On Tue, Jan 11, 2011 at 12:29:41PM +0200, Avi Kivity wrote:
> On 01/11/2011 11:49 AM, Avi Kivity wrote:
> > On 01/10/2011 09:31 PM, Kirill A. Shutemov wrote:
> >> On Mon, Jan 10, 2011 at 10:52:05AM -0600, Christoph Lameter wrote:
> >> >
> >> > On Mon, 10 Jan 2011, Kirill A. Shutemov wrote:
> >> >
> >> > > Every time I run qemu with KVM enabled I get this in dmesg:
> >> > >
> >> > > [ 182.878328] BUG: sleeping function called from invalid
> >> context at mm/slub.c:793
> >> > > [ 182.878339] in_atomic(): 1, irqs_disabled(): 0, pid: 4992,
> >> name: qemu
> >> > > [ 182.878355] Pid: 4992, comm: qemu Not tainted 2.6.37+ #31
> >> > > [ 182.878361] Call Trace:
> >> > > [ 182.878381] [<c104e317>] ? __might_sleep+0xd0/0xd7
> >> > > [ 182.878394] [<c10ec337>] ?
> >> slab_pre_alloc_hook.clone.39+0x23/0x27
> >> > > [ 182.878404] [<c10ece27>] ? kmem_cache_alloc+0x22/0xc8
> >> > > [ 182.878414] [<c1030221>] ? init_fpu+0x44/0x7b
> >> >
> >> > fpu_alloc() does call kmem_cache_alloc with GFP_KERNEL although we
> >> are in
> >> > an atomic context.
> >>
> >> Something like this?
> >>
> >> ---
> >> From 7c6fbfed72e7d22cbdf7393f9711d521e0fbb4a6 Mon Sep 17 00:00:00 2001
> >> From: Kirill A. Shutemov<kirill@xxxxxxxxxxxxx>
> >> Date: Mon, 10 Jan 2011 21:24:23 +0200
> >> Subject: [PATCH] x86, fpu_alloc(): call kmem_cache_alloc() with
> >> GFP_ATOMIC
> >>
> >> [ 182.878328] BUG: sleeping function called from invalid context at
> >> mm/slub.c:793
> >> [ 182.878339] in_atomic(): 1, irqs_disabled(): 0, pid: 4992, name: qemu
> >> [ 182.878355] Pid: 4992, comm: qemu Not tainted 2.6.37+ #31
> >> [ 182.878361] Call Trace:
> >> [ 182.878381] [<c104e317>] ? __might_sleep+0xd0/0xd7
> >> [ 182.878394] [<c10ec337>] ? slab_pre_alloc_hook.clone.39+0x23/0x27
> >> [ 182.878404] [<c10ece27>] ? kmem_cache_alloc+0x22/0xc8
> >> [ 182.878414] [<c1030221>] ? init_fpu+0x44/0x7b
> >> [ 182.878426] [<c130cc29>] ? do_device_not_available+0x0/0x1b
> >> [ 182.878435] [<c1030221>] ? init_fpu+0x44/0x7b
> >> [ 182.878444] [<c102a588>] ? math_state_restore+0x24/0x47
> >> [ 182.878453] [<c130cc39>] ? do_device_not_available+0x10/0x1b
> >> [ 182.878462] [<c130c4ab>] ? error_code+0x67/0x6c
> >> [ 182.878475] [<c1012340>] ? kvm_load_guest_fpu+0xa1/0xaa
> >> [ 182.878484] [<c1013364>] ? kvm_arch_vcpu_ioctl_run+0x798/0xbe8
> >> [ 182.878496] [<c1004523>] ? kvm_vcpu_ioctl+0x105/0x46e
> >> [ 182.878508] [<c107dce0>] ? get_futex_key+0x73/0x132
> >> [ 182.878517] [<c107e352>] ? futex_wake+0xb6/0xc0
> >> [ 182.878527] [<c107f8d6>] ? do_futex+0x87/0x669
> >> [ 182.878535] [<c100441e>] ? kvm_vcpu_ioctl+0x0/0x46e
> >> [ 182.878545] [<c1101ebf>] ? do_vfs_ioctl+0x4a0/0x4d1
> >> [ 182.878554] [<c130e348>] ? do_page_fault+0x2eb/0x316
> >> [ 182.878564] [<c1101f36>] ? sys_ioctl+0x46/0x68
> >> [ 182.878572] [<c130bdc0>] ? syscall_call+0x7/0xb
> >> [ 182.878585] [<c1300000>] ? aer_probe+0x1da/0x274
> >>
> >> Signed-off-by: Kirill A. Shutemov<kirill@xxxxxxxxxxxxx>
> >> ---
> >> arch/x86/include/asm/i387.h | 2 +-
> >> 1 files changed, 1 insertions(+), 1 deletions(-)
> >>
> >> diff --git a/arch/x86/include/asm/i387.h b/arch/x86/include/asm/i387.h
> >> index ef32890..8b896dd 100644
> >> --- a/arch/x86/include/asm/i387.h
> >> +++ b/arch/x86/include/asm/i387.h
> >> @@ -426,7 +426,7 @@ static inline int fpu_alloc(struct fpu *fpu)
> >> {
> >> if (fpu_allocated(fpu))
> >> return 0;
> >> - fpu->state = kmem_cache_alloc(task_xstate_cachep, GFP_KERNEL);
> >> + fpu->state = kmem_cache_alloc(task_xstate_cachep, GFP_ATOMIC);
> >> if (!fpu->state)
> >> return -ENOMEM;
> >> WARN_ON((unsigned long)fpu->state& 15);
> >
> > If this fails, a task will be killed. I'll patch kvm to ensure that
> > the fpu is initialized.
> >
>
> Please try out the attached patch.
It helps.
Reported-and-tested-by: Kirill A. Shutemov <kas@xxxxxxxxxx>
Thanks.
>
> --
> error compiling committee.c: too many arguments to function
>
> From f3a6041b5bb3bf7c88f9694a66d7f34be2f78845 Mon Sep 17 00:00:00 2001
> From: Avi Kivity <avi@xxxxxxxxxx>
> Date: Tue, 11 Jan 2011 12:15:54 +0200
> Subject: [PATCH] KVM: Initialize fpu state in preemptible context
>
> init_fpu() (which is indirectly called by the fpu switching code) assumes
> it is in process context. Rather than makeing init_fpu() use an atomic
> allocation, which can cause a task to be killed, make sure the fpu is
> already initialized when we enter the run loop.
>
> Signed-off-by: Avi Kivity <avi@xxxxxxxxxx>
> ---
> arch/x86/kernel/i387.c | 1 +
> arch/x86/kvm/x86.c | 3 +++
> 2 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
> index 58bb239..e60c38c 100644
> --- a/arch/x86/kernel/i387.c
> +++ b/arch/x86/kernel/i387.c
> @@ -169,6 +169,7 @@ int init_fpu(struct task_struct *tsk)
> set_stopped_child_used_math(tsk);
> return 0;
> }
> +EXPORT_SYMBOL_GPL(init_fpu);
>
> /*
> * The xstateregs_active() routine is the same as the fpregs_active() routine,
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 8652643..fd93cda 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -5351,6 +5351,9 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
> int r;
> sigset_t sigsaved;
>
> + if (!tsk_used_math(current) && init_fpu(current))
> + return -ENOMEM;
> +
> if (vcpu->sigset_active)
> sigprocmask(SIG_SETMASK, &vcpu->sigset, &sigsaved);
>
> --
> 1.7.1
>
--
Kirill A. Shutemov
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
