Eric Blake [Mon, 26 Mar 2018 18:41:47 +0000 (13:41 -0500)]
maint: Add .mailmap entries for patches claiming list authorship
The list did not author any patches, but it does rewrite the
'From:' header of messages sent from any domain with restrictive
SPF policies that would otherwise prevent the message from reaching
all list recipients. If a maintainer is not careful to undo the
list header rewrite, and the author did not include a manual
'From:' line in the body to fix the munged header, then 'git am'
happily attributes the patch to the list. Add some mailmap
entries to correct the few that have escaped our attention; while
we also work on improving the tooling to catch the problem in
the future before a merge is even made.
Also improve the comments occurring in the file, including line
length improvements.
Eric Blake [Tue, 27 Mar 2018 20:21:51 +0000 (15:21 -0500)]
dump: Fix build with newer gcc
gcc 8 on rawhide is picky enough to complain:
/home/dummy/qemu/dump.c: In function 'create_header32':
/home/dummy/qemu/dump.c:817:5: error: 'strncpy' output truncated before terminating nul copying 8 bytes from a string of the same length [-Werror=stringop-truncation]
strncpy(dh->signature, KDUMP_SIGNATURE, strlen(KDUMP_SIGNATURE));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
But we already have SIG_LEN defined as the right length without needing
to do a strlen(), and memcpy() is better than strncpy() when we know
we do not want a trailing NUL byte.
qemu-pr-helper: Daemonize before dropping privileges
After we've dropped privileges it might be not possible to write
pidfile. For instance, if this binary is run as root (because
user wants it to write pidfile to some privileged location)
writing pidfile fails because privileges are dropped before we
even get to that.
linzhecheng [Wed, 28 Mar 2018 13:34:35 +0000 (21:34 +0800)]
virtio-serial: fix heapover-flow
Check device having the feature of VIRTIO_CONSOLE_F_EMERG_WRITE before
get config->emerg_wr. It is neccessary because sizeof(virtio_console_config)
is 8 byte if VirtIOSerial doesn't have the feature of
VIRTIO_CONSOLE_F_EMERG_WRITE(see virtio_serial_device_realize),
read/write emerg_wr will lead to heap-over-flow.
Michael Chapman [Fri, 6 Apr 2018 05:34:06 +0000 (15:34 +1000)]
kvmclock: fix clock_is_reliable on migration from QEMU < 2.9
When migrating from a pre-2.9 QEMU, no clock_is_reliable flag is
transferred. We should assume that the source host has an unreliable
KVM_GET_CLOCK, rather than using whatever was determined locally, to
ensure that any drift from the TSC-based value calculated by the guest
is corrected.
The 40p machine type already creates the device i82374. If specified in the
command line, it will try to create it again, hence generating the error. The
function isa_bus_dma() isn't supposed to be called twice for the same bus.
Check the bus doesn't already have a DMA controller registered before creating
the device.
QEMU SCSI code makes assumptions about how the PROTECT and BYTCHK
works in the protocol, denying support for PI (Protection
Information) in case the guest OS requests it. However, in SCSI versions 2
and older, there is no PI concept in the protocol.
This means that when dealing with such devices:
- there is no PROTECT bit in byte 5 of the standard INQUIRY response. The
whole byte is marked as "Reserved";
- there is no RDPROTECT in byte 2 of READ. We have 'Logical Unit Number'
in this field instead;
- there is no VRPROTECT in byte 2 of VERIFY. We have 'Logical Unit Number'
in this field instead. This also means that the BYTCHK bit in this case
is not related to PI.
Since QEMU does not consider these changes, a SCSI passthrough using
a SCSI-2 device will not work. It will mistake these fields with
PI information and return Illegal Request SCSI SENSE thinking
that the driver is asking for PI support.
This patch fixes it by adding a new attribute called 'scsi_version'
that is read from the standard INQUIRY response of passthrough
devices. This allows for a version verification before applying
conditions related to PI that doesn't apply for older versions.
