This adds vhost net device support in qemu. Will be tied to tap device
and virtio by following patches. Raw backend is currently missing,
will be worked on/submitted separately.
vhost needs physical addresses for ring and other queue fields,
so add APIs for these. In particular, add binding API to set
host/guest notifiers. Will be used by vhost.
event notifiers are slightly generalized eventfd descriptors. Current
implementation depends on eventfd because vhost is the only user, and
vhost depends on eventfd anyway, but a stub is provided for non-eventfd
case.
We'll be able to further generalize this when another user comes along
and we see how to best do this.
Comment on kvm usage: rather than require users to do if (kvm_enabled())
and/or ifdefs, this patch adds an API that, internally, is defined to
stub function on non-kvm build, and checks kvm_enabled for non-kvm
run.
While rest of qemu code still uses if (kvm_enabled()), I think this
approach is cleaner, and we should convert rest of code to it
long term.
Shahar Havivi [Sun, 14 Mar 2010 20:41:15 +0000 (22:41 +0200)]
Restore terminal monitor attributes - addition
Patch 2d753894c7553d6a05e8fdbed5f4704398919a35 was missing this check,
when running monitor as /dev/tty and other serial device, i.e:
qemu -monitor /dev/tty -serial /dev/pts/1
Without this patch any serial device will override the monitor stored
attributes. (monitor is called in main() before any serial device).
Alexander Graf [Tue, 16 Mar 2010 18:18:07 +0000 (19:18 +0100)]
Don't check for bus master for old guests
Older Linux guests don't activate the bus master enable bit. So for those we
can just try to be clever and track if they set the DEVICE_OK bit even though
bus mastering is still disabled.
Under that condition we can disable the windows safety check. With that logic
in place both guests should work just fine. Without PCI hotplug breaks
virtio-net in Linux < 2.6.34 guests.
Max Reitz [Sun, 14 Mar 2010 11:19:03 +0000 (12:19 +0100)]
usb: class specific interface requests
Mass Storage Reset and Get Max LUN are class specific requests, but
they were not marked as such in hw/usb-msd.c, moved therefore
ClassInterfaceRequest and ClassInterfaceOutRequest from hw/usb-net.c
to hw/usb.h.
Furthermore there was a problem in hw/usb-ohci.c when using DEBUG
concerning systems where size_t is a 32 bit integer (printf resulted
in a segmentation fault).
Juergen Lock [Thu, 25 Mar 2010 21:32:16 +0000 (22:32 +0100)]
Get bsd-user host page protection code working on FreeBSD hosts
Use kinfo_getvmmap(3) on FeeBSD >= 7.x and /compat/linux/proc on older
FreeBSD. (kinfo_getvmmap is preferred since /compat/linux/proc is
usually only mounted on hosts also using the Linuxolator.)
This patch is a bit hacky because the includes needed for kinfo_getvmmap
conflict with other definitions in exec.c by default so I had to `trick
around' a little, but I built the result in FreeBSD 6.4-stable and
7.2-stable tbs and on 8-stable on the host so the hacks at least
should be stable. (If this is a problem maybe we could also move the
kinfo_getvmmap invocations into a seperate source file but that would
be more work...)
Blue Swirl [Mon, 29 Mar 2010 19:23:52 +0000 (19:23 +0000)]
Refactor target specific handling, compile vl.c only once
Move target specific functions and RAM handling to arch_init.c.
Add a flag to QEMUOptions structure to indicate for which
architectures the option is allowed, check the flag
in run time and remove conditional code in option handling.
Now that no target dependencies remain, compile vl.c only once
for all targets.
Rob Landley [Sun, 28 Mar 2010 14:51:43 +0000 (16:51 +0200)]
linux-user/ppc: use the Linux register layout
The dynamic linker converts the Linux layout to the AIX layout and is
reentrant so it won't do it a second time if it's already been
converted. In short it work just fine with either register layout.
OTOH, statically linked binaries expect a Linux layout.
Remove code converting the Linux layout to AIX layout so that all
binaries are presented the Linux Layout.
Blue Swirl [Sat, 27 Mar 2010 18:41:08 +0000 (18:41 +0000)]
Fix BSD and win32 builds
CC net/tap-bsd.o
/src/qemu/net/tap-bsd.c: In function `tap_open':
/src/qemu/net/tap-bsd.c:93: warning: implicit declaration of function `error_report'
CC sparc-softmmu/../net/tap-win32.o
cc1: warnings being treated as errors
/src/qemu/target-sparc/../net/tap-win32.c: In function 'net_init_tap':
/src/qemu/target-sparc/../net/tap-win32.c:709: warning: implicit declaration of function 'error_report'
error: Link qemu-img, qemu-nbd, qemu-io with qemu-error.o
The location tracking interface is used by code shared with qemi-img,
qemu-nbd and qemu-io, so it needs to be available there. Commit 827b0813 provides it in a rather hamfisted way: it adds a dummy
implementation to qemu-tool.c.
It's cleaner to provide the real thing, and put a few more dummy
monitor functions into qemu-tool.c.
linux-user: Use RLIMIT_STACK for default stack size.
The current default stack limit of 512kB is far too small; a fair
number of gcc testsuite failures (for all guests) are directly
attributable to this. Using the -s option in every invocation of
the emulator is annoying to say the least.
A reasonable compromise seems to be to honor the system rlimit.
At least on two Linux distributions, this is set to 8MB and 10MB
respectively. If the system does not limit the stack, then we're
no worse off than before.
At the same time, rename the variable from x86_stack_size and
change the ultimate fallback size from 512kB to 8MB.
Paolo Bonzini [Fri, 19 Mar 2010 10:30:35 +0000 (11:30 +0100)]
fix race between timer firing vs. alarm_timer->pending = 0
The period for Win32 timers is very short and always the same
independent of dynticks, so it's possible that the timer fires
before qemu_run_all_timers has reset alarm_timer->pending to zero.
Reset alarm_timer->pending before rearming.
Adam Litke [Thu, 25 Mar 2010 13:58:17 +0000 (08:58 -0500)]
balloon: Fix overflow when reporting actual memory size
Beginning with its introduction, the virtio balloon has had an overflow error
that causes 'info balloon' to misreport the actual memory size when the balloon
itself becomes larger than 4G. Use a cast when converting dev->actual from
pages to kB to prevent overflows.
Before:
(qemu) info balloon
balloon: actual=5120
(qemu) balloon 1025
(qemu) info balloon
balloon: actual=1025
(qemu) balloon 1024
(qemu) info balloon
balloon: actual=5120
After:
(qemu) info balloon
balloon: actual=5120
(qemu) balloon 1025
(qemu) info balloon
balloon: actual=1025
(qemu) balloon 1024
(qemu) info balloon
balloon: actual=1024
Kevin Wolf [Fri, 12 Mar 2010 12:52:31 +0000 (13:52 +0100)]
raw-posix: Better error return values for hdev_create
Now that we output an error message according to the returned error code in
qemu-img, let's return the real error codes. "Input/output error" for
everything isn't helpful.
Alexander Graf [Thu, 25 Mar 2010 13:59:02 +0000 (14:59 +0100)]
target-s390: Don't compile in virtio-pci
As soon as virtio-pci.c gets compiled and used on S390 the internal qdev magic
gets confused and tries to give us PCI devices instead of S390 virtio devices.
Since we don't have PCI on S390, we can safely not compile virtio-pci at all.
In order to do this I added a new config option "CONFIG_VIRTIO_PCI" that I
enabled for every platform except S390. Thanks to this the change should be a
complete nop for every other platform.
Dmitry Ilyevsky [Fri, 26 Mar 2010 00:25:36 +0000 (03:25 +0300)]
target-ppc: generic PowerPC TBL
Time base SPRs TBL/TBU should be accessible in user/priv modes for reading
as specified in POWER ISA documentation. Therefore SPRs permissions were
changed in gen_tbl function.
tcg: Disambiguate qemu_ld32u with 32-bit and 64-bit outputs.
Some targets (e.g. Alpha and MIPS64) need to keep 32-bit operands
sign-extended in 64-bit registers (regardless of the "real" sign
of the operand). For that, we need to be able to distinguish
between a 32-bit load with a 32-bit result and a 32-bit load with
a given extension to a 64-bit result. This distinction already
exists for the ld* loads, but not the qemu_ld* loads.
Reserve qemu_ld32u for 64-bit outputs and introduce qemu_ld32 for
32-bit outputs. Adjust all code generators to match.
CC qemu-error.o
cc1: warnings being treated as errors
/home/edgar/src/c/qemu/git/qemu/qemu-error.c: In function 'error_print_loc':
/home/edgar/src/c/qemu/git/qemu/qemu-error.c:191: error: format not a string literal and no format arguments
make: *** [qemu-error.o] Error 1
projects / qemu.git / log