From: <ch...@su...> - 2014-04-30 13:14:18
|
Hi! > > I do recall a testcase, which had special setup for ia64, because > > it didn't work as expected for NULL. When looking at process map, > > there was a single page mapped 0, which didn't come from test: > > > > # cat /proc/self/maps > > 00000000-00004000 r--p 00000000 00:00 0 > > > > See this email thread: > > https://www.mail-archive.com/ltp...@li.../msg17855.html > > Hmm, so I guess that only portable way to cause EFAULT would be mapping > PROT_NONE page. And we can parse /proc/self/maps and figure out if NULL is mapped and try more than one address, however I haven't had figured out a reasonably simple way how to change the current testcases to cope with something between 1 and 3 tests for EFAULT yet. -- Cyril Hrubis ch...@su... |