Paolo Bonzini [Fri, 8 Feb 2013 13:06:11 +0000 (14:06 +0100)]
block: implement BDRV_O_UNMAP
It is better to present homogeneous hardware independent of the storage
technology that is chosen on the host, hence we make discard a host
parameter; the user can choose whether to pass it down to the image
format and protocol, or to ignore it.
Using DISCARD with filesystems can cause very severe fragmentation, so it
is left default-off for now. This can change later when we implement the
"anchor" operation for efficient management of preallocated files.
There is still one choice to make: whether DISCARD has an effect on the
dirty bitmap or not. I chose yes, though there is a disadvantage: if
the guest is buggy and issues discards for data that is in use, there
will be no way to migrate storage for that guest without downgrading
the machine type to an older one.
Paolo Bonzini [Tue, 19 Feb 2013 10:59:10 +0000 (11:59 +0100)]
coroutine: trim down nesting level in perf_nesting test
20000 nested coroutines require 20 GB of virtual address space.
Only nest 1000 of them so that the test (only enabled with
"-m perf" on the command line) runs on 32-bit machines too.
Paolo Bonzini [Tue, 19 Feb 2013 10:59:09 +0000 (11:59 +0100)]
coroutine: move pooling to common code
The coroutine pool code is duplicated between the ucontext and
sigaltstack backends, and absent from the win32 backend. But the
code can be shared easily by moving it to qemu-coroutine.c.
block: Add synchronous wrapper for bdrv_co_is_allocated_above
There's no synchronous wrapper for bdrv_co_is_allocated_above function
so it's not possible to check for sector allocation in an image with
a backing file.
Stefan Hajnoczi [Wed, 13 Feb 2013 15:53:42 +0000 (16:53 +0100)]
block: use Error in do_check_io_limits()
The do_check_io_limits() function returns false when I/O limits are
invalid but it doesn't set an Error to indicate why. The two
do_check_io_limits() callers duplicate error reporting. Solve this by
passing an Error pointer into do_check_io_limits().
Note that the two callers report slightly different errors: drive_init()
prints a custom error message while qmp_block_set_io_throttle() does
error_set(errp, QERR_INVALID_PARAMETER_COMBINATION).
QERR_INVALID_PARAMETER_COMBINATION is a generic error, see
include/qapi/qmp/qerror.h:
Stefan Hajnoczi [Thu, 7 Feb 2013 16:15:03 +0000 (17:15 +0100)]
qemu-img: fix missing space in qemu-img check output
The qemu-img check fragmentation printf() is missing a space before the
'=' sign. The human output is not guaranteed to be stable and we are
not aware of screen scrapers, so add the missing space.
Also fix the missing indentation of the printf() arguments.
Stefan Hajnoczi [Thu, 7 Feb 2013 16:15:02 +0000 (17:15 +0100)]
qcow2: record fragmentation statistics during check
The qemu-img check command can display fragmentation statistics:
* Total number of clusters in virtual disk
* Number of allocated clusters
* Number of fragmented clusters
This patch adds fragmentation statistics support to qcow2.
Compressed and normal clusters count as allocated. Zero clusters are
not counted as allocated unless their L2 entry has a non-zero offset
(e.g. preallocation).
Only the current L1 table counts towards the statistics - snapshots are
ignored.
Stefan Hajnoczi [Thu, 7 Feb 2013 16:15:01 +0000 (17:15 +0100)]
qcow2: introduce check_refcounts_l1/l2() flags
The check_refcounts_l1/l2() functions have a check_copied argument to
check that the QCOW_O_COPIED flag is consistent with refcount == 1.
This should be a bool, not an int.
However, the next patch introduces qcow2 fragmentation statistics and
also needs to pass an option to check_refcounts_l1/l2(). This is a good
opportunity to use an int flags field.
This patch adds the support for reporting the image end offset (in
bytes). This is particularly useful after a conversion (or a rebase)
where the destination is a block device in order to find the first
unused byte at the end of the image.
Anthony Liguori [Fri, 22 Feb 2013 14:40:30 +0000 (08:40 -0600)]
ui/gtk: require at least GTK 2.18 and VTE 0.26
This gives us the bare amount of features we need. We can add work arounds
for older versions and lower the requirement but this should be a good
starting point.
Suggested-by: Daniel Berrange <[email protected]> Signed-off-by: Anthony Liguori <[email protected]>
---
v1 -> v2
- tremendous simplification suggested by danpb
Anthony Liguori [Wed, 20 Feb 2013 13:43:26 +0000 (07:43 -0600)]
gtk: suppress accelerators from the File menu when grab is active
If you're full screen, you probably expect Ctrl-Q to go to the guest,
not the host. I think restricting certain menus is the right way to
handle this generally speaking.
Anthony Liguori [Wed, 20 Feb 2013 13:43:23 +0000 (07:43 -0600)]
gtk: add support for screen scaling and full screen (v5)
Basic menu items to enter full screen mode and zoom in/out. Unlike SDL, we
don't allow arbitrary scaling based on window resizing. The current behavior
with SDL causes a lot of problems for me.
Sometimes I accidentally resize the window a tiny bit while trying to move it
(Ubuntu's 1-pixel window decorations don't help here). After that, scaling is
now active and if the screen changes size again, badness ensues since the
aspect ratio is skewed.
Allowing zooming by 25% in and out should cover most use cases. We can add a
more flexible scaling later but for now, I think this is a more friendly
behavior.
Anthony Liguori [Wed, 20 Feb 2013 13:43:22 +0000 (07:43 -0600)]
gtk: add support for input grabbing (v2)
There is a small deviation from SDL's behavior here. Instead of Ctrl+Alt
triggering grab, we now use Ctrl-Alt-g to trigger grab.
GTK will not accept Ctrl+Alt as an accelerator since it just consists of
modifiers. Having grab as a proper accelerator is important as it allows a user
to override the accelerator for accessibility purposes.
We also are not automatically grabbing on left-click. Besides the inability to
tie mouse clicks to an accelerator, I think this behavior is hard to discover
and since it only happens depending on the guest state, it can lead to confusing
behavior.
This can be changed in the future if there's a strong resistence to dropping
left-click-to-grab, but I think we're better off dropping it.
Anthony Liguori [Wed, 20 Feb 2013 13:43:21 +0000 (07:43 -0600)]
gtk: add virtual console support (v2)
This enables VteTerminal to be used to render the text consoles. VteTerminal is
the same widget used by gnome-terminal which means it's VT100 emulation is as
good as they come.
It's also screen reader accessible, supports copy/paste, proper scrolling and
most of the other features you would expect from a terminal widget.
Anthony Liguori [Wed, 20 Feb 2013 13:43:20 +0000 (07:43 -0600)]
ui: add basic GTK gui (v5)
This is minimalistic and just contains the basic widget infrastructure. The GUI
consists of a menu and a GtkNotebook. To start with, the notebook has its tabs
hidden which provides a UI that looks very similar to SDL with the exception of
the menu bar.
The menu bar allows a user to toggle the visibility of the tabs. Cairo is used
for rendering.
I used gtk-vnc as a reference. gtk-vnc solves the same basic problems as QEMU
since it was originally written as a remote display for QEMU. So for the most
part, the approach to rendering and keyboard handling should be pretty solid for
GTK.
Anthony Liguori [Wed, 20 Feb 2013 13:43:18 +0000 (07:43 -0600)]
build: disable Wstrict-prototypes
GTK won't build with strict-prototypes due to gtkitemfactory.h:
/* We use () here to mean unspecified arguments. This is deprecated
* as of C99, but we can't change it without breaking compatibility.
* (Note that if we are included from a C++ program () will mean
* (void) so an explicit cast will be needed.)
*/
typedef void (*GtkItemFactoryCallback) ();
num_interfaces only tells you how many interfaces the concrete child class has
(as defined in the TypeInfo). This means if you have a child class which defines
no interfaces of its own, but its parent has interfaces you cannot cast to those
parent interfaces.
Fixed changing the guard to check the class->interfaces list instead (which is
a complete flattened list of implemented interfaces).
The QOM framework will attempt the recreate a classes interface list from
scratch for each class. This means that a child class should zero out the
list of interfaces when cloned from the parent class.
Currently the list is memcpy()d from the parent to the child. As the interface
list is just a pointer to a list, this means the parent and child will share
the same list of interfaces. When the child inits, it will append its own
interfaces to the parents list. This is incorrect as the parent should not pick
up its childs interfaces.
This actually causes an infinite loop at class init time, as the child will
iterate through the parent interface list adding each itf to its own list(in
type_initialize()). As the list is (erroneously) shared, the new interface
instances for the child are appended to the parent, and the iterator never hits
the tail and loops forever.
Gerd Hoffmann [Wed, 20 Feb 2013 08:37:12 +0000 (09:37 +0100)]
vga: fix byteswapping.
In case host and guest endianness differ the vga code first creates
a shared surface (using qemu_create_displaysurface_from), then goes
patch the surface format to indicate that the bytes must be swapped.
The switch to pixman broke that hack as the format patching isn't
propagated into the pixman image, so ui code using the pixman image
directly (such as vnc) uses the wrong format.
Fix that by adding a byteswap parameter to
qemu_create_displaysurface_from, so we'll use the correct format
when creating the surface (and the pixman image) and don't have
to patch the format afterwards.
Peter Maydell [Wed, 20 Feb 2013 16:24:22 +0000 (16:24 +0000)]
Remove elderly top level TODO file
The top level TODO file hasn't been touched since 2008, so it's now
an unhelpful and out of date mix of things that have already been done,
things that don't make sense any more and things which could in theory
be done but are not in practice important enough (or we'd have done
them some time in the last five years). Remove it. The bug tracking
system is probably a better place to track TODO items if we want to
do so.
Stefan Hajnoczi [Wed, 20 Feb 2013 10:28:33 +0000 (11:28 +0100)]
aio: support G_IO_HUP and G_IO_ERR
aio-posix.c could not take advantage of G_IO_HUP and G_IO_ERR because
select(2) does not have equivalent events. Now that g_poll(3) is used
we can support G_IO_HUP and G_IO_ERR.
I came up with this table by looking at Linux fs/select.c which maps
select(2) to poll(2) internally.
Another detail to watch out for are the global variables that reference
rfds/wfds/xfds during slirp_select_poll(). sofcantrcvmore() and
sofcantsendmore() use these globals to clear fd_set bits. When
sofcantrcvmore() is called, the wfds bit is cleared so that the write
handler will no longer be run for this iteration of the event loop.
This actually seems buggy to me since TCP connections can be half-closed
and we'd still want to handle data in half-duplex fashion. I think the
real intention is to avoid running the read/write handler when the
socket has been fully closed. This is indicated with the SS_NOFDREF
state bit so we now check for it before invoking the TCP write handler.
Note that UDP/ICMP code paths don't care because they are
connectionless.
Note that slirp/ has a lot of tabs and sometimes mixed tabs with spaces.
I followed the style of the surrounding code.
Stefan Hajnoczi [Wed, 20 Feb 2013 10:28:25 +0000 (11:28 +0100)]
main-loop: switch to g_poll() on POSIX hosts
Use g_poll(3) instead of select(2). Well, this is kind of a cheat.
It's true that we're now using g_poll(3) on POSIX hosts but the *_fill()
and *_poll() functions are still using rfds/wfds/xfds.
We've set the scene to start converting *_fill() and *_poll() functions
step-by-step until no more rfds/wfds/xfds users remain. Then we'll drop
the temporary gpollfds_from_select() and gpollfds_to_select() functions
and be left with native g_poll(2).
