Dave Hansen [Mon, 11 Aug 2008 17:01:45 +0000 (10:01 -0700)]
KVM: Reduce kvm stack usage in kvm_arch_vm_ioctl()
On my machine with gcc 3.4, kvm uses ~2k of stack in a few
select functions. This is mostly because gcc fails to
notice that the different case: statements could have their
stack usage combined. It overflows very nicely if interrupts
happen during one of these large uses.
This patch uses two methods for reducing stack usage.
1. dynamically allocate large objects instead of putting
on the stack.
2. Use a union{} member for all of the case variables. This
tricks gcc into combining them all into a single stack
allocation. (There's also a comment on this)
Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Based on a patch from: Amit Shah <amit.shah@qumranet.com>
This patch adds support for handling PCI devices that are assigned to
the guest.
The device to be assigned to the guest is registered in the host kernel
and interrupt delivery is handled. If a device is already assigned, or
the device driver for it is still loaded on the host, the device
assignment is failed by conveying a -EBUSY reply to the userspace.
Devices that share their interrupt line are not supported at the moment.
By itself, this patch will not make devices work within the guest.
The VT-d extension is required to enable the device to perform DMA.
Another alternative is PVDMA.
Signed-off-by: Amit Shah <amit.shah@qumranet.com> Signed-off-by: Ben-Ami Yassour <benami@il.ibm.com> Signed-off-by: Weidong Han <weidong.han@intel.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Userspace may specify memory slots that are backed by mmio pages rather than
normal RAM. In some cases it is not enough to identify these mmio pages
by pfn_valid(). This patch adds checking the PageReserved as well.
Glauber Costa [Mon, 28 Jul 2008 14:47:53 +0000 (11:47 -0300)]
x86: KVM guest: use paravirt function to calculate cpu khz
We're currently facing timing problems in guests that do
calibration under heavy load, and then the load vanishes.
This means we'll have a much lower lpj than we actually should,
and delays end up taking less time than they should, which is a
nasty bug.
Solution is to pass on the lpj value from host to guest, and have it
preset.
Signed-off-by: Glauber Costa <gcosta@redhat.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
The PIT injection logic is problematic under the following cases:
1) If there is a higher priority vector to be delivered by the time
kvm_pit_timer_intr_post is invoked ps->inject_pending won't be set.
This opens the possibility for missing many PIT event injections (say if
guest executes hlt at this point).
2) ps->inject_pending is racy with more than two vcpus. Since there's no locking
around read/dec of pt->pending, two vcpu's can inject two interrupts for a single
pt->pending count.
Fix 1 by using an irq ack notifier: only reinject when the previous irq
has been acked. Fix 2 with appropriate locking around manipulation of
pending count and irq_ack by the injection / ack paths.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
KVM: powerpc: Map guest userspace with TID=0 mappings
When we use TID=N userspace mappings, we must ensure that kernel mappings have
been destroyed when entering userspace. Using TID=1/TID=0 for kernel/user
mappings and running userspace with PID=0 means that userspace can't access the
kernel mappings, but the kernel can directly access userspace.
The net is that we don't need to flush the TLB on privilege switches, but we do
on guest context switches (which are far more infrequent). Guest boot time
performance improvement: about 30%.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
KVM: ppc: Write only modified shadow entries into the TLB on exit
Track which TLB entries need to be written, instead of overwriting everything
below the high water mark. Typically only a single guest TLB entry will be
modified in a single exit.
Guest boot time performance improvement: about 15%.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
We're saving the host TLB state to memory on every exit, but never using it.
Originally I had thought that we'd want to restore host TLB for heavyweight
exits, but that could actually hurt when context switching to an unrelated host
process (i.e. not qemu).
Since this decreases the performance penalty of all exits, this patch improves
guest boot time by about 15%.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Alexander Graf [Tue, 22 Jul 2008 06:00:45 +0000 (08:00 +0200)]
KVM: Ignore DEBUGCTL MSRs with no effect
Netware writes to DEBUGCTL and reads from the DEBUGCTL and LAST*IP MSRs
without further checks and is really confused to receive a #GP during that.
