qemu.git
2 years agotarget/i386: sev: Rename QSevGuestInfo
David Gibson [Thu, 4 Jun 2020 06:42:13 +0000 (16:42 +1000)] 
target/i386: sev: Rename QSevGuestInfo

At the moment this is a purely passive object which is just a container for
information used elsewhere, hence the name.  I'm going to change that
though, so as a preliminary rename it to SevGuestState.

That name risks confusion with both SEVState and SevState, but I'll be
working on that in following patches.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200604064219.436242-4-david@gibson.dropbear.id.au>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: sev: Move local structure definitions into .c file
David Gibson [Thu, 4 Jun 2020 06:42:12 +0000 (16:42 +1000)] 
target/i386: sev: Move local structure definitions into .c file

Neither QSevGuestInfo nor SEVState (not to be confused with SevState) is
used anywhere outside target/i386/sev.c, so they might as well live in
there rather than in a (somewhat) exposed header.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200604064219.436242-3-david@gibson.dropbear.id.au>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: sev: Remove unused QSevGuestInfoClass
David Gibson [Thu, 4 Jun 2020 06:42:11 +0000 (16:42 +1000)] 
target/i386: sev: Remove unused QSevGuestInfoClass

This structure is nothing but an empty wrapper around the parent class,
which by QOM conventions means we don't need it at all.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200604064219.436242-2-david@gibson.dropbear.id.au>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoxen: fix build without pci passthrough
Anthony PERARD [Wed, 3 Jun 2020 16:04:42 +0000 (17:04 +0100)] 
xen: fix build without pci passthrough

Xen PCI passthrough support may not be available and thus the global
variable "has_igd_gfx_passthru" might be compiled out. Common code
should not access it in that case.

Unfortunately, we can't use CONFIG_XEN_PCI_PASSTHROUGH directly in
xen-common.c so this patch instead move access to the
has_igd_gfx_passthru variable via function and those functions are
also implemented as stubs. The stubs will be used when QEMU is built
without passthrough support.

Now, when one will want to enable igd-passthru via the -machine
property, they will get an error message if QEMU is built without
passthrough support.

Fixes: 46472d82322d0 ('xen: convert "-machine igd-passthru" to an accelerator property')
Reported-by: Roger Pau Monné <roger.pau@citrix.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Message-Id: <20200603160442.3151170-1-anthony.perard@citrix.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop HVFX86EmulatorState
Roman Bolshakov [Thu, 28 May 2020 19:37:58 +0000 (22:37 +0300)] 
i386: hvf: Drop HVFX86EmulatorState

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-14-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Move mmio_buf into CPUX86State
Roman Bolshakov [Thu, 28 May 2020 19:37:57 +0000 (22:37 +0300)] 
i386: hvf: Move mmio_buf into CPUX86State

There's no similar field in CPUX86State, but it's needed for MMIO traps.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-13-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Move lazy_flags into CPUX86State
Roman Bolshakov [Thu, 28 May 2020 19:37:56 +0000 (22:37 +0300)] 
i386: hvf: Move lazy_flags into CPUX86State

The lazy flags are still needed for instruction decoder.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-12-r.bolshakov@yadro.com>
[Move struct to target/i386/cpu.h - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop regs in HVFX86EmulatorState
Roman Bolshakov [Thu, 28 May 2020 19:37:55 +0000 (22:37 +0300)] 
i386: hvf: Drop regs in HVFX86EmulatorState

HVFX86EmulatorState carries it's own copy of x86 registers. It can be
dropped in favor of regs in generic CPUX86State.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-11-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop copy of RFLAGS defines
Roman Bolshakov [Thu, 28 May 2020 19:37:54 +0000 (22:37 +0300)] 
i386: hvf: Drop copy of RFLAGS defines

Use the ones provided in target/i386/cpu.h instead.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-10-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop rflags from HVFX86EmulatorState
Roman Bolshakov [Thu, 28 May 2020 19:37:53 +0000 (22:37 +0300)] 
i386: hvf: Drop rflags from HVFX86EmulatorState

HVFX86EmulatorState carries it's own copy of x86 flags. It can be
dropped in favor of eflags in generic CPUX86State.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-9-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop fetch_rip from HVFX86EmulatorState
Roman Bolshakov [Thu, 28 May 2020 19:37:52 +0000 (22:37 +0300)] 
i386: hvf: Drop fetch_rip from HVFX86EmulatorState

The field is used to print address of instructions that have no parser
in decode_invalid(). RIP from VMCS is saved into fetch_rip before
decoding starts but it's also saved into env->eip in load_regs().
Therefore env->eip can be used instead of fetch_rip.

While at it, correct address printed in decode_invalid(). It prints an
address before the unknown instruction.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-8-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Use IP from CPUX86State
Roman Bolshakov [Thu, 28 May 2020 19:37:51 +0000 (22:37 +0300)] 
i386: hvf: Use IP from CPUX86State

Drop and replace rip field from HVFX86EmulatorState in favor of eip from
common CPUX86State.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-7-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
2 years agoi386: hvf: Use ins_len to advance IP
Roman Bolshakov [Thu, 28 May 2020 19:37:50 +0000 (22:37 +0300)] 
i386: hvf: Use ins_len to advance IP

There's no need to read VMCS twice, instruction length is already
available in ins_len.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-6-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
2 years agoi386: hvf: Drop unused variable
Roman Bolshakov [Thu, 28 May 2020 19:37:49 +0000 (22:37 +0300)] 
i386: hvf: Drop unused variable

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-5-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
2 years agoi386: hvf: Clean stray includes in sysemu
Roman Bolshakov [Thu, 28 May 2020 19:37:48 +0000 (22:37 +0300)] 
i386: hvf: Clean stray includes in sysemu

They have no use.

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Message-Id: <20200528193758.51454-4-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: hvf: Drop useless declarations in sysemu
Roman Bolshakov [Thu, 28 May 2020 19:37:47 +0000 (22:37 +0300)] 
i386: hvf: Drop useless declarations in sysemu

They're either declared elsewhere or have no use.

While at it, rename _hvf_cpu_synchronize_post_init() to
do_hvf_cpu_synchronize_post_init().

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-3-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
2 years agoi386: hvf: Move HVFState definition into hvf
Roman Bolshakov [Thu, 28 May 2020 19:37:46 +0000 (22:37 +0300)] 
i386: hvf: Move HVFState definition into hvf

"sysemu/hvf.h" is intended for inclusion in generic code. However it
also contains several hvf definitions and declarations, including
HVFState that are used only inside "hvf.c". "hvf-i386.h" would be more
appropriate place to define HVFState as it's only included by "hvf.c"
and "x86_task.c".

Signed-off-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200528193758.51454-2-r.bolshakov@yadro.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/ppc: Restrict PPCVirtualHypervisorClass to system-mode
Philippe Mathieu-Daudé [Tue, 26 May 2020 17:24:24 +0000 (19:24 +0200)] 
target/ppc: Restrict PPCVirtualHypervisorClass to system-mode

The code related to PPC Virtual Hypervisor is pointless in user-mode.

Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200526172427.17460-5-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agosysemu/hvf: Only declare hvf_allowed when HVF is available
Philippe Mathieu-Daudé [Tue, 26 May 2020 17:24:23 +0000 (19:24 +0200)] 
sysemu/hvf: Only declare hvf_allowed when HVF is available

When HVF is not available, the hvf_allowed variable does not exist.

Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200526172427.17460-4-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agosysemu/tcg: Only declare tcg_allowed when TCG is available
Philippe Mathieu-Daudé [Tue, 26 May 2020 17:24:22 +0000 (19:24 +0200)] 
sysemu/tcg: Only declare tcg_allowed when TCG is available

When TCG is not available, the tcg_allowed variable does not exist.

Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200526172427.17460-3-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agosysemu/accel: Restrict machine methods to system-mode
Philippe Mathieu-Daudé [Tue, 26 May 2020 17:24:21 +0000 (19:24 +0200)] 
sysemu/accel: Restrict machine methods to system-mode

Restrict init_machine(), setup_post() and has_memory()
to system-mode.

Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Message-Id: <20200526172427.17460-2-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: correct fix for pcmpxstrx substring search
Joseph Myers [Fri, 12 Jun 2020 13:45:23 +0000 (13:45 +0000)] 
target/i386: correct fix for pcmpxstrx substring search

This corrects a bug introduced in my previous fix for SSE4.2 pcmpestri
/ pcmpestrm / pcmpistri / pcmpistrm substring search, commit
ae35eea7e4a9f21dd147406dfbcd0c4c6aaf2a60.

That commit fixed a bug that showed up in four GCC tests with one libc
implementation.  The tests in question generate random inputs to the
intrinsics and compare results to a C implementation, but they only
test 1024 possible random inputs, and when the tests use the cases of
those instructions that work with word rather than byte inputs, it's
easy to have problematic cases that show up much less frequently than
that.  Thus, testing with a different libc implementation, and so a
different random number generator, showed up a problem with the
previous patch.

When investigating the previous test failures, I found the description
of these instructions in the Intel manuals (starting from computing a
16x16 or 8x8 set of comparison results) confusing and hard to match up
with the more optimized implementation in QEMU, and referred to AMD
manuals which described the instructions in a different way.  Those
AMD descriptions are very explicit that the whole of the string being
searched for must be found in the other operand, not running off the
end of that operand; they say "If the prototype and the SUT are equal
in length, the two strings must be identical for the comparison to be
TRUE.".  However, that statement is incorrect.

In my previous commit message, I noted:

  The operation in this case is a search for a string (argument d to
  the helper) in another string (argument s to the helper); if a copy
  of d at a particular position would run off the end of s, the
  resulting output bit should be 0 whether or not the strings match in
  the region where they overlap, but the QEMU implementation was
  wrongly comparing only up to the point where s ends and counting it
  as a match if an initial segment of d matched a terminal segment of
  s.  Here, "run off the end of s" means that some byte of d would
  overlap some byte outside of s; thus, if d has zero length, it is
  considered to match everywhere, including after the end of s.

The description "some byte of d would overlap some byte outside of s"
is accurate only when understood to refer to overlapping some byte
*within the 16-byte operand* but at or after the zero terminator; it
is valid to run over the end of s if the end of s is the end of the
16-byte operand.  So the fix in the previous patch for the case of d
being empty was correct, but the other part of that patch was not
correct (as it never allowed partial matches even at the end of the
16-byte operand).  Nor was the code before the previous patch correct
for the case of d nonempty, as it would always have allowed partial
matches at the end of s.

Fix with a partial revert of my previous change, combined with
inserting a check for the special case of s having maximum length to
determine where it is necessary to check for matches.

In the added test, test 1 is for the case of empty strings, which
failed before my 2017 patch, test 2 is for the bug introduced by my
2017 patch and test 3 deals with the case where a match of an initial
segment at the end of the string is not valid when the string ends
before the end of the 16-byte operand (that is, the case that would be
broken by a simple revert of the non-empty-string part of my 2017
patch).

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2006121344290.9881@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix IEEE x87 floating-point exception raising
Joseph Myers [Fri, 15 May 2020 21:21:24 +0000 (21:21 +0000)] 
target/i386: fix IEEE x87 floating-point exception raising

Most x87 instruction implementations fail to raise the expected IEEE
floating-point exceptions because they do nothing to convert the
exception state from the softfloat machinery into the exception flags
in the x87 status word.  There is special-case handling of division to
raise the divide-by-zero exception, but that handling is itself buggy:
it raises the exception in inappropriate cases (inf / 0 and nan / 0,
which should not raise any exceptions, and 0 / 0, which should raise
"invalid" instead).

Fix this by converting the floating-point exceptions raised during an
operation by the softfloat machinery into exceptions in the x87 status
word (passing through the existing fpu_set_exception function for
handling related to trapping exceptions).  There are special cases
where some functions convert to integer internally but exceptions from
that conversion are not always correct exceptions for the instruction
to raise.

There might be scope for some simplification if the softfloat
exception state either could always be assumed to be in sync with the
state in the status word, or could always be ignored at the start of
each instruction and just set to 0 then; I haven't looked into that in
detail, and it might run into interactions with the various ways the
emulation does not yet handle trapping exceptions properly.  I think
the approach taken here, of saving the softfloat state, setting
exceptions there to 0 and then merging the old exceptions back in
after carrying out the operation, is conservatively safe.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005152120280.3469@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoexec: set map length to zero when returning NULL
Prasad J Pandit [Tue, 26 May 2020 11:17:43 +0000 (16:47 +0530)] 
exec: set map length to zero when returning NULL

When mapping physical memory into host's virtual address space,
'address_space_map' may return NULL if BounceBuffer is in_use.
Set and return '*plen = 0' to avoid later NULL pointer dereference.

Reported-by: Alexander Bulekov <alxndr@bu.edu>
Fixes: https://bugs.launchpad.net/qemu/+bug/1878259
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Message-Id: <20200526111743.428367-1-ppandit@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoconfigure: Do not ignore malloc value
Leonid Bloch [Sun, 24 May 2020 22:12:04 +0000 (01:12 +0300)] 
configure: Do not ignore malloc value

Not checking the value of malloc will cause a warning with GCC 10.1,
which may result in configuration failure, with the following line in
config.log:

config-temp/qemu-conf.c:2:18: error: ignoring return value of â€˜malloc’ declared with attribute â€˜warn_unused_result’ [-Werror=unused-result]
    2 | int main(void) { malloc(1); return 0; }
      |                  ^~~~~~~~~

Signed-off-by: Leonid Bloch <lb.workbox@gmail.com>
Message-Id: <20200524221204.9791-1-lb.workbox@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqemu/thread: Mark qemu_thread_exit() with 'noreturn' attribute
Philippe Mathieu-Daudé [Mon, 1 Jun 2020 09:38:08 +0000 (11:38 +0200)] 
qemu/thread: Mark qemu_thread_exit() with 'noreturn' attribute

After upgrading to Ubuntu 20.04 LTS, GCC 9.3 complains:

  util/qemu-thread-posix.c: In function ‘qemu_thread_exit’:
  util/qemu-thread-posix.c:577:6: error: function might be candidate for attribute ‘noreturn’ [-Werror=suggest-attribute=noreturn]
    577 | void qemu_thread_exit(void *retval)
        |      ^~~~~~~~~~~~~~~~

Fix by marking the qemu_thread_exit function with QEMU_NORETURN
to set the 'noreturn' attribute.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agomemory: Make 'info mtree' not display disabled regions by default
Philippe Mathieu-Daudé [Fri, 29 May 2020 12:53:25 +0000 (14:53 +0200)] 
memory: Make 'info mtree' not display disabled regions by default

We might have many disabled memory regions, making the 'info mtree'
output too verbose to be useful.
Remove the disabled regions in the default output, but allow the
monitor user to display them using the '-D' option.

Before:

  (qemu) info mtree
  memory-region: system
    0000000000000000-ffffffffffffffff (prio 0, i/o): system
      0000000000000000-0000000007ffffff (prio 0, ram): alias ram-below-4g @pc.ram 0000000000000000-0000000007ffffff
      0000000000000000-ffffffffffffffff (prio -1, i/o): pci
        00000000000a0000-00000000000bffff (prio 1, i/o): vga-lowmem
        00000000000c0000-00000000000dffff (prio 1, rom): pc.rom
        00000000000e0000-00000000000fffff (prio 1, rom): alias isa-bios @pc.bios 0000000000020000-000000000003ffff
        00000000fffc0000-00000000ffffffff (prio 0, rom): pc.bios
      00000000000a0000-00000000000bffff (prio 1, i/o): alias smram-region @pci 00000000000a0000-00000000000bffff
      00000000000c0000-00000000000c3fff (prio 1, ram): alias pam-ram @pc.ram 00000000000c0000-00000000000c3fff [disabled]
      00000000000c0000-00000000000c3fff (prio 1, ram): alias pam-pci @pc.ram 00000000000c0000-00000000000c3fff [disabled]
      00000000000c0000-00000000000c3fff (prio 1, ram): alias pam-rom @pc.ram 00000000000c0000-00000000000c3fff [disabled]
      00000000000c0000-00000000000c3fff (prio 1, i/o): alias pam-pci @pci 00000000000c0000-00000000000c3fff
      00000000000c4000-00000000000c7fff (prio 1, ram): alias pam-ram @pc.ram 00000000000c4000-00000000000c7fff [disabled]
      00000000000c4000-00000000000c7fff (prio 1, ram): alias pam-pci @pc.ram 00000000000c4000-00000000000c7fff [disabled]
      00000000000c4000-00000000000c7fff (prio 1, ram): alias pam-rom @pc.ram 00000000000c4000-00000000000c7fff [disabled]
      00000000000c4000-00000000000c7fff (prio 1, i/o): alias pam-pci @pci 00000000000c4000-00000000000c7fff
      00000000000c8000-00000000000cbfff (prio 1, ram): alias pam-ram @pc.ram 00000000000c8000-00000000000cbfff [disabled]
      00000000000c8000-00000000000cbfff (prio 1, ram): alias pam-pci @pc.ram 00000000000c8000-00000000000cbfff [disabled]
      00000000000c8000-00000000000cbfff (prio 1, ram): alias pam-rom @pc.ram 00000000000c8000-00000000000cbfff [disabled]
      00000000000c8000-00000000000cbfff (prio 1, i/o): alias pam-pci @pci 00000000000c8000-00000000000cbfff
      00000000000cc000-00000000000cffff (prio 1, ram): alias pam-ram @pc.ram 00000000000cc000-00000000000cffff [disabled]
      00000000000cc000-00000000000cffff (prio 1, ram): alias pam-pci @pc.ram 00000000000cc000-00000000000cffff [disabled]
      00000000000cc000-00000000000cffff (prio 1, ram): alias pam-rom @pc.ram 00000000000cc000-00000000000cffff [disabled]
      00000000000cc000-00000000000cffff (prio 1, i/o): alias pam-pci @pci 00000000000cc000-00000000000cffff
      00000000000d0000-00000000000d3fff (prio 1, ram): alias pam-ram @pc.ram 00000000000d0000-00000000000d3fff [disabled]
      00000000000d0000-00000000000d3fff (prio 1, ram): alias pam-pci @pc.ram 00000000000d0000-00000000000d3fff [disabled]
      00000000000d0000-00000000000d3fff (prio 1, ram): alias pam-rom @pc.ram 00000000000d0000-00000000000d3fff [disabled]
      00000000000d0000-00000000000d3fff (prio 1, i/o): alias pam-pci @pci 00000000000d0000-00000000000d3fff
      00000000000d4000-00000000000d7fff (prio 1, ram): alias pam-ram @pc.ram 00000000000d4000-00000000000d7fff [disabled]
      00000000000d4000-00000000000d7fff (prio 1, ram): alias pam-pci @pc.ram 00000000000d4000-00000000000d7fff [disabled]
      00000000000d4000-00000000000d7fff (prio 1, ram): alias pam-rom @pc.ram 00000000000d4000-00000000000d7fff [disabled]
      00000000000d4000-00000000000d7fff (prio 1, i/o): alias pam-pci @pci 00000000000d4000-00000000000d7fff
      00000000000d8000-00000000000dbfff (prio 1, ram): alias pam-ram @pc.ram 00000000000d8000-00000000000dbfff [disabled]
      00000000000d8000-00000000000dbfff (prio 1, ram): alias pam-pci @pc.ram 00000000000d8000-00000000000dbfff [disabled]
      00000000000d8000-00000000000dbfff (prio 1, ram): alias pam-rom @pc.ram 00000000000d8000-00000000000dbfff [disabled]
      00000000000d8000-00000000000dbfff (prio 1, i/o): alias pam-pci @pci 00000000000d8000-00000000000dbfff
      00000000000dc000-00000000000dffff (prio 1, ram): alias pam-ram @pc.ram 00000000000dc000-00000000000dffff [disabled]
      00000000000dc000-00000000000dffff (prio 1, ram): alias pam-pci @pc.ram 00000000000dc000-00000000000dffff [disabled]
      00000000000dc000-00000000000dffff (prio 1, ram): alias pam-rom @pc.ram 00000000000dc000-00000000000dffff [disabled]
      00000000000dc000-00000000000dffff (prio 1, i/o): alias pam-pci @pci 00000000000dc000-00000000000dffff
      00000000000e0000-00000000000e3fff (prio 1, ram): alias pam-ram @pc.ram 00000000000e0000-00000000000e3fff [disabled]
      00000000000e0000-00000000000e3fff (prio 1, ram): alias pam-pci @pc.ram 00000000000e0000-00000000000e3fff [disabled]
      00000000000e0000-00000000000e3fff (prio 1, ram): alias pam-rom @pc.ram 00000000000e0000-00000000000e3fff [disabled]
      00000000000e0000-00000000000e3fff (prio 1, i/o): alias pam-pci @pci 00000000000e0000-00000000000e3fff
      00000000000e4000-00000000000e7fff (prio 1, ram): alias pam-ram @pc.ram 00000000000e4000-00000000000e7fff [disabled]
      00000000000e4000-00000000000e7fff (prio 1, ram): alias pam-pci @pc.ram 00000000000e4000-00000000000e7fff [disabled]
      00000000000e4000-00000000000e7fff (prio 1, ram): alias pam-rom @pc.ram 00000000000e4000-00000000000e7fff [disabled]
      00000000000e4000-00000000000e7fff (prio 1, i/o): alias pam-pci @pci 00000000000e4000-00000000000e7fff
      00000000000e8000-00000000000ebfff (prio 1, ram): alias pam-ram @pc.ram 00000000000e8000-00000000000ebfff [disabled]
      00000000000e8000-00000000000ebfff (prio 1, ram): alias pam-pci @pc.ram 00000000000e8000-00000000000ebfff [disabled]
      00000000000e8000-00000000000ebfff (prio 1, ram): alias pam-rom @pc.ram 00000000000e8000-00000000000ebfff [disabled]
      00000000000e8000-00000000000ebfff (prio 1, i/o): alias pam-pci @pci 00000000000e8000-00000000000ebfff
      00000000000ec000-00000000000effff (prio 1, ram): alias pam-ram @pc.ram 00000000000ec000-00000000000effff [disabled]
      00000000000ec000-00000000000effff (prio 1, ram): alias pam-pci @pc.ram 00000000000ec000-00000000000effff [disabled]
      00000000000ec000-00000000000effff (prio 1, ram): alias pam-rom @pc.ram 00000000000ec000-00000000000effff [disabled]
      00000000000ec000-00000000000effff (prio 1, i/o): alias pam-pci @pci 00000000000ec000-00000000000effff
      00000000000f0000-00000000000fffff (prio 1, ram): alias pam-ram @pc.ram 00000000000f0000-00000000000fffff [disabled]
      00000000000f0000-00000000000fffff (prio 1, ram): alias pam-pci @pc.ram 00000000000f0000-00000000000fffff [disabled]
      00000000000f0000-00000000000fffff (prio 1, ram): alias pam-rom @pc.ram 00000000000f0000-00000000000fffff [disabled]
      00000000000f0000-00000000000fffff (prio 1, i/o): alias pam-pci @pci 00000000000f0000-00000000000fffff
      00000000fec00000-00000000fec00fff (prio 0, i/o): ioapic
      00000000fed00000-00000000fed003ff (prio 0, i/o): hpet
      00000000fee00000-00000000feefffff (prio 4096, i/o): apic-msi

After:

  (qemu) info mtree
  memory-region: system
    0000000000000000-ffffffffffffffff (prio 0, i/o): system
      0000000000000000-0000000007ffffff (prio 0, ram): alias ram-below-4g @pc.ram 0000000000000000-0000000007ffffff
      0000000000000000-ffffffffffffffff (prio -1, i/o): pci
        00000000000a0000-00000000000bffff (prio 1, i/o): vga-lowmem
        00000000000c0000-00000000000dffff (prio 1, rom): pc.rom
        00000000000e0000-00000000000fffff (prio 1, rom): alias isa-bios @pc.bios 0000000000020000-000000000003ffff
        00000000fffc0000-00000000ffffffff (prio 0, rom): pc.bios
      00000000000a0000-00000000000bffff (prio 1, i/o): alias smram-region @pci 00000000000a0000-00000000000bffff
      00000000000c0000-00000000000c3fff (prio 1, i/o): alias pam-pci @pci 00000000000c0000-00000000000c3fff
      00000000000c4000-00000000000c7fff (prio 1, i/o): alias pam-pci @pci 00000000000c4000-00000000000c7fff
      00000000000c8000-00000000000cbfff (prio 1, i/o): alias pam-pci @pci 00000000000c8000-00000000000cbfff
      00000000000cc000-00000000000cffff (prio 1, i/o): alias pam-pci @pci 00000000000cc000-00000000000cffff
      00000000000d0000-00000000000d3fff (prio 1, i/o): alias pam-pci @pci 00000000000d0000-00000000000d3fff
      00000000000d4000-00000000000d7fff (prio 1, i/o): alias pam-pci @pci 00000000000d4000-00000000000d7fff
      00000000000d8000-00000000000dbfff (prio 1, i/o): alias pam-pci @pci 00000000000d8000-00000000000dbfff
      00000000000dc000-00000000000dffff (prio 1, i/o): alias pam-pci @pci 00000000000dc000-00000000000dffff
      00000000000e0000-00000000000e3fff (prio 1, i/o): alias pam-pci @pci 00000000000e0000-00000000000e3fff
      00000000000e4000-00000000000e7fff (prio 1, i/o): alias pam-pci @pci 00000000000e4000-00000000000e7fff
      00000000000e8000-00000000000ebfff (prio 1, i/o): alias pam-pci @pci 00000000000e8000-00000000000ebfff
      00000000000ec000-00000000000effff (prio 1, i/o): alias pam-pci @pci 00000000000ec000-00000000000effff
      00000000000f0000-00000000000fffff (prio 1, i/o): alias pam-pci @pci 00000000000f0000-00000000000fffff
      00000000fec00000-00000000fec00fff (prio 0, i/o): ioapic
      00000000fed00000-00000000fed003ff (prio 0, i/o): hpet
      00000000fee00000-00000000feefffff (prio 4096, i/o): apic-msi

The old behavior is preserved using 'info mtree -D'.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoutil/oslib: Returns the real thread identifier on FreeBSD and NetBSD
David Carlier [Tue, 26 May 2020 07:25:26 +0000 (08:25 +0100)] 
util/oslib: Returns the real thread identifier on FreeBSD and NetBSD

getpid is good enough in a mono thread context, however thr_self/_lwp_self
reflects the real current thread identifier from a given process.