On Windows things are a little crazy: convert from rfds/wfds/xfds to
GPollFDs, back to rfds/wfds/xfds, call select(2), rfds/wfds/xfds back to
GPollFDs, and finally back to rfds/wfds/xfds again. This is only
temporary and keeps the Windows build working through the following
patches. We'll drop this excessive conversion later and be left with a
single GPollFDs -> select(2) -> GPollFDs sequence that allows Windows to
use select(2) while the rest of QEMU only knows about GPollFD.
Anthony Liguori [Thu, 21 Feb 2013 15:38:27 +0000 (09:38 -0600)]
Merge remote-tracking branch 'stefanha/trivial-patches' into staging
# By Alin Tomescu (1) and others
# Via Stefan Hajnoczi
* stefanha/trivial-patches:
.gitignore: Ignore optionrom/*.asm
ppc: fix bamboo >256MB RAM initialization in hw/ppc4xx_devs.c
Add some missing qtest binaries to .gitignore
xilinx_axienet.c: Assert no error when making link
Remove forward declaration of non-existant variable
Alin Tomescu [Thu, 21 Feb 2013 02:36:09 +0000 (21:36 -0500)]
ppc: fix bamboo >256MB RAM initialization in hw/ppc4xx_devs.c
I was trying to launch a PowerPC "bamboo" machine with more than 256MB of RAM
with qemu-system-ppc -M bamboo -kernel $kernel -initrd $ramdisk -m 512, but QEMU
would just hang. However, when I used -m 256, the machine would boot.
I looked through the code in hw/ and it seems there is an error when the
RAM memory is setup (if my understanding is correct).
After patching it, the machine launched and booted successfully with 512MB of
RAM.
Gerd Hoffmann [Mon, 28 Jan 2013 14:52:57 +0000 (15:52 +0100)]
usb: fix endpoint descriptor ordering
Fix the ordering of the endpoint descriptors for superspeed endpoints:
The superspeed companion must come first, possible additional
descriptors for the endpoint after that.
Gerd Hoffmann [Fri, 1 Feb 2013 09:49:43 +0000 (10:49 +0100)]
make usb devices configurable
Leave the core usb devices (usb hub, tablet, mouse, keyboard)
enabled unconditionally. Make the other ones configurable.
Exceptions:
- bluetooth: not qdevified yet, has a vl.c dependency because
of that, thus disabling isn't as easy as not linking the
object file.
- smardcard: ccid-card-emulated depends on that one *and*
CONFIG_SMARTCARD_NSS. So it isn't a one-liner and comes
as separate patch because of that.
Gerd Hoffmann [Fri, 1 Feb 2013 09:48:34 +0000 (10:48 +0100)]
fix scripts/make_device_config.sh
Make it handle multiple include statements in a file:
(1) The printf needs a space so the include files will be separated.
(2) Also $f can contain multiple failes, so redirection will not work
and we have to use cat to process all files.
Andre Przywara [Thu, 18 Oct 2012 09:16:58 +0000 (11:16 +0200)]
vnc-tls: Fix compilation with newer versions of GNU-TLS
In my installation of GNU-TLS (v3.0.23) the type
gnutls_anon_server_credentials is marked deprecated, so -Werror
breaks compilation.
Simply replacing it with the newer ..._t version fixed the compilation
on my machine (Slackware 14.0). I cannot tell how far back this "new"
type goes, at least the header file in RHEL 5.0 (v1.4.1) seems to have
it already. If someone finds a broken distribution, tell me and I
insert some compat code.
End tables before headings, start new ones afterwards. Fixes
incorrect indentation of headings "File system options" and "Virtual
File system pass-through options" in manual page and qemu-doc.
Normalize markup some to increase chances it survives future edits.
Andreas Färber [Sat, 16 Feb 2013 21:44:01 +0000 (22:44 +0100)]
libqtest: Convert macros to functions and clean up documentation
libqtest.h provides a number of shortcut macros to avoid tests feeding
it the QTestState they operate on. Most of these can easily be turned
into static inline functions, so let's do that for clarity.
