Messages in this thread | | | Subject | Re: [PATCH] printk: don't unconditionally shortcut print_time() | From | Tetsuo Handa <> | Date | Tue, 4 Dec 2018 00:00:22 +0900 |
| |
On 2018/12/03 23:14, Petr Mladek wrote: > Well, IMHO, it does not explain the pagefault above. The copy_to_user() > could come either from syslog_print() or from syslog_print_all(). They > both have their own checks that prevent the buf overflow. > > The code is tricky but it looks safe now. Is it possible that > the crash was with some old, still broken, code? Or did you > add some extra changes that you used to trigger a testing > scenario?
I think I was using unmodified current linux.git, but I failed to save the reproducer because I was editing under /tmp directory which does not create a backup.
Anyway, after this patch, no problem is occurring.
| |