Signed-off-by: David Carlier <devnexen@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: David Carlier <devnexen@gmail.com>
2 years agotarget/i386: define a new MSR based feature word - FEAT_PERF_CAPABILITIES
Like Xu [Fri, 29 May 2020 07:43:47 +0000 (15:43 +0800)] 
target/i386: define a new MSR based feature word - FEAT_PERF_CAPABILITIES

The Perfmon and Debug Capability MSR named IA32_PERF_CAPABILITIES is
a feature-enumerating MSR, which only enumerates the feature full-width
write (via bit 13) by now which indicates the processor supports IA32_A_PMCx
interface for updating bits 32 and above of IA32_PMCx.

The existence of MSR IA32_PERF_CAPABILITIES is enumerated by CPUID.1:ECX[15].

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: qemu-devel@nongnu.org
Signed-off-by: Like Xu <like.xu@linux.intel.com>
Message-Id: <20200529074347.124619-5-like.xu@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: Remove unused define's from hax and hvf
Julio Faracco [Mon, 23 Mar 2020 20:05:38 +0000 (17:05 -0300)] 
i386: Remove unused define's from hax and hvf

Commit acb9f95a removed boundary checks for ID and VCPU ID. After that,
the max definitions of that boundaries are not required anymore. This
commit is only a code cleanup.

Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
Message-Id: <20200323200538.202164-1-jcfaracco@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoreplay: implement fair mutex
Pavel Dovgalyuk [Thu, 30 Apr 2020 09:13:49 +0000 (12:13 +0300)] 
replay: implement fair mutex

In record/replay icount mode main loop thread and vCPU thread
do not perform simultaneously. They take replay mutex to synchronize
the actions. Sometimes vCPU thread waits for locking the mutex for
very long time, because main loop releases the mutex and takes it
back again. Standard qemu mutex do not provide the ordering
capabilities.

This patch adds a "queue" for replay mutex. Therefore thread ordering
becomes more "fair". Threads are executed in the same order as
they are trying to take the mutex.

Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru>
Message-Id: <158823802979.28101.9340462887738957616.stgit@pasha-ThinkPad-X280>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/amd_iommu: Fix the reserved bits definition of IOMMU commands
Wei Huang [Sat, 18 Apr 2020 04:28:45 +0000 (23:28 -0500)] 
hw/i386/amd_iommu: Fix the reserved bits definition of IOMMU commands

Many reserved bits of amd_iommu commands are defined incorrectly in QEMU.
Because of it, QEMU incorrectly injects lots of illegal commands into guest
VM's IOMMU event log.

Signed-off-by: Wei Huang <wei.huang2@amd.com>
Message-Id: <20200418042845.596457-1-wei.huang2@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotests: machine-none-test: Enable MicroBlaze testing
Edgar E. Iglesias [Thu, 16 Apr 2020 19:33:03 +0000 (21:33 +0200)] 
tests: machine-none-test: Enable MicroBlaze testing

Enable MicroBlaze testing.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-Id: <20200416193303.23674-2-edgar.iglesias@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agochardev/char-socket: Properly make qio connections non blocking
Sai Pavan Boddu [Sun, 19 Apr 2020 09:51:40 +0000 (15:21 +0530)] 
chardev/char-socket: Properly make qio connections non blocking

In tcp_chr_sync_read function, there is a possibility of socket
disconnection during blocking read, then tcp_chr_hup function would clean up
the qio channel pointers(i.e ioc, sioc).

Signed-off-by: Sai Pavan Boddu <sai.pavan.boddu@xilinx.com>
Message-Id: <1587289900-29485-1-git-send-email-sai.pavan.boddu@xilinx.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoKVM: Kick resamplefd for split kernel irqchip
Peter Xu [Wed, 18 Mar 2020 14:52:03 +0000 (10:52 -0400)] 
KVM: Kick resamplefd for split kernel irqchip

This is majorly only for X86 because that's the only one that supports
split irqchip for now.

When the irqchip is split, we face a dilemma that KVM irqfd will be
enabled, however the slow irqchip is still running in the userspace.
It means that the resamplefd in the kernel irqfds won't take any
effect and it will miss to ack INTx interrupts on EOIs.

One example is split irqchip with VFIO INTx, which will break if we
use the VFIO INTx fast path.

This patch can potentially supports the VFIO fast path again for INTx,
that the IRQ delivery will still use the fast path, while we don't
need to trap MMIOs in QEMU for the device to emulate the EIOs (see the
callers of vfio_eoi() hook).  However the EOI of the INTx will still
need to be done from the userspace by caching all the resamplefds in
QEMU and kick properly for IOAPIC EOI broadcast.

This is tricky because in this case the userspace ioapic irr &
remote-irr will be bypassed.  However such a change will greatly boost
performance for assigned devices using INTx irqs (TCP_RR boosts 46%
after this patch applied).

When the userspace is responsible for the resamplefd kickup, don't
register it on the kvm_irqfd anymore, because on newer kernels (after
commit 654f1f13ea56, 5.2+) the KVM_IRQFD will fail if with both split
irqchip and resamplefd.  This will make sure that the fast path will
work for all supported kernels.

https://patchwork.kernel.org/patch/10738541/#22609933

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20200318145204.74483-5-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoKVM: Pass EventNotifier into kvm_irqchip_assign_irqfd
Peter Xu [Wed, 18 Mar 2020 14:52:02 +0000 (10:52 -0400)] 
KVM: Pass EventNotifier into kvm_irqchip_assign_irqfd

So that kvm_irqchip_assign_irqfd() can have access to the
EventNotifiers, especially the resample event.  It is needed in follow
up patch to cache and kick resamplefds from QEMU.

Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20200318145204.74483-4-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agovfio/pci: Use kvm_irqchip_add_irqfd_notifier_gsi() for irqfds
Peter Xu [Wed, 18 Mar 2020 14:52:01 +0000 (10:52 -0400)] 
vfio/pci: Use kvm_irqchip_add_irqfd_notifier_gsi() for irqfds

VFIO is currently the only one left that is not using the generic
function (kvm_irqchip_add_irqfd_notifier_gsi()) to register irqfds.
Let VFIO use the common framework too.

Follow up patches will introduce extra features for kvm irqfd, so that
VFIO can easily leverage that after the switch.

Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20200318145204.74483-3-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agox86/cpu: Enable AVX512_VP2INTERSECT cpu feature
Cathy Zhang [Mon, 13 Apr 2020 06:52:38 +0000 (14:52 +0800)] 
x86/cpu: Enable AVX512_VP2INTERSECT cpu feature

AVX512_VP2INTERSECT compute vector pair intersection to a pair
of mask registers, which is introduced with intel Tiger Lake,
defining as CPUID.(EAX=7,ECX=0):EDX[bit 08].

Refer to the following release spec:
https://software.intel.com/sites/default/files/managed/c5/15/\
architecture-instruction-set-extensions-programming-reference.pdf

Signed-off-by: Cathy Zhang <cathy.zhang@intel.com>
Message-Id: <1586760758-13638-1-git-send-email-cathy.zhang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Allow QTest use without crashing
Philippe Mathieu-Daudé [Mon, 18 May 2020 10:31:13 +0000 (12:31 +0200)] 
hw/i386/vmport: Allow QTest use without crashing

Trying libFuzzer on the vmport device, we get:

  AddressSanitizer:DEADLYSIGNAL
  =================================================================
  ==29476==ERROR: AddressSanitizer: SEGV on unknown address 0x000000008840 (pc 0x56448bec4d79 bp 0x7ffeec9741b0 sp 0x7ffeec9740e0 T0)
  ==29476==The signal is caused by a READ memory access.
    #0 0x56448bec4d78 in vmport_ioport_read (qemu-fuzz-i386+0x1260d78)
    #1 0x56448bb5f175 in memory_region_read_accessor (qemu-fuzz-i386+0xefb175)
    #2 0x56448bb30c13 in access_with_adjusted_size (qemu-fuzz-i386+0xeccc13)
    #3 0x56448bb2ea27 in memory_region_dispatch_read1 (qemu-fuzz-i386+0xecaa27)
    #4 0x56448bb2e443 in memory_region_dispatch_read (qemu-fuzz-i386+0xeca443)
    #5 0x56448b961ab1 in flatview_read_continue (qemu-fuzz-i386+0xcfdab1)
    #6 0x56448b96336d in flatview_read (qemu-fuzz-i386+0xcff36d)
    #7 0x56448b962ec4 in address_space_read_full (qemu-fuzz-i386+0xcfeec4)

This is easily reproducible using:

  $ echo inb 0x5658 | qemu-system-i386 -M isapc,accel=qtest -qtest stdio
  [I 1589796572.009763] OPENED
  [R +0.008069] inb 0x5658
  Segmentation fault (core dumped)

