace: Add a proper API to manage auto-generated events from the 'tcg' property

Formalizes the existence of the 'event_trans' and 'event_exec' event
attributes, which until now were monkey-patched only when necessary.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145640558759.20978.6374959404425591089.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: Add 'vcpu' event property to trace guest vCPU
Lluís Vilanova [Thu, 25 Feb 2016 16:43:38 +0000 (17:43 +0100)]
trace: Add 'vcpu' event property to trace guest vCPU

This property identifies events that trace vCPU-specific information.

It adds a "CPUState*" argument to events with the property, identifying
the vCPU raising the event. TCG translation events also have a
"TCGv_env" implicit argument that is later used as the "CPUState*"
argument at execution time.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145641861797.30295.6991314023181842105.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotypedefs: Add CPUState
Lluís Vilanova [Thu, 25 Feb 2016 16:43:32 +0000 (17:43 +0100)]
typedefs: Add CPUState

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145641861239.30295.8564457138934628740.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: Add helper function to cast event arguments
Lluís Vilanova [Thu, 25 Feb 2016 16:43:27 +0000 (17:43 +0100)]
trace: Add helper function to cast event arguments

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145641860680.30295.1873612736245870753.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotcg: Move definition of type TCGv
Lluís Vilanova [Thu, 25 Feb 2016 16:43:21 +0000 (17:43 +0100)]
tcg: Move definition of type TCGv

The target-dependant type TCGv must be defined in "tcg/tcg.h" before
including the tracing helper wrappers in "tcg/tcg-op.h".

It also makes more sense to define it here, where other TCG types are
defined too.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145641860129.30295.17554707227384022653.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotcg: Add type for vCPU pointers
Lluís Vilanova [Thu, 25 Feb 2016 16:43:15 +0000 (17:43 +0100)]
tcg: Add type for vCPU pointers

Adds the 'TCGv_env' type for pointers to 'CPUArchState' objects. The
tracing infrastructure later needs to differentiate between regular
pointers and pointers to vCPUs.

Also changes all targets to use the new 'TCGv_env' type instead of the
generic 'TCGv_ptr'. As of now, the change is merely cosmetic ('TCGv_env'
translates into 'TCGv_ptr'), but that could change in the future to
enforce the difference.

Note that a 'TCGv_env' type (for 'CPUState') is not added, since all
helpers currently receive the architecture-specific
pointer ('CPUArchState').

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Acked-by: Richard Henderson <rth@twiddle.net>
Message-id: 145641859552.30295.7821536833590725201.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: Remove unnecessary intermediate event copies
Lluís Vilanova [Thu, 25 Feb 2016 16:43:10 +0000 (17:43 +0100)]
trace: Remove unnecessary intermediate event copies

The current code forces the use of a chain of ".original" dereferences,
which looks odd.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Message-id: 145641858988.30295.7223459456488075843.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: Extend API to manage event arguments
Lluís Vilanova [Thu, 25 Feb 2016 16:43:04 +0000 (17:43 +0100)]
trace: Extend API to manage event arguments

Lets the user manage event arguments as a list, and simplifies argument
concatenation.

Signed-off-by: Lluís Vilanova <vilanova@ac.upc.edu>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 145641858432.30295.3069911069472672646.stgit@localhost
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agovl: fix tracing initialization
Denis V. Lunev [Tue, 9 Feb 2016 16:49:05 +0000 (19:49 +0300)]
vl: fix tracing initialization

we should call trace_init_backends() before trace_init_file() for
CONFIG_TRACE_SIMPLE There is no difference for other cases.

This problem was introduced by the commit
    commit 41fc57e44ed64cd4ab5393d83624afd897dabd4f
    Author: Paolo Bonzini <pbonzini@redhat.com>
    Date:   Thu Jan 7 16:55:24 2016 +0300

    trace: split trace_init_file out of trace_init_backends

'make check' was failed as a result if configured with
  --enable-trace-backends=simple

Spotted by Alex Bennée.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Message-id: 1455036545-14870-1-git-send-email-den@openvz.org
CC: Alex Bennée <alex.bennee@linaro.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: use addresses instead of offsets in memory tracepoints
Hollis Blanchard [Tue, 9 Feb 2016 00:03:05 +0000 (16:03 -0800)]
trace: use addresses instead of offsets in memory tracepoints

When memory_region_ops tracepoints are enabled, calculate and record the
absolute address being accessed. Otherwise, we only get offsets into the
memory region instead of addresses.

[Fixed "offset" -> "addr" in trace event format strings.
--Stefan]

Signed-off-by: Hollis Blanchard <hollis_blanchard@mentor.com>
Message-id: 1454976185-30095-3-git-send-email-hollis_blanchard@mentor.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: split subpage MMIOs into their own trace events.
Hollis Blanchard [Tue, 9 Feb 2016 00:03:04 +0000 (16:03 -0800)]
trace: split subpage MMIOs into their own trace events.

Previously, a single MMIO could trigger the memory_region_ops tracepoint twice:
once on its way into subpage ops, then later on its way into the model's ops.

Also, the fields previously called "addr" are actually offsets into the memory
region. Rename them to "offset" while we're editing the tracepoint definitions.

Signed-off-by: Hollis Blanchard <hollis_blanchard@mentor.com>
Message-id: 1454976185-30095-2-git-send-email-hollis_blanchard@mentor.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: docs: "simple" backend does support strings
Hollis Blanchard [Tue, 9 Feb 2016 00:03:03 +0000 (16:03 -0800)]
trace: docs: "simple" backend does support strings

The simple tracing backend has supported strings for more than three years
(62bab73213ba885426a781eb2741670b9f3cae36).

Signed-off-by: Hollis Blanchard <hollis_blanchard@mentor.com>
Message-id: 1454976185-30095-1-git-send-email-hollis_blanchard@mentor.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agotrace: drop trailing empty strings
Greg Kurz [Mon, 11 Jan 2016 17:30:36 +0000 (18:30 +0100)]
trace: drop trailing empty strings

Also fix a typo in the virtio_balloon_handle_output() trace while here.

[The double-quoting was a limitation of the old tracetool.sh script.
The modern tracetool.py script does not require double-quotes at the end
of the line.  See commit cf85cf8e972f3ad79f203be4edb7968d6e052293
("trace: Format strings must begin/end with double quotes").
--Stefan]

Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-id: 20160111173036.24764.59878.stgit@bahia.huguette.org
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
8 years agoMerge remote-tracking branch 'remotes/cohuck/tags/s390x-20160301' into staging
Peter Maydell [Tue, 1 Mar 2016 13:09:55 +0000 (13:09 +0000)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20160301' into staging

Assorted fixes, cleanups and enhancements.

# gpg: Signature made Tue 01 Mar 2016 11:45:12 GMT using RSA key ID C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"

* remotes/cohuck/tags/s390x-20160301:
  s390x/css: only suspend when enabled by orb
  MAINTAINERS: Remove entry for hw/s390x/s390-virtio-bus.[ch]
  MAINTAINERS: Remove the old s390-virtio machine
  s390x/pci: use PCI_MSIX_FLAGS on retrieving the MSIX entries
  s390x/css: Use static initialization for channel_subsys fields
  s390x/css: Allocate channel_subsys statically
  s390x/pci: fix reg/dereg irq functions
  s390x/css: introduce indicator refcounting interfaces
  s390x/virtio: old machine leftovers
  watchdog/diag288: avoid race condition on expired watchdog
  s390x: remove {kvm_}s390_virtio_irq()
  s390x: fix debug statement in trigger_page_fault()
  s390x/kvm: sync fprs via kvm_run
  linux-headers: update against kvm/next

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-seabios-20160301-1' into staging
Peter Maydell [Tue, 1 Mar 2016 12:18:23 +0000 (12:18 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-seabios-20160301-1' into staging

seabios: update to 1.9.1 stable release

# gpg: Signature made Tue 01 Mar 2016 08:39:53 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-seabios-20160301-1:
  seabios: update to 1.9.1 stable release

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agos390x/css: only suspend when enabled by orb
Cornelia Huck [Mon, 22 Feb 2016 15:43:04 +0000 (16:43 +0100)]
s390x/css: only suspend when enabled by orb

We must not allow a channel program to suspend if the suspend
control bit in the orb had not been specified.

Reviewed-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agoMAINTAINERS: Remove entry for hw/s390x/s390-virtio-bus.[ch]
Thomas Huth [Thu, 25 Feb 2016 10:45:00 +0000 (11:45 +0100)]
MAINTAINERS: Remove entry for hw/s390x/s390-virtio-bus.[ch]

The files have been deleted recently, no need to keep these entries
anymore.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1456397100-22746-1-git-send-email-thuth@redhat.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agoMAINTAINERS: Remove the old s390-virtio machine
Thomas Huth [Thu, 25 Feb 2016 09:57:54 +0000 (10:57 +0100)]
MAINTAINERS: Remove the old s390-virtio machine

The old s390-virtio machine has been removed last year, so we don't
need the corresponding section in the MAINTAINERS file anymore.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1456394274-21082-1-git-send-email-thuth@redhat.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/pci: use PCI_MSIX_FLAGS on retrieving the MSIX entries
Wei Yang [Fri, 19 Feb 2016 15:18:11 +0000 (15:18 +0000)]
s390x/pci: use PCI_MSIX_FLAGS on retrieving the MSIX entries

Even PCI_CAP_FLAGS has the same value as PCI_MSIX_FLAGS, the later one is
the more proper on retrieving MSIX entries.

This patch uses PCI_MSIX_FLAGS to retrieve the MSIX entries.

Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
CC: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: Christian Borntraeger <borntraeger@de.ibm.com>
Message-Id: <1455895091-7589-3-git-send-email-richard.weiyang@gmail.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/css: Use static initialization for channel_subsys fields
Eduardo Habkost [Tue, 16 Feb 2016 20:59:06 +0000 (18:59 -0200)]
s390x/css: Use static initialization for channel_subsys fields

machine_init() will be gone, but we don't need it if we just
initialize the channel_subsys fields statically.

Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <1455656347-29033-4-git-send-email-ehabkost@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
[adapted on top of indicator changes]
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/css: Allocate channel_subsys statically
Eduardo Habkost [Tue, 16 Feb 2016 20:59:05 +0000 (18:59 -0200)]
s390x/css: Allocate channel_subsys statically

There's no need to use g_malloc0() to allocate the channel_subsys
struct, just use a static variable.

Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <1455656347-29033-3-git-send-email-ehabkost@redhat.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
[adapted on top of indicator changes]
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/pci: fix reg/dereg irq functions
Yi Min Zhao [Thu, 28 Jan 2016 05:26:43 +0000 (13:26 +0800)]
s390x/pci: fix reg/dereg irq functions

Indicator refcounting interfaces are introduced. This patch fixes
introducing unneeded indicator mappings and failure to release
AISB mappings on deregistration.

Signed-off-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/css: introduce indicator refcounting interfaces
Yi Min Zhao [Wed, 27 Jan 2016 08:05:26 +0000 (16:05 +0800)]
s390x/css: introduce indicator refcounting interfaces

Currently, virtio-ccw uses its own interfaces to keep indicators mapped
just once even if the same address has been registered multiple times.
These interfaces fit the PCI use case as well. Therefore, move them to
css and make them generic interfaces.

Signed-off-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/virtio: old machine leftovers
Cornelia Huck [Wed, 10 Feb 2016 12:22:32 +0000 (13:22 +0100)]
s390x/virtio: old machine leftovers

Remove some now unused #defines.

Reviewed-By: Sascha Silbe <silbe@linux.vnet.ibm.com>
Reviewed-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agowatchdog/diag288: avoid race condition on expired watchdog
Sascha Silbe [Fri, 29 Jan 2016 14:51:45 +0000 (15:51 +0100)]
watchdog/diag288: avoid race condition on expired watchdog

When configured to inject an NMI, watchdog_perform_action() may cause
the BQL to be temporarily relinquished (inject_nmi() → ... →
s390_nmi() → s390_cpu_restart() → run_on_cpu()). When the guest issues
diag 288 again in response to the NMI, the diag 288 operation will
race against wdt_diag288_reset(). Depending on scheduler behaviour,
wdt_diag288_reset() may be run after the guest issued a diag 288
Init. As a result, we will cancel the timer the guest just set up. The
effect observed by the guest is that a second expiry does not trigger
the watchdog action and diag 288 Change operations fail.

Fix this by resetting the timer _before_ invoking the action.

