Everything
php
Learn Community Releases The PHP Project
Home » Project » Newsgroups » php.internals » [RFC] Platform and extension-independent API to the system CSPRNG »

[RFC] Platform and extension-independent API to the system CSPRNG

php.internals

Tom Worster

14 years ago
Hi, I added the new RFC https://wiki.php.net/rfc/csrandombytes which is in its first draft. I attempted to be fair with respect to the discussion that has already taken place here. How does this get listed on the https://wiki.php.net/rfc page? Tom

Ferenc Kovacs

14 years ago
On Mon, Jan 9, 2012 at 1:42 AM, Tom Worster <fsb@thefