WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

[Xen-devel] Re: [Xen-users] 2.6.23 oops

To: Mark Williamson <mark.williamson@xxxxxxxxxxxx>
Subject: [Xen-devel] Re: [Xen-users] 2.6.23 oops
From: Jeremy Fitzhardinge <jeremy@xxxxxxxx>
Date: Thu, 11 Oct 2007 09:41:44 -0700
Cc: xen-devel@xxxxxxxxxxxxxxxxxxx, Morten Bøgeskov <xen-users@xxxxxxxxxxxxxxxxxx>
Delivery-date: Thu, 11 Oct 2007 09:43:09 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
In-reply-to: <200710111629.05457.mark.williamson@xxxxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
References: <20071011144817.n5x1drwcgug44sg0@xxxxxxxxxxxx> <200710111629.05457.mark.williamson@xxxxxxxxxxxx>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Thunderbird 2.0.0.5 (X11/20070727)
Mark Williamson wrote:
> I'm bringing this discussion onto Xen-devel as it smells like it needs some 
> more specific developer input than I can give.
>
>   
>>   I've tried to run 2.6.23 on xen 3.1.0, my hardware is a Dual Athlon MP.
>> my DomU is running 2.6.23 SMP (vcpus = 1), without SMP it hang during
>> boot, with it panics. I've attached my config, if somebody thinks I've
>> left something out
>>     

Oh, I just fixed this.  Try these patches.

    J
Subject: xen: add batch completion callbacks

This adds a mechanism to register a callback function to be called once
a batch of hypercalls has been issued.  This is typically used to unlock
things which must remain locked until the hypercall has taken place.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xxxxxxxxxxxxx>
---
 arch/i386/xen/multicalls.c |   29 ++++++++++++++++++++++++++---
 arch/i386/xen/multicalls.h |    3 +++
 2 files changed, 29 insertions(+), 3 deletions(-)

===================================================================
--- a/arch/i386/xen/multicalls.c
+++ b/arch/i386/xen/multicalls.c
@@ -32,7 +32,11 @@ struct mc_buffer {
 struct mc_buffer {
        struct multicall_entry entries[MC_BATCH];
        u64 args[MC_ARGS];
-       unsigned mcidx, argidx;
+       struct callback {
+               void (*fn)(void *);
+               void *data;
+       } callbacks[MC_BATCH];
+       unsigned mcidx, argidx, cbidx;
 };
 
 static DEFINE_PER_CPU(struct mc_buffer, mc_buffer);
@@ -43,6 +47,7 @@ void xen_mc_flush(void)
        struct mc_buffer *b = &__get_cpu_var(mc_buffer);
        int ret = 0;
        unsigned long flags;
+       int i;
 
        BUG_ON(preemptible());
 
@@ -51,8 +56,6 @@ void xen_mc_flush(void)
        local_irq_save(flags);
 
        if (b->mcidx) {
-               int i;
-
                if (HYPERVISOR_multicall(b->entries, b->mcidx) != 0)
                        BUG();
                for (i = 0; i < b->mcidx; i++)
@@ -64,6 +67,13 @@ void xen_mc_flush(void)
                BUG_ON(b->argidx != 0);
 
        local_irq_restore(flags);
+
+       for(i = 0; i < b->cbidx; i++) {
+               struct callback *cb = &b->callbacks[i];
+
+               (*cb->fn)(cb->data);
+       }
+       b->cbidx = 0;
 
        BUG_ON(ret);
 }
@@ -88,3 +98,16 @@ struct multicall_space __xen_mc_entry(si
 
        return ret;
 }
+
+void xen_mc_callback(void (*fn)(void *), void *data)
+{
+       struct mc_buffer *b = &__get_cpu_var(mc_buffer);
+       struct callback *cb;
+
+       if (b->cbidx == MC_BATCH)
+               xen_mc_flush();
+
+       cb = &b->callbacks[b->cbidx++];
+       cb->fn = fn;
+       cb->data = data;
+}
===================================================================
--- a/arch/i386/xen/multicalls.h
+++ b/arch/i386/xen/multicalls.h
@@ -42,4 +42,7 @@ static inline void xen_mc_issue(unsigned
        local_irq_restore(x86_read_percpu(xen_mc_irq_flags));
 }
 
+/* Set up a callback to be called when the current batch is flushed */
+void xen_mc_callback(void (*fn)(void *), void *data);
+
 #endif /* _XEN_MULTICALLS_H */
Subject: xen: deal with stale cr3 values when unpinning pagetables

When a pagetable is no longer in use, it must be unpinned so that its
pages can be freed.  However, this is only possible if there are no
stray uses of the pagetable.  The code currently deals with all the
usual cases, but there's a rare case where a vcpu is changing cr3, but
is doing so lazily, and the change hasn't actually happened by the time
the pagetable is unpinned, even though it appears to have been completed.

This change adds a second per-cpu cr3 variable - xen_current_cr3 -
which tracks the actual state of the vcpu cr3.  It is only updated once
the actual hypercall to set cr3 has been completed.  Other processors
wishing to unpin a pagetable can check other vcpu's xen_current_cr3
values to see if any cross-cpu IPIs are needed to clean things up.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xxxxxxxxxxxxx>

---
 arch/i386/xen/enlighten.c |   63 ++++++++++++++++++++++++++++++---------------
 arch/i386/xen/mmu.c       |   33 ++++++++++++++++++++---
 arch/i386/xen/xen-ops.h   |    1 
 3 files changed, 71 insertions(+), 26 deletions(-)

===================================================================
--- a/arch/i386/xen/enlighten.c
+++ b/arch/i386/xen/enlighten.c
@@ -55,7 +55,23 @@ DEFINE_PER_CPU(enum paravirt_lazy_mode, 
 
 DEFINE_PER_CPU(struct vcpu_info *, xen_vcpu);
 DEFINE_PER_CPU(struct vcpu_info, xen_vcpu_info);
-DEFINE_PER_CPU(unsigned long, xen_cr3);
+
+/*
+ * Note about cr3 (pagetable base) values:
+ *
+ * xen_cr3 contains the current logical cr3 value; it contains the
+ * last set cr3.  This may not be the current effective cr3, because
+ * its update may be being lazily deferred.  However, a vcpu looking
+ * at its own cr3 can use this value knowing that it everything will
+ * be self-consistent.
+ *
+ * xen_current_cr3 contains the actual vcpu cr3; it is set once the
+ * hypercall to set the vcpu cr3 is complete (so it may be a little
+ * out of date, but it will never be set early).  If one vcpu is
+ * looking at another vcpu's cr3 value, it should use this variable.
+ */
+DEFINE_PER_CPU(unsigned long, xen_cr3);         /* cr3 stored as physaddr */
+DEFINE_PER_CPU(unsigned long, xen_current_cr3);         /* actual vcpu cr3 */
 
 struct start_info *xen_start_info;
 EXPORT_SYMBOL_GPL(xen_start_info);
@@ -631,32 +647,36 @@ static unsigned long xen_read_cr3(void)
        return x86_read_percpu(xen_cr3);
 }
 
+static void set_current_cr3(void *v)
+{
+       x86_write_percpu(xen_current_cr3, (unsigned long)v);
+}
+
 static void xen_write_cr3(unsigned long cr3)
 {
+       struct mmuext_op *op;
+       struct multicall_space mcs;
+       unsigned long mfn = pfn_to_mfn(PFN_DOWN(cr3));
+
        BUG_ON(preemptible());
 
-       if (cr3 == x86_read_percpu(xen_cr3)) {
-               /* just a simple tlb flush */
-               xen_flush_tlb();
-               return;
-       }
-
+       mcs = xen_mc_entry(sizeof(*op));  /* disables interrupts */
+
+       /* Update while interrupts are disabled, so its atomic with
+          respect to ipis */
        x86_write_percpu(xen_cr3, cr3);
 
