This makes lguest able to use the virtio devices.
We change the device descriptor page from a simple array to a variable
length "type, config_len, status, config data..." format, and
implement virtio_config_ops to read from that config data.
We use the virtio ring implementation for an efficient Guest <-> Host
virtqueue mechanism, and the new LHCALL_NOTIFY hypercall to kick the
host when it changes.
We also use LHCALL_NOTIFY on kernel addresses for very very early
console output. We could have another hypercall, but this hack works
quite well.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This patch gets rid of the old lguest host I/O infrastructure and
replaces it with a single hypercall "LHCALL_NOTIFY" which takes an
address.
The main change is the removal of io.c: that mainly did inter-guest
I/O, which virtio doesn't yet support.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This gets rid of the lguest bus, drivers and DMA mechanism, to make
way for a generic virtio mechanism.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1) This allows us to get alot closer to booting bzImages.
2) It means we don't have to know page_offset.
3) The Guest needs to modify the boot pagetables to create the
PAGE_OFFSET mapping before jumping to C code.
4) guest_pa() walks the page tables rather than using page_offset.
5) We don't use page_offset to figure out whether to emulate: it was
always kinda quesationable, and won't work for instructions done
before remapping (bzImage unpacking in particular).
6) We still want the kernel address for tlb flushing: have the initial
hypercall give us that, too.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
(Based on Ron Minnich's LGUEST_PLAN9_SYSCALL patch).
This patch allows Guests to specify what system call vector they want,
and we try to reserve it. We only allow one non-Linux system call
vector, to try to avoid DoS on the Host.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is my first step in the migration of page_tables.c to the kernel
types and functions/macros (2.6.23-rc3). Seems to be working OK.
Signed-off-by: Matias Zabaljauregui <matias.zabaljauregui@cern.ch>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Move setup_regs() to lguest_arch_setup_regs() in i386_core.c given
that this is very architecture specific.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Apply Clue 2x4 to lguest userland<->kernel handling code and the
lguest launcher. Pointers are not to be passed in u32's!
Basic rule of thumb: Anything passing u32's back and forth should be
passing unsigned longs to be portable to 64 bit archs.
For those who forgotten already, I repeat: NO POINTERS IN u32!
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Clean up the hypercall code to make the code in hypercalls.c
architecture independent. First process the common hypercalls and
then call lguest_arch_do_hcall() if the call hasn't been handled.
Rename struct hcall_ring to hcall_args.
This patch requires the previous patch which reorganize the layout of
struct lguest_regs on i386 so they match the layout of struct
hcall_args.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Currently we look at the "trapnum" to see if the Guest wants a
hypercall. But once the hypercall is done we have to reset trapnum to
a bogus value, otherwise if we exit to userspace and return, we'd run
the same hypercall twice (that was a nasty bug to find!).
This has two main effects:
1) When Jes's patch changes the hypercall args to be a generic "struct
hcall_args" we simply change the type of "lg->hcall". It's set by
arch code, so if it has to copy args or something it can do so, and
point "hcall" into lg->arch somewhere.
2) Async hypercalls only get run when an actual hypercall is pending.
This simplfies the code a little and is a more logical semantic.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Move eax next to ebx/ecx/edx in struct lguest_regs on i386, so they
will be located together and allow it to map directly to a struct
hcall_ring entry (which will be renamed struct hcall_args as in a
subsequent patch).
This is in preparation for making the code hcall code architecture
independent.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Separate i386 architecture specific from core.c and move it to
x86/core.c and add x86/lguest.h header file to match.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This simplifies the code a little, in preparation for allowing
alternate system call vectors in guests (Plan 9 uses 0x40).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Back when we had all the Guest state in the switcher, we had a fixed
array of them. This is no longer necessary.
If we switch the network code to using random_ether_addr (46 bits is
enough to avoid clashes), we can get rid of the concept of "guest id"
altogether.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
In order to avoid problematic special linking of the Launcher, we give
the Host an offset: this means we can use any memory region in the
Launcher as Guest memory rather than insisting on mmap() at 0.
The result is quite pleasing: a number of casts are replaced with
simple additions.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
lguest uses a "switcher" shim mapped high to bounce between host and
guest. As lguest becomes less i386-centric, we separate this code
into a subdir.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Lguest has two sides: host support (to launch guests) and guest
support (replacement boot path and paravirt_ops). This moves the
guest side to arch/x86/lguest where it's closer to related code.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Currently lguest will spend a lot of of time waking up the host, as it
cannot go tickless (if the [host] TSC has been marked unstable). On my
laptop I was getting ~40% of wakeups from lguest.
With this patch applied, my laptop is much happier!
Signed-off-by: Tony Breeds <tony@bakeyournoodle.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Use copy_to_user() when copying a struct timespec to the guest -
put_user() cannot handle two long's in one go on a 64bit arch.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Jes Sorensen <jes@sgi.com>
Cc: Al Viro <viro@ftp.linux.org.uk>
1) Group all the "guest OS" support options together, under a PARAVIRT_GUEST
menu.