To make it happy we should just make them stubs, which is exactly what SVM
already does.
Writes to DEBUGCTL that are vendor-specific are resembled to behave as if the
virtual CPU does not know them.
Signed-off-by: Alexander Graf <agraf@suse.de> Signed-off-by: Avi Kivity <avi@qumranet.com>
Jerone Young [Mon, 14 Jul 2008 12:00:03 +0000 (14:00 +0200)]
KVM: ppc: adds trace points for ppc tlb activity
This patch adds trace points to track powerpc TLB activities using the
KVM_TRACE infrastructure.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com> Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Jerone Young [Mon, 14 Jul 2008 12:00:02 +0000 (14:00 +0200)]
KVM: ppc: enable KVM_TRACE building for powerpc
This patch enables KVM_TRACE to build for PowerPC arch. This means just
adding sections to Kconfig and Makefile.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com> Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
KVM: kvmtrace: replace get_cycles with ktime_get v3
The current kvmtrace code uses get_cycles() while the interpretation would be
easier using using nanoseconds. ktime_get() should give at least the same
accuracy as get_cycles on all architectures (even better on 32bit archs) but
at a better unit (e.g. comparable between hosts with different frequencies.
[avi: avoid ktime_t in public header]
Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
KVM: kvmtrace: Remove use of bit fields in kvm trace structure
This patch fixes kvmtrace use on big endian systems. When using bit fields the
compiler will lay data out in the wrong order expected when laid down into a
file.
This fixes it by using one variable instead of using bit fields.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com> Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Jan Kiszka [Mon, 14 Jul 2008 10:28:51 +0000 (12:28 +0200)]
KVM: VMX: Reinject real mode exception
As we execute real mode guests in VM86 mode, exception have to be
reinjected appropriately when the guest triggered them. For this purpose
the patch adopts the real-mode injection pattern used in vmx_inject_irq
to vmx_queue_exception, additionally taking care that the IP is set
correctly for #BP exceptions. Furthermore it extends
handle_rmode_exception to reinject all those exceptions that can be
raised in real mode.
This fixes the execution of himem.exe from FreeDOS and also makes its
debug.com work properly.
Note that guest debugging in real mode is broken now. This has to be
fixed by the scheduled debugging infrastructure rework (will be done
once base patches for QEMU have been accepted).
Signed-off-by: Jan Kiszka <jan.kiszka@web.de> Signed-off-by: Avi Kivity <avi@qumranet.com>
Avi Kivity [Thu, 3 Jul 2008 13:14:28 +0000 (16:14 +0300)]
KVM: VMX: Move interrupt post-processing to vmx_complete_interrupts()
Instead of looking at failed injections in the vm entry path, move
processing to the exit path in vmx_complete_interrupts(). This simplifes
the logic and removes any state that is hidden in vmx registers.
Avi Kivity [Thu, 3 Jul 2008 11:50:12 +0000 (14:50 +0300)]
KVM: VMX: Fix pending exception processing
The vmx code assumes that IDT-Vectoring can only be set when an exception
is injected due to the exception in question. That's not true, however:
if the exception is injected correctly, and later another exception occurs
but its delivery is blocked due to a fault, then we will incorrectly assume
the first exception was not delivered.
Fix by unconditionally dequeuing the pending exception, and requeuing it
(or the second exception) if we see it in the IDT-Vectoring field.
Avi Kivity [Thu, 3 Jul 2008 11:59:22 +0000 (14:59 +0300)]
KVM: Clear exception queue before emulating an instruction
If we're emulating an instruction, either it will succeed, in which case
any previously queued exception will be spurious, or we will requeue the
same exception.
Avi Kivity [Wed, 2 Jul 2008 06:28:55 +0000 (09:28 +0300)]
KVM: VMX: Move nmi injection failure processing to vm exit path
Instead of processing nmi injection failure in the vm entry path, move
it to the vm exit path (vm_complete_interrupts()). This separates nmi
injection from nmi post-processing, and moves the nmi state from the VT
state into vcpu state (new variable nmi_injected specifying an injection
in progress).
