From dacd45f4e793e46e8299c9a580e400866ffe0770 Mon Sep 17 00:00:00 2001 From: Konrad Rzeszutek Wilk Date: Mon, 22 Oct 2012 11:35:16 -0400 Subject: xen/smp: Move the common CPU init code a bit to prep for PVH patch. The PV and PVH code CPU init code share some functionality. The PVH code ("xen/pvh: Extend vcpu_guest_context, p2m, event, and XenBus") sets some of these up, but not all. To make it easier to read, this patch removes the PV specific out of the generic way. No functional change - just code movement. Acked-by: Stefano Stabellini [v2: Fixed compile errors noticed by Fengguang Wu build system] Signed-off-by: Konrad Rzeszutek Wilk --- arch/x86/xen/smp.c | 42 +++++++++++++++++++++++------------------- 1 file changed, 23 insertions(+), 19 deletions(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index 34bc4ce..09ea61d 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -300,8 +300,6 @@ cpu_initialize_context(unsigned int cpu, struct task_struct *idle) gdt = get_cpu_gdt_table(cpu); ctxt->flags = VGCF_IN_KERNEL; - ctxt->user_regs.ds = __USER_DS; - ctxt->user_regs.es = __USER_DS; ctxt->user_regs.ss = __KERNEL_DS; #ifdef CONFIG_X86_32 ctxt->user_regs.fs = __KERNEL_PERCPU; @@ -310,35 +308,41 @@ cpu_initialize_context(unsigned int cpu, struct task_struct *idle) ctxt->gs_base_kernel = per_cpu_offset(cpu); #endif ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle; - ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */ memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt)); - xen_copy_trap_info(ctxt->trap_ctxt); + { + ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */ + ctxt->user_regs.ds = __USER_DS; + ctxt->user_regs.es = __USER_DS; - ctxt->ldt_ents = 0; + xen_copy_trap_info(ctxt->trap_ctxt); - BUG_ON((unsigned long)gdt & ~PAGE_MASK); + ctxt->ldt_ents = 0; - gdt_mfn = arbitrary_virt_to_mfn(gdt); - make_lowmem_page_readonly(gdt); - make_lowmem_page_readonly(mfn_to_virt(gdt_mfn)); + BUG_ON((unsigned long)gdt & ~PAGE_MASK); - ctxt->gdt_frames[0] = gdt_mfn; - ctxt->gdt_ents = GDT_ENTRIES; + gdt_mfn = arbitrary_virt_to_mfn(gdt); + make_lowmem_page_readonly(gdt); + make_lowmem_page_readonly(mfn_to_virt(gdt_mfn)); - ctxt->user_regs.cs = __KERNEL_CS; - ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs); + ctxt->gdt_frames[0] = gdt_mfn; + ctxt->gdt_ents = GDT_ENTRIES; - ctxt->kernel_ss = __KERNEL_DS; - ctxt->kernel_sp = idle->thread.sp0; + ctxt->kernel_ss = __KERNEL_DS; + ctxt->kernel_sp = idle->thread.sp0; #ifdef CONFIG_X86_32 - ctxt->event_callback_cs = __KERNEL_CS; - ctxt->failsafe_callback_cs = __KERNEL_CS; + ctxt->event_callback_cs = __KERNEL_CS; + ctxt->failsafe_callback_cs = __KERNEL_CS; #endif - ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback; - ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback; + ctxt->event_callback_eip = + (unsigned long)xen_hypervisor_callback; + ctxt->failsafe_callback_eip = + (unsigned long)xen_failsafe_callback; + } + ctxt->user_regs.cs = __KERNEL_CS; + ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs); per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir); ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir)); -- cgit v1.1 From 76eaca031f0af2bb303e405986f637811956a422 Mon Sep 17 00:00:00 2001 From: Stefan Bader Date: Fri, 15 Feb 2013 09:48:52 +0100 Subject: xen: Send spinlock IPI to all waiters There is a loophole between Xen's current implementation of pv-spinlocks and the scheduler. This was triggerable through a testcase until v3.6 changed the TLB flushing code. The problem potentially is still there just not observable in the same way. What could happen was (is): 1. CPU n tries to schedule task x away and goes into a slow wait for the runq lock of CPU n-# (must be one with a lower number). 2. CPU n-#, while processing softirqs, tries to balance domains and goes into a slow wait for its own runq lock (for updating some records). Since this is a spin_lock_irqsave in softirq context, interrupts will be re-enabled for the duration of the poll_irq hypercall used by Xen. 3. Before the runq lock of CPU n-# is unlocked, CPU n-1 receives an interrupt (e.g. endio) and when processing the interrupt, tries to wake up task x. But that is in schedule and still on_cpu, so try_to_wake_up goes into a tight loop. 4. The runq lock of CPU n-# gets unlocked, but the message only gets sent to the first waiter, which is CPU n-# and that is busily stuck. 5. CPU n-# never returns from the nested interruption to take and release the lock because the scheduler uses a busy wait. And CPU n never finishes the task migration because the unlock notification only went to CPU n-#. To avoid this and since the unlocking code has no real sense of which waiter is best suited to grab the lock, just send the IPI to all of them. This causes the waiters to return from the hyper- call (those not interrupted at least) and do active spinlocking. BugLink: http://bugs.launchpad.net/bugs/1011792 Acked-by: Jan Beulich Signed-off-by: Stefan Bader Cc: stable@vger.kernel.org Signed-off-by: Konrad Rzeszutek Wilk --- arch/x86/xen/spinlock.c | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/spinlock.c b/arch/x86/xen/spinlock.c index 83e866d..f7a080e 100644 --- a/arch/x86/xen/spinlock.c +++ b/arch/x86/xen/spinlock.c @@ -328,7 +328,6 @@ static noinline void xen_spin_unlock_slow(struct xen_spinlock *xl) if (per_cpu(lock_spinners, cpu) == xl) { ADD_STATS(released_slow_kicked, 1); xen_send_IPI_one(cpu, XEN_SPIN_UNLOCK_VECTOR); - break; } } } -- cgit v1.1