Messages in this thread Patch in this message | | | From | Tetsuo Handa <> | Subject | [PATCH v2] /dev/mem: Bail out upon SIGKILL. | Date | Fri, 23 Aug 2019 13:11:19 +0900 |
| |
syzbot found that a thread can stall for minutes inside read_mem() or write_mem() after that thread was killed by SIGKILL [1]. Reading 2GB at one read() is legal, but delaying termination of killed thread for minutes is bad. Let's insert cond_resched() and SIGKILL check into iteration loop of reading/writing /dev/mem and /dev/kmem.
[ 1335.912419][T20577] read_mem: sz=4096 count=2134565632 [ 1335.943194][T20577] read_mem: sz=4096 count=2134561536 [ 1335.978280][T20577] read_mem: sz=4096 count=2134557440 [ 1336.011147][T20577] read_mem: sz=4096 count=2134553344 [ 1336.041897][T20577] read_mem: sz=4096 count=2134549248
[1] https://syzkaller.appspot.com/bug?id=a0e3436829698d5824231251fad9d8e998f94f5e
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Reported-by: syzbot <syzbot+8ab2d0f39fb79fe6ca40@syzkaller.appspotmail.com> --- drivers/char/mem.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+)
diff --git a/drivers/char/mem.c b/drivers/char/mem.c index b08dc50..cb8e653 100644 --- a/drivers/char/mem.c +++ b/drivers/char/mem.c @@ -140,6 +140,10 @@ static ssize_t read_mem(struct file *file, char __user *buf, int allowed, probe; sz = size_inside_page(p, count); + cond_resched(); + err = -EINTR; + if (fatal_signal_pending(current)) + goto failed; err = -EPERM; allowed = page_is_allowed(p >> PAGE_SHIFT); @@ -218,6 +222,9 @@ static ssize_t write_mem(struct file *file, const char __user *buf, int allowed; sz = size_inside_page(p, count); + cond_resched(); + if (fatal_signal_pending(current)) + return -EINTR; allowed = page_is_allowed(p >> PAGE_SHIFT); if (!allowed) @@ -451,6 +458,9 @@ static ssize_t read_kmem(struct file *file, char __user *buf, #endif while (low_count > 0) { sz = size_inside_page(p, low_count); + cond_resched(); + if (fatal_signal_pending(current)) + return -EINTR; /* * On ia64 if a page has been mapped somewhere as @@ -477,6 +487,11 @@ static ssize_t read_kmem(struct file *file, char __user *buf, return -ENOMEM; while (count > 0) { sz = size_inside_page(p, count); + cond_resched(); + if (fatal_signal_pending(current)) { + err = -EINTR; + break; + } if (!is_vmalloc_or_module_addr((void *)p)) { err = -ENXIO; break; @@ -523,6 +538,9 @@ static ssize_t do_write_kmem(unsigned long p, const char __user *buf, void *ptr; sz = size_inside_page(p, count); + cond_resched(); + if (fatal_signal_pending(current)) + return -EINTR; /* * On ia64 if a page has been mapped somewhere as uncached, then @@ -581,6 +599,11 @@ static ssize_t write_kmem(struct file *file, const char __user *buf, unsigned long sz = size_inside_page(p, count); unsigned long n; + cond_resched(); + if (fatal_signal_pending(current)) { + err = -EINTR; + break; + } if (!is_vmalloc_or_module_addr((void *)p)) { err = -ENXIO; break; -- 1.8.3.1
| |