Avi Kivity [Tue, 1 Jul 2008 13:20:21 +0000 (16:20 +0300)]
KVM: Move NMI IRET fault processing to new vmx_complete_interrupts()
Currently most interrupt exit processing is handled on the entry path,
which is confusing. Move the NMI IRET fault processing to a new function,
vmx_complete_interrupts(), which is called on the vmexit path.
Move KVM trace definitions from x86 specific kvm headers to common kvm
headers to create a cross-architecture numbering scheme for trace
events. This means the kvmtrace_format userspace tool won't need to know
which architecture produced the log file being processed.
Signed-off-by: Jerone Young <jyoung5@us.ibm.com> Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Marcelo Tosatti [Fri, 27 Jun 2008 17:58:02 +0000 (14:58 -0300)]
KVM: x86: accessors for guest registers
As suggested by Avi, introduce accessors to read/write guest registers.
This simplifies the ->cache_regs/->decache_regs interface, and improves
register caching which is important for VMX, where the cost of
vmcs_read/vmcs_write is significant.
[avi: fix warnings]
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com> Signed-off-by: Avi Kivity <avi@qumranet.com>
Linus Torvalds [Tue, 14 Oct 2008 23:53:02 +0000 (16:53 -0700)]
Merge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
* 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6:
i2c-viapro: Add support for SMBus Process Call transactions
i2c: Restore i2c_smbus_process_call function
i2c: Do earlier driver model init
i2c: Only build Tyan SMBus mux drivers on x86
i2c: Guard against oopses from bad init sequences
i2c: Document the implementation details of the /dev interface
i2c: Improve dev-interface documentation
i2c-parport-light: Don't register a platform device resource
hwmon: (dme1737) Convert to a new-style i2c driver
hwmon: (dme1737) Be less i2c-centric
i2c/tps65010: Vibrator hookup to gpiolib
i2c-viapro: Add VX800/VX820 support
i2c: Renesas Highlander FPGA SMBus support
i2c-pca-isa: Don't grab arbitrary resources
i2c/isp1301_omap: Convert to a new-style i2c driver, part 2
i2c/isp1301_omap: Convert to a new-style i2c driver, part 1
Linus Torvalds [Tue, 14 Oct 2008 23:35:43 +0000 (16:35 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid: (55 commits)
HID: build drivers for all quirky devices by default
HID: add missing blacklist entry for Apple ATV ircontrol
HID: add support for Bright ABNT2 brazilian device
HID: Don't let Avermedia Radio FM800 be handled by usb hid drivers
HID: fix numlock led on Dell device 0x413c/0x2105
HID: remove warn() macro from usb hid drivers
HID: remove info() macro from usb HID drivers
HID: add appletv IR receiver quirk
HID: fix a lockup regression when using force feedback on a PID device
HID: hiddev.h: Fix example code.
HID: hiddev.h: Fix mixed space and tabs in example code.
HID: convert to dev_* prints
HID: remove hid-ff
HID: move zeroplus FF processing
HID: move thrustmaster FF processing
HID: move pantherlord FF processing
HID: fix incorrent length condition in hidraw_write()
HID: fix tty<->hid deadlock
HID: ignore iBuddy devices
HID: report descriptor fix for remaining MacBook JIS keyboards
...
Linus Torvalds [Tue, 14 Oct 2008 23:34:11 +0000 (16:34 -0700)]
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2: (56 commits)
ocfs2: Make cached block reads the common case.
ocfs2: Kill the last naked wait_on_buffer() for cached reads.
ocfs2: Move ocfs2_bread() into dir.c
ocfs2: Simplify ocfs2_read_block()
ocfs2: Require an inode for ocfs2_read_block(s)().
ocfs2: Separate out sync reads from ocfs2_read_blocks()
ocfs2: Refactor xattr list and remove ocfs2_xattr_handler().
ocfs2: Calculate EA hash only by its suffix.
ocfs2: Move trusted and user attribute support into xattr.c
ocfs2: Uninline ocfs2_xattr_name_hash()
ocfs2: Don't check for NULL before brelse()
ocfs2: use smaller counters in ocfs2_remove_xattr_clusters_from_cache
ocfs2: Documentation update for user_xattr / nouser_xattr mount options
ocfs2: make la_debug_mutex static
ocfs2: Remove pointless !!
ocfs2: Add empty bucket support in xattr.
ocfs2/xattr.c: Fix a bug when inserting xattr.
ocfs2: Add xattr mount option in ocfs2_show_options()
ocfs2: Switch over to JBD2.
ocfs2: Add the 'inode64' mount option.
