Right now, kvm keeps the memory allocation split, so we can
handle different areas in different ways. This schema works with qemu
too, so it appears to be the common ground.
This patch proposes using this common ground for everyone, by spliting
raw qemu.
Signed-off-by: Glauber Costa <glommer@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5228 c046a42c-6fe2-441c-8c8c-71466251a162
Make page_find() return 0 for too-large addresses (Eduardo Habkost)
On some cases, such as under KVM, tb_invalidate_phys_page_range()
may be called for large addresses, when qemu is configured to more than
4GB of RAM.
On these cases, qemu was crashing because it was using an index too
large for l1_map[], that supports only 32-bit addresses when compiling
without CONFIG_USER_ONLY.
Right now, we sprinkle #if defined(QEMU_IMG) && defined(QEMU_NBD) all over the
code. It's ugly and causes us to have to build multiple object files for
linking against qemu and the tools.
This patch introduces a new file, qemu-tool.c which contains enough for
qemu-img, qemu-nbd, and QEMU to all share the same objects.
This also required getting qemu-nbd to be a bit more Windows friendly. I also
changed the Windows block-raw to use normal IO instead of overlapping IO since
we don't actually do AIO yet on Windows. I changed the various #if 0's to
#if WIN32_AIO to make it easier for someone to eventually fix AIO on Windows.
After this patch, there are no longer any #ifdef's related to qemu-img and
qemu-nbd.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5226 c046a42c-6fe2-441c-8c8c-71466251a162
according to the alpha arch reference, the literal field of an operate
instruction is unsigned:
If bit <12> of the instruction is 1, an 8-bit zero-extended literal
constant is formed by bits
<20:13> of the instruction. The l teral is interpreted as a positive
integer bet ween 0 and 255
and is zero-extended to 64 bits.
This patch fixes the mis-interpretation of the literal field.
usb: Support for removing device by host addr, improved auto filter syntax (Max Krasnyansky)
This patch adds support for removing USB devices by host address.
Which is usefull for things like libvirtd because there is no easy way to
find guest USB address of the host device.
In other words you can now do:
usb_add host:3.5
...
usb_del host:3.5
Before the patch 'usb_del' did not support 'host:' notation.
----
Syntax for specifying auto connect filters has been improved.
Old syntax was
host:bus.dev
host:pid:vid
New syntax is
host:auto:bus.dev[:pid:vid]
In both the cases any attribute can be set to "*".
New syntax is more flexible and lets you do things like
host:3.*:5533:* /* grab any device on bus 3 with vendor id 5533 */
It's now possible to remove auto filters. For example:
usb_del host:auto:3.*:5533:*
Active filters are printed after all host devices in 'info usb' output.
Which now looks like this:
Device 1.1, speed 480 Mb/s
Hub: USB device 1d6b:0002, EHCI Host Controller
Device 1.4, speed 480 Mb/s
Class 00: USB device 1058:0704, External HDD
Auto filters:
Device 3.* ID *:*
Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5205 c046a42c-6fe2-441c-8c8c-71466251a162
husb: Make control transactions asynchronous (Max Krasnyansky)
USB is 99.8% async now :). 0.2% is the three control requests that
we need to execute synchronously. We could off-load that to a thread
or something but it's not worth the pain since those requests are
performed only during device initialization (ie when device is
connected to the VM).
The change is a bit bigger than I wanted due to the fact that generic
handle_packet()/handle_control() interface was not designed for
async transactions. So I ended up adding custom handle_packet()
code to usb-linux. We can make that generic if/when some other
component needs it.
Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5204 c046a42c-6fe2-441c-8c8c-71466251a162
uhci: Change default transaction lifetime to 32 frames (Max Krasnyansky)
Transaction lifetime was originally set to 10 frames. That was an arbitrary
number I picked without much thinking :).
I'm changing that to 32 frames because things like interrupt transfers
and such are scheduled at that rate. It seems like 1/32 is accepted as
lowest supported rate. OHCI, for example, defines exactly 32 interrupt
heads.
While testing USB webcam under XP I noticed that interrupt transactions were
being canceled and then resubmitted on a regular basis, which works but is a
waste of CPU cycles. This change fixes that.
All other devices I have are not affected.