  $ coredumpctl gdb -q
  Program terminated with signal SIGSEGV, Segmentation fault.
  #0  0x00005605b54d0f21 in vmport_ioport_read (opaque=0x5605b7531ce0, addr=0, size=4) at hw/i386/vmport.c:77
  77          eax = env->regs[R_EAX];
  (gdb) p cpu
  $1 = (X86CPU *) 0x0
  (gdb) bt
  #0  0x00005605b54d0f21 in vmport_ioport_read (opaque=0x5605b7531ce0, addr=0, size=4) at hw/i386/vmport.c:77
  #1  0x00005605b53db114 in memory_region_read_accessor (mr=0x5605b7531d80, addr=0, value=0x7ffc9d261a30, size=4, shift=0, mask=4294967295, attrs=...) at memory.c:434
  #2  0x00005605b53db5d4 in access_with_adjusted_size (addr=0, value=0x7ffc9d261a30, size=1, access_size_min=4, access_size_max=4, access_fn=
      0x5605b53db0d2 <memory_region_read_accessor>, mr=0x5605b7531d80, attrs=...) at memory.c:544
  #3  0x00005605b53de156 in memory_region_dispatch_read1 (mr=0x5605b7531d80, addr=0, pval=0x7ffc9d261a30, size=1, attrs=...) at memory.c:1396
  #4  0x00005605b53de228 in memory_region_dispatch_read (mr=0x5605b7531d80, addr=0, pval=0x7ffc9d261a30, op=MO_8, attrs=...) at memory.c:1424
  #5  0x00005605b537c80a in flatview_read_continue (fv=0x5605b7650290, addr=22104, attrs=..., ptr=0x7ffc9d261b4b, len=1, addr1=0, l=1, mr=0x5605b7531d80) at exec.c:3200
  #6  0x00005605b537c95d in flatview_read (fv=0x5605b7650290, addr=22104, attrs=..., buf=0x7ffc9d261b4b, len=1) at exec.c:3239
  #7  0x00005605b537c9e6 in address_space_read_full (as=0x5605b5f74ac0 <address_space_io>, addr=22104, attrs=..., buf=0x7ffc9d261b4b, len=1) at exec.c:3252
  #8  0x00005605b53d5a5d in address_space_read (len=1, buf=0x7ffc9d261b4b, attrs=..., addr=22104, as=0x5605b5f74ac0 <address_space_io>) at include/exec/memory.h:2401
  #9  0x00005605b53d5a5d in cpu_inb (addr=22104) at ioport.c:88

X86CPU is NULL because QTest accelerator does not use CPU.
Fix by returning default values when QTest accelerator is used.

Reported-by: Clang AddressSanitizer
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fisttpl, fisttpll handling of out-of-range values
Joseph Myers [Fri, 15 May 2020 21:20:25 +0000 (21:20 +0000)] 
target/i386: fix fisttpl, fisttpll handling of out-of-range values

The fist / fistt family of instructions should all store the most
negative integer in the destination format when the rounded /
truncated integer result is out of range or the input is an invalid
encoding, infinity or NaN.  The fisttpl and fisttpll implementations
(32-bit and 64-bit results, truncate towards zero) failed to do this,
producing the most positive integer in some cases instead.  Fix this
by copying the code used to handle this issue for fistpl and fistpll,
adjusted to use the _round_to_zero functions for the actual
conversion (but without any other changes to that code).

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005152119160.3469@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fbstp handling of out-of-range values
Joseph Myers [Wed, 13 May 2020 23:51:42 +0000 (23:51 +0000)] 
target/i386: fix fbstp handling of out-of-range values

The fbstp implementation fails to check for out-of-range and invalid
values, instead just taking the result of conversion to int64_t and
storing its sign and low 18 decimal digits.  Fix this by checking for
an out-of-range result (invalid conversions always result in INT64_MAX
or INT64_MIN from the softfloat code, which are large enough to be
considered as out-of-range by this code) and storing the packed BCD
indefinite encoding in that case.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005132351110.11687@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fbstp handling of negative zero
Joseph Myers [Wed, 13 May 2020 23:51:09 +0000 (23:51 +0000)] 
target/i386: fix fbstp handling of negative zero

The fbstp implementation stores +0 when the rounded result should be
-0 because it compares an integer value with 0 to determine the sign.
Fix this by checking the sign bit of the operand instead.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005132350230.11687@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fxam handling of invalid encodings
Joseph Myers [Wed, 13 May 2020 23:50:19 +0000 (23:50 +0000)] 
target/i386: fix fxam handling of invalid encodings

The fxam implementation does not check for invalid encodings, instead
treating them like NaN or normal numbers depending on the exponent.
Fix it to check that the high bit of the significand is set before
treating an encoding as NaN or normal, thus resulting in correct
handling (all of C0, C2 and C3 cleared) for invalid encodings.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005132349311.11687@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix floating-point load-constant rounding
Joseph Myers [Wed, 13 May 2020 23:49:27 +0000 (23:49 +0000)] 
target/i386: fix floating-point load-constant rounding

The implementations of the fldl2t, fldl2e, fldpi, fldlg2 and fldln2
instructions load fixed constants independent of the rounding mode.
Fix them to load a value correctly rounded for the current rounding
mode (but always rounded to 64-bit precision independent of the
precision control, and without setting "inexact") as specified.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <alpine.DEB.2.21.2005132348310.11687@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/elf_ops: Do not ignore write failures when loading ELF
Philippe Mathieu-Daudé [Mon, 18 May 2020 15:53:05 +0000 (17:53 +0200)] 
hw/elf_ops: Do not ignore write failures when loading ELF

Do not ignore the MemTxResult error type returned by
address_space_write().

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agodisas: Let disas::read_memory() handler return EIO on error
Philippe Mathieu-Daudé [Mon, 18 May 2020 15:53:04 +0000 (17:53 +0200)] 
disas: Let disas::read_memory() handler return EIO on error

Both cpu_memory_rw_debug() and address_space_read() return
an error on failed transaction. Check the returned value,
and return EIO in case of error.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoexec: Propagate cpu_memory_rw_debug() error
Philippe Mathieu-Daudé [Mon, 18 May 2020 15:53:03 +0000 (17:53 +0200)] 
exec: Propagate cpu_memory_rw_debug() error

Do not ignore the MemTxResult error type returned by
the address_space_rw() API.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoexec: Let address_space_read/write_cached() propagate MemTxResult
Philippe Mathieu-Daudé [Mon, 18 May 2020 15:53:02 +0000 (17:53 +0200)] 
exec: Let address_space_read/write_cached() propagate MemTxResult

Both address_space_read_cached_slow() and
address_space_write_cached_slow() return a MemTxResult type.
Do not discard it, return it to the caller.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fscale handling of rounding precision
Joseph Myers [Thu, 7 May 2020 00:46:28 +0000 (00:46 +0000)] 
target/i386: fix fscale handling of rounding precision

The fscale implementation uses floatx80_scalbn for the final scaling
operation.  floatx80_scalbn ends up rounding the result using the
dynamic rounding precision configured for the FPU.  But only a limited
set of x87 floating-point instructions are supposed to respect the
dynamic rounding precision, and fscale is not in that set.  Fix the
implementation to save and restore the rounding precision around the
call to floatx80_scalbn.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005070045430.18350@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fscale handling of infinite exponents
Joseph Myers [Thu, 7 May 2020 00:45:38 +0000 (00:45 +0000)] 
target/i386: fix fscale handling of infinite exponents

The fscale implementation passes infinite exponents through to generic
code that rounds the exponent to a 32-bit integer before using
floatx80_scalbn.  In round-to-nearest mode, and ignoring exceptions,
this works in many cases.  But it fails to handle the special cases of
scaling 0 by a +Inf exponent or an infinity by a -Inf exponent, which
should produce a NaN, and because it produces an inexact result for
finite nonzero numbers being scaled, the result is sometimes incorrect
in other rounding modes.  Add appropriate handling of infinite
exponents to produce a NaN or an appropriately signed exact zero or
infinity as a result.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005070045010.18350@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fscale handling of invalid exponent encodings
Joseph Myers [Thu, 7 May 2020 00:44:57 +0000 (00:44 +0000)] 
target/i386: fix fscale handling of invalid exponent encodings

The fscale implementation does not check for invalid encodings in the
exponent operand, thus treating them like INT_MIN (the value returned
for invalid encodings by floatx80_to_int32_round_to_zero).  Fix it to
treat them similarly to signaling NaN exponents, thus generating a
quiet NaN result.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005070044190.18350@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix fscale handling of signaling NaN
Joseph Myers [Thu, 7 May 2020 00:44:14 +0000 (00:44 +0000)] 
target/i386: fix fscale handling of signaling NaN

The implementation of the fscale instruction returns a NaN exponent
unchanged.  Fix it to return a quiet NaN when the provided exponent is
a signaling NaN.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Message-Id: <alpine.DEB.2.21.2005070043330.18350@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: implement special cases for fxtract
Joseph Myers [Thu, 7 May 2020 00:43:30 +0000 (00:43 +0000)] 
target/i386: implement special cases for fxtract

The implementation of the fxtract instruction treats all nonzero
operands as normal numbers, so yielding incorrect results for invalid
formats, infinities, NaNs and subnormal and pseudo-denormal operands.
Implement appropriate handling of all those cases.

Signed-off-by: Joseph Myers <joseph@codesourcery.com>
Acked-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <alpine.DEB.2.21.2005070042360.18350@digraph.polyomino.org.uk>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agomegasas: use unsigned type for positive numeric fields
Prasad J Pandit [Wed, 13 May 2020 19:25:40 +0000 (00:55 +0530)] 
megasas: use unsigned type for positive numeric fields

Use unsigned type for the MegasasState fields which hold positive
numeric values.

Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20200513192540.1583887-4-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agomegasas: avoid NULL pointer dereference
Prasad J Pandit [Wed, 13 May 2020 19:25:39 +0000 (00:55 +0530)] 
megasas: avoid NULL pointer dereference

While in megasas_handle_frame(), megasas_enqueue_frame() may
set a NULL frame into MegasasCmd object for a given 'frame_addr'
address. Add check to avoid a NULL pointer dereference issue.

Reported-by: Alexander Bulekov <alxndr@bu.edu>
Fixes: https://bugs.launchpad.net/qemu/+bug/1878259
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Acked-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20200513192540.1583887-3-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agomegasas: use unsigned type for reply_queue_head and check index
Prasad J Pandit [Wed, 13 May 2020 19:25:38 +0000 (00:55 +0530)] 
megasas: use unsigned type for reply_queue_head and check index

A guest user may set 'reply_queue_head' field of MegasasState to
a negative value. Later in 'megasas_lookup_frame' it is used to
index into s->frames[] array. Use unsigned type to avoid OOB
access issue.

Also check that 'index' value stays within s->frames[] bounds
through the while() loop in 'megasas_lookup_frame' to avoid OOB
access.

Reported-by: Ren Ding <rding@gatech.edu>
Reported-by: Hanqing Zhao <hanqing@gatech.edu>
Reported-by: Alexander Bulekov <alxndr@bu.edu>
Signed-off-by: Prasad J Pandit <pjp@fedoraproject.org>
Acked-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20200513192540.1583887-2-ppandit@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386/kvm: fix a use-after-free when vcpu plug/unplug
Pan Nengyuan [Wed, 13 May 2020 13:26:30 +0000 (09:26 -0400)] 
i386/kvm: fix a use-after-free when vcpu plug/unplug

When we hotplug vcpus, cpu_update_state is added to vm_change_state_head
in kvm_arch_init_vcpu(). But it forgot to delete in kvm_arch_destroy_vcpu() after
unplug. Then it will cause a use-after-free access. This patch delete it in
kvm_arch_destroy_vcpu() to fix that.

Reproducer:
    virsh setvcpus vm1 4 --live
    virsh setvcpus vm1 2 --live
    virsh suspend vm1
    virsh resume vm1

The UAF stack:
==qemu-system-x86_64==28233==ERROR: AddressSanitizer: heap-use-after-free on address 0x62e00002e798 at pc 0x5573c6917d9e bp 0x7fff07139e50 sp 0x7fff07139e40
WRITE of size 1 at 0x62e00002e798 thread T0
    #0 0x5573c6917d9d in cpu_update_state /mnt/sdb/qemu/target/i386/kvm.c:742
    #1 0x5573c699121a in vm_state_notify /mnt/sdb/qemu/vl.c:1290
    #2 0x5573c636287e in vm_prepare_start /mnt/sdb/qemu/cpus.c:2144
    #3 0x5573c6362927 in vm_start /mnt/sdb/qemu/cpus.c:2150
    #4 0x5573c71e8304 in qmp_cont /mnt/sdb/qemu/monitor/qmp-cmds.c:173
    #5 0x5573c727cb1e in qmp_marshal_cont qapi/qapi-commands-misc.c:835
    #6 0x5573c7694c7a in do_qmp_dispatch /mnt/sdb/qemu/qapi/qmp-dispatch.c:132
    #7 0x5573c7694c7a in qmp_dispatch /mnt/sdb/qemu/qapi/qmp-dispatch.c:175
    #8 0x5573c71d9110 in monitor_qmp_dispatch /mnt/sdb/qemu/monitor/qmp.c:145
    #9 0x5573c71dad4f in monitor_qmp_bh_dispatcher /mnt/sdb/qemu/monitor/qmp.c:234

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200513132630.13412-1-pannengyuan@huawei.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohax: Dynamic allocate vcpu state structure
WangBowen [Sat, 9 May 2020 03:59:52 +0000 (11:59 +0800)] 
hax: Dynamic allocate vcpu state structure

Dynamic allocating vcpu state structure according to smp value to be
more precise and safe. Previously it will alloccate array of fixed size
HAX_MAX_VCPU.

This is achieved by using g_new0 to dynamic allocate the array. The
allocated size is obtained from smp.max_cpus in MachineState. Also, the
size is compared with HAX_MAX_VCPU when creating the vm. The reason for
choosing dynamic array over linked list is because the status is visited
by index all the time.

This will lead to QEMU checking whether the smp value is larger than the
HAX_MAX_VCPU when creating vm, if larger, the process will terminate,
otherwise it will allocate array of size smp to store the status.

V2: Check max_cpus before open vm. (Philippe)

Signed-off-by: WangBowen <bowen.wang@intel.com>
Signed-off-by: Colin Xu <colin.xu@intel.com>
Message-Id: <20200509035952.187615-1-colin.xu@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agocpus: Fix botched configure_icount() error API violation fix
Markus Armbruster [Fri, 15 May 2020 04:22:31 +0000 (06:22 +0200)] 
cpus: Fix botched configure_icount() error API violation fix

Before recent commit abc9bf69a66, configure_icount() returned early
when option "shift" was absent: succeed when option "align" was also
absent, else fail.

Since then, it still errors out when only "align" is present, but
continues when both are absent.  Crashes when examining the value of
"shift" further.  Reproducer: -icount "".

Revert this erroneous part of the commit.

Fixes: abc9bf69a66a11499a801ff545b8fe7adbb3a04c
Fixes: Coverity CID 1428754
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200515042231.18201-1-armbru@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/container: remove .instance_size initializer from container_info
Masahiro Yamada [Wed, 13 May 2020 03:36:00 +0000 (12:36 +0900)] 
qom/container: remove .instance_size initializer from container_info

You can omit .instance_size if it is the same as that of the parent.

    .class_size = sizeof(ObjectClass)

... is omitted here, so removing .instance_size is more consistent.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200513033600.2709646-1-masahiroy@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/object: pass (Object *) to object_initialize_with_type()
Masahiro Yamada [Tue, 12 May 2020 17:31:04 +0000 (02:31 +0900)] 
qom/object: pass (Object *) to object_initialize_with_type()

object_new_with_type() already passes (Object *) pointer.
Avoid casting back and forth.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200512173104.2293073-1-masahiroy@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/object: simplify type_initialize_interface()
Masahiro Yamada [Tue, 12 May 2020 18:25:01 +0000 (03:25 +0900)] 
qom/object: simplify type_initialize_interface()

iface_impl->class is the same as new_iface. Make it more readable.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200512182501.2300530-1-masahiroy@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/object: factor out the initialization of hash table of properties
Masahiro Yamada [Tue, 12 May 2020 17:26:15 +0000 (02:26 +0900)] 
qom/object: factor out the initialization of hash table of properties

Properties are not related to the initialization of interfaces.

The initialization of the hash table can be moved after the if-block,
and unified.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200512172615.2291999-1-masahiroy@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom: remove index from object_resolve_abs_path()
Masahiro Yamada [Sun, 10 May 2020 01:32:35 +0000 (10:32 +0900)] 
qom: remove index from object_resolve_abs_path()

You can advance 'parts' to track the current path fragment.
The 'index' parameter is unneeded.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200510013235.954906-1-masahiroy@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoaccel: Move Xen accelerator code under accel/xen/
Philippe Mathieu-Daudé [Fri, 8 May 2020 10:02:22 +0000 (12:02 +0200)] 
accel: Move Xen accelerator code under accel/xen/

This code is not related to hardware emulation.
Move it under accel/ with the other hypervisors.

Reviewed-by: Paul Durrant <paul@xen.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200508100222.7112-1-philmd@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Assert vmport initialized before registering commands
Liran Alon [Thu, 12 Mar 2020 16:54:31 +0000 (18:54 +0200)] 
hw/i386/vmport: Assert vmport initialized before registering commands

vmport_register() is also called from other modules such as vmmouse.
Therefore, these modules rely that vmport is realized before those call
sites. If this is violated, vmport_register() will NULL-deref.

To make such issues easier to debug, assert in vmport_register() that
vmport is already realized.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-17-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Add support for CMD_GETHZ
Liran Alon [Thu, 12 Mar 2020 16:54:30 +0000 (18:54 +0200)] 
hw/i386/vmport: Add support for CMD_GETHZ

This command returns to guest information on LAPIC bus frequency and TSC
frequency.

One can see how this interface is used by Linux vmware_platform_setup()
introduced in Linux commit 88b094fb8d4f ("x86: Hypervisor detection and
get tsc_freq from hypervisor").

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-16-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386/cpu: Store LAPIC bus frequency in CPU structure
Liran Alon [Thu, 12 Mar 2020 16:54:29 +0000 (18:54 +0200)] 
i386/cpu: Store LAPIC bus frequency in CPU structure

No functional change.
This information will be used by following patches.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-15-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Allow x2apic without IR
Liran Alon [Thu, 12 Mar 2020 16:54:28 +0000 (18:54 +0200)] 
hw/i386/vmport: Allow x2apic without IR

Signal to guest that hypervisor supports x2apic without VT-d/IOMMU
Interrupt-Remapping support. This allows guest to use x2apic in
case all APIC IDs fits in 8-bit (i.e. Max APIC ID < 255).

See Linux kernel commit 4cca6ea04d31 ("x86/apic: Allow x2apic
without IR on VMware platform") and Linux try_to_enable_x2apic()
function.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-14-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Add support for CMD_GET_VCPU_INFO
Liran Alon [Thu, 12 Mar 2020 16:54:27 +0000 (18:54 +0200)] 
hw/i386/vmport: Add support for CMD_GET_VCPU_INFO

Command currently returns that it is unimplemented by setting
the reserved-bit in it's return value.

Following patches will return various useful vCPU information
to guest.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-13-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Add support for CMD_GETBIOSUUID
Liran Alon [Thu, 12 Mar 2020 16:54:24 +0000 (18:54 +0200)] 
hw/i386/vmport: Add support for CMD_GETBIOSUUID

This is VMware documented functionallity that some guests rely on.
Returns the BIOS UUID of the current virtual machine.

Note that we also introduce a new compatability flag "x-cmds-v2" to
make sure to expose new VMPort commands only to new machine-types.
This flag will also be used by the following patches that will introduce
additional VMPort commands.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-10-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Define enum for all commands
Liran Alon [Thu, 12 Mar 2020 16:54:23 +0000 (18:54 +0200)] 
hw/i386/vmport: Define enum for all commands

No functional change.

Defining an enum for all VMPort commands have the following advantages:
* It gets rid of the error-prone requirement to update VMPORT_ENTRIES
when new VMPort commands are added to QEMU.
* It makes it clear to know by looking at one place at the source, what
are all the VMPort commands supported by QEMU.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200312165431.82118-9-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Introduce vmport.h
Liran Alon [Thu, 12 Mar 2020 16:54:22 +0000 (18:54 +0200)] 
hw/i386/vmport: Introduce vmport.h

No functional change. This is mere refactoring.

Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200312165431.82118-8-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Report vmware-vmx-type in CMD_GETVERSION
Liran Alon [Thu, 12 Mar 2020 16:54:21 +0000 (18:54 +0200)] 
hw/i386/vmport: Report vmware-vmx-type in CMD_GETVERSION

As can be seen from VmCheck_GetVersion() in open-vm-tools code,
CMD_GETVERSION should return vmware-vmx-type in ECX register.

Default is to fake host as VMware ESX server. But user can control
this value by "-global vmport.vmware-vmx-type=X".

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-7-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Introduce vmware-vmx-version property
Liran Alon [Thu, 12 Mar 2020 16:54:20 +0000 (18:54 +0200)] 
hw/i386/vmport: Introduce vmware-vmx-version property

vmware-vmx-version is a number returned from CMD_GETVERSION which specifies
to guest VMware Tools the the host VMX version. If the host reports a number
that is different than what the guest VMware Tools expects, it may force
guest to upgrade VMware Tools. (See comment above VERSION_MAGIC and
VmCheck_IsVirtualWorld() function in open-vm-tools open-source code).

For better readability and allow maintaining compatability for guests
which may expect different vmware-vmx-version, make vmware-vmx-version a
VMPort object property. This would allow user to control it's value via
"-global vmport.vmware-vmx-version=X".

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200312165431.82118-6-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Set EAX to -1 on failed and unsupported commands
Liran Alon [Thu, 12 Mar 2020 16:54:19 +0000 (18:54 +0200)] 
hw/i386/vmport: Set EAX to -1 on failed and unsupported commands

This is used as a signal for VMware Tools to know if a command it
attempted to invoke, failed or is unsupported. As a result, VMware Tools
will either report failure to user or fallback to another backdoor command
in attempt to perform some operation.

A few examples:
* open-vm-tools TimeSyncReadHost() function fallbacks to
CMD_GETTIMEFULL command when CMD_GETTIMEFULL_WITH_LAG
fails/unsupported.
* open-vm-tools Hostinfo_NestingSupported() function verifies
EAX != -1 to check for success.
* open-vm-tools Hostinfo_VCPUInfoBackdoor() functions checks
if reserved-bit is set to indicate command is unimplemented.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-5-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Propagate IOPort read to vCPU EAX register
Liran Alon [Thu, 12 Mar 2020 16:54:18 +0000 (18:54 +0200)] 
hw/i386/vmport: Propagate IOPort read to vCPU EAX register

vmport_ioport_read() returns the value that should propagate to vCPU EAX
register when guest reads VMPort IOPort (i.e. By x86 IN instruction).

However, because vmport_ioport_read() calls cpu_synchronize_state(), the
returned value gets overridden by the value in QEMU vCPU EAX register.
i.e. cpu->env.regs[R_EAX].

To fix this issue, change vmport_ioport_read() to explicitly override
cpu->env.regs[R_EAX] with the value it wish to propagate to vCPU EAX
register.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-4-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Add device properties
Liran Alon [Thu, 12 Mar 2020 16:54:17 +0000 (18:54 +0200)] 
hw/i386/vmport: Add device properties

No functional change.

This is done as a preparation for the following patches that will
introduce several device properties.

Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200312165431.82118-3-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohw/i386/vmport: Add reference to VMware open-vm-tools
Liran Alon [Thu, 12 Mar 2020 16:54:16 +0000 (18:54 +0200)] 
hw/i386/vmport: Add reference to VMware open-vm-tools

This official VMware open-source project can be used as reference to
understand how guest code interacts with VMPort virtual device. Thus,
providing understanding on how device is expected to behave.

Signed-off-by: Liran Alon <liran.alon@oracle.com>
Message-Id: <20200312165431.82118-2-liran.alon@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: fix phadd* with identical destination and source register
Janne Grunau [Wed, 1 Apr 2020 22:52:53 +0000 (00:52 +0200)] 
target/i386: fix phadd* with identical destination and source register

Detected by asm test suite failures in dav1d
(https://code.videolan.org/videolan/dav1d). Can be reproduced by
`qemu-x86_64 -cpu core2duo ./tests/checkasm --test=mc_8bpc 1659890620`.

Signed-off-by: Janne Grunau <j@jannau.net>
Message-Id: <20200401225253.30745-1-j@jannau.net>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Fix the CPUID leaf CPUID_Fn80000008
Babu Moger [Fri, 17 Apr 2020 21:55:13 +0000 (16:55 -0500)] 
target/i386: Fix the CPUID leaf CPUID_Fn80000008

CPUID leaf CPUID_Fn80000008_ECX provides information about the
number of threads supported by the processor. It was found that
the field ApicIdSize(bits 15-12) was not set correctly.

ApicIdSize is defined as the number of bits required to represent
all the ApicId values within a package.

Valid Values: Value Description
3h-0h Reserved.
4h up to 16 threads.
5h up to 32 threads.
6h up to 64 threads.
7h up to 128 threads.
Fh-8h Reserved.

Fix the bit appropriately.

This came up during following thread.
https://lore.kernel.org/qemu-devel/158643709116.17430.15995069125716778943.malonedeb@wampee.canonical.com/#t

Refer the Processor Programming Reference (PPR) for AMD Family 17h
Model 01h, Revision B1 Processors. The documentation is available
from the bugzilla Link below.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=206537
Reported-by: Philipp Eppelt <1871842@bugs.launchpad.net>
Signed-off-by: Babu Moger <babu.moger@amd.com>
Message-Id: <20200417215345.64800.73351.stgit@localhost.localdomain>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agovmbus: add infrastructure to save/load vmbus requests
Jon Doron [Fri, 24 Apr 2020 12:34:44 +0000 (15:34 +0300)] 
vmbus: add infrastructure to save/load vmbus requests

This can be allow to include controller-specific data while
saving/loading in-flight scsi requests of the vmbus scsi controller.

Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-7-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386: Hyper-V VMBus ACPI DSDT entry
Jon Doron [Fri, 24 Apr 2020 12:34:43 +0000 (15:34 +0300)] 
i386: Hyper-V VMBus ACPI DSDT entry

Guest OS uses ACPI to discover VMBus presence.  Add a corresponding
entry to DSDT in case VMBus has been enabled.

Experimentally Windows guests were found to require this entry to
include two IRQ resources. They seem to never be used but they still
have to be there.

Make IRQ numbers user-configurable via corresponding properties; use 7
and 13 by default.

Signed-off-by: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-6-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoi386:pc: whitelist dynamic vmbus-bridge
Jon Doron [Fri, 24 Apr 2020 12:34:42 +0000 (15:34 +0300)] 
i386:pc: whitelist dynamic vmbus-bridge

As vmbus-bridge is derived from sysbus device, it has to be whitelisted
to be allowed to be created with -device.

Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-5-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agovmbus: vmbus implementation
Jon Doron [Fri, 24 Apr 2020 12:34:41 +0000 (15:34 +0300)] 
vmbus: vmbus implementation

Add the VMBus infrastructure -- bus, devices, root bridge, vmbus state
machine, vmbus channel interactions, etc.