...
Bjorn Helgaas [Tue, 14 Oct 2008 23:01:59 +0000 (17:01 -0600)]
rtc-cmos: look for PNP RTC first, then for platform RTC
We shouldn't rely on "pnp_platform_devices" to tell us whether there
is a PNP RTC device.
I introduced "pnp_platform_devices", but I think it was a mistake.
All it tells us is whether we found any PNPBIOS or PNPACPI devices.
Many machines have some PNP devices, but do not describe the RTC
via PNP. On those machines, we need to do the platform driver probe
to find the RTC.
We should just register the PNP driver and see whether it claims anything.
If we don't find a PNP RTC, fall back to the platform driver probe.
This (in conjunction with the arch/x86/kernel/rtc.c patch to add
a platform RTC device when PNP doesn't have one) should resolve
these issues:
Jiri Kosina [Tue, 14 Oct 2008 21:37:33 +0000 (23:37 +0200)]
HID: build drivers for all quirky devices by default
Once kernel configuration has CONFIG_HID turned on, let also all the
specialized drivers for quirky devices to be built (unless CONFIG_EMBEDDED is
specified), as usually users don't care that much which driver
gives them the functionality, but when they want generic support, they
probably want to have support for all the quirky devices as well.
USB should not be having it's own printk macros, so remove warn() and
use the system-wide standard of dev_warn() wherever possible. In the
few places that will not work out, use a basic printk().
Anssi Hannula [Sat, 4 Oct 2008 12:44:06 +0000 (14:44 +0200)]
HID: fix a lockup regression when using force feedback on a PID device
Commit 8006479c9b75fb6594a7b746af3d7f1fbb68f18f introduced a spinlock in
input_dev->event_lock, which is locked when handling input events.
However, the hid-pidff driver sleeps when handling events as it waits for
reports being sent to the device before changing the report contents
again.
This causes a system lockup when trying to use force feedback with a PID
device, a regression introduced in 2.6.24 and 2.6.23.15.
Fix it by extracting the raw report data from struct hid_report
immediately when hid_submit_report() is called, therefore allowing
drivers to change the contents of struct hid_report immediately without
affecting the already-queued transfer.
In hid-pidff, re-add the removed usbhid_wait_io() to
pidff_erase_effect() instead, to prevent a full report queue from causing
the submission to fail, thus not freeing up device memory.
pidff_erase_effect() is not called while dev->event_lock is held.
hid_compat_load() runs on the default workqueue, it request_module(), it
execs modprobe, it exits, tty flushes default workqueue, it hangs, because
we are still in it.
HID: report descriptor fix for remaining MacBook JIS keyboards
This patch fixes a problem that MacBook JIS keyboard sends wrong report
descriptors. Although it has already been fixed in the first Core 2 Duo model,
it still remains in other models of MacBook.
HID: Autocentering support for Logitech MOMO Racing Wheel
Current kernel has no support for autocentering for Logitech wheels. By
default autocentering enabled in wheel and constant effect does not work
properly. Using USB sniffer I found command which change autocentering
settings: 0xFE, 0x0D, 0x0R, 0x0L, 0x80, 0x00, 0x00, where R - clockwise force,
L - counter-clockwise (0x0-0xF, 0xC = 100%).
Richard Hughes [Wed, 27 Aug 2008 09:19:37 +0000 (11:19 +0200)]
HID: remove ignore quirk for MGE UPS devices
This patch reverts the change made four years ago here:
http://www.vg.kernel.org/pub/linux/kernel/people/gregkh/usb/2.4/usb-hid-2.4.25-pre7.patch
UPS's made by MGE can be used with usbhid just fine, and by removing the
ignore quirk allows them to be used with HAL so they just work when plugged
in, without needing to be manually configured.
