Nathan Froyd [Fri, 5 Jun 2009 02:02:28 +0000 (19:02 -0700)]
gdb-xml: fix hacks in powerpc register numbering
The powerpc xml files contained a hack--an empty, non-existent
register--for getting the register numbers to line up for
newer (XML-aware) and older (non-XML-aware) GDB. While this hack worked
in some cases, it didn't work in all cases, notably when the user used
`finish' or `continue': GDB would attempt to read the non-existent
register and QEMU would complain.
This patch fixes things up properly. Instead of inserting a fake
register, we explicitly declare the floating-point and SPE registers to
start at 71. This action accomplishes the same thing as the nasty hack,
except that now GDB never tries to fetch the non-existant register 70.
Use parameter 'next' to fix the hdecr case.
Also pass 'next' by value instead of pointer (more easy to read and no
performance issue for an always_inline function).
Igor Kovalenko [Sun, 12 Jul 2009 08:35:31 +0000 (12:35 +0400)]
sparc64: trap handling corrections
On Sun, Jul 12, 2009 at 12:09 PM, Blue Swirl<[email protected]> wrote:
> On 7/12/09, Igor Kovalenko <[email protected]> wrote:
>> Good trap handling is required to process interrupts.
>> This patch fixes the following:
>>
>> - sparc64 has no wim register
>> - sparc64 has no psret register, use IE bit of pstate
>> extract IE checking code to cpu_interrupts_enabled
>> - alternate globals are not available if cpu has GL feature
>> in this case bit AG of pstate is constant zero
>> - write to pstate must actually write pstate
>> even if cpu has GL feature
>>
>> Also timer interrupt is handled using do_interrupt.
>
> A bit too much for one patch. Please also remove the code instead of
> commenting out.
I now excluded timer interrupt related part.
To my mind other changes are essentially tied together.
> PUT_PSR for Sparc64 needs CC_OP = CC_OP_FLAGS; like Sparc32.
Igor Kovalenko [Sat, 11 Jul 2009 20:57:03 +0000 (00:57 +0400)]
sparc64: fix helper_st_asi little endian case typo
On Sun, Jul 12, 2009 at 12:43 AM, Stuart Brady<[email protected]> wrote:
> On Sat, Jul 11, 2009 at 10:22:18PM +0400, Igor Kovalenko wrote:
>> It is clear that intention is to byte-swap value to be written, not
>> the target address.
>
> @@ -1949,13 +1949,13 @@ void helper_st_asi(target_ulong addr, ta
> case 0x89: // Secondary LE
> switch(size) {
> case 2:
> - addr = bswap16(addr);
> + addr = bswap16(val);
> ^^^^
> Shouldn't that be 'val = bswap16(val)' (and likewise for the 32-bit and
> 64-bit cases)? Also needs a 'signed-off-by:'...
>
> Cheers,
> --
> Stuart Brady
>
Thanks, that part I did not runtime-tested.
Not sure if those asi stores are of any use for user-mode emulator.
Avi Kivity [Tue, 23 Jun 2009 13:20:36 +0000 (16:20 +0300)]
block: Clean up after deleting BHs
Commit 6a7ad299 ("Call qemu_bh_delete at bdrv_aio_bh_cb") deletes emulated
aio bottom halves to prevent endless accumulation. However, it leaves a
stale ->bh pointer, which is then waited on when the aio is reused.
Zeroing the pointer fixes the issue, allowing vmdk format images to be used.
Glauber Costa [Mon, 6 Jul 2009 13:32:09 +0000 (09:32 -0400)]
flush pending aio requests
When we finish migration, there may be pending async io requests
in flight. If we don't flush it before stage3 starting, it might be
the case that the guest loses it.
Clean up msix vector usage state on load. Since guest might have control
over it through the device, the device will have to load this state from
file.
Kevin Wolf [Tue, 7 Jul 2009 16:09:42 +0000 (18:09 +0200)]
qcow2: Fix L1 table memory allocation
Contrary to what one could expect, the size of L1 tables is not cluster
aligned. So as we're writing whole sectors now instead of single entries,
we need to ensure that the L1 table in memory is large enough; otherwise
write would access memory after the end of the L1 table.
Mark McLoughlin [Fri, 3 Jul 2009 08:28:02 +0000 (09:28 +0100)]
Prefer sysfs for USB host devices
Scanning for devices via /sys/bus/usb/devices/ and using them via the
/dev/bus/usb/<bus>/<device> character devices is the prefered method
on modern kernels, so try that first.
When using SELinux and libvirt, qemu will have access to /sys/bus/usb
but not /proc/bus/usb, so although the current code will work just
fine, it will generate SELinux AVC warnings.
Add a -g flag to the open command and the main qemu-io command line to
allow opening a file growable. This is only allowed for protocols,
mirroring the limitation exposed through bdrv_file_open.
qemu-io: better input validation for vector-based commands
Fix up a couple of issues with validating the input of the various
length arguments for the vectored I/O commands:
- do the alignment check on each length instead the always 0 count argument
- use a long long varibale for the cvtnum return value so that we can check
wether it wasn't a number
- check for a too large argument instead of truncating it
Also refactor it into a common helper for all four calers and avoid parsing
the numbers twice.