Paolo Bonzini [Thu, 5 Apr 2018 16:09:51 +0000 (18:09 +0200)]
scsi-disk: allow customizing the SCSI version
We would like to have different behavior for passthrough devices
depending on the SCSI version they expose. To prepare for that,
allow the user of emulated devices to specify the desired SCSI
level, and adjust the emulation according to the property value.
The next patch will set the level for scsi-block and scsi-generic
devices.
Fam Zheng [Tue, 27 Mar 2018 16:41:41 +0000 (00:41 +0800)]
scsi-disk: Don't enlarge min_io_size to max_io_size
Some backends report big max_io_sectors. Making min_io_size the same
value in this case will make it impossible for guest to align memory,
therefore the disk may not be usable at all.
Roman Kagan [Fri, 30 Mar 2018 17:02:09 +0000 (20:02 +0300)]
i386/hyperv: error out if features requested but unsupported
In order to guarantee compatibility on migration, QEMU should have
complete control over the features it announces to the guest via CPUID.
However, for a number of Hyper-V-related cpu properties, if the
corresponding feature is not supported by the underlying KVM, the
propery is silently ignored and the feature is not announced to the
guest.
Roman Kagan [Fri, 30 Mar 2018 17:02:08 +0000 (20:02 +0300)]
i386/hyperv: add hv-frequencies cpu property
In order to guarantee compatibility on migration, QEMU should have
complete control over the features it announces to the guest via CPUID.
However, the availability of Hyper-V frequency MSRs
(HV_X64_MSR_TSC_FREQUENCY and HV_X64_MSR_APIC_FREQUENCY) depends solely
on the support for them in the underlying KVM.
Introduce "hv-frequencies" cpu property (off by default) which gives
QEMU full control over whether these MSRs are announced.
While at this, drop the redundant check of the cpu tsc frequency, and
decouple this feature from hv-time.
Implements the CPUID trap for CPUID 1 to include the
CPUID_EXT_HYPERVISOR flag in the ECX results. This was preventing some
older linux kernels from booting when trying to access MSR's that dont
make sense when virtualized.
memfd: fix vhost-user-test on non-memfd capable host
On RHEL7, memfd is not supported, and vhost-user-test fails:
TEST: tests/vhost-user-test... (pid=10248)
/x86_64/vhost-user/migrate:
qemu-system-x86_64: -object memory-backend-memfd,id=mem,size=2M,: failed to create memfd
FAIL
There is a qemu_memfd_check() to prevent running memfd path, but it
also checks for fallback implementation. Let's specialize
qemu_memfd_check() to check memfd only, while qemu_memfd_alloc_check()
checks for the qemu_memfd_alloc() API.
Su Hang [Mon, 26 Mar 2018 02:06:22 +0000 (10:06 +0800)]
scripts/checkpatch.pl: Bug fix
Commit 2b9aef6fcd96ba7ed8c1ee723e391901852d344c introduced a regression:
checkpatch.pl started complaining about the following valid pattern:
do {
/* something */
} while (condition);
In commit 7073fbada733c8d10992f00772c9b9299d740e9b, the `andn` instruction
was implemented via `tcg_gen_andc` but passes the operands in the wrong
order:
- X86 defines `andn dest,src1,src2` as: dest = ~src1 & src2
- TCG defines `andc dest,src1,src2` as: dest = src1 & ~src2
Jeff Cody [Wed, 4 Apr 2018 15:40:45 +0000 (11:40 -0400)]
block/rbd: remove processed options from qdict
Commit 4bfb274 added some QAPIfication of option parsing in
qemu_rbd_open(). We need to remove all the options we processed,
otherwise in bdrv_open_inherit() we will think the remaining options are
invalid.
(This needs to go in 2.12 to avoid a regression that prevents rbd
from being opened.)
