lkml.org 
[lkml]   [2007]   [May]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[patch 11/28]xen: fix multicall batching
Disable interrupts between allocating a multicall entry and actually
issuing it, to prevent an interrupt from coming in, allocating and
initializing further multicall entries, and then issuing them all,
including the partially completed one.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Acked-by: Chris Wright <chrisw@sous-sol.org>

---
arch/i386/xen/enlighten.c | 44 +++++++++++++++++++++++++++-----------------
arch/i386/xen/mmu.c | 18 ++++++++++++------
arch/i386/xen/multicalls.c | 9 ++++++++-
arch/i386/xen/multicalls.h | 27 +++++++++++++++++++++++----
arch/i386/xen/xen-ops.h | 5 +----
5 files changed, 71 insertions(+), 32 deletions(-)
===================================================================
--- a/arch/i386/xen/enlighten.c
+++ b/arch/i386/xen/enlighten.c
@@ -187,13 +187,25 @@ static void xen_halt(void)

static void xen_set_lazy_mode(enum paravirt_lazy_mode mode)
{
- enum paravirt_lazy_mode *lazy = &get_cpu_var(xen_lazy_mode);
+ switch(mode) {
+ case PARAVIRT_LAZY_NONE:
+ BUG_ON(x86_read_percpu(xen_lazy_mode) == PARAVIRT_LAZY_NONE);
+ break;
+
+ case PARAVIRT_LAZY_MMU:
+ case PARAVIRT_LAZY_CPU:
+ BUG_ON(x86_read_percpu(xen_lazy_mode) != PARAVIRT_LAZY_NONE);
+ break;
+
+ case PARAVIRT_LAZY_FLUSH:
+ /* flush if necessary, but don't change state */
+ if (x86_read_percpu(xen_lazy_mode) != PARAVIRT_LAZY_NONE)
+ xen_mc_flush();
+ return;
+ }

xen_mc_flush();
-
- *lazy = mode;
-
- put_cpu_var(xen_lazy_mode);
+ x86_write_percpu(xen_lazy_mode, mode);
}

static unsigned long xen_store_tr(void)
@@ -220,7 +232,7 @@ static void xen_set_ldt(const void *addr

MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

- xen_mc_issue();
+ xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void xen_load_gdt(const struct Xgt_desc_struct *dtr)
@@ -248,7 +260,7 @@ static void xen_load_gdt(const struct Xg

MULTI_set_gdt(mcs.mc, frames, size / sizeof(struct desc_struct));

- xen_mc_issue();
+ xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void load_TLS_descriptor(struct thread_struct *t,
@@ -256,18 +268,20 @@ static void load_TLS_descriptor(struct t
{
struct desc_struct *gdt = get_cpu_gdt_table(cpu);
xmaddr_t maddr = virt_to_machine(&gdt[GDT_ENTRY_TLS_MIN+i]);
- struct multicall_space mc = xen_mc_entry(0);
+ struct multicall_space mc = __xen_mc_entry(0);

MULTI_update_descriptor(mc.mc, maddr.maddr, t->tls_array[i]);
}

static void xen_load_tls(struct thread_struct *t, unsigned int cpu)
{
+ xen_mc_batch();
+
load_TLS_descriptor(t, cpu, 0);
load_TLS_descriptor(t, cpu, 1);
load_TLS_descriptor(t, cpu, 2);

- xen_mc_issue();
+ xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void xen_write_ldt_entry(struct desc_struct *dt, int entrynum, u32 low, u32 high)
@@ -387,13 +401,9 @@ static void xen_load_esp0(struct tss_str
static void xen_load_esp0(struct tss_struct *tss,
struct thread_struct *thread)
{
- if (xen_get_lazy_mode() != PARAVIRT_LAZY_CPU) {
- if (HYPERVISOR_stack_switch(__KERNEL_DS, thread->esp0))
- BUG();
- } else {
- struct multicall_space mcs = xen_mc_entry(0);
- MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->esp0);
- }
+ struct multicall_space mcs = xen_mc_entry(0);
+ MULTI_stack_switch(mcs.mc, __KERNEL_DS, thread->esp0);
+ xen_mc_issue(PARAVIRT_LAZY_CPU);
}

static void xen_set_iopl_mask(unsigned mask)
@@ -485,7 +495,7 @@ static void xen_write_cr3(unsigned long

MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

- xen_mc_issue();
+ xen_mc_issue(PARAVIRT_LAZY_CPU);
}
}

===================================================================
--- a/arch/i386/xen/mmu.c
+++ b/arch/i386/xen/mmu.c
@@ -366,7 +366,7 @@ static int pin_page(struct page *page, u
else {
void *pt = lowmem_page_address(page);
unsigned long pfn = page_to_pfn(page);
- struct multicall_space mcs = xen_mc_entry(0);
+ struct multicall_space mcs = __xen_mc_entry(0);

flush = 0;

@@ -386,10 +386,12 @@ void xen_pgd_pin(pgd_t *pgd)
struct multicall_space mcs;
struct mmuext_op *op;

+ xen_mc_batch();
+
if (pgd_walk(pgd, pin_page, TASK_SIZE))
kmap_flush_unused();

- mcs = xen_mc_entry(sizeof(*op));
+ mcs = __xen_mc_entry(sizeof(*op));
op = mcs.args;

#ifdef CONFIG_X86_PAE
@@ -400,7 +402,7 @@ void xen_pgd_pin(pgd_t *pgd)
op->arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(pgd)));
MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);

- xen_mc_flush();
+ xen_mc_issue(0);
}

/* The init_mm pagetable is really pinned as soon as its created, but
@@ -424,7 +426,7 @@ static int unpin_page(struct page *page,
if (pgfl && !PageHighMem(page)) {
void *pt = lowmem_page_address(page);
unsigned long pfn = page_to_pfn(page);
- struct multicall_space mcs = xen_mc_entry(0);
+ struct multicall_space mcs = __xen_mc_entry(0);

MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
pfn_pte(pfn, PAGE_KERNEL),
@@ -438,7 +440,11 @@ static void xen_pgd_unpin(pgd_t *pgd)
static void xen_pgd_unpin(pgd_t *pgd)
{
struct mmuext_op *op;
- struct multicall_space mcs = xen_mc_entry(sizeof(*op));
+ struct multicall_space mcs;
+
+ xen_mc_batch();
+
+ mcs = __xen_mc_entry(sizeof(*op));

op = mcs.args;
op->cmd = MMUEXT_UNPIN_TABLE;
@@ -448,7 +454,7 @@ static void xen_pgd_unpin(pgd_t *pgd)

pgd_walk(pgd, unpin_page, TASK_SIZE);

- xen_mc_flush();
+ xen_mc_issue(0);
}


===================================================================
--- a/arch/i386/xen/multicalls.c
+++ b/arch/i386/xen/multicalls.c
@@ -35,11 +35,17 @@ struct mc_buffer {
};

static DEFINE_PER_CPU(struct mc_buffer, mc_buffer);
+DEFINE_PER_CPU(unsigned long, xen_mc_irq_flags);

void xen_mc_flush(void)
{
struct mc_buffer *b = &get_cpu_var(mc_buffer);
int ret = 0;
+ unsigned long flags;
+
+ /* Disable interrupts in case someone comes in and queues
+ something in the middle */
+ local_irq_save(flags);

if (b->mcidx) {
int i;
@@ -55,11 +61,12 @@ void xen_mc_flush(void)
BUG_ON(b->argidx != 0);

put_cpu_var(mc_buffer);
+ local_irq_restore(flags);

BUG_ON(ret);
}

-struct multicall_space xen_mc_entry(size_t args)
+struct multicall_space __xen_mc_entry(size_t args)
{
struct mc_buffer *b = &get_cpu_var(mc_buffer);
struct multicall_space ret;
===================================================================
--- a/arch/i386/xen/multicalls.h
+++ b/arch/i386/xen/multicalls.h
@@ -11,16 +11,35 @@ struct multicall_space
};

/* Allocate room for a multicall and its args */
-struct multicall_space xen_mc_entry(size_t args);
+struct multicall_space __xen_mc_entry(size_t args);
+
+DECLARE_PER_CPU(unsigned long, xen_mc_irq_flags);
+
+/* Call to start a batch of multiple __xen_mc_entry()s. Must be
+ paired with xen_mc_issue() */
+static inline void xen_mc_batch(void)
+{
+ /* need to disable interrupts until this entry is complete */
+ local_irq_save(__get_cpu_var(xen_mc_irq_flags));
+}
+
+static inline struct multicall_space xen_mc_entry(size_t args)
+{
+ xen_mc_batch();
+ return __xen_mc_entry(args);
+}

/* Flush all pending multicalls */
void xen_mc_flush(void);

-/* Issue a multicall if we're not in lazy mode */
-static inline void xen_mc_issue(void)
+/* Issue a multicall if we're not in a lazy mode */
+static inline void xen_mc_issue(unsigned mode)
{
- if (xen_get_lazy_mode() == PARAVIRT_LAZY_NONE)
+ if ((xen_get_lazy_mode() & mode) == 0)
xen_mc_flush();
+
+ /* restore flags saved in xen_mc_batch */
+ local_irq_restore(x86_read_percpu(xen_mc_irq_flags));
}

#endif /* _XEN_MULTICALLS_H */
===================================================================
--- a/arch/i386/xen/xen-ops.h
+++ b/arch/i386/xen/xen-ops.h
@@ -26,10 +26,7 @@ DECLARE_PER_CPU(enum paravirt_lazy_mode,

static inline unsigned xen_get_lazy_mode(void)
{
- unsigned ret = get_cpu_var(xen_lazy_mode);
- put_cpu_var(xen_lazy_mode);
-
- return ret;
+ return x86_read_percpu(xen_lazy_mode);
}


--

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2007-05-11 21:13    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site