RE: [PHP-CVS] com php-src: Remove arc4random: ext/standard/config.m4 ext/standard/random.c

Hi, > -----Original Message----- > From: Leigh [mailto:leigh@php.net] > Sent: Monday, October 26, 2015 9:40 PM > To: php-cvs@lists.php.net > Subject: [PHP-CVS] com php-src: Remove arc4random: ext/standard/config.m4 > ext/standard/random.c > > Commit: a53510239f629b36f1fdccc860670872463eeb77 > Author: Leigh <leigh@php.net> Mon, 26 Oct 2015 20:40:28 +0000 > Parents: fff11a1e0709fb670012b0fac4fc21cb7fc14291 > Branches: PHP-7.0 > > Link: http://git.php.net/?p=php- > src.git;a=commitdiff;h=a53510239f629b36f1fdccc860670872463eeb77 > > Log: > Remove arc4random > > There has been a lot of discussion around whether arc4random should be > included. Given how many different impementations of it are in the wild, we > can't guarantee a secure implementation on all platforms. > > Changed paths: > M ext/standard/config.m4 > M ext/standard/random.c > > > Diff: > diff --git a/ext/standard/config.m4 b/ext/standard/config.m4 index > c435f96..f41df7e 100644 > --- a/ext/standard/config.m4 > +++ b/ext/standard/config.m4 > @@ -593,11 +593,6 @@ dnl > AC_CHECK_HEADERS([atomic.h]) > > dnl > -dnl Check for arc4random on BSD systems -dnl > -AC_CHECK_DECLS([arc4random_buf]) > - > -dnl > dnl Check for getrandom on newer Linux kernels dnl > AC_CHECK_DECLS([getrandom]) > diff --git a/ext/standard/random.c b/ext/standard/random.c index > f46c4e0..fd730d1 100644 > --- a/ext/standard/random.c > +++ b/ext/standard/random.c > @@ -87,8 +87,6 @@ static int php_random_bytes(void *bytes, size_t size) > zend_throw_exception(zend_ce_exception, "Could not gather > sufficient random data", 0); > return FAILURE; > } > -#elif HAVE_DECL_ARC4RANDOM_BUF > - arc4random_buf(bytes, size); > #elif HAVE_DECL_GETRANDOM > /* Linux getrandom(2) syscall */ > size_t read_bytes = 0; > Which discussions do you mean, could you please link to them? The one I remember right now is https://github.com/php/php-src/pull/1513 , but there's actually no obvious conclusion. Removing arc4random is quite a radical move, but OpenBSD and NetBSD could still profit from not using /dev/urandom and it would cost just extending the macro condition. Thanks Anatol