Migration pull 2018-10-11
With one bonus HMP fix.
# gpg: Signature made Thu 11 Oct 2018 20:23:12 BST
# gpg: using RSA key 0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>"
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A 9FA9 0516 331E BC5B FDE7
* remotes/dgilbert/tags/pull-migration-20181011a:
migration-test: Only generate a single target architecture
qmp, hmp: make subsystem/system-vendor identities optional
vhost-user: Don't ask for reply on postcopy mem table set
vhost-user: Fix userfaultfd leak
migration: Stop postcopy fault thread before notifying
tests/migration: Enable the migration test on s390x, too
tests: Add migration test for aarch64
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Several changes:
- We only allow generate header "inside" the tree. Why? Because we
need to connit the result, so it makes no sense to generate them on
the build dir.
- We only generate a single target each time. Getting all the
cross-compilers correctly is an impossible task. So know you do:
make -C tests/migration $target (native)
make CROSS_PREFIX=foo- -C tests/migratiion $target (cross)
And you are done.
- If we are building out of tree, we have no data about if we are
cross-compile or whatever. So instead of guess what is happening,
just do what I pointed on previous point.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20180913132313.11370-1-quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
According to PCI specification, subsystem id and subsystem vendor id
are present only in type 0 and type 2 headers (at different offsets),
but not in type 1 headers.
Thus we should make this data optional in struct PciDeviceId and skip
reporting them via HMP if the information is not available.
Additional (wrong information) about PCI bridges (Type1 devices) has been
added in 5383a705 and fortunately not released. This patch fixes that
problem. The problem was spotted by Markus.
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
Message-Id: <20181002135538.12113-1-den@openvz.org>
Reported-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
According to documentation, NEED_REPLY_MASK should not be set
for VHOST_USER_SET_MEM_TABLE request in postcopy mode.
This restriction was mistakenly applied to 'reply_supported'
variable, which is local and used only for non-postcopy case.
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
Fixes: 9bb3801994 ("vhost+postcopy: Send address back to qemu")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Message-Id: <20181002140947.4107-1-i.maximets@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
We can re-use the s390-ccw bios code to implement a small firmware
for a s390x guest which prints out the "A" and "B" characters and
modifies the memory, as required for the migration test.
[quintela: Converted the compile script to Makefile rules]
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <1539078677-25396-1-git-send-email-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Fixed up Makefile since the aarch patch sneaked in first
This patch adds migration test support for aarch64. The test code, which
implements the same functionality as x86, is booted as a kernel in qemu.
Here are the design choices we make for aarch64:
* We choose this -kernel approach because aarch64 QEMU doesn't provide a
built-in fw like x86 does. So instead of relying on a boot loader, we
use -kernel approach for aarch64.
* The serial output is sent to PL011 directly.
* The physical memory base for mach-virt machine is 0x40000000. We change
the start_address and end_address for aarch64.
In addition to providing the binary, this patch also includes the source
code and the build script in tests/migration/aarch64. So users can change
the source and/or re-compile the binary as they wish.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Andrew Jones <drjones@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Wei Huang <wei@redhat.com>
Message-Id: <1538669326-28135-1-git-send-email-wei@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Commit 2aece63c8a "hostmem: detect host backend memory is being used
properly" fixed "ivshmem-plain" to reject memory backends that are
already in use, and to block their deletion while in use. Two bugs
escaped review:
* New ivshmem_plain_exit() fails to call ivshmem_exit(). This breaks
unplug. Reproducer: migration after unplug still fails with
"Migration is disabled when using feature 'peer mode' in device
'ivshmem'".
* It failed to update legacy "ivshmem". Harmless, because it creates
the memory backend itself, and nothing else should use it.
Fix by moving the two host_memory_backend_set_mapped() calls into
ivshmem_common_realize() and ivshmem_exit(), guarded by s->hostmem.
Fixes: 2aece63c8a
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180926163709.22876-1-armbru@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Coverity complains (CID 1395628) that the multiply in the calculation
of the framebuffer base is performed as 32x32 but then used in a
context that takes a 64-bit hwaddr. This can't actually ever
overflow the 32-bit result, because of the constraints placed on
the s->config values in bcm2835_fb_validate_config(). But we
can placate Coverity anyway, by explicitly casting one of the
inputs to a hwaddr, so the whole expression is calculated with
64-bit arithmetic.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20181005133012.26490-1-peter.maydell@linaro.org
Add the v8M stack checks for the VLDM/VSTM
(aka VPUSH/VPOP) instructions. This code is currently
unreachable because we haven't yet implemented M profile
floating point support, but since the change is simple,
we add it now because otherwise we're likely to forget to
do it later.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-13-peter.maydell@linaro.org
Add the v8M stack checks for:
* LDM (T2 encoding)
* STM (T2 encoding)
This includes the 32-bit encodings of the instructions listed
in v8M ARM ARM rule R_YVWT as
* LDM, LDMIA, LDMFD
* LDMDB, LDMEA
* POP (multiple registers)
* PUSH (muliple registers)
* STM, STMIA, STMEA
* STMDB, STMFD
We perform the stack limit before doing any other part
of the load or store.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-10-peter.maydell@linaro.org
Add the v8M stack checks for:
* LDRD (immediate)
* STRD (immediate)
Loads and stores are more complicated than ADD/SUB/MOV, because we
must ensure that memory accesses below the stack limit are not
performed, so we can't simply do the check when we actually update
SP.
For these instructions, if the stack limit check triggers
we must not:
* perform any memory access below the SP limit
* update PC, SP or the load/store base register
but it is IMPDEF whether we:
* perform any accesses above or equal to the SP limit
* update destination registers for loads
For QEMU we choose to always check the limit before doing any other
part of the load or store, so we won't update any registers or
perform any memory accesses.
It is UNKNOWN whether the limit check triggers for a load or store
where the initial SP value is below the limit and one of the stores
would be below the limit, but the writeback moves SP to above the
limit. For QEMU we choose to trigger the check in this situation.
Note that limit checks happen only for loads and stores which update
SP via writeback; they do not happen for loads and stores which
simply use SP as a base register.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-9-peter.maydell@linaro.org
Check the v8M stack limits when pushing the frame for a
non-secure function call via BLXNS.
In order to be able to generate the exception we need to
promote raise_exception() from being local to op_helper.c
so we can call it from helper.c.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-8-peter.maydell@linaro.org
Add checks for breaches of the v8M stack limit when the
stack pointer is decremented to push the exception frame
for exception entry.
Note that the exception-entry case is unique in that the
stack pointer is updated to be the limit value if the limit
is hit (per rule R_ZLZG).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-7-peter.maydell@linaro.org
Add some comments to the Thumb decoder indicating what bits
of the instruction have been decoded at various points in
the code.
This is not an exhaustive set of comments; we're gradually
adding comments as we work with particular bits of the code.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-6-peter.maydell@linaro.org
Add code to insert calls to a helper function to do the stack
limit checking when we handle these forms of instruction
that write to SP:
* ADD (SP plus immediate)
* ADD (SP plus register)
* SUB (SP minus immediate)
* SUB (SP minus register)
* MOV (register)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20181002163556.10279-5-peter.maydell@linaro.org
The Arm v8M architecture includes hardware stack limit checking.
When certain instructions update the stack pointer, if the new
value of SP is below the limit set in the associated limit register
then an exception is taken. Add a TB flag that tracks whether
the limit-checking code needs to be emitted.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20181002163556.10279-2-peter.maydell@linaro.org
In commit c79c0a314c we enabled emulation of external aborts
when the guest attempts to access a physical address with no
mapped device. In commit 4672cbd7be we suppress this for
most legacy boards to prevent breakage of previously working
guests, but we didn't suppress it in the 'virt' board, with
the rationale "we know that guests won't try to prod devices
that we don't describe in the device tree or ACPI tables". This
is mostly true, but we've had a report of a Linux guest image
that this did break. The problem seems to be that the guest
is (incorrectly) configured with a DEBUG_UART_PHYS value that
tells it there is a uart at 0x10009000 (which is true for
vexpress but not for virt), so in early bootup the kernel
probes this bogus address.
This is a misconfigured guest, so we don't need to worry
about it too much, but we can arrange that guests that ran
on QEMU v2.10 (before c79c0a314c) will still run on
the "virt-2.10" board model, by suppressing external aborts
only for that version and earlier. This seems a reasonable
compromise: "virt-2.10" is supposed to behave the same way
that "virt" did in the 2.10 release, and making it do that
provides a usable workaround for guests with bugs like this.
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20180925144127.31965-1-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Peter Maydell