From 1cefe28f95da307287a05a6a0c10213f01055e2a Mon Sep 17 00:00:00 2001 From: Kautuk Consul Date: Thu, 31 May 2012 16:26:03 -0700 Subject: um/kernel/trap.c: port OOM changes to handle_page_fault() Commit d065bd810b6d ("mm: retry page fault when blocking on disk transfer") and commit 37b23e0525d3 ("x86,mm: make pagefault killable") introduced changes into the x86 pagefault handler for making the page fault handler retryable as well as killable. These changes reduce the mmap_sem hold time, which is crucial during OOM killer invocation. Port these changes to um. Signed-off-by: Kautuk Consul Cc: Jeff Dike Cc: Richard Weinberger Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/um/kernel/trap.c | 24 +++++++++++++++++++----- 1 file changed, 19 insertions(+), 5 deletions(-) (limited to 'arch/um/kernel') diff --git a/arch/um/kernel/trap.c b/arch/um/kernel/trap.c index dafc94715950..3be60765c0e2 100644 --- a/arch/um/kernel/trap.c +++ b/arch/um/kernel/trap.c @@ -30,6 +30,8 @@ int handle_page_fault(unsigned long address, unsigned long ip, pmd_t *pmd; pte_t *pte; int err = -EFAULT; + unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE | + (is_write ? FAULT_FLAG_WRITE : 0); *code_out = SEGV_MAPERR; @@ -40,6 +42,7 @@ int handle_page_fault(unsigned long address, unsigned long ip, if (in_atomic()) goto out_nosemaphore; +retry: down_read(&mm->mmap_sem); vma = find_vma(mm, address); if (!vma) @@ -65,7 +68,11 @@ good_area: do { int fault; - fault = handle_mm_fault(mm, vma, address, is_write ? FAULT_FLAG_WRITE : 0); + fault = handle_mm_fault(mm, vma, address, flags); + + if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current)) + goto out_nosemaphore; + if (unlikely(fault & VM_FAULT_ERROR)) { if (fault & VM_FAULT_OOM) { goto out_of_memory; @@ -75,10 +82,17 @@ good_area: } BUG(); } - if (fault & VM_FAULT_MAJOR) - current->maj_flt++; - else - current->min_flt++; + if (flags & FAULT_FLAG_ALLOW_RETRY) { + if (fault & VM_FAULT_MAJOR) + current->maj_flt++; + else + current->min_flt++; + if (fault & VM_FAULT_RETRY) { + flags &= ~FAULT_FLAG_ALLOW_RETRY; + + goto retry; + } + } pgd = pgd_offset(mm, address); pud = pud_offset(pgd, address); -- cgit v1.2.3 From 9bd0a077122a6b15069aab7a59c8f50f736fa1dd Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Thu, 31 May 2012 16:26:25 -0700 Subject: um: should hold tasklist_lock while traversing processes Traversing the tasks requires holding tasklist_lock, otherwise it is unsafe. p.s. However, I'm not sure that calling os_kill_ptraced_process() in the atomic context is correct. It seem to work, but please take a closer look. Signed-off-by: Anton Vorontsov Cc: Richard Weinberger Cc: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/um/kernel/reboot.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/um/kernel') diff --git a/arch/um/kernel/reboot.c b/arch/um/kernel/reboot.c index 4d93dff6b371..66d754c46e7b 100644 --- a/arch/um/kernel/reboot.c +++ b/arch/um/kernel/reboot.c @@ -4,6 +4,7 @@ */ #include "linux/sched.h" +#include "linux/spinlock.h" #include "linux/slab.h" #include "kern_util.h" #include "os.h" @@ -22,6 +23,7 @@ static void kill_off_processes(void) struct task_struct *p; int pid; + read_lock(&tasklist_lock); for_each_process(p) { if (p->mm == NULL) continue; @@ -29,6 +31,7 @@ static void kill_off_processes(void) pid = p->mm->context.id.u.pid; os_kill_ptraced_process(pid, 1); } + read_unlock(&tasklist_lock); } } -- cgit v1.2.3 From 137d1a26c842a384c98dd2a6a605014fddf5c777 Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Thu, 31 May 2012 16:26:25 -0700 Subject: um: fix possible race on task->mm Checking for task->mm is dangerous as ->mm might disappear (exit_mm() assigns NULL under task_lock(), so tasklist lock is not enough). We can't use get_task_mm()/mmput() pair as mmput() might sleep, so let's take the task lock while we care about its mm. Note that we should also use find_lock_task_mm() to check all process' threads for a valid mm, but for uml we'll do it in a separate patch. Signed-off-by: Anton Vorontsov Cc: Richard Weinberger Cc: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/um/kernel/reboot.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'arch/um/kernel') diff --git a/arch/um/kernel/reboot.c b/arch/um/kernel/reboot.c index 66d754c46e7b..1411f4e79f9a 100644 --- a/arch/um/kernel/reboot.c +++ b/arch/um/kernel/reboot.c @@ -25,10 +25,13 @@ static void kill_off_processes(void) read_lock(&tasklist_lock); for_each_process(p) { - if (p->mm == NULL) + task_lock(p); + if (!p->mm) { + task_unlock(p); continue; - + } pid = p->mm->context.id.u.pid; + task_unlock(p); os_kill_ptraced_process(pid, 1); } read_unlock(&tasklist_lock); -- cgit v1.2.3 From 2c922c51e6924298f8271d75732964f1ffdecaf2 Mon Sep 17 00:00:00 2001 From: Anton Vorontsov Date: Thu, 31 May 2012 16:26:26 -0700 Subject: um: properly check all process' threads for a live mm kill_off_processes() might miss a valid process, this is because checking for process->mm is not enough. Process' main thread may exit or detach its mm via use_mm(), but other threads may still have a valid mm. To catch this we use find_lock_task_mm(), which walks up all threads and returns an appropriate task (with task lock held). Suggested-by: Oleg Nesterov Signed-off-by: Anton Vorontsov Cc: Richard Weinberger Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/um/kernel/reboot.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'arch/um/kernel') diff --git a/arch/um/kernel/reboot.c b/arch/um/kernel/reboot.c index 1411f4e79f9a..3d15243ce692 100644 --- a/arch/um/kernel/reboot.c +++ b/arch/um/kernel/reboot.c @@ -6,6 +6,7 @@ #include "linux/sched.h" #include "linux/spinlock.h" #include "linux/slab.h" +#include "linux/oom.h" #include "kern_util.h" #include "os.h" #include "skas.h" @@ -25,13 +26,13 @@ static void kill_off_processes(void) read_lock(&tasklist_lock); for_each_process(p) { - task_lock(p); - if (!p->mm) { - task_unlock(p); + struct task_struct *t; + + t = find_lock_task_mm(p); + if (!t) continue; - } - pid = p->mm->context.id.u.pid; - task_unlock(p); + pid = t->mm->context.id.u.pid; + task_unlock(t); os_kill_ptraced_process(pid, 1); } read_unlock(&tasklist_lock); -- cgit v1.2.3