2) Make those options select CONFIG_PARAVIRT, as suggested by Andi.
3) Make kconfig help titles consistent.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Cc: Zach Amsden <zach@vmware.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Chris Wright <chrisw@sous-sol.org>
* 'xen-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/jeremy/xen:
xfs: eagerly remove vmap mappings to avoid upsetting Xen
xen: add some debug output for failed multicalls
xen: fix incorrect vcpu_register_vcpu_info hypercall argument
xen: ask the hypervisor how much space it needs reserved
xen: lock pte pages while pinning/unpinning
xen: deal with stale cr3 values when unpinning pagetables
xen: add batch completion callbacks
xen: yield to IPI target if necessary
Clean up duplicate includes in arch/i386/xen/
remove dead code in pgtable_cache_init
paravirt: clean up lazy mode handling
paravirt: refactor struct paravirt_ops into smaller pv_*_ops
Currently, the set_lazy_mode pv_op is overloaded with 5 functions:
1. enter lazy cpu mode
2. leave lazy cpu mode
3. enter lazy mmu mode
4. leave lazy mmu mode
5. flush pending batched operations
This complicates each paravirt backend, since it needs to deal with
all the possible state transitions, handling flushing, etc. In
particular, flushing is quite distinct from the other 4 functions, and
seems to just cause complication.
This patch removes the set_lazy_mode operation, and adds "enter" and
"leave" lazy mode operations on mmu_ops and cpu_ops. All the logic
associated with enter and leaving lazy states is now in common code
(basically BUG_ONs to make sure that no mode is current when entering
a lazy mode, and make sure that the mode is current when leaving).
Also, flush is handled in a common way, by simply leaving and
re-entering the lazy mode.
The result is that the Xen, lguest and VMI lazy mode implementations
are much simpler.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Andi Kleen <ak@suse.de>
Cc: Zach Amsden <zach@vmware.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Anthony Liguory <aliguori@us.ibm.com>
Cc: "Glauber de Oliveira Costa" <glommer@gmail.com>
Cc: Jun Nakajima <jun.nakajima@intel.com>
This patch refactors the paravirt_ops structure into groups of
functionally related ops:
pv_info - random info, rather than function entrypoints
pv_init_ops - functions used at boot time (some for module_init too)
pv_misc_ops - lazy mode, which didn't fit well anywhere else
pv_time_ops - time-related functions
pv_cpu_ops - various privileged instruction ops
pv_irq_ops - operations for managing interrupt state
pv_apic_ops - APIC operations
pv_mmu_ops - operations for managing pagetables
There are several motivations for this:
1. Some of these ops will be general to all x86, and some will be
i386/x86-64 specific. This makes it easier to share common stuff
while allowing separate implementations where needed.
2. At the moment we must export all of paravirt_ops, but modules only
need selected parts of it. This allows us to export on a case by case
basis (and also choose which export license we want to apply).
3. Functional groupings make things a bit more readable.
Struct paravirt_ops is now only used as a template to generate
patch-site identifiers, and to extract function pointers for inserting
into jmp/calls when patching. It is only instantiated when needed.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Cc: Zach Amsden <zach@vmware.com>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Anthony Liguory <aliguori@us.ibm.com>
Cc: "Glauber de Oliveira Costa" <glommer@gmail.com>
Cc: Jun Nakajima <jun.nakajima@intel.com>
The assembly templates for lguest guest patching are in the .init.text
section. This means that modules get patched with "cc cc cc cc" or similar
junk.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
One of the very first things lguest_init() does is a memcpy. On
Athlon/Duron/K7 or CyrixIII/VIA-C3 or Geode GX/LX, this tries to use
MMX.
memcpy -> _mmx_memcpy -> kernel_fpu_begin -> clts -> paravirt_ops.clts
But we haven't set paravirt_ops.clts yet, so we do the native version
and crash. The simplest solution is to use __memcpy.
Thanks to Michael Rasenberger for the bug report.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If the stack pointer is 0xc057a000, then the first stack page is at
0xc0579000 (the stack pointer is decremented before use). Not
calculating this correctly caused guests with CONFIG_DEBUG_PAGEALLOC=y
to be killed with a "bad stack page" message: the initial kernel stack
was just proceeding the .smp_locks section which
CONFIG_DEBUG_PAGEALLOC marks read-only when freeing.
Thanks to Frederik Deweerdt for the bug report!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 19d36ccdc3 "x86: Fix alternatives
and kprobes to remap write-protected kernel text" uses code which is
being patched for patching.
In particular, paravirt_ops does patching in two stages: first it
calls paravirt_ops.patch, then it fills any remaining instructions
with nop_out(). nop_out calls text_poke() which calls
lookup_address() which calls pgd_val() (aka paravirt_ops.pgd_val):
that call site is one of the places we patch.
If we always do patching as one single call to text_poke(), we only
need make sure we're not patching the memcpy in text_poke itself.
This means the prototype to paravirt_ops.patch needs to change, to
marshal the new code into a buffer rather than patching in place as it
does now. It also means all patching goes through text_poke(), which
is known to be safe (apply_alternatives is also changed to make a
single patch).
AK: fix compilation on x86-64 (bad rusty!)
AK: fix boot on x86-64 (sigh)
AK: merged with other patches
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Files using bits from paravirt.h should explicitly include it rather than
relying on it being pulled in by something else.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If a Guest makes hypercall which sets a GDT entry to not present, we
currently set any segment registers using that GDT entry to 0.
Unfortunately, this is not sufficient: there are other ways of
altering GDT entries which will cause a fault.
The correct solution to do what Linux does: let them set any GDT value
they want and handle the #GP when popping causes a fault. This has
the added benefit of making our Switcher slightly more robust in the
case of any other bugs which cause it to fault.
We kill the Guest if it causes a fault in the Switcher: it's the
Guest's responsibility to make sure it's not using segments when it
changes them.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
lguest uses a host-supplied wallclock-based clocksource when the TSC
is not reliable. As this is already in nanoseconds, I naively used a
multiplier of 1 and a shift of 0.
But update_wall_time() in its infinite wisdom decides to adjust the
clock a little (where does it think it's getting a more accurate time
from?)
It will happily tweak the multiplier... to 0, then -1.
So the "fix" is to use a shift of 22 like everyone else, and a
multiplier of 1 << 22.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Lguest drivers need to default to "Y" otherwise they're never selected
for new builds. (We don't bother prompting, because they're less than
4k combined, and implied by selecting lguest support).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Gabriel C reports lguest doesn't compile with CONFIG_BLOCK=n. Fix this
by introducing a config var for the block device, which depends on
LGUEST && BLOCK. Do the same for the net driver, rather then depending
gratuitously on CONFIG_NET.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Gabriel C <nix.or.die@googlemail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A non-periodic clock_event_device and the "jiffies" clock don't mix well:
tick_handle_periodic() can go into an infinite loop.
Currently lguest guests use the jiffies clock when the TSC is
unusable. Instead, make the Host write the current time into the lguest
page on every interrupt. This doesn't cost much but is more precise
and at least as accurate as the jiffies clock. It also gets rid of
the GET_WALLCLOCK hypercall.
Also, delay setting sched_clock until our clock is set up, otherwise
the early printk timestamps can go backwards (not harmful, just ugly).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jason Yeh sent his crashing .config: bzImages made with
CONFIG_RELOCATABLE=y put the relocs where the BSS is expected, and we
crash with unusual results such as:
lguest: unhandled trap 14 at 0xc0122ae1 (0xa9)
Relying on BSS being zero was merely laziness on my part, and
unfortunately, lguest doesn't go through the normal startup path (which
does this in asm).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The FIXMEs
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The Switcher
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The Host
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The Launcher
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The Drivers
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation: The Guest
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The netfilter code had very good documentation: the Netfilter Hacking HOWTO.
Noone ever read it.
So this time I'm trying something different, using a bit of Knuthiness.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We need to make sure, that the clockevent devices are resumed, before
the tick is resumed. The current resume logic does not guarantee this.
Add CLOCK_EVT_MODE_RESUME and call the set mode functions of the clock
event devices before resuming the tick / oneshot functionality.
Fixup the existing users.
Thanks to Nigel Cunningham for tracking down a long standing thinko,
which affected the jinxed VAIO.
[akpm@linux-foundation.org: xen build fix]
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Guests currently use the default scheduler clock: this means they
always use jiffies even if TSC is actually available. It doesn't make
any noticeable difference here, but it's a better thing to do.
Also remove commented-out asm/sched-clock.h from -mm tree.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The sense of the IF bit is backwards in the host interrupt handling.
This means we always save "IF=1" on the stack when injecting an
interrupt. It turns out this is almost always correct (unless the
guest is taking a page fault in an interrupt due to an unpopulated
vmalloc mapping), so went unnoticed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the Kconfig and Makefile to allow lguest to actually be
compiled.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the code for the "lg.ko" module, which allows lguest guests to
be launched.
[akpm@linux-foundation.org: update for futex-new-private-futexes]
[akpm@linux-foundation.org: build fix]
[jmorris@namei.org: lguest: use hrtimers]
[akpm@linux-foundation.org: x86_64 build fix]
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
lguest is a simple hypervisor for Linux on Linux. Unlike kvm it doesn't need
VT/SVM hardware. Unlike Xen it's simply "modprobe and go". Unlike both, it's
5000 lines and self-contained.
Performance is ok, but not great (-30% on kernel compile). But given its
hackability, I expect this to improve, along with the paravirt_ops code which
it supplies a complete example for. There's also a 64-bit version being
worked on and other craziness.
But most of all, lguest is awesome fun! Too much of the kernel is a big ball
of hair. lguest is simple enough to dive into and hack, plus has some warts
which scream "fork me!".
This patch:
This is the code and headers required to make an i386 kernel an lguest guest.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