Peter Maydell [Wed, 4 Apr 2018 13:00:07 +0000 (14:00 +0100)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Block layer patches
# gpg: Signature made Tue 03 Apr 2018 16:48:53 BST
# gpg: using RSA key 7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <[email protected]>"
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74 56FE 7F09 B272 C88F 2FD6
* remotes/kevin/tags/for-upstream:
iotests: Test abnormally large size in compressed cluster descriptor
qemu-iotests: Use ppc64 qemu_arch on ppc64le host
iotests: Test preallocated truncate of 2G image
block/file-posix: Fix fully preallocated truncate
iotests: fix 208 for luks format
iotests: Update 186 after commit ac64273c66ab136c44043259162
iotests: Update 051 and 186 after commit 1454509726719e0933c
block: handle invalid lseek returns gracefully
gluster: Fix blockdev-add with server.N.type=unix
* remotes/cody/tags/block-pull-request:
gluster: Fix blockdev-add with server.N.type=unix
blockjob: use qapi enum helpers
blockjob: leak fix, remove from txn when failing early
Peter Maydell [Tue, 3 Apr 2018 22:23:58 +0000 (23:23 +0100)]
Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20180403' into staging
- fix a memory leak in the ipl code introduced with this release
- increase timeout in the bios to avoid hangs during migration (and
rebuild bios to activate the change)
Peter Maydell [Tue, 3 Apr 2018 18:02:46 +0000 (19:02 +0100)]
Merge remote-tracking branch 'remotes/xtensa/tags/20180402-xtensa' into staging
xtensa-specific fixes for linux-user:
- fix flushing registers for signal processing in call8 and call12 frames;
- fix PC value for restarted syscalls;
- fix sysv IPC structures;
- fix fadvise64 syscall;
generic fixes for linux-user:
- fix QEMU assertion in multithreaded application by calling cpu_copy
under clone_lock;
- fix mq_getsetattr implementation;
- fix error propagation in clock_gettime;
- implement clock_settime.
Kevin Wolf [Tue, 3 Apr 2018 15:48:45 +0000 (17:48 +0200)]
Merge remote-tracking branch 'mreitz/tags/pull-block-2018-04-03' into queue-block
A fix for preallocated truncation, a new iotest, and a fix to make the iotests work more comfortably on ppc64
# gpg: Signature made Tue Apr 3 17:40:57 2018 CEST
# gpg: using RSA key F407DB0061D5CF40
# gpg: Good signature from "Max Reitz <[email protected]>"
# Primary key fingerprint: 91BE B60A 30DB 3E88 57D1 1829 F407 DB00 61D5 CF40
* mreitz/tags/pull-block-2018-04-03:
iotests: Test abnormally large size in compressed cluster descriptor
qemu-iotests: Use ppc64 qemu_arch on ppc64le host
iotests: Test preallocated truncate of 2G image
block/file-posix: Fix fully preallocated truncate
Alberto Garcia [Thu, 29 Mar 2018 12:07:45 +0000 (15:07 +0300)]
iotests: Test abnormally large size in compressed cluster descriptor
L2 entries for compressed clusters have a field that indicates the
number of sectors used to store the data in the image.
That's however not the size of the compressed data itself, just the
number of sectors where that data is located. The actual data size is
usually not a multiple of the sector size, and therefore cannot be
represented with this field.
The way it works is that QEMU reads all the specified sectors and
starts decompressing the data until there's enough to recover the
original uncompressed cluster. If there are any bytes left that
haven't been decompressed they are simply ignored.
One consequence of this is that even if the size field is larger than
it needs to be QEMU can handle it just fine: it will read more data
from disk but it will ignore the extra bytes.
This test creates an image with two compressed clusters that use 5
sectors (2.5 KB) each, increases the size field to the maximum (8192
sectors, or 4 MB) and verifies that the data can be read without
problems.
This test is important because while the decompressed data takes
exactly one cluster, the maximum value allowed in the compressed size
field is twice the cluster size. So although QEMU won't produce images
with such large values we need to make sure that it can handle them.
Another effect of increasing the size field is that it can make
it include data from the following host cluster(s). In this case
'qemu-img check' will detect that the refcounts are not correct, and
we'll need to rebuild them.
Additionally, this patch also tests that decreasing the size corrupts
the image since the original data can no longer be recovered. In this
case QEMU returns an error when trying to read the compressed data,
but 'qemu-img check' doesn't see anything wrong if the refcounts are
consistent.
One possible task for the future is to make 'qemu-img check' verify
the sizes of the compressed clusters, by trying to decompress the data
and checking that the size stored in the L2 entry is correct.
Lukáš Doktor [Thu, 29 Mar 2018 11:20:53 +0000 (13:20 +0200)]
qemu-iotests: Use ppc64 qemu_arch on ppc64le host
The qemu target does not always correspond to the host machine type. For
example ppc64le machine target is ppc64. Let's introduce "qemu_arch"
variable to store the matching qemu architecture related to the current
architecture and use it when auto-detecting the default qemu binary.
Max Reitz [Wed, 28 Feb 2018 13:13:14 +0000 (14:13 +0100)]
block/file-posix: Fix fully preallocated truncate
Storing the lseek() result in an int results in it overflowing when the
file is at least 2 GB big. Then, we have a 50 % chance of the result
being "negative" and thus thinking an error occurred when actually
everything went just fine.
So we should use the correct type for storing the result: off_t.
Peter Maydell [Tue, 3 Apr 2018 14:59:50 +0000 (15:59 +0100)]
Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2018-04-02' into staging
nbd patches for 2018-04-02
- Eric Blake: nbd: Fix 32-bit compilation on BLOCK_STATUS
- Eric Blake: nbd/client: Correctly handle bad server REP_META_CONTEXT
- Eric Blake: nbd: trace meta context negotiation
* remotes/ericb/tags/pull-nbd-2018-04-02:
nbd: trace meta context negotiation
nbd/client: Correctly handle bad server REP_META_CONTEXT
nbd: Fix 32-bit compilation on BLOCK_STATUS
Commit ac64273c66ab136c44 modified the output of iotest 186, changing
the QOM path of floppy drives from /machine/unattached/device[17] to
/machine/unattached/device[13].
Instead of updating the test output to reflect this change, this patch
adds a new filter that hides all QOM paths from the 'Attached to:'
line of the 'info block' command.
Kevin Wolf [Tue, 3 Apr 2018 11:08:10 +0000 (13:08 +0200)]
gluster: Fix blockdev-add with server.N.type=unix
The legacy command line interface gets the socket path from an option
called 'socket'. QAPI in contract uses SocketAddress, where the
corresponding option is called 'path'.
Fix the gluster block driver to accept both 'socket' and 'path', with
'path' being the preferred syntax.
blockjob: leak fix, remove from txn when failing early
This fixes leaks found by ASAN such as:
GTESTER tests/test-blockjob
=================================================================
==31442==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 24 byte(s) in 1 object(s) allocated from:
#0 0x7f88483cba38 in __interceptor_calloc (/lib64/libasan.so.4+0xdea38)
#1 0x7f8845e1bd77 in g_malloc0 ../glib/gmem.c:129
#2 0x7f8845e1c04b in g_malloc0_n ../glib/gmem.c:360
#3 0x5584d2732498 in block_job_txn_new /home/elmarco/src/qemu/blockjob.c:172
#4 0x5584d2739b28 in block_job_create /home/elmarco/src/qemu/blockjob.c:973
#5 0x5584d270ae31 in mk_job /home/elmarco/src/qemu/tests/test-blockjob.c:34
#6 0x5584d270b1c1 in do_test_id /home/elmarco/src/qemu/tests/test-blockjob.c:57
#7 0x5584d270b65c in test_job_ids /home/elmarco/src/qemu/tests/test-blockjob.c:118
#8 0x7f8845e40b69 in test_case_run ../glib/gtestutils.c:2255
#9 0x7f8845e40f29 in g_test_run_suite_internal ../glib/gtestutils.c:2339
#10 0x7f8845e40fd2 in g_test_run_suite_internal ../glib/gtestutils.c:2351
#11 0x7f8845e411e9 in g_test_run_suite ../glib/gtestutils.c:2426
#12 0x7f8845e3fe72 in g_test_run ../glib/gtestutils.c:1692
#13 0x5584d270d6e2 in main /home/elmarco/src/qemu/tests/test-blockjob.c:377
#14 0x7f8843641f29 in __libc_start_main (/lib64/libc.so.6+0x20f29)
Add an assert to make sure that the job doesn't have associated txn before free().
[Jeff Cody: N.B., used updated patch provided by John Snow]
Jeff Cody [Tue, 3 Apr 2018 04:37:26 +0000 (00:37 -0400)]
block: handle invalid lseek returns gracefully
In commit 223a23c198787328ae75bc65d84edf5fde33c0b6, we implemented a
workaround in the gluster driver to handle invalid values returned for
SEEK_DATA or SEEK_HOLE.
In some instances, these same invalid values can be seen in the posix
file handler as well - for example, it has been reported on FUSE gluster
mounts.
Calling assert() for these invalid values is overly harsh; we can safely
return -EIO and allow this case to be treated as a "learned nothing"
case (e.g., D4 / H4, as commented in the code).
This patch does the same thing that 223a23c198787 did for gluster.c,
except in file-posix.c
Kevin Wolf [Tue, 3 Apr 2018 10:58:32 +0000 (12:58 +0200)]
gluster: Fix blockdev-add with server.N.type=unix
The legacy command line interface gets the socket path from an option
called 'socket'. QAPI in contract uses SocketAddress, where the
corresponding option is called 'path'.
Fix the gluster block driver to accept both 'socket' and 'path', with
'path' being the preferred syntax.
Cornelia Huck [Thu, 29 Mar 2018 14:13:37 +0000 (10:13 -0400)]
pc-bios/s390-ccw: update image
Contains the following commits:
- pc-bios/s390-ccw: Move string arrays from bootmap header to .c file
- pc-bios/s390-ccw: Increase virtio timeout to 30 seconds
Thomas Huth [Thu, 29 Mar 2018 09:37:31 +0000 (11:37 +0200)]
pc-bios/s390-ccw: Increase virtio timeout to 30 seconds
The current timeout is set to only three seconds - and considering that
vring_wait_reply() or rather get_second() is not doing any rounding,
the real timeout is likely rather 2 seconds in most cases. When the
host is really badly loaded, it's possible that we hit this timeout by
mistake; it's even more likely if we run the guest in TCG mode instead
of KVM.
So let's increase the timeout to 30 seconds instead to ease this situation
(30 seconds is also the timeout that is used by the Linux SCSI subsystem
for example, so this seems to be a sane value for block IO timeout).
Eric Blake [Thu, 29 Mar 2018 23:18:37 +0000 (18:18 -0500)]
nbd/client: Correctly handle bad server REP_META_CONTEXT
It's never a good idea to blindly read for size bytes as
returned by the server without first validating that the size
is within bounds; a malicious or buggy server could cause us
to hang or get out of sync from reading further messages.
It may be smarter to try and teach the client to cope with
unexpected context ids by silently ignoring them instead of
hanging up on the server, but for now, if the server doesn't
reply with exactly the one context we expect, it's easier to
just give up - however, if we give up for any reason other
than an I/O failure, we might as well try to politely tell
the server we are quitting rather than continuing.
Eric Blake [Tue, 27 Mar 2018 21:05:17 +0000 (16:05 -0500)]
nbd: Fix 32-bit compilation on BLOCK_STATUS
iotests 123 and 209 fail on 32-bit platforms. The culprit:
sizeof(extent) is wrong; we want sizeof(*extent). But since
the struct is 8 bytes, it happened to work on 64-bit platforms
where the pointer is also 8 bytes (nasty).
Max Filippov [Sun, 1 Apr 2018 20:14:04 +0000 (13:14 -0700)]
linux-user: fix error propagation in clock_gettime
host_to_target_timespec may return error if target address could not be
locked, but it is ignored.
Propagate return value of host_to_target_timespec to the caller of
clock_gettime.
- make target_ipc_perm fields match kernel definitions for xtensa;
- add target_semid64_ds with proper order of times and reserved fields
for little/big endian specific for xtensa;
- add missing reserved fields after time fields to the target_shmid_ds;
- fix types of shm_cpid, shm_lpid and shm_nattch fields of
target_shmid_ds to match kernel definitions for xtensa.