Mark McLoughlin [Wed, 1 Jul 2009 15:46:38 +0000 (16:46 +0100)]
Don't leak VLANClientState on PCI hot remove
destroy_nic() requires that NICInfo::private by a PCIDevice pointer,
but then goes on to require that the same pointer matches
VLANClientState::opaque.
That is no longer the case for virtio-net since qdev and wasn't
previously the case for rtl8139, ne2k_pci or eepro100.
Make the situation a lot more clear by maintaining a VLANClientState
pointer in NICInfo.
Alexander Graf [Wed, 1 Jul 2009 20:08:21 +0000 (22:08 +0200)]
Replace signrom with shell script v3
In order to not execute code we just compiled, let's replace signrom
with a shell script that does the same thing while staying compatible
to pretty much every system available.
This should make cross-compilation for windows easier.
Kevin Wolf [Tue, 30 Jun 2009 11:06:04 +0000 (13:06 +0200)]
qcow2: Make cache=writethrough default
The performance of qcow2 has improved meanwhile, so we don't need to
special-case it any more. Switch the default to write-through caching
like all other block drivers.
Mark McLoughlin [Tue, 30 Jun 2009 09:02:57 +0000 (10:02 +0100)]
net: set a default value for sndbuf=
On reflection, perhaps it does make sense to set a default value for
the sndbuf= tap parameter.
For best effect, sndbuf= should be set to just below the capacity of
the physical NIC.
Setting it higher will cause packets to be dropped before the limit
is hit. Setting it much lower will not cause any problems unless
you set it low enough such that the guest cannot queue up new packets
before the NIC has emptied its queue.
In Linux, txqueuelen=1000 by default for ethernet NICs. Given a 1500
byte MTU, 1Mb is a good choice for sndbuf.
If it turns out that txqueuelen is actually much lower than this, then
sndbuf is essentially disabled. In the event that txqueuelen is much
higher, it's unlikely that the NIC will be able to empty a 1Mb queue.
Huang Ying [Tue, 23 Jun 2009 02:05:14 +0000 (10:05 +0800)]
QEMU: MCE: Add MCE simulation to qemu/tcg
- MCE features are initialized when VCPU is intialized according to CPUID.
- A monitor command "mce" is added to inject a MCE.
- A new interrupt mask: CPU_INTERRUPT_MCE is added to inject the MCE.
Gerd Hoffmann [Tue, 30 Jun 2009 12:12:11 +0000 (14:12 +0200)]
qdev/pci: misc fixes.
* fix secondary bus setup.
* use base->name instead of "FIXME" for device name.
Yes, the device name is redundant. Only for drivers converted
to qdev already though. Once all drivers are converted we can
and should kill it.
Gerd Hoffmann [Tue, 30 Jun 2009 12:12:09 +0000 (14:12 +0200)]
qdev: remove DeviceType
The only purpose DeviceType serves is creating a linked list of
DeviceInfo structs. This removes DeviceType and add a next field to
DeviceInfo instead, so the DeviceInfo structs can be changed that way.
Elimitates a pointless extra level of indirection.
Gerd Hoffmann [Tue, 30 Jun 2009 12:12:08 +0000 (14:12 +0200)]
qdev: replace bus_type enum with bus_info struct.
BusInfo is filled with name and size (pretty much like I did for
DeviceInfo as well). There is also a function pointer to print
bus-specific device information to the monitor. sysbus is hooked
up there, I've also added a print function for PCI.
Device creation is slightly modified as well: The device type search
loop now also checks the bus type while scanning the list instead of
complaining thereafter in case of a mismatch. This effectively gives
each bus a private namespace for device names.
Gerd Hoffmann [Tue, 30 Jun 2009 12:12:07 +0000 (14:12 +0200)]
qdev: update pci device registration.
Makes pci_qdev_register take a PCIDeviceInfo struct instead of a bunch
of parameters. Also adds config_read and config_write callbacks to
PCIDeviceInfo, so drivers needing these can be converted to the qdev
device API too.
vibi sreenivasan [Fri, 19 Jun 2009 08:04:39 +0000 (13:34 +0530)]
linux-user/syscall.c: remove warning: ‘array’ may be used uninitialized in this function
Removes the following warning
CC i386-linux-user/syscall.o
cc1: warnings being treated as errors
/media/nfs/qemu/linux-user/syscall.c: In function ‘do_syscall’:
/media/nfs/qemu/linux-user/syscall.c:2219: warning: ‘array’ may be used uninitialized in this function
Arnaud Patard [Fri, 19 Jun 2009 07:44:45 +0000 (10:44 +0300)]
linux-user: check some parameters for some socket syscalls.
This patch is fixing following issues :
- commit 8fea36025b9d6d360ff3b78f88a84ccf221807e8 was applied to
do_getsockname instead of do_accept.
- Some syscalls were not checking properly the memory addresses passed
as argument
- Add check before syscalls made for cases like do_getpeername() where
we're using the address parameter after doing the syscall
- Fix do_accept to return EINVAL instead of EFAULT when parameters
invalid to match with linux behaviour
Stefan Weil [Wed, 1 Jul 2009 21:13:34 +0000 (23:13 +0200)]
Fix hxtool.
When converting from hx to texi format, hxtool (or to be
more precise sh which interprets hxtool) used standard
shell expansion of wildcards while writing lines to the
output.
Thus, something like "Password: ********"
looked very different in the generated documentation.
The patch disables this unwanted wildcard expansion.
projects / qemu.git / log