Peter Maydell [Fri, 24 May 2013 15:26:54 +0000 (16:26 +0100)]
configure: Require libfdt for arm, ppc, microblaze softmmu targets
A number of our softmmu targets (PPC, ARM, Microblaze) now more or
less require flattened device tree support for various board models
to work correctly. Make libfdt mandatory if the target list includes
these, rather than building unhelpful half-functional binaries.
Currently QEMU provides a local clone of the file libfdt_env.h in
/include. This file is supposed to come with the libfdt package and is
only needed for broken installs of libfdt. Now that we have submodule
dtc, just ignore these broken installs and prompt for the dtc submodule
install instead. QEMU's local libfdt_env.h is removed accordingly.
Manifests as a bug when building QEMU with modern libfdt. The new
version of libfdt does not compile when QEMUs libfdt_env.h takes
precedence over the hosts.
Paolo Bonzini [Tue, 4 Jun 2013 12:45:26 +0000 (14:45 +0200)]
build: do not use TARGET_ARCH
TARGET_ARCH is generally wrong to use, there are better variables
provided in config-target.mak. The right one is usually TARGET_NAME
(previously TARGET_ARCH2), but for bsd-user we can also use TARGET_ABI_DIR
for consistency with linux-user.
Alon Levy [Fri, 14 Jun 2013 14:19:07 +0000 (15:19 +0100)]
Add a stp file for usage from build directory
For systemtap the location of the process being tapped is crucial, as a
result the existing stp file requires installation for use.
There are now two files:
$(TARGET_DIR)/$(QEMU_PROG).stp-installed: copied to $(tapdir)/$(QEMU_PROG).stp
$(TARGET_DIR)/$(QEMU_PROG).stp: pointing to the built binary, usable
without installation
Anthony Liguori [Fri, 14 Jun 2013 12:51:45 +0000 (07:51 -0500)]
Merge remote-tracking branch 'mjt/trivial-patches-next' into staging
# By Michael Tokarev (4) and others
# Via Michael Tokarev
* mjt/trivial-patches-next: (26 commits)
piix: fix some printf errors when debug is enabled
cputlb: fix debug logs
create qemu_openpty_raw() helper function and move it to a separate file
main-loop: do not include slirp/slirp.h, use libslirp.h instead
libcacard/vscclient: fix leakage of socket on error paths
linux-headers: Update to v3.10-rc5
KVM: PPC: Add dummy kvm_arch_init_irq_routing()
KVM: S390: Add dummy kvm_arch_init_irq_routing()
KVM: ARM: Add dummy kvm_arch_init_irq_routing()
ivshmem: add missing error exit(2)
hw/xen: Use g_free instead of free and fix potential memory leaks
target-sparc: Replace free by g_free
hw/scsi: Don't increment a boolean value
device tree: Fix cppcheck warning
Makefile: Install qemu-img and qemu-nbd man pages only if built
Unbreak -no-quit for GTK, validate SDL options
gtk: implement -full-screen
char/serial: serial_ioport_write: Factor out common code
char/serial: Use generic Fifo8
char/serial: cosmetic fixes.
...
Anthony Liguori [Fri, 14 Jun 2013 12:51:30 +0000 (07:51 -0500)]
Merge remote-tracking branch 'afaerber/qom-cpu' into staging
# By Andreas Färber (12) and others
# Via Andreas Färber
* afaerber/qom-cpu:
spapr_rtas: Abstract rtas_start_cpu() with qemu_get_cpu()
spapr_rtas: Abstract rtas_query_cpu_stopped_state() with qemu_get_cpu()
memory_mapping: Improve qemu_get_guest_memory_mapping() error reporting
dump: Abstract dump_init() with cpu_synchronize_all_states()
cpu: Change default for CPUClass::get_paging_enabled()
dump: Drop qmp_dump_guest_memory() stub and build for all targets
memory_mapping: Drop qemu_get_memory_mapping() stub
cpu: Turn cpu_get_memory_mapping() into a CPUState hook
memory_mapping: Move MemoryMappingList typedef to qemu/typedefs.h
cpu: Turn cpu_paging_enabled() into a CPUState hook
monitor: Simplify do_inject_mce() with qemu_get_cpu()
target-i386: cpu: Fix potential buffer overrun in get_register_name_32()
target-i386: Set level=4 on Conroe/Penryn/Nehalem
target-i386: Update model values on Conroe/Penryn/Nehalem CPU models
pc: Create pc-*-1.6 machine-types
pc: Fix crash when attempting to hotplug CPU with negative ID
dump: Move stubs into libqemustub.a
Anthony Liguori [Fri, 14 Jun 2013 12:51:05 +0000 (07:51 -0500)]
Merge remote-tracking branch 'pmaydell/tcg-aarch64.next' into staging
# By Claudio Fontana (9) and others
# Via Peter Maydell
* pmaydell/tcg-aarch64.next:
MAINTAINERS: add tcg/aarch64 maintainer
configure: permit compilation on arm aarch64
tcg/aarch64: implement user mode qemu ld/st
user-exec.c: aarch64 initial implementation of cpu_signal_handler
tcg/aarch64: implement sign/zero extend operations
tcg/aarch64: implement byte swap operations
tcg/aarch64: implement AND/TEST immediate pattern
tcg/aarch64: improve arith shifted regs operations
tcg/aarch64: implement new TCG target for aarch64
include/elf.h: add aarch64 ELF machine and relocs
configure: Drop CONFIG_ATFILE test
linux-user: Drop direct use of openat etc syscalls
linux-user: Allow getdents to be provided by getdents64
Anthony Liguori [Fri, 14 Jun 2013 12:50:36 +0000 (07:50 -0500)]
Merge remote-tracking branch 'rth/tcg-for-anthony' into staging
# By Richard Henderson
# Via Richard Henderson
* rth/tcg-for-anthony:
tcg: Remove redundant tcg_target_init checks
tcg: Use QEMU_BUILD_BUG_ON for CPU_TLB_ENTRY_BITS
Michael Tokarev [Wed, 5 Jun 2013 14:44:54 +0000 (18:44 +0400)]
create qemu_openpty_raw() helper function and move it to a separate file
In two places qemu uses openpty() which is very system-dependent,
and in both places the pty is switched to raw mode as well.
Make a wrapper function which does both steps, and move all the
system-dependent complexity into a separate file, together
with static/local implementations of openpty() and cfmakeraw()
from qemu-char.c.
It is in a separate file, not part of oslib-posix.c, because
openpty() often resides in -lutil which is not linked to
every program qemu builds.
This change removes #including of <pty.h>, <termios.h>
and other rather specific system headers out of qemu-common.h,
which isn't a place for such specific headers really.
This version has been verified to build correctly on Linux,
OpenBSD, FreeBSD and OpenIndiana. On the latter it lets qemu
to be built with gtk gui which were not possible there due to
missing openpty() and cfmakeraw().
Andreas Färber [Wed, 29 May 2013 19:06:39 +0000 (21:06 +0200)]
spapr_rtas: Abstract rtas_start_cpu() with qemu_get_cpu()
Instead of looping over all CPUArchState, use a helper to obtain the
desired CPUState.
Free the "cpu" variable for PowerPCCPU, to access its CPUPPCState.
Peter Maydell [Wed, 12 Jun 2013 15:20:21 +0000 (16:20 +0100)]
linux-user: Drop direct use of openat etc syscalls
The linux-user syscall emulation layer currently supports the
openat family of syscalls via two mechanisms: simply calling
the corresponding libc functions, and making direct syscalls.
Since glibc has supported these functions since at least glibc
2.5, there's no real need to retain the (essentially untested)
direct syscall fallback code, so simply delete it. This allows
us to remove some ifdeffery that was attempting to disable
provision of some of the syscalls if the host didn't seem to
support them, which in some cases was actually wrong (eg where
there are several flavours of the syscall and we only need
one of them, not necessarily the exact one the guest has,
as with the fstatat* calls).
Peter Maydell [Wed, 12 Jun 2013 15:20:21 +0000 (16:20 +0100)]
linux-user: Allow getdents to be provided by getdents64
Newer architectures may only implement the getdents64 syscall, not
getdents. Provide an implementation of getdents in terms of getdents64
so that we can run getdents-using targets on a getdents64-only host.
