I am able to reproduce the oops reported by Simon in __switch_to() with
lguest.
My debug showed that there is at least one lguest specific
issue (which should be present in 2.6.25 and before aswell) and it got
exposed with a kernel oops with the recent fpu dynamic allocation patches.
In addition to the previous possible scenario (with fpu_counter), in the
presence of lguest, it is possible that the cpu's TS bit it still set and the
lguest launcher task's thread_info has TS_USEDFPU still set.
This is because of the way the lguest launcher handling the guest's TS bit.
(look at lguest_set_ts() in lguest_arch_run_guest()). This can result
in a DNA fault while doing unlazy_fpu() in __switch_to(). This will
end up causing a DNA fault in the context of new process thats
getting context switched in (as opossed to handling DNA fault in the context
of lguest launcher/helper process).
This is wrong in both pre and post 2.6.25 kernels. In the recent
2.6.26-rc series, this is showing up as NULL pointer dereferences or
sleeping function called from atomic context(__switch_to()), as
we free and dynamically allocate the FPU context for the newly
created threads. Older kernels might show some FPU corruption for processes
running inside of lguest.
With the appended patch, my test system is running for more than 50 mins
now. So atleast some of your oops (hopefully all!) should get fixed.
Please give it a try. I will spend more time with this fix tomorrow.
Reported-by: Simon Holm Thøgersen <odie@cs.aau.dk>
Reported-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Anthony Liguori points out that three different transports use the virtio code,
but each one keeps its own counter to set the virtio_device's index field. In
theory (though not in current practice) this means that names could be
duplicated, and that risk grows as more transports are created.
So we move the selection of the unique virtio_device.index into the common code
in virtio.c, which has the side-benefit of removing duplicate code.
The only complexity is that lguest and S/390 use the index to uniquely identify
the device in case of catastrophic failure before register_virtio_device() is
called: now we use the offset within the descriptor page as a unique identifier
for the printks.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Carsten Otte <cotte@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Chris Lalancette <clalance@redhat.com>
Cc: Anthony Liguori <anthony@codemonkey.ws>
Thanks to Jon Corbet & LWN. Only took me a day to join the dots.
Host->Guest netcat before (with unnecessily large receive buffers):
1073741824 bytes (1.1 GB) copied, 24.7528 seconds, 43.4 MB/s
After:
1073741824 bytes (1.1 GB) copied, 17.6369 seconds, 60.9 MB/s
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This brings us closer to Real Life, where we'd examine the device
features once it's set the DRIVER_OK status bit.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
NR_CPUS (being a host number) is an arbitrary limit for the Guest.
Using the array size directly (which currently happes to be NR_CPUS)
is more futureproof.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
A recent proposed feature addition to the virtio block driver revealed
some flaws in the API: in particular, we assume that feature
negotiation is complete once a driver's probe function returns.
There is nothing in the API to require this, however, and even I
didn't notice when it was violated.
So instead, we require the driver to specify what features it supports
in a table, we can then move the feature negotiation into the virtio
core. The intersection of device and driver features are presented in
a new 'features' bitmap in the struct virtio_device.
Note that this highlights the difference between Linux unsigned-long
bitmaps where each unsigned long is in native endian, and a
straight-forward little-endian array of bytes.
Drivers can still remove feature bits in their probe routine if they
really have to.
API changes:
- dev->config->feature() no longer gets and acks a feature.
- drivers should advertise their features in the 'feature_table' field
- use virtio_has_feature() for extra sanity when checking feature bits
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
None of these files use any of the functionality promised by
asm/semaphore.h. It's possible that they rely on it dragging in some
unrelated header file, but I can't build all these files, so we'll have
fix any build failures as they come up.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Took some cycles to re-read the Lguest Journey end-to-end, fix some
rot and tighten some phrases.
Only comments change. No new jokes, but a couple of recycled old jokes.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Ahmed managed to crash the Host in release_pgd(), which cannot be a Guest
bug, and indeed it wasn't.
The bug was that handing a 0 as the address of the toplevel page table
being manipulated can cause the lookup code in find_pgdir() to return
an uninitialized cache entry (we shadow up to 4 top level page tables
for each Guest).
