From 47cb2ed9df2789fc4a3fe1201e475078f93c4839 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Thu, 6 Nov 2008 13:48:24 -0800 Subject: x86, xen: fix use of pgd_page now that it really does return a page Impact: fix 32-bit Xen guest boot crash On 32-bit PAE, pud_page, for no good reason, didn't really return a struct page *. Since Jan Beulich's fix "i386/PAE: fix pud_page()", pud_page does return a struct page *. Because PAE has 3 pagetable levels, the pud level is folded into the pgd level, so pgd_page() is the same as pud_page(), and now returns a struct page *. Update the xen/mmu.c code which uses pgd_page() accordingly. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/xen/mmu.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c index aba77b2b7d1..49697d86c6a 100644 --- a/arch/x86/xen/mmu.c +++ b/arch/x86/xen/mmu.c @@ -874,7 +874,7 @@ static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd) #else /* CONFIG_X86_32 */ #ifdef CONFIG_X86_PAE /* Need to make sure unshared kernel PMD is pinnable */ - xen_pin_page(mm, virt_to_page(pgd_page(pgd[pgd_index(TASK_SIZE)])), + xen_pin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]), PT_PMD); #endif xen_do_pin(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(pgd))); @@ -991,7 +991,7 @@ static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd) #ifdef CONFIG_X86_PAE /* Need to make sure unshared kernel PMD is unpinned */ - xen_unpin_page(mm, virt_to_page(pgd_page(pgd[pgd_index(TASK_SIZE)])), + xen_unpin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]), PT_PMD); #endif -- cgit v1.2.3 From d05fdf316067cd311d5e7add08da26ded8a58080 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 28 Oct 2008 19:23:06 +1100 Subject: xen: make sure stray alias mappings are gone before pinning Xen requires that all mappings of pagetable pages are read-only, so that they can't be updated illegally. As a result, if a page is being turned into a pagetable page, we need to make sure all its mappings are RO. If the page had been used for ioremap or vmalloc, it may still have left over mappings as a result of not having been lazily unmapped. This change makes sure we explicitly mop them all up before pinning the page. Unlike aliases created by kmap, the there can be vmalloc aliases even for non-high pages, so we must do the flush unconditionally. Signed-off-by: Jeremy Fitzhardinge Cc: Linux Memory Management List Cc: Nick Piggin Signed-off-by: Ingo Molnar --- arch/x86/xen/enlighten.c | 5 +++-- arch/x86/xen/mmu.c | 9 ++++++--- 2 files changed, 9 insertions(+), 5 deletions(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c index b61534c7a4c..5e4686d70f6 100644 --- a/arch/x86/xen/enlighten.c +++ b/arch/x86/xen/enlighten.c @@ -863,15 +863,16 @@ static void xen_alloc_ptpage(struct mm_struct *mm, unsigned long pfn, unsigned l if (PagePinned(virt_to_page(mm->pgd))) { SetPagePinned(page); + vm_unmap_aliases(); if (!PageHighMem(page)) { make_lowmem_page_readonly(__va(PFN_PHYS((unsigned long)pfn))); if (level == PT_PTE && USE_SPLIT_PTLOCKS) pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn); - } else + } else { /* make sure there are no stray mappings of this page */ kmap_flush_unused(); - vm_unmap_aliases(); + } } } diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c index aba77b2b7d1..89f3b6edc65 100644 --- a/arch/x86/xen/mmu.c +++ b/arch/x86/xen/mmu.c @@ -850,13 +850,16 @@ static int xen_pin_page(struct mm_struct *mm, struct page *page, read-only, and can be pinned. */ static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd) { + vm_unmap_aliases(); + xen_mc_batch(); - if (xen_pgd_walk(mm, xen_pin_page, USER_LIMIT)) { - /* re-enable interrupts for kmap_flush_unused */ + if (xen_pgd_walk(mm, xen_pin_page, USER_LIMIT)) { + /* re-enable interrupts for flushing */ xen_mc_issue(0); + kmap_flush_unused(); - vm_unmap_aliases(); + xen_mc_batch(); } -- cgit v1.2.3 From 86bbc2c235e500957b213e7e64ce2e0ccb8bc131 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Fri, 21 Nov 2008 10:21:33 +0000 Subject: xen: pin correct PGD on suspend Impact: fix Xen guest boot failure commit eefb47f6a1e855653d275cb90592a3587ea93a09 ("xen: use spin_lock_nest_lock when pinning a pagetable") changed xen_pgd_walk to walk over mm->pgd rather than taking pgd as an argument. This breaks xen_mm_(un)pin_all() because it makes init_mm.pgd readonly instead of the pgd we are interested in and therefore the pin subsequently fails. (XEN) mm.c:2280:d15 Bad type (saw 00000000e8000001 != exp 0000000060000000) for mfn bc464 (pfn 21ca7) (XEN) mm.c:2665:d15 Error while pinning mfn bc464 [ 14.586913] 1 multicall(s) failed: cpu 0 [ 14.586926] Pid: 14, comm: kstop/0 Not tainted 2.6.28-rc5-x86_32p-xenU-00172-gee2f6cc #200 [ 14.586940] Call Trace: [ 14.586955] [] ? printk+0x18/0x1e [ 14.586972] [] xen_mc_flush+0x163/0x1d0 [ 14.586986] [] __xen_pgd_pin+0xa1/0x110 [ 14.587000] [] ? stop_cpu+0x0/0xf0 [ 14.587015] [] xen_mm_pin_all+0x4b/0x70 [ 14.587029] [] xen_suspend+0x39/0xe0 [ 14.587042] [] ? stop_cpu+0x0/0xf0 [ 14.587054] [] stop_cpu+0x9d/0xf0 [ 14.587067] [] run_workqueue+0x8d/0x150 [ 14.587080] [] ? _spin_unlock_irqrestore+0x23/0x40 [ 14.587094] [] ? prepare_to_wait+0x3a/0x70 [ 14.587107] [] worker_thread+0x88/0xf0 [ 14.587120] [] ? autoremove_wake_function+0x0/0x50 [ 14.587133] [] ? worker_thread+0x0/0xf0 [ 14.587146] [] kthread+0x3c/0x70 [ 14.587157] [] ? kthread+0x0/0x70 [ 14.587170] [] kernel_thread_helper+0x7/0x10 [ 14.587181] call 1/3: op=14 arg=[c0415000] result=0 [ 14.587192] call 2/3: op=14 arg=[e1ca2000] result=0 [ 14.587204] call 3/3: op=26 arg=[c1808860] result=-22 Signed-off-by: Ian Campbell Acked-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/xen/mmu.c | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c index 688936044dc..636ef4caa52 100644 --- a/arch/x86/xen/mmu.c +++ b/arch/x86/xen/mmu.c @@ -661,12 +661,11 @@ void xen_set_pgd(pgd_t *ptr, pgd_t val) * For 64-bit, we must skip the Xen hole in the middle of the address * space, just after the big x86-64 virtual hole. */ -static int xen_pgd_walk(struct mm_struct *mm, - int (*func)(struct mm_struct *mm, struct page *, - enum pt_level), - unsigned long limit) +static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd, + int (*func)(struct mm_struct *mm, struct page *, + enum pt_level), + unsigned long limit) { - pgd_t *pgd = mm->pgd; int flush = 0; unsigned hole_low, hole_high; unsigned pgdidx_limit, pudidx_limit, pmdidx_limit; @@ -753,6 +752,14 @@ out: return flush; } +static int xen_pgd_walk(struct mm_struct *mm, + int (*func)(struct mm_struct *mm, struct page *, + enum pt_level), + unsigned long limit) +{ + return __xen_pgd_walk(mm, mm->pgd, func, limit); +} + /* If we're using split pte locks, then take the page's lock and return a pointer to it. Otherwise return NULL. */ static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm) @@ -854,7 +861,7 @@ static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd) xen_mc_batch(); - if (xen_pgd_walk(mm, xen_pin_page, USER_LIMIT)) { + if (__xen_pgd_walk(mm, pgd, xen_pin_page, USER_LIMIT)) { /* re-enable interrupts for flushing */ xen_mc_issue(0); @@ -998,7 +1005,7 @@ static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd) PT_PMD); #endif - xen_pgd_walk(mm, xen_unpin_page, USER_LIMIT); + __xen_pgd_walk(mm, pgd, xen_unpin_page, USER_LIMIT); xen_mc_issue(0); } -- cgit v1.2.3 From 37af46efa5413c6f4c25d9a24b4c43f2cc718eed Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 22 Nov 2008 17:37:54 +0000 Subject: xen_setup_vcpu_info_placement() is not init on x86 ... so get xen-ops.h in agreement with xen/smp.c Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- arch/x86/xen/xen-ops.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h index d7422dc2a55..9e1afae8461 100644 --- a/arch/x86/xen/xen-ops.h +++ b/arch/x86/xen/xen-ops.h @@ -49,7 +49,7 @@ bool xen_vcpu_stolen(int vcpu); void xen_mark_init_mm_pinned(void); -void __init xen_setup_vcpu_info_placement(void); +void xen_setup_vcpu_info_placement(void); #ifdef CONFIG_SMP void xen_smp_init(void); -- cgit v1.2.3 From df6b07949b6cab9d119363d02ef63379160f6c82 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 22 Nov 2008 17:38:04 +0000 Subject: xen_play_dead() is __cpuinit Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- arch/x86/xen/smp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/xen') diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index d77da613b1d..acd9b6705e0 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -362,7 +362,7 @@ static void xen_cpu_die(unsigned int cpu) alternatives_smp_switch(0); } -static void xen_play_dead(void) +static void __cpuinit xen_play_dead(void) /* used only with CPU_HOTPLUG */ { play_dead_common(); HYPERVISOR_vcpu_op(VCPUOP_down, smp_processor_id(), NULL); -- cgit v1.2.3