Signed-off-by: Sascha Silbe <silbe@linux.vnet.ibm.com>
Acked-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x: remove {kvm_}s390_virtio_irq()
Cornelia Huck [Wed, 27 Jan 2016 12:39:53 +0000 (13:39 +0100)]
s390x: remove {kvm_}s390_virtio_irq()

This interface was only used by the old virtio machine and therefore
is not needed anymore.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Halil Pasic <pasic@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x: fix debug statement in trigger_page_fault()
David Hildenbrand [Mon, 25 Jan 2016 08:03:57 +0000 (09:03 +0100)]
s390x: fix debug statement in trigger_page_fault()

When mmu_translate debugging output is enabled, code won't compile.
Let's just use the same statement as in trigger_prot_fault().

Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agos390x/kvm: sync fprs via kvm_run
David Hildenbrand [Tue, 2 Feb 2016 08:08:58 +0000 (09:08 +0100)]
s390x/kvm: sync fprs via kvm_run

We can now also sync the fprs via kvm_run, avoiding one ioctl.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agolinux-headers: update against kvm/next
Cornelia Huck [Mon, 22 Feb 2016 12:36:17 +0000 (13:36 +0100)]
linux-headers: update against kvm/next

Update against commit efef127c, but keep userfaultd.h.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-input-20160301-1' into staging
Peter Maydell [Tue, 1 Mar 2016 11:15:00 +0000 (11:15 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-input-20160301-1' into staging

qapi: fix input-send-event and promote to stable

# gpg: Signature made Tue 01 Mar 2016 08:19:52 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-input-20160301-1:
  qapi: promote input-send-event to stable
  qapi: rename InputAxis values.
  qapi: rename input buttons
  qapi: switch x-input-send-event from console to device+head
  console: add & use qemu_console_lookup_by_device_name

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/kraxel/tags/pull-vga-20160301-1' into staging
Peter Maydell [Tue, 1 Mar 2016 10:34:19 +0000 (10:34 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/pull-vga-20160301-1' into staging

vga: minor cirrus/qxl bugfixes.

# gpg: Signature made Tue 01 Mar 2016 07:16:22 GMT using RSA key ID D3E87138
# gpg: Good signature from "Gerd Hoffmann (work) <kraxel@redhat.com>"
# gpg:                 aka "Gerd Hoffmann <gerd@kraxel.org>"
# gpg:                 aka "Gerd Hoffmann (private) <kraxel@gmail.com>"

* remotes/kraxel/tags/pull-vga-20160301-1:
  qxl: lock current_async update in qxl_soft_reset
  cirrus_vga: fix off-by-one in blit_region_is_unsafe

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoMerge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging
Peter Maydell [Tue, 1 Mar 2016 09:54:53 +0000 (09:54 +0000)]
Merge remote-tracking branch 'remotes/cody/tags/block-pull-request' into staging

# gpg: Signature made Mon 29 Feb 2016 20:08:16 GMT using RSA key ID C0DE3057
# gpg: Good signature from "Jeffrey Cody <jcody@redhat.com>"
# gpg:                 aka "Jeffrey Cody <jeff@codyprime.org>"
# gpg:                 aka "Jeffrey Cody <codyprime@gmail.com>"

* remotes/cody/tags/block-pull-request:
  iotests/124: Add cluster_size mismatch test
  block/backup: avoid copying less than full target clusters
  block/backup: make backup cluster size configurable
  mirror: Add mirror_wait_for_io
  mirror: Rewrite mirror_iteration
  vhdx: Simplify vhdx_set_shift_bits()
  vhdx: DIV_ROUND_UP() in vhdx_calc_bat_entries()
  iscsi: add support for getting CHAP password via QCryptoSecret API
  curl: add support for HTTP authentication parameters
  rbd: add support for getting password from QCryptoSecret object
  sheepdog: allow to delete snapshot
  block/nfs: add support for setting debug level

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8 years agoseabios: update to 1.9.1 stable release
Gerd Hoffmann [Tue, 1 Mar 2016 08:36:14 +0000 (09:36 +0100)]
seabios: update to 1.9.1 stable release

git shortlog rel-1.9.0..rel-1.9.1
=================================

Cole Robinson (1):
      biostables: Support SMBIOS 2.6+ UUID format

Kevin O'Connor (7):
      xhci: Check for device disconnects during USB2 reset polling
      xhci: Wait for port enable even for USB3 devices
      sdcard: Only enable error_irq_enable for bits defined in SDHCI v1 spec
      sdcard: fix typo causing 32bit write to 16bit block_size field
      nmi: Don't try to switch onto extra stack in NMI handler
      scsi: Do not call printf() from scsi_is_ready()
      coreboot: Check for unaligned cbfs header