These changes fix guest ipcs output and fix glibc testsuite tests
sysvipc/test-sysvsem and sysvipc/test-sysvshm.
Max Filippov [Sat, 31 Mar 2018 15:20:15 +0000 (08:20 -0700)]
linux-user: fix mq_getsetattr implementation
mq_getsetattr implementation does not set errno correctly in case of
error. Also in the presence of both 2nd and 3rd arguments it calls both
mq_getattr and mq_setattr, whereas only the latter call would suffice.
Don't call mq_getattr in the presence of the 2nd argument. Don't copy
output back to user in case of error. Use get_errno to set errno value.
This fixes test rt/tst-mqueue2 from the glibc testsuite.
Max Filippov [Fri, 30 Mar 2018 13:19:58 +0000 (06:19 -0700)]
linux-user: call cpu_copy under clone_lock
cpu_copy adds newly created CPU object to container/machine/unattached,
but does it w/o proper locking. As a result when multiple threads create
threads rapidly QEMU may abort with the following message:
Max Filippov [Wed, 28 Mar 2018 07:25:22 +0000 (00:25 -0700)]
target/xtensa: fix flush_window_regs
flush_window_regs uses wrong stack frame to save overflow registers in
call8 and call12 frames, which results in wrong register values in
callers of a function that received a signal.
Reimplement flush_window_regs closely following window overflow
sequence.
Peter Maydell [Sat, 31 Mar 2018 08:42:33 +0000 (09:42 +0100)]
Merge remote-tracking branch 'remotes/riscv/tags/riscv-qemu-2.12-critical-fixes' into staging
RISC-V: Critical fixes for QEMU 2.12
This series includes changes that are considered release critical,
such as floating point register file corruption under SMP Linux
due to incorrect handling of mstatus.FS.
This workaround will be replaced with a more comprehensive fix
for mstatus.FS handling in QEMU 2.13.
# gpg: Signature made Thu 29 Mar 2018 18:22:42 BST
# gpg: using DSA key 6BF1D7B357EF3E4F
# gpg: Good signature from "Michael Clark <[email protected]>"
# gpg: aka "Michael Clark <[email protected]>"
# gpg: aka "Michael Clark <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 7C99 930E B17C D8BA 073D 5EFA 6BF1 D7B3 57EF 3E4F
* remotes/riscv/tags/riscv-qemu-2.12-critical-fixes:
RISC-V: Workaround for critical mstatus.FS bug
Peter Maydell [Sat, 31 Mar 2018 07:39:08 +0000 (08:39 +0100)]
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging
# gpg: Signature made Fri 30 Mar 2018 04:49:42 BST
# gpg: using RSA key EF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F 3562 EF04 965B 398D 6211
* remotes/jasowang/tags/net-pull-request:
qemu-doc: Rework the network options chapter to make "-net" less prominent
Peter Maydell [Fri, 30 Mar 2018 22:05:19 +0000 (23:05 +0100)]
Merge remote-tracking branch 'remotes/stefanberger/tags/pull-tpm-2018-03-29-1' into staging
Merge tpm 2018/03/29 v1
# gpg: Signature made Fri 30 Mar 2018 01:04:47 BST
# gpg: using RSA key 75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B818 B9CA DF90 89C2 D5CE C66B 75AD 6580 2A0B 4211
* remotes/stefanberger/tags/pull-tpm-2018-03-29-1:
tests: Tests more flags of the CRB interface
tpm: CRB: Enforce locality is requested before processing buffer
tpm: CRB: Reset Granted flag when relinquishing locality
tpm: CRB: set the Idle flag by default
Thomas Huth [Mon, 12 Mar 2018 12:20:50 +0000 (13:20 +0100)]
qemu-doc: Rework the network options chapter to make "-net" less prominent
"-net" is clearly a legacy option. Yet we still use it in almost all
examples in the qemu documentation, and many other spots in the network
chapter. We should make it less prominent that users are not lured into
using it so often anymore. So instead of starting the network chapter with
"-net nic" and documenting "-net <backend>" below "-netdev <backend>"
everywhere, all the "-net" related documentation is now moved to the end
of the chapter. The new "-nic" option is moved to the beginning of the
chapter instead, with a new example that should demonstrate how "-nic"
can be used to shortcut "-device" with "-netdev". The examples in this
chapter are changed to use the "-device" and "-netdev" options or
"-nic" instead of "-net nic -net <backend>".
While we're at it, also remove a legacy remark about very old Linux
distributions. Also remove the "[...]" from the examples in this chapter
since we are not using this ellipsis in any other examples in our docu-
mentation.
Stefan Berger [Wed, 28 Mar 2018 19:39:17 +0000 (15:39 -0400)]
tpm: CRB: Enforce locality is requested before processing buffer
Section 5.5.3.2.2 of the CRB specs states that use of the TPM
through the localty control method must first be requested,
otherwise the command will be dropped.
Michael Clark [Fri, 23 Mar 2018 23:49:07 +0000 (16:49 -0700)]
RISC-V: Workaround for critical mstatus.FS bug
This change is a workaround for a bug where mstatus.FS
is not correctly reporting dirty after operations that
modify floating point registers. This a critical bug
or RISC-V in QEMU as it results in floating point
register file corruption when running SMP Linux due to
task migration and possibly uniprocessor Linux if
more than one process is using the FPU.
This workaround will return dirty if mstatus.FS is
switched from off to initial or clean. According to
the specification it is legal for an implementation
to return only off, or dirty.
migration: Don't activate block devices if using -S
Activating the block devices causes the locks to be taken on
the backing file. If we're running with -S and the destination libvirt
hasn't started the destination with 'cont', it's expecting the locks are
still untaken.
Don't activate the block devices if we're not going to autostart the VM;
'cont' already will do that anyway.
Direct leak of 16 byte(s) in 1 object(s) allocated from:
#0 0x7fe1abb80a38 in __interceptor_calloc (/lib64/libasan.so.4+0xdea38)
#1 0x7fe1aaf1bf75 in g_malloc0 ../glib/gmem.c:124
#2 0x7fe1aaf1c249 in g_malloc0_n ../glib/gmem.c:355
#3 0x55f4841cfaa9 in postcopy_ram_fault_thread /home/elmarco/src/qemu/migration/postcopy-ram.c:596
#4 0x55f48479447b in qemu_thread_start /home/elmarco/src/qemu/util/qemu-thread-posix.c:504
#5 0x7fe1a043550a in start_thread (/lib64/libpthread.so.0+0x750a)
Peter Maydell [Wed, 28 Mar 2018 21:13:38 +0000 (22:13 +0100)]
Merge remote-tracking branch 'remotes/riscv/tags/riscv-qemu-2.12-important-fixes' into staging
RISC-V: Important fixes for QEMU 2.12
This series includes changes that are considered important.
i.e. correct user-visible bugs that are exercised by common
operations such as -cpu list (CPU model changes) or -d in_asm
(fix for disassembly of addiw)
# gpg: Signature made Wed 28 Mar 2018 21:34:57 BST
# gpg: using DSA key 6BF1D7B357EF3E4F
# gpg: Good signature from "Michael Clark <[email protected]>"
# gpg: aka "Michael Clark <[email protected]>"
# gpg: aka "Michael Clark <[email protected]>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 7C99 930E B17C D8BA 073D 5EFA 6BF1 D7B3 57EF 3E4F
* remotes/riscv/tags/riscv-qemu-2.12-important-fixes:
RISC-V: Fix incorrect disassembly for addiw
RISC-V: Convert cpu definition to future model
Michael Clark [Fri, 23 Mar 2018 08:07:01 +0000 (01:07 -0700)]
RISC-V: Fix incorrect disassembly for addiw
This fixes a bug in the disassembler constraints used
to lift instructions into pseudo-instructions, whereby
addiw instructions are always lifted to sext.w instead
of just lifting addiw with a zero immediate.