Michael Tokarev [Wed, 12 Jun 2013 12:31:44 +0000 (16:31 +0400)]
main-loop: do not include slirp/slirp.h, use libslirp.h instead
The header slirp/slirp.h is an internal header for slirp, and
main-loop.c does not use internals from there. Instead, it uses
public functions (slirp_update_timeout(), slirp_pollfds_fill()
etc) which are declared in slirp/libslirp.h.
Including slirp/slirp.h is somewhat dangerous since it redefines
errno on WIN32, so any file including it may misbehave wrt errno.
Unfortunately libslirp isn't self-contained, it needs declaration
of struct in_addr, which is provided by qemu/sockets.h. Maybe
instead of #including qemu/sockets.h before libslirp.h, it is
better to make the latter self-contained.
Scott Wood [Wed, 12 Jun 2013 07:26:54 +0000 (17:26 +1000)]
KVM: PPC: Add dummy kvm_arch_init_irq_routing()
The common KVM code insists on calling kvm_arch_init_irq_routing()
as soon as it sees kernel header support for it (regardless of whether
QEMU supports it). Provide a dummy function to satisfy this.
Unlike x86, PPC does not have one default irqchip, so there's no common
code that we'd stick here. Even if you ignore the routes themselves,
which even on x86 are not set up in this function, the initial XICS
kernel implementation will not support IRQ routing, so it's best to
leave even the general feature flags up to the specific irqchip code.
The common KVM code insists on calling kvm_arch_init_irq_routing()
as soon as it sees kernel header support for it (regardless of whether
QEMU supports it). Provide a dummy function to satisfy this.
The common KVM code insists on calling kvm_arch_init_irq_routing()
as soon as it sees kernel header support for it (regardless of whether
QEMU supports it). Provide a dummy function to satisfy this.
Stefan Hajnoczi [Mon, 3 Jun 2013 08:58:31 +0000 (10:58 +0200)]
ivshmem: add missing error exit(2)
If the user fails to specify 'chardev' or 'shm' then we cannot continue.
Exit right away so that we don't invoke shm_open(3) with a NULL pointer.
It would be nice to replace exit(1) with error returns in the PCI device
.init() function, but leave that for another patch since exit(1) is
currently used elsewhere.
Andreas Färber [Tue, 11 Jun 2013 11:13:58 +0000 (13:13 +0200)]
Makefile: Install qemu-img and qemu-nbd man pages only if built
When splitting openSUSE's qemu and qemu-linux-user packages we noticed
that for linux-user-only builds unrelated man pages got installed.
It's surely possible to delete them before packaging, but not installing
them in the first place seems more logical.
Peter Wu [Tue, 11 Jun 2013 08:55:08 +0000 (10:55 +0200)]
Unbreak -no-quit for GTK, validate SDL options
Certain options (-no-frame, -alt-grab, -ctrl-grab) only make sense with SDL.
When compiling without SDL, these options (and -no-quit) print an error message
and exit qemu.
In case QEMU is compiled with SDL support, the three aforementioned options
still do not make sense with other display types. This patch addresses that
issue by printing a warning. I have chosen not to exit QEMU afterwards because
having the option is not harmful and before this patch it would be ignored
anyway.
By delaying the sanity check from compile-time with some ifdefs to run-time,
-no-quit is now also properly supported when compiling without SDL.
Peter Wu [Mon, 10 Jun 2013 18:04:43 +0000 (20:04 +0200)]
gtk: implement -full-screen
Aiming for GTK as replacement for SDL, a feature like -full-screen should also
be implemented.
Bringing the window into full-screen mode is done by activating the "Fullscreen"
menu item. This is done after showing the windows to make the cursor and menu
hidden.
v2: drop -no-frame implementation, use booleans instead of ints and ensure
consistency between ui state and menu.
char/serial: serial_ioport_write: Factor out common code
These three lines are common to both FIFO and regular mode. Just factor
them out to outside the if rather than replicate the same lines inside
both if and else.
I'm not sure how this went in. The thing is that
ignoring *.patch, in my opinion, is just wrong.
Especially for downstreams who apply patches for
real.
Michael Tokarev [Wed, 5 Jun 2013 15:08:59 +0000 (19:08 +0400)]
qemu-char: remove a few needless #includes
This removes <syslog.h> since we don't use
syslogging, and removes second, solaris-specific,
include of <net/if.h> (which is included in
a common part of the file)
Pass any Error out into dump_init() and have it actually stop on errors.
Whether it is unsupported on a certain CPU can be checked by looking for
a NULL CPUClass::get_memory_mapping field.
Andreas Färber [Sun, 9 Jun 2013 14:03:46 +0000 (16:03 +0200)]
cpu: Change default for CPUClass::get_paging_enabled()
qemu_get_guest_memory_mapping() uses cpu_paging_enabled() to determine
whether to use cpu_get_memory_mapping() to return mappings or whether to
fall back to a simple identity map.
Since by default CPUClass::get_memory_mapping() is not implemented,
change the default to false to use the identity map by default.
Andreas Färber [Fri, 17 May 2013 09:54:40 +0000 (11:54 +0200)]
dump: Drop qmp_dump_guest_memory() stub and build for all targets
qmp_dump_guest_memory() calls dump_init() and returns an Error when
cpu_get_dump_info() returns an error, as done by the stub.
So there is no need to have a stub for qmp_dump_guest_memory().
Enable the documentation of the always-present dump-guest-memory command.
That way we can drop CONFIG_HAVE_CORE_DUMP and leave configure
completely out of the picture for target CPU features.
Andreas Färber [Tue, 28 May 2013 12:53:32 +0000 (14:53 +0200)]
memory_mapping: Drop qemu_get_memory_mapping() stub
dump.c:dump_init() never checked for the return code anyway.
If paging is not enabled, it will fall back to an identity map.
If paging is enabled and getting memory mapping list is not
implemented, qemu_get_guest_memory_mapping() will return an error.
Since the targets not implementing memory mapping also don't implement
dump support, we will not reach this code today and can worry about
changing cpu_paging_enabled() default when the need arises.
This allows us to drop CONFIG_HAVE_GET_MEMORY_SUPPORT.
Eduardo Habkost [Mon, 27 May 2013 20:23:55 +0000 (17:23 -0300)]
target-i386: Set level=4 on Conroe/Penryn/Nehalem
The CPUID level value on Conroe, Penryn, and Nehalem are too low. This
causes at least one known problem: the -smp "threads" option doesn't
work as expect if level is < 4, because thread count information is
provided to the guest on CPUID[EAX=4,ECX=2].EAX
Eduardo Habkost [Mon, 27 May 2013 20:23:54 +0000 (17:23 -0300)]
target-i386: Update model values on Conroe/Penryn/Nehalem CPU models
The CPUID model values on Conroe, Penryn, and Nehalem are too
conservative and don't reflect the values found on real Conroe, Penryn,
and Nehalem CPUs.
This causes at least one known problems: Windows XP disables sysenter
when (family == 6 && model <= 2), but Skype tries to use the sysenter
instruction anyway because it is reported as available on CPUID, making
it crash.
This patch sets appropriate model values that correspond to real Conroe,
Penryn, and Nehalem CPUs.
Since for PC in current implementation id should be in range [0...maxcpus)
and maxcpus is already checked, add check for lower bound and error out
on incorrect value.
Which we basically used to print out a greeting/prompt for the
monitor.
AFAICT the only reason this was ever done in a BH was because in
some cases we'd modify the chr_write handler for a new chardev
backend *after* the site where we issued the reset (see:
86e94d:qemu_chr_open_stdio())
At some point this event was renamed to CHR_EVENT_OPENED, and we've
maintained the use of this BH ever since.
However, due to 9f939df955a4152aad69a19a77e0898631bb2c18, we schedule
the BH via g_idle_add(), which is causing events to sometimes be
delivered after we've already begun processing data from backends,
leading to:
known bugs:
QMP:
session negotation resets with OPENED event, in some cases this
is causing new sessions to get sporadically reset
potential bugs:
hw/usb/redirect.c:
can_read handler checks for dev->parser != NULL, which may be
true if CLOSED BH has not been executed yet. In the past, OPENED
quiesced outstanding CLOSED events prior to us reading client
data. If it's delayed, our check may allow reads to occur even
though we haven't processed the OPENED event yet, and when we
do finally get the OPENED event, our state may get reset.
qtest.c:
can begin session before OPENED event is processed, leading to
a spurious reset of the system and irq_levels
gdbstub.c:
may start a gdb session prior to the machine being paused
To fix these, let's just drop the BH.
Since the initial reasoning for using it still applies to an extent,
work around that by deferring the delivery of CHR_EVENT_OPENED until
after the chardevs have been fully initialized, toward the end of
qmp_chardev_add() (or some cases, qemu_chr_new_from_opts()). This
defers delivery long enough that we can be assured a CharDriverState
is fully initialized before CHR_EVENT_OPENED is sent.
Also, rather than requiring each chardev to do an explicit open, do it
automatically, and allow the small few who don't desire such behavior to
suppress the OPENED-on-init behavior by setting a 'explicit_be_open'
flag.
We additionally add missing OPENED events for stdio backends on w32,
which were previously not being issued, causing us to not recieve the
banner and initial prompts for qmp/hmp.
Anthony Liguori [Mon, 10 Jun 2013 16:37:39 +0000 (11:37 -0500)]
Merge remote-tracking branch 'stefanha/net' into staging
# By Jason Wang (1) and Stefan Hajnoczi (1)
# Via Stefan Hajnoczi
* stefanha/net:
tap: fix NULL dereference when passing invalid parameters to tap
vmxnet3: fix NICState cleanup
Peter Maydell [Sun, 2 Jun 2013 15:17:49 +0000 (16:17 +0100)]
softfloat: Fix shift128Right for shift counts 64..127
shift128Right would give the wrong result for a shift count
between 64 and 127. This was never noticed because all of
our uses of this function are guaranteed not to use shift
counts in this range.
Paolo Bonzini [Mon, 3 Jun 2013 15:06:55 +0000 (17:06 +0200)]
gdbstub: let the debugger resume from guest panicked state
While in general we forbid a "continue" from the guest panicked
state, it makes sense to have an exception for that when continuing
in the debugger. Perhaps the guest entered that state due to a bug,
for example, and we want to continue no matter what.
This commit used the wrong check to prevent an assertion failure.
After this commit, you need to start a guest in the monitor, you
cannot use anymore the "c" command in the debugger. This is
undesirable. The commit's aim was to prevent a restart
after a KVM internal error or something like that; use
runstate_needs_reset() for that.
Wendy Liang [Fri, 7 Jun 2013 03:05:38 +0000 (13:05 +1000)]
xilinx_axidma: Do not set DMA .notify to NULL after notify
If a stream notify function is not ready, it may re-populate the notify call-
back to indicate it should be re-polled later. This break in this usage, as
immediately following the notify() call, .notify is set to NULL. reverse the
ordering of the notify call and NULL assignment accordingly.
Luiz Capitulino [Mon, 3 Jun 2013 15:14:26 +0000 (11:14 -0400)]
MAINTAINERS: split Monitor (QMP/HMP) entry
This entry doesn't reflect reality for a few years now. This commit
splits it into Human Monitor (HMP), QAPI and QMP. Markus is dropped
as a maintainer.
This is what we have been for the last few years. Also, it's going
to help me to offload some of this work to someone else in the near
future.
Luiz Capitulino [Mon, 3 Jun 2013 15:11:53 +0000 (11:11 -0400)]
MAINTAINERS: new maintainers for qapi-schema.json
I'm facing two problems lately wrt QMP patch review: increasingly
lack of bandwidth and lack of background in so many different areas
that are getting new QMP commands almost every week.
In order to help me mitigate this problem, I'm adding Eric and Markus
(besides me) as maintainers of the qapi-schema.json file.
Markus has been an old timer reviewer. Eric is being the most active
and prolific reviewer of QMP patches for some time now.
I believe Markus and Eric will keep doing their work as before, but
starting now I'll require the ACK of at least one of them before
appling a patch/series that touches the qapi-schema.json file.
Jason Wang [Tue, 4 Jun 2013 05:18:17 +0000 (13:18 +0800)]
tap: fix NULL dereference when passing invalid parameters to tap
This patch forbid the following invalid parameters to tap:
1) fd and vhostfds were specified but vhostfd were not specified
2) vhostfds were specified but fds were not specified
3) fds and vhostfd were specified
For 1 and 2, net_init_tap_one() will still pass NULL as vhostfdname to
monitor_handle_fd_param(), which may crash the qemu.
Anthony Liguori [Fri, 7 Jun 2013 13:40:52 +0000 (08:40 -0500)]
Merge remote-tracking branch 'stefanha/block' into staging
# By Kevin Wolf (19) and others
# Via Stefan Hajnoczi
* stefanha/block: (26 commits)
hmp: add parameters device and -v for info block
hmp: show ImageInfo in 'info block'
qmp: add ImageInfo in BlockDeviceInfo used by query-block
block: add image info query function bdrv_query_image_info()
block: add snapshot info query function bdrv_query_snapshot_info_list()
ide-test: Add FLUSH CACHE test case
ide: Set BSY bit during FLUSH
ide-test: Add enum value for DEV
blkdebug: Add BLKDBG_FLUSH_TO_OS/DISK events
Make qemu-io commands available in HMP
qemu-io: Use the qemu version for -V
qemu-io: Interface cleanup
qemu-io: Move remaining helpers from cmd.c
qemu-io: Move command_loop() and friends
qemu-io: Move functions for registering and running commands
qemu-io: Move qemu_strsep() to cutils.c
qemu-io: Move 'quit' function
qemu-io: Move 'help' function
qemu-io: Factor out qemuio_command
qemu-io: Split off commands to qemu-io-cmds.c
...
Wenchao Xia [Thu, 6 Jun 2013 04:27:59 +0000 (12:27 +0800)]
qmp: add ImageInfo in BlockDeviceInfo used by query-block
Now image info will be retrieved as an embbed json object inside
BlockDeviceInfo, backing chain info and all related internal snapshot
info can be got in the enhanced recursive structure of ImageInfo. New
recursive member *backing-image is added to reflect the backing chain
status.
Wenchao Xia [Thu, 6 Jun 2013 04:27:58 +0000 (12:27 +0800)]
block: add image info query function bdrv_query_image_info()
This patch adds function bdrv_query_image_info(), which will
retrieve image info in qmp object format. The implementation is
based on the code moved from qemu-img.c, but uses block layer
function to get snapshot info.
Wenchao Xia [Thu, 6 Jun 2013 04:27:57 +0000 (12:27 +0800)]
block: add snapshot info query function bdrv_query_snapshot_info_list()
This patch adds function bdrv_query_snapshot_info_list(), which will
retrieve snapshot info of an image in qmp object format. The implementation
is based on the code moved from qemu-img.c with modification to fit more
for qmp based block layer API.
Andreas Färber [Wed, 5 Jun 2013 13:17:57 +0000 (15:17 +0200)]
ide: Set BSY bit during FLUSH
The implementation of the ATA FLUSH command invokes a flush at the block
layer, which may on raw files on POSIX entail a synchronous fdatasync().
This may in some cases take so long that the SLES 11 SP1 guest driver
reports I/O errors and filesystems get corrupted or remounted read-only.
Avoid this by setting BUSY_STAT, so that the guest is made aware we are
in the middle of an operation and no ATA commands are attempted to be
processed concurrently.
Kevin Wolf [Wed, 5 Jun 2013 12:19:41 +0000 (14:19 +0200)]
Make qemu-io commands available in HMP
It was decided to not make this command available in QMP in order to
make clear that this is not supposed to be a stable API and should be
used only for testing and debugging purposes.
Kevin Wolf [Wed, 5 Jun 2013 12:19:40 +0000 (14:19 +0200)]
qemu-io: Use the qemu version for -V
Always printing 0.0.1 and never updating the version number wasn't very
useful. qemu-io is released with qemu, so using the same version number
makes most sense.
projects / qemu.git / log