Commit 37cc8d7f96 introduced this
behaviour in the Guest, uncovering the bug.
The patch which he submitted (which removed the /4 from the index
calculation) simply ensured that these high-indexed entries hit the
early exit path of guest_set_pmd(). But you get lots of segfaults in
guest userspace as the PMDs aren't being updated.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Robert Bragg's 5dc3318528 tightened
(ie. fixed) the checking in __get_vm_area, and it broke lguest.
lguest should pass the exact "end" it wants, not some random constant
(it was possible previously that it would actually get an address
different from SWITCHER_ADDR).
Also, Fabio Checconi pointed out that we should make sure we're not
hitting the fixmap area.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Robert Bragg <robert@sixbynine.org>
If req is LHREQ_INITIALIZE, and the guest has been initialized before
(unlikely), it will attempt to access cpu->tsk even though cpu is not yet
initialized.
Signed-off-by: Eugene Teo <eugeneteo@kernel.sg>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Beginning from commit 4138cc3418, ioremap_nocache() sets the _PAGE_PWT
flag.
Lguest doesn't accept a guest pte with a _PWT flag and reports a "bad
page table entry" in that case.
Accept guest _PAGE_PWT page table entries.
Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Using "attr" twice is not OK, because it effectively prohibits such
container_of() on variables not named "attr".
Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
Acked-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>
A reset function solves three problems:
1) It allows us to renegotiate features, eg. if we want to upgrade a
guest driver without rebooting the guest.
2) It gives us a clean way of shutting down virtqueues: after a reset,
we know that the buffers won't be used by the host, and
3) It helps the guest recover from messed-up drivers.
So we remove the ->shutdown hook, and the only way we now remove
feature bits is via reset.
We leave it to the driver to do the reset before it deletes queues:
the balloon driver, for example, needs to chat to the host in its
remove function.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It seems that virtio_net wants to disable callbacks (interrupts) before
calling netif_rx_schedule(), so we can't use the return value to do so.
Rename "restart" to "cb_enable" and introduce "cb_disable" hook: callback
now returns void, rather than a boolean.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Previously we used a type/len pair within the config space, but this
seems overkill. We now simply define a structure which represents the
layout in the config space: the config space can now only be extended
at the end.
The main driver-visible changes:
1) We indicate what fields are present with an explicit feature bit.
2) Virtqueues are explicitly numbered, and not in the config space.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
drivers/lguest/x86/core.c: In function ‘copy_in_guest_info’:
drivers/lguest/x86/core.c:97: error: ‘struct x86_hw_tss’ has no member named ‘esp1’
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus: (27 commits)
lguest: use __PAGE_KERNEL instead of _PAGE_KERNEL
lguest: Use explicit includes rateher than indirect
lguest: get rid of lg variable assignments
lguest: change gpte_addr header
lguest: move changed bitmap to lg_cpu
lguest: move last_pages to lg_cpu
lguest: change last_guest to last_cpu
lguest: change spte_addr header
lguest: per-vcpu lguest pgdir management
lguest: make pending notifications per-vcpu
lguest: makes special fields be per-vcpu
lguest: per-vcpu lguest task management
lguest: replace lguest_arch with lg_cpu_arch.
lguest: make registers per-vcpu
lguest: make emulate_insn receive a vcpu struct.
lguest: map_switcher_in_guest() per-vcpu
lguest: per-vcpu interrupt processing.
lguest: per-vcpu lguest timers
lguest: make hypercalls use the vcpu struct
lguest: make write() operation smp aware
...
Manual conflict resolved (maybe even correctly, who knows) in
drivers/lguest/x86/core.c
This changes size-specific register names (eip/rip, esp/rsp, etc.) to
generic names in the thread and tss structures.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
x86_64 don't expose the intermediate representation with one underline,
_PAGE_KERNEL, just the double-underlined one.
Use it, to get a common ground between 32 and 64-bit
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
explicitly use ktime.h include
explicitly use hrtimer.h include
explicitly use sched.h include
This patch adds headers explicitly to lguest sources file,
to avoid depending on them being included somewhere else.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We can save some lines of code by getting rid of
*lg = cpu... lines of code spread everywhere by now.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
gpte_addr() does not depend on any guest information. So we wipe out
the lg parameter from it completely.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
events represented in the 'changed' bitmap are per-cpu, not per-guest.
move it to the lg_cpu structure
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
in our new model, pages are assigned to a virtual cpu, not to a guest.