With the ignore quirk in place a user would have to configure NUT before the
UPS could be used, as NUT uses it's own internal USB matching framework
to match against the USB devices, do low level control messages on the
device and then parse the HID tables all in userspace.
This is not needed, as allowing the device to be claimed as a usbhid device
allows it to be used like any other USB UPS device. The devices correctly
advertise the power device page which can be queried for the device state.
I assume the quirk was changed so that people using < libusb 0.1.8 could
still use NUT's internal HID code to manage the UPS.
libusb 0.1.8 was released quite some time ago: 2004-02-11.
This patch does not break NUT as in drivers/libusb.c the device is force
unbound from the kernel driver using usb_detach_kernel_driver_np () where
it can be controlled like normal.
[jkosina@suse.cz: adapt to the new hidbus code]
Signed-off-by: Richard Hughes <rhughes@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Jiri Kosina [Wed, 20 Aug 2008 17:13:52 +0000 (19:13 +0200)]
HID: introduce list for hiddev creation forcing
Introduce a list of devices for which there is need to
force a creation of the hiddev interface, but still they
are operated by generic driver (i.e. certain UPS).
Jiri Slaby [Fri, 16 May 2008 09:49:20 +0000 (11:49 +0200)]
HID: move ignore quirks
Move ignore quirks from usbhid-quirks into hid-core code. Also don't output
warning when ENODEV is error code in usbhid and try ordinal input in hidp
when that error is returned.
Jiri Slaby [Fri, 16 May 2008 09:49:16 +0000 (11:49 +0200)]
HID: hid, make parsing event driven
Next step for complete hid bus, this patch includes:
- call parser either from probe or from hid-core if there is no probe.
- add ll_driver structure and centralize some stuff there (open, close...)
- split and merge usb_hid_configure and hid_probe into several functions
to allow hooks/fixes between them
Jiri Slaby [Fri, 16 May 2008 09:49:15 +0000 (11:49 +0200)]
HID: make a bus from hid code
Make a bus from hid core. This is the first step for converting all the
quirks and separate almost-drivers into real drivers attached to this bus.
It's implemented to change behaviour in very tiny manner, so that no driver
needs to be changed this time.
Also add generic drivers for both usb and bt into usbhid or hidp
respectively which will bind all non-blacklisted device. Those blacklisted
will be either grabbed by special drivers or by nobody if they are broken at
the very rude base.
Linus Torvalds [Tue, 14 Oct 2008 19:31:14 +0000 (12:31 -0700)]
Merge branch 'for-2.6.28' of git://linux-nfs.org/~bfields/linux
* 'for-2.6.28' of git://linux-nfs.org/~bfields/linux: (59 commits)
svcrdma: Fix IRD/ORD polarity
svcrdma: Update svc_rdma_send_error to use DMA LKEY
svcrdma: Modify the RPC reply path to use FRMR when available
svcrdma: Modify the RPC recv path to use FRMR when available
svcrdma: Add support to svc_rdma_send to handle chained WR
svcrdma: Modify post recv path to use local dma key
svcrdma: Add a service to register a Fast Reg MR with the device
svcrdma: Query device for Fast Reg support during connection setup
svcrdma: Add FRMR get/put services
NLM: Remove unused argument from svc_addsock() function
NLM: Remove "proto" argument from lockd_up()
NLM: Always start both UDP and TCP listeners
lockd: Remove unused fields in the nlm_reboot structure
lockd: Add helper to sanity check incoming NOTIFY requests
lockd: change nlmclnt_grant() to take a "struct sockaddr *"
lockd: Adjust nlmsvc_lookup_host() to accomodate AF_INET6 addresses
lockd: Adjust nlmclnt_lookup_host() signature to accomodate non-AF_INET
lockd: Support non-AF_INET addresses in nlm_lookup_host()
NLM: Convert nlm_lookup_host() to use a single argument
svcrdma: Add Fast Reg MR Data Types
...
Code Review / linux-2.6.git / log