VMBus is a collection of technologies.  At its lowest layer, it's a message
passing and signaling mechanism, allowing efficient passing of messages to and
from guest VMs.  A layer higher, it's a mechanism for defining channels of
communication, where each channel is tagged with a type (which implies a
protocol) and a instance ID.  A layer higher than that, it's a bus driver,
serving as the basis of device enumeration within a VM, where a channel can
optionally be exposed as a paravirtual device.  When a server-side (paravirtual
back-end) component wishes to offer a channel to a guest VM, it does so by
specifying a channel type, a mode, and an instance ID.  VMBus then exposes this
in the guest.

More information about VMBus can be found in the file
vmbuskernelmodeclientlibapi.h in Microsoft's WDK.

TODO:
 - split into smaller palatable pieces
 - more comments
 - check and handle corner cases

Kudos to Evgeny Yakovlev (formerly eyakovlev@virtuozzo.com) and Andrey
Smetatin (formerly asmetanin@virtuozzo.com) for research and
prototyping.

Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-4-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agovmbus: add vmbus protocol definitions
Jon Doron [Fri, 24 Apr 2020 12:34:40 +0000 (15:34 +0300)] 
vmbus: add vmbus protocol definitions

Add a header with data structures and constants used in Hyper-V VMBus
hypervisor <-> guest interactions.

Based on the respective stuff from Linux kernel.

Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-3-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agohyperv: expose API to determine if synic is enabled
Jon Doron [Fri, 24 Apr 2020 12:34:39 +0000 (15:34 +0300)] 
hyperv: expose API to determine if synic is enabled

Signed-off-by: Jon Doron <arilou@gmail.com>
Message-Id: <20200424123444.3481728-2-arilou@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoMakefile: Let the 'help' target list the helper targets
Philippe Mathieu-Daudé [Thu, 23 Apr 2020 10:43:45 +0000 (12:43 +0200)] 
Makefile: Let the 'help' target list the helper targets

List the name of the helper targets when calling 'make help',
along with the tool targets:

  $ make help
  [...]

  Helper targets:
    fsdev/virtfs-proxy-helper      - Build virtfs-proxy-helper
    scsi/qemu-pr-helper            - Build qemu-pr-helper
    qemu-bridge-helper             - Build qemu-bridge-helper
    vhost-user-gpu                 - Build vhost-user-gpu
    virtiofsd                      - Build virtiofsd

  Tools targets:
    qemu-ga                        - Build qemu-ga tool
    qemu-keymap                    - Build qemu-keymap tool
    elf2dmp                        - Build elf2dmp tool
    ivshmem-client                 - Build ivshmem-client tool
    ivshmem-server                 - Build ivshmem-server tool
    qemu-nbd                       - Build qemu-nbd tool
    qemu-storage-daemon            - Build qemu-storage-daemon tool
    qemu-img                       - Build qemu-img tool
    qemu-io                        - Build qemu-io tool
    qemu-edid                      - Build qemu-edid tool

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoio/task: Move 'qom/object.h' header to source
Philippe Mathieu-Daudé [Mon, 4 May 2020 11:56:55 +0000 (13:56 +0200)] 
io/task: Move 'qom/object.h' header to source

We need "qom/object.h" to call object_ref()/object_unref(),
and to test the TYPE_DUMMY.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200504115656.6045-3-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/object: Move Object typedef to 'qemu/typedefs.h'
Philippe Mathieu-Daudé [Mon, 4 May 2020 11:56:54 +0000 (13:56 +0200)] 
qom/object: Move Object typedef to 'qemu/typedefs.h'

We use the Object type all over the place.
Forward declare it in "qemu/typedefs.h".

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200504115656.6045-2-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agotarget/i386: Fix OUTL debug output
Philippe Mathieu-Daudé [Sun, 17 May 2020 11:01:47 +0000 (13:01 +0200)] 
target/i386: Fix OUTL debug output

Fix OUTL instructions incorrectly displayed as OUTW.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200517110147.26026-1-f4bug@amsat.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoqom/object: Fix object_child_foreach_recursive() return value
Cédric Le Goater [Sat, 4 Apr 2020 15:33:40 +0000 (17:33 +0200)] 
qom/object: Fix object_child_foreach_recursive() return value

When recursing, the return value of do_object_child_foreach() is not
taken into account.

Cc: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Fixes: d714b8de7747 ("qom: Add recursive version of object_child_for_each")
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200404153340.164861-1-clg@kaod.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoicount: fix shift=auto for record/replay
Pavel Dovgalyuk [Tue, 19 May 2020 10:43:20 +0000 (13:43 +0300)] 
icount: fix shift=auto for record/replay

This patch fixes shift=auto when record/replay is enabled.
Now user does not need to guess the best shift value.

Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru>
--

v2:
  moved icount_time_shift to vmstate subsection
Message-Id: <158988500050.15192.692077802469400393.stgit@pasha-ThinkPad-X280>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agonuma: prevent usage of -M memory-backend and -numa memdev at the same time
Igor Mammedov [Mon, 11 May 2020 14:11:03 +0000 (10:11 -0400)] 
numa: prevent usage of -M memory-backend and -numa memdev at the same time

Options -M memory-backend and -numa memdev are mutually exclusive,
and if used together, it might lead to a crash in the worst case.
For example when the same backend is used with these options together:
  -m 4G \
  -object memory-backend-ram,id=mem0,size=4G \
  -M pc,memory-backend=mem0 \
  -numa node,memdev=mem0
QEMU will abort with:
   exec.c:2006: qemu_ram_set_idstr: Assertion `!new_block->idstr[0]' failed.

and following backtrace:
    abort ()
    qemu_ram_set_idstr ()
    vmstate_register_ram ()
    vmstate_register_ram_global ()
    machine_consume_memdev ()
    numa_init_memdev_container ()
    numa_complete_configuration ()
    machine_run_board_init ()

add a check to error out in case the user tries to use both options at
the same time.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20200511141103.43768-3-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agovl.c: run preconfig loop before creating default RAM backend
Igor Mammedov [Mon, 11 May 2020 14:11:02 +0000 (10:11 -0400)] 
vl.c: run preconfig loop before creating default RAM backend

Default RAM backend depends on numa_uses_legacy_mem(), which is
infulenced by -numa options on CLI or set-numa-node QMP command
at preconfig time. If QEMU is started with  '-preconfig'
without -numa, it will lead to creating default RAM backend
even if later set-numa-node is used to assing RAM to NUMA nodes
using 'memdev' NUMA option.
That at best will waste RAM object created by default and with
next patch adding a check to prevent usage of conflicting
 '-M memory-backend' and '-numa memdev'
options, it will make QEMU error out if user tries to configure
NUMA at preconfig time with memdev option, making set-numa-node
unusable.

To fix issue, move preconfig loop before default RAM backend is
created, so that numa_uses_legacy_mem() would take into account
effects of set-numa-node commands executed at preconfig time.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20200511141103.43768-2-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agorun-coverity-scan: support --update-tools-only --docker
Paolo Bonzini [Wed, 22 Apr 2020 14:38:57 +0000 (10:38 -0400)] 
run-coverity-scan: support --update-tools-only --docker

Just build the container when run-coverity-scan is invoked with
--update-tools-only --docker.  This requires moving the "docker build"
logic into the update_coverity_tools function.

The only snag is that --update-tools-only --docker requires access to
the dockerfile.  For now just report an error for --src-tarball, and
"docker build" will fail if not in a source tree.  Another possibility
could be to host our container images on a public registry, and use
"FROM qemu:fedora" to make the Dockerfile small enough that it can be
included directly in the run-coverity-scan script.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agorun-coverity-scan: download tools outside the container
Paolo Bonzini [Wed, 22 Apr 2020 15:49:40 +0000 (11:49 -0400)] 
run-coverity-scan: download tools outside the container

This lets us look at coverity_tool.md5 across executions of run-coverity-scan
and skip the download.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agorun-coverity-scan: use --no-update-tools in docker run
Paolo Bonzini [Wed, 22 Apr 2020 15:43:14 +0000 (11:43 -0400)] 
run-coverity-scan: use --no-update-tools in docker run

Tools are already updated via the docker build.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agorun-coverity-scan: add --no-update-tools option
Paolo Bonzini [Wed, 22 Apr 2020 15:37:55 +0000 (11:37 -0400)] 
run-coverity-scan: add --no-update-tools option

Provide a quick way to skip building the container while we figure out how
to get caching right.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agorun-coverity-scan: use docker.py
Paolo Bonzini [Wed, 22 Apr 2020 14:38:57 +0000 (10:38 -0400)] 
run-coverity-scan: use docker.py

Our trusted docker wrapper allows run-coverity-scan to run with both
docker and podman.

For the "run" phase this is transparent; for the "build" phase however
scripts are replaced with a bind mount (-v).  This is not an issue
because the secret option is meant for secrets stored globally in the
system and bind mounts are a valid substitute for secrets that are known
to whoever builds the container.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>