Signed-off-by: Max Krasnyansky <maxk@kernel.org> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5199 c046a42c-6fe2-441c-8c8c-71466251a162
Only build compatfd when using AIO and make sure to always init AIO
OpenBSD doesn't use AIO so don't try to build compatfd when not using AIO.
Also make sure to call qemu_aio_init() from bdrv_init. Everything that uses
bdrv calls bdrv_init so it makes sense to init aio from there instead of
in every single tool.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5197 c046a42c-6fe2-441c-8c8c-71466251a162
It turns out, we're never reading from the signalfd() which is causing it to
remain readable forever. I'll fix this up but I thought I'd commit this fix
in the interim.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5194 c046a42c-6fe2-441c-8c8c-71466251a162
This patch introduces signalfd() to work around the signal/select race in
checking for AIO completions. For platforms that don't support signalfd(), we
emulate it with threads.
There was a long discussion about this approach. I don't believe there are any
fundamental problems with this approach and I believe eliminating the use of
signals is a good thing.
I've tested Windows and Linux using Windows and Linux guests. I've also checked
for disk IO performance regressions.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5187 c046a42c-6fe2-441c-8c8c-71466251a162
While QEMU officially uses SVN, there are a number of unofficial git
repositories that many developers use. Adding a .gitignore (derived from the
svn:ignore) will make their lives a lot easier.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5183 c046a42c-6fe2-441c-8c8c-71466251a162
As discussed in
http://lists.gnu.org/archive/html/qemu-devel/2008-08/msg00667.html,
current pxe boot is broken for some use cases. The problem
goes away if we reduce the number of allowed bits in the address space
to 32 (which has the side effect of reducing guest max mem size to 4Gb).
After digging for a while, it turns out that it happens because pxelinux
tries to access address 0x10009e9a6, which does not fit a 32-bit address.
A closer look, however, reveals this access is totally valid: It's just
0x9e9a6 with an add carry.
To avoid this, this patch casts the address passed to the POPL macro to
a 32-bit value. This is also done, although just theorectically, for
PUSHL too.
Signed-off-by: Glauber Costa <glommer@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> Reported-by: Chris Lalancette <clalance@redhat.com> CC: Eduardo Habkost <ehabkost@redhat.com>
git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5182 c046a42c-6fe2-441c-8c8c-71466251a162
Replace op_load_gpr_{T0,T1,T2} and op_store_{T0,T1,T2} with tcg_gen_mov_tl.
Introduce TCG variables cpu_gpr[0..31].
For the SPE extension, assure that ppc_gpr_t is only uint64_t for ppc64.
Introduce TCG variables cpu_gprh[0..31] for upper 32 bits on ppc and helpers
gen_{load,store}_gpr64. Based on suggestions by Aurelien, Thiemo and Blue.
ETRAX-FS: Add support for DMA channel resets, needed for recent linux kernels.
* Correct numeric value for the RST state.
* Add emulation for reseting a DMA channel.
* Add a few sanity checks.
* Make it compile with debug enabled.
This patch enhances the pcnet NIC emulation with better loopback mode
support, including CRC generation for looped-back packets in "raw" mode.
The patch has practically no impact on the normal RX and TX path.
Successfully tested against an ancient proprietary pcnet driver that
does a lot of hardware checks on boot-up and now works fine over qemu as
well.
This adds trivial support for the R2D-PLUS FPGA, mostly just for the
versioning information that the kernel uses for IRL mappings, in addition
to handling the heartbeat and poweroff writes.
Trivial patch adding CPU listing and the ability to do per-subtype
CVR/PVR/PRR values. Presently SH7750R and SH7751R definitions are
provided, as these are the ones in present use in-tree.
The CVR value for SH7751R is intentionally restricted so the kernel
boots, though this will want to be switched to the proper CVR value
once system emulation has sufficiently stabilized.
This also makes it trivial to abstract subtype specific registers like
MMU_PTEA and to set up feature bits in line with the kernel probing for
things like conditionalizing FPU/DSP context.
Trivial fix for a corner case: system_shutdown on isapc machines causes
qemu to segfaults due to accessing the uninitialized pm_state. Issue a
system shutdown instead.
projects / qemu-xen-4.0-testing.git / log