We move it to the lg_cpu structure.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
in our model, a guest does not run in a cpu anymore: a virtual cpu
does. So we change last_guest to last_cpu
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
spte_addr does not depend on any guest information, so we
wipe out the lg parameter completely.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
this patch makes the pgdir management per-vcpu. The pgdirs pool
is still guest-wide (although it'll probably need to grow when we
are really executing more vcpus), but the pgdidx index is gone,
since it makes no sense anymore. Instead, we use a per-vcpu
index.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
this patch makes the pending_notify field, used to control
pending notifications, per-vcpu, instead of per-guest
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
lguest struct have room for some fields, namely, cr2, ts, esp1
and ss1, that are not really guest-wide, but rather, vcpu-wide.
This patch puts it in the vcpu struct
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
lguest uses tasks to control its running behaviour (like sending
breaks, controlling halted state, etc). In a per-vcpu environment,
each vcpu will have its own underlying task. So this patch
makes the infrastructure for that possible
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The fields found in lguest_arch are not really per-guest,
but per-cpu (gdt, idt, etc). So this patch turns lguest_arch
into lg_cpu_arch.
It makes sense to have a per-guest per-arch struct, but this
can be addressed later, when the need arrives.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the most obvious per-vcpu field: registers.
So this patch moves it from struct lguest to struct vcpu,
and patch the places in which they are used, accordingly
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
emulate_insn() needs to know about current eip, which will be,
in the future, a per-vcpu thing. So in this patch, the function
prototype is modified to receive a vcpu struct
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The switcher needs to be mapped per-vcpu, because different vcpus
will potentially have different page tables (they don't have to,
because threads will share the same).
So our first step is the make the function receive a vcpu struct
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This patch adapts interrupt processing for using the vcpu struct.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Here, I introduce per-vcpu timers. With this, we can have
local expiries, needed for accounting time in smp guests
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
this patch changes do_hcall() and do_async_hcall() interfaces (and obviously their
callers) to get a vcpu struct. Again, a vcpu services the hypercall, not the whole
guest
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This patch makes the write() file operation smp aware. Which means, receiving
the vcpu_id value through the offset parameter, and being well aware to which
vcpu we're talking to.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This patch makes the run_guest() routine use the lg_cpu struct.
This is required since in a smp guest environment, there's no
more the notion of "running the guest", but rather, it is "running the vcpu"
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
this patch initializes the first vcpu in the initialize() routing,
which is responsible for starting the process of putting the guest up.
right now, as much of the fields are still not per-vcpu, it does not
do much.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
this patch introduces a vcpu struct for lguest. In upcoming patches,
more and more fields will be moved from the lguest struct to the vcpu
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Currently, lguest module can't be compiled without the PARAVIRT flag being
on. This is a fake dependency, since the module itself shouldn't need any
paravirt override. Reason for that is the reference to pv_info structure
in initial loading tests.
This patch removes it in favour of a more generic error message.
Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Replace all lock_cpu_hotplug/unlock_cpu_hotplug from the kernel and use
get_online_cpus and put_online_cpus instead as it highlights the
refcount semantics in these operations.
The new API guarantees protection against the cpu-hotplug operation, but
it doesn't guarantee serialized access to any of the local data
structures. Hence the changes needs to be reviewed.
In case of pseries_add_processor/pseries_remove_processor, use
cpu_maps_update_begin()/cpu_maps_update_done() as we're modifying the
cpu_present_map there.
Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
It was moved to arch/x86/lguest/Kconfig, but I lost the deletion part in a
patch suffle. My confused one-liner "fix" to turn it on is also reverted:
84f7466ee2
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There's currently no way to turn on Lguest guest support; the planned
Kconfig virtualization reorg didn't get into 2.6.25.
This was unnoticed because if you already had CONFIG_LGUEST_GUEST=y in
your config, it worked. Too bad about new users...
Also, the Kconfig help was wrong now the virtio drivers are merged.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Suresh Siddha