Marcel Apfelbaum (1):
      fw/pci: do not automatically allocate IO region for PCIe bridges

Roger Pau Monne (1):
      build: fix typo in buildversion.py

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoqapi: promote input-send-event to stable
Gerd Hoffmann [Tue, 12 Jan 2016 11:39:33 +0000 (12:39 +0100)]
qapi: promote input-send-event to stable

With all fixups being in place now, we can promote input-send-event
to stable abi by removing the x- prefix.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoqapi: rename InputAxis values.
Gerd Hoffmann [Tue, 12 Jan 2016 11:34:20 +0000 (12:34 +0100)]
qapi: rename InputAxis values.

Lowercase them.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoqapi: rename input buttons
Gerd Hoffmann [Tue, 12 Jan 2016 11:14:12 +0000 (12:14 +0100)]
qapi: rename input buttons

All lowercase, use-dash instead of CamelCase.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
8 years agoqapi: switch x-input-send-event from console to device+head
Gerd Hoffmann [Tue, 12 Jan 2016 11:11:14 +0000 (12:11 +0100)]
qapi: switch x-input-send-event from console to device+head

Use display device qdev id and head number instead of console index to
specify the QemuConsole.  This makes things consistent with input
devices (for input routing) and vnc server configuration, which both use
display and head too.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
8 years agoconsole: add & use qemu_console_lookup_by_device_name
Gerd Hoffmann [Tue, 12 Jan 2016 10:45:43 +0000 (11:45 +0100)]
console: add & use qemu_console_lookup_by_device_name

We have two places needing this, and a third one will come shortly.
So factor things out into a helper function to reduce code duplication.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
8 years agoqxl: lock current_async update in qxl_soft_reset
Gerd Hoffmann [Tue, 20 Oct 2015 07:57:30 +0000 (09:57 +0200)]
qxl: lock current_async update in qxl_soft_reset

This should fix a defect report from Coverity.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agocirrus_vga: fix off-by-one in blit_region_is_unsafe
Paolo Bonzini [Wed, 10 Feb 2016 16:17:39 +0000 (17:17 +0100)]
cirrus_vga: fix off-by-one in blit_region_is_unsafe

The "max" value is being compared with >=, but addr + width points to
the first byte that will _not_ be copied.  Laszlo suggested using a
"greater than" comparison, instead of subtracting one like it is
already done above for the height, so that max remains always positive.

The mistake is "safe"---it will reject some blits, but will never cause
out-of-bounds writes.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Message-id: 1455121059-18280-1-git-send-email-pbonzini@redhat.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
8 years agoiotests/124: Add cluster_size mismatch test
John Snow [Thu, 25 Feb 2016 20:58:31 +0000 (15:58 -0500)]
iotests/124: Add cluster_size mismatch test

If a backing file isn't specified in the target image and the
cluster_size is larger than the bitmap granularity, we run the risk of
creating bitmaps with allocated clusters but empty/no data which will
prevent the proper reading of the backup in the future.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1456433911-24718-4-git-send-email-jsnow@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>
8 years agoblock/backup: avoid copying less than full target clusters
John Snow [Thu, 25 Feb 2016 20:58:30 +0000 (15:58 -0500)]
block/backup: avoid copying less than full target clusters

During incremental backups, if the target has a cluster size that is
larger than the backup cluster size and we are backing up to a target
that cannot (for whichever reason) pull clusters up from a backing image,
we may inadvertantly create unusable incremental backup images.

For example:

If the bitmap tracks changes at a 64KB granularity and we transmit 64KB
of data at a time but the target uses a 128KB cluster size, it is
possible that only half of a target cluster will be recognized as dirty
by the backup block job. When the cluster is allocated on the target
image but only half populated with data, we lose the ability to
distinguish between zero padding and uninitialized data.

This does not happen if the target image has a backing file that points
to the last known good backup.

Even if we have a backing file, though, it's likely going to be faster
to just buffer the redundant data ourselves from the live image than
fetching it from the backing file, so let's just always round up to the
target granularity.

The same logic applies to backup modes top, none, and full. Copying
fractional clusters without the guarantee of COW is dangerous, but even
if we can rely on COW, it's likely better to just re-copy the data.

Reported-by: Fam Zheng <famz@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 1456433911-24718-3-git-send-email-jsnow@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>