This avoids getting off-by-one error messages when passing wrong args.
Some macros had a val argument but documented @value argument. Fix this.
While touching things, enforce gtk-doc markup for return values and for
referencing types.
Anthony Liguori [Mon, 18 Feb 2013 14:37:29 +0000 (08:37 -0600)]
Merge remote-tracking branch 'afaerber/qom-cpu' into staging
# By Andreas Färber
# Via Andreas Färber
* afaerber/qom-cpu: (47 commits)
target-i386: Split command line parsing out of cpu_x86_register()
target-i386: Move cpu_x86_init()
target-lm32: Drop unused cpu_lm32_close() prototype
target-s390x: Drop unused cpu_s390x_close() prototype
spapr_hcall: Replace open-coded CPU loop with qemu_get_cpu()
ppce500_spin: Replace open-coded CPU loop with qemu_get_cpu()
e500: Replace open-coded loop with qemu_get_cpu()
cpu: Add CPUArchState pointer to CPUState
cputlb: Pass CPUState to cpu_unlink_tb()
cpu: Move current_tb field to CPUState
cpu: Move exit_request field to CPUState
cpu: Move running field to CPUState
cpu: Move host_tid field to CPUState
target-cris: Introduce CRISCPU subclasses
target-m68k: Pass M68kCPU to m68k_set_irq_level()
mcf_intc: Pass M68kCPU to mcf_intc_init()
mcf5206: Pass M68kCPU to mcf5206_init()
target-m68k: Return M68kCPU from cpu_m68k_init()
ppc405_uc: Pass PowerPCCPU to ppc40x_{core,chip,system}_reset()
target-xtensa: Move TCG initialization to XtensaCPU initfn
...
The new formulation makes better use of add-with-carry type insns
that the host may have. Use gcc's sign adjustment trick to avoid
having to perform a 128-bit negation.
Replace some x86_64 specific inline assembly with something that
all 64-bit hosts ought to optimize well. At worst this becomes
a call to the gcc __multi3 routine, which is no worse than our
implementation in util/host-utils.c.
With gcc 4.7, we get identical code generation for x86_64. We
now get native multiplication on ia64 and s390x hosts. With minor
improvements to gcc we can get it for ppc64 as well.
Andreas Färber [Sat, 16 Feb 2013 22:21:24 +0000 (23:21 +0100)]
tcg/ppc: Fix build of tcg_qemu_tb_exec()
Commit 0b0d3320db74cde233ee7855ad32a9c121d20eb4 (TCG: Final globals
clean-up) moved code_gen_prologue but forgot to update ppc code.
This broke the build on 32-bit ppc. ppc64 is unaffected.
Andreas Färber [Fri, 15 Feb 2013 14:21:13 +0000 (15:21 +0100)]
e500: Replace open-coded loop with qemu_get_cpu()
Since we still need env for ppc-specific fields, obtain it via the new
env_ptr fields to avoid "cpu" name conflicts between CPUState and
PowerPCCPU for now.
This fixes a potential issue with env being NULL at the end of the loop
but cpu still being a valid pointer corresponding to a previous env.
Andreas Färber [Thu, 17 Jan 2013 11:13:41 +0000 (12:13 +0100)]
cpu: Add CPUArchState pointer to CPUState
The target-specific ENV_GET_CPU() macros have allowed us to navigate
from CPUArchState to CPUState. The reverse direction was not supported.
Avoid introducing CPU_GET_ENV() macros by initializing an untyped
pointer that is initialized in derived instance_init functions.
The field may not be called "env" due to it being poisoned.
Andreas Färber [Wed, 16 Jan 2013 18:29:31 +0000 (19:29 +0100)]
cpu: Move current_tb field to CPUState
Explictly NULL it on CPU reset since it was located before breakpoints.
Change vapic_report_tpr_access() argument to CPUState. This also
resolves the use of void* for cpu.h independence.
Change vAPIC patch_instruction() argument to X86CPU.
projects / qemu.git / log