An associated fix has been made to the metadata used to
machine generate the disseasembler:
Michael Clark [Thu, 8 Mar 2018 22:12:31 +0000 (11:12 +1300)]
RISC-V: Convert cpu definition to future model
- Model borrowed from target/sh4/cpu.c
- Rewrote riscv_cpu_list to use object_class_get_list
- Dropped 'struct RISCVCPUInfo' and used TypeInfo array
- Replaced riscv_cpu_register_types with DEFINE_TYPES
- Marked base class as abstract
- Fixes -cpu list
Failure to do so results in the tcg optimizer sign-extending
any constant fold from 32-bits. This turns out to be visible
in the RISC-V testsuite using a host that emits these opcodes
(e.g. any non-x86_64).
GDB crashes when qemu exits (at least on sparc-softmmu):
Remote communication error. Target disconnected.: Connection reset by peer.
Quitting: putpkt: write failed: Broken pipe.
So send a packet to exit GDB before we exit QEMU:
[Inferior 1 (Thread 0) exited normally]
Commit 137b5cb6 refactored 'info cpus' output, changing
'thread_id' to 'thread-id'. While HMP is not a stable
interface, it is trivial to keep the spelling consistent
for test frameworks that have not yet updated to using QMP.
This patch just reverts back output format to 'thread_id'.
Peter Xu [Mon, 26 Mar 2018 06:39:01 +0000 (14:39 +0800)]
tests: qmp-test: add test for new "x-oob"
Test the new OOB capability. It's mostly the reverted OOB test
(see commit 4fd78ad7), but differs in that:
- It uses the new qtest_init_without_qmp_handshake() parameter to
create the monitor with "x-oob"
- Squashed the capability tests on greeting message
- Don't use qtest_global any more, instead use self-maintained
QTestState, which is the trend
Eric Blake [Tue, 27 Mar 2018 01:36:19 +0000 (20:36 -0500)]
tests: Add parameter to qtest_init_without_qmp_handshake
Allow callers to choose whether to allow OOB support during a test;
for now, all existing callers pass false, but the next patch will
add a new caller. Also, rewrite the monitor setup to be generic
(using the -qmp shorthand is insufficient for honoring the parameter).
It is not allowed if either MUX or non-QMP is detected, since
Out-Of-Band is currently only for QMP, and non-MUX chardev backends.
Note that the client STILL has to request 'oob' during qmp_capabilities;
in part because the x-oob command line option may disappear in the
future if we decide the capabilities negotiation is sufficient.
Peter Xu [Mon, 26 Mar 2018 06:38:55 +0000 (14:38 +0800)]
qmp: cleanup qmp queues properly
Marc-André Lureau reported that we can have this happen:
1. client1 connects, send command C1
2. client1 disconnects before getting response for C1
3. client2 connects, who might receive response of C1
However client2 should not receive remaining responses for client1.
Basically, we should clean up the request/response queue elements when:
- after a session is closed
- before destroying the queues
Some helpers are introduced to achieve that. We need to make sure we're
with the lock when operating on those queues. This also needed the
declaration of QMPRequest moved earlier.
Peter Xu [Mon, 26 Mar 2018 06:38:58 +0000 (14:38 +0800)]
tests: let qapi-schema tests detect oob
The allow_oob parameter was passed in but not used in tests. Now
reflect that in the tests, so we need to touch up other command testers
with that new change.
Eric Blake [Fri, 23 Mar 2018 20:43:41 +0000 (15:43 -0500)]
tests: Silence false positive warning on generated test name
Running 'make check' on rawhide with gcc 8.0.1 fails:
tests/test-visitor-serialization.c: In function 'main':
tests/test-visitor-serialization.c:1127:34: error: '/primitives/' directive writing 12 bytes into a region of size between 1 and 128 [-Werror=format-overflow=]
The warning is a false positive (we have two buffers of size 128,
so yes, if we FULLY used the first buffer, then sprint'ing it into
the second will overflow the second). But in practice, our first
buffer will not be longer than "/visitor/serialization/String",
so sizing it smaller is enough to let gcc see that we don't
overflow the second.
projects / qemu.git / log