-
-       {
-               struct mmuext_op *op;
-               struct multicall_space mcs = xen_mc_entry(sizeof(*op));
-               unsigned long mfn = pfn_to_mfn(PFN_DOWN(cr3));
-
-               op = mcs.args;
-               op->cmd = MMUEXT_NEW_BASEPTR;
-               op->arg1.mfn = mfn;
-
-               MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
-
-               xen_mc_issue(PARAVIRT_LAZY_CPU);
-       }
+       op = mcs.args;
+       op->cmd = MMUEXT_NEW_BASEPTR;
+       op->arg1.mfn = mfn;
+
+       MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+       /* Update xen_update_cr3 once the batch has actually
+          been submitted. */
+       xen_mc_callback(set_current_cr3, (void *)cr3);
+
+       xen_mc_issue(PARAVIRT_LAZY_CPU);  /* interrupts restored */
 }
 
 /* Early in boot, while setting up the initial pagetable, assume
@@ -1124,6 +1144,7 @@ asmlinkage void __init xen_start_kernel(
        /* keep using Xen gdt for now; no urgent need to change it */
 
        x86_write_percpu(xen_cr3, __pa(pgd));
+       x86_write_percpu(xen_current_cr3, __pa(pgd));
 
 #ifdef CONFIG_SMP
        /* Don't do the full vcpu_info placement stuff until we have a
===================================================================
--- a/arch/i386/xen/mmu.c
+++ b/arch/i386/xen/mmu.c
@@ -564,20 +564,43 @@ static void drop_other_mm_ref(void *info
 
        if (__get_cpu_var(cpu_tlbstate).active_mm == mm)
                leave_mm(smp_processor_id());
+
+       /* If this cpu still has a stale cr3 reference, then make sure
+          it has been flushed. */
+       if (x86_read_percpu(xen_current_cr3) == __pa(mm->pgd)) {
+               load_cr3(swapper_pg_dir);
+               arch_flush_lazy_cpu_mode();
+       }
 }
 
 static void drop_mm_ref(struct mm_struct *mm)
 {
+       cpumask_t mask;
+       unsigned cpu;
+
        if (current->active_mm == mm) {
                if (current->mm == mm)
                        load_cr3(swapper_pg_dir);
                else
                        leave_mm(smp_processor_id());
-       }
-
-       if (!cpus_empty(mm->cpu_vm_mask))
-               xen_smp_call_function_mask(mm->cpu_vm_mask, drop_other_mm_ref,
-                                          mm, 1);
+               arch_flush_lazy_cpu_mode();
+       }
+
+       /* Get the "official" set of cpus referring to our pagetable. */
+       mask = mm->cpu_vm_mask;
+
+       /* It's possible that a vcpu may have a stale reference to our
+          cr3, because its in lazy mode, and it hasn't yet flushed
+          its set of pending hypercalls yet.  In this case, we can
+          look at its actual current cr3 value, and force it to flush
+          if needed. */
+       for_each_online_cpu(cpu) {
+               if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
+                       cpu_set(cpu, mask);
+       }
+
+       if (!cpus_empty(mask))
+               xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
 }
 #else
 static void drop_mm_ref(struct mm_struct *mm)
===================================================================
--- a/arch/i386/xen/xen-ops.h
+++ b/arch/i386/xen/xen-ops.h
@@ -11,6 +11,7 @@ void xen_copy_trap_info(struct trap_info
 
 DECLARE_PER_CPU(struct vcpu_info *, xen_vcpu);
 DECLARE_PER_CPU(unsigned long, xen_cr3);
+DECLARE_PER_CPU(unsigned long, xen_current_cr3);
 
 extern struct start_info *xen_start_info;
 extern struct shared_info *HYPERVISOR_shared_info;
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel