Alexander Graf [Mon, 21 Dec 2009 11:24:17 +0000 (12:24 +0100)]
PPC64: Fix timebase
On PPC we have a 64-bit time base. Usually (PPC32) this is accessed using
two separate 32 bit SPR accesses to SPR_TBU and SPR_TBL.
On PPC64 the SPR_TBL register acts as 64 bit though, so we get the full
64 bits as return value. If we only take the lower ones, fine. But Linux
wants to see all 64 bits or it breaks.
This patch makes PPC64 Linux work even after TB crossed the 32-bit boundary,
which usually happened a few seconds after bootup.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Alpha always used 32-bit uids, but never renamed the syscalls
to match i386 when 32-bit uids were added there. This enables
the proper bits in syscall.c.
Signed-off-by: Richard Henderson <rth@twiddle.net> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Kevin Wolf [Sat, 19 Dec 2009 20:15:20 +0000 (21:15 +0100)]
Multiboot support: Fix rom_copy
ROMs need to be loaded if they are anywhere in the requested area, not
only at the very beginning. This fixes Multiboot with ELF kernels that
have more than one program header.
Signed-off-by: Kevin Wolf <mail@kevin-wolf.de> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Nathan Froyd [Fri, 11 Dec 2009 17:04:46 +0000 (09:04 -0800)]
linux-user: commonify definitions of target typedefs
There's no sense in separately declaring target_{elf_greg,uid,gid,pid}_t
for every architecture. Just declare them once with appropriate
USE_UID16 handling.
Riku Voipio [Fri, 4 Dec 2009 13:16:32 +0000 (15:16 +0200)]
linux-user: cleanup force_sig() calls
Force_sig should be always called with TARGET_ signals.
Not that it really matters with SEGV, so this patch is
just for cleanup and improving consistency.
Riku Voipio [Fri, 4 Dec 2009 13:16:31 +0000 (15:16 +0200)]
linux-user: Fix mmap_lock ordering
mmap_lock() can be called while tb_lock() is being held. To
avoid deadlock when one thread is holding mmap_lock and another
tb_lock, _always_ lock first tb_lock().
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name> Cc: Blue Swirl <blauwirbel@gmail.com> Signed-off-by: Juan Quintela <quintela@redhat.com> Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
do_migrate_set_speed() accepts a suffix for the 'value' argument and this is
not good for QMP. We will have to add a new argument type to handle that and
this will have to wait for 0.13.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Luiz Capitulino [Fri, 18 Dec 2009 15:25:03 +0000 (13:25 -0200)]
QMP: Update spec file
- Remove "draft" status
- Change default success response to be json-object
- Change error and event data member to be a json-object
- Update examples
- Add new section "Compatibility Considerations"
- Other fixes and clarifications
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Riku Voipio [Thu, 3 Dec 2009 13:56:08 +0000 (15:56 +0200)]
usb-musb: convert fifo to 8bit and add more registers
Convert musb fifo to 8bit to allow 8/16/32bit access
MUSB allows reading and writing to the fifo in 32/16/8 bit
width. The Linux kernel does this sometimes, most usually at
the end of writing the packet to allow packet to end at a
odd bytecount.
Convert the fifo to 8bit allows removing lots of shifts
which shows that the fifo is more natural as 8bit.
While at it, add multiple missing register definitions and
and cleanup debug prints.
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:12 +0000 (12:01 +0100)]
roms: remove option rom packing logic
Now that we load the option roms via fw_cfg, we can stop copying
them to the 0xc000 -> 0xe000. The patch does just that.
Also the rom loader gets simplified as all remaining users of the
rom loader load the bits at a fixed address so the packing and
aligning logic can go away.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:10 +0000 (12:01 +0100)]
fw_cfg: add API for file transfer.
This patch adds a file transfer interface to fw_cfg. Intended to be
used for passing non-pci option roms and vgabios to seabios. Namespace
is modeled after the existing cbfs filesystem support in seabios.
Reading the new FW_CFG_FILE_DIR entry returns a file list.
Fields there are in network byte order (aka bigendian).
aliguori: fix fw_cfg.h for multiboot.bin, add proper fw_cfg.h declarations,
quiet fprintf() in fw_cfg.c
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Gerd Hoffmann [Fri, 18 Dec 2009 11:01:08 +0000 (12:01 +0100)]
pci romfiles: add property, add default to PCIDeviceInfo
This patch adds a romfile property to the pci bus. It allows to specify
a romfile to load into the rom bar of the pci device. The default value
comes from a new field in PCIDeviceInfo. The property allows to change
the file and also to disable the rom loading using an empty string.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Anthony Liguori [Fri, 18 Dec 2009 11:01:07 +0000 (12:01 +0100)]
Support PCI based option rom loading
Currently, we preload option roms into the option rom space in memory. This
prevents DDIM from functioning correctly which severely limits the number
of roms we can support.
This patch introduces a pci_add_option_rom() which registers the
PCI_ROM_ADDRESS bar which points to our option rom. It also converts over
the cirrus vga adapter, the rtl8139, virtio, and the e1000 to use this
new mechanism.
The result is that PXE boot functions even with three unique types of cards.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Anthony Liguori [Thu, 17 Dec 2009 22:08:10 +0000 (08:08 +1000)]
VMware VGA: Only enable dirty log tracking when fifo is disabled
This patch enables dirty log tracking whenever it's needed and disables it
when it is not.
We unconditionally enable dirty log tracking on reset, restart dirty log
tracking when PCI IO regions are remapped, and disable/enable it based on
commands from the guest.
Rebased-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Anthony Liguori [Thu, 17 Dec 2009 22:08:09 +0000 (08:08 +1000)]
Fix VMware VGA depth computation
VMware VGA requires that the depth presented to the guest is the same as the
DisplaySurface that it renders to. This is because it performs a very simple
memcpy() to blit from one surface to another.
We currently hardcode a 24-bit depth. The surface allocator for SDL may, and
usually will, allocate a surface with a different depth causing screen
corruption.
This changes the code to allocate the DisplaySurface before initializing the
device which allows the depth of the DisplaySurface to be used instead of
hardcoding something.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Anthony Liguori [Thu, 17 Dec 2009 22:08:07 +0000 (08:08 +1000)]
Make sure to enable dirty tracking of VBE vram mapping
Apparently, VBE maps the VGA vram to a fixed physical location. KVM requires
that all mappings of the VGA vram have dirty tracking enabled on them. Any
access to the VGA vram through the VBE mapping currently fails to result in
dirty page tracking updates causing a black screen.
This is the true root cause of VMware VGA not working correctly under KVM and
likely also an issue with some of the std-vga black screen issues too.
Cirrus does not enable VBE so it would not be a problem when using Cirrus.
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> Rebased-by: Dave Airlie <airlied@redhat.com>
Gerd Hoffmann [Wed, 16 Dec 2009 13:22:11 +0000 (14:22 +0100)]
qdev: improve property error reporting.
Add a error message in case we fail to parse a qdev property.
Also make qemu not abort() in case setting a global property can't be
set. This used to be a clear programming error. The introduction of
the -global switch changed that though, so better exit instead (after
printing the new error message).
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Luiz Capitulino [Mon, 14 Dec 2009 20:53:24 +0000 (18:53 -0200)]
monitor: Catch printing to non-existent monitor
The monitor_vprintf() function now touches the 'mon' pointer
before calling monitor_puts(), this causes block migration
to segfault as its functions call monitor_printf() with a
NULL 'mon'.
To fix the problem this commit moves the 'mon' NULL check
from monitor_puts() to monitor_vprintf().
This can potentially hide bugs, but for some reason this has
been the behavior for a long time.
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Avoid permanently disabled QEMU monitor when UNIX migration fails
If a UNIX migration command is attempt to a UNIX socket which does
not exist, then the monitor is suspended, but never resumed. This
prevents any further use of the monitor
* migration-unix.c: Only call migrate_fd_monitor_suspend() once
connected to the UNIX socket.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Kevin Wolf [Fri, 4 Dec 2009 16:19:25 +0000 (17:19 +0100)]
Fix loading of ELF multiboot kernels
The multiboot implementation assumed that there is only one program header
(which contains the entry point) and that the entry point is at the start of
the code. This doesn't hold true generally and caused too little data to be
loaded.
Fix the loading code to pass the whole loaded data to the Multiboot Option ROM.
Signed-off-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Kevin Wolf [Fri, 4 Dec 2009 11:06:32 +0000 (12:06 +0100)]
Revert "Rename DriveInfo.onerror to on_write_error" (fix mismerge)
Part of the first patch of the -drive rerror series has been merged once more
on top of the rest of the series. This effectively disables the rerror option
and always goes with the default value. Reverting the commit re-enables the
option.
Jan Kiszka [Mon, 14 Dec 2009 11:26:17 +0000 (12:26 +0100)]
kvm: x86: Use separate exception_injected CPUState field
Marcelo correctly remarked that there are usage conflicts between QEMU
core code and KVM /wrt exception_index. So spend a separate field and
also save/restore it properly.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Alexander Graf [Thu, 17 Dec 2009 12:06:09 +0000 (13:06 +0100)]
S390: Don't tell guest we're updating config space
Currently we always set the "config space changed" bit to 1 when triggering
any virtio interrupt. While that worked in 2.6.27, newer kernels interpret
that value as "only the config space changed and nothing else happened".
Since we usually trigger interrupts to tell the guest that something did
happen, we just not tell it the config space changed for now until we
implement the correct callback for that.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Alexander Graf [Thu, 17 Dec 2009 12:06:07 +0000 (13:06 +0100)]
S390: Loop through virtio console devices
We used to always create one single virtio console device. This breaks when
either zero of multiple virtio console devices are requested, so let's use
the same code as on x86.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Alexander Graf [Thu, 17 Dec 2009 12:56:47 +0000 (13:56 +0100)]
target-s390: Fail on unknown instructions
We were being a bit too nice and didn't give the guest an invalid instruction
interrupt.
While that works, it's not exactly the fastest thing to do, since now the
guest doesn't know that we're not really implementing that instruction, so it
continues doing it.
We run into this with the set_page_unstable hint instruction. So let's bail out
in these cases.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Andre Przywara [Fri, 18 Dec 2009 09:45:07 +0000 (10:45 +0100)]
osdep: Fix runtime failure on older Linux kernels
If QEMU finds newer kernel header files on compilation time, it will use
advertised features like pipe2 or SOCK_CLOEXEC by just doing a compile test.
If later the executables are executed on an older kernel (<2.6.27,
like Xen Dom0 2.6.18), then QEMU will fail on opening sockets and creating
pipes and returns the rather unspecific "qemu_init_main_loop failed".
This patch fixes this by checking the return values of these calls
for EINVAL and ENOSYS and falling back to the older versions automatically.
Signed-off-by: Andre Przywara <andre.przywara@amd.com> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Juergen Lock [Mon, 14 Dec 2009 20:54:43 +0000 (21:54 +0100)]
Fix a make -j race
Make libuser.a depend on $(GENERATED_HEADERS) too so make -j won't start
building it before the headers exist. (There may be more bugs like this
but at least this makes (g)make -j4 started from scratch on a quadcore
now always complete here again.)
There was a bug in float32_to_s that incorrectly mapped a zero exponent
to 0x38. This meant 0.0f != 0. At the same time, fix a generic type
punning bug in helper_memory_to_s and helper_s_to_memory.
Signed-off-by: Richard Henderson <rth@twiddle.net> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The old fcmov implementation had a typo:
- tcg_gen_mov_i64(cpu_fir[rc], cpu_fir[ra]);
which moved the condition, not the second source, to the destination.
But it's also easy to implement the simplified fp comparison inline.
Signed-off-by: Richard Henderson <rth@twiddle.net> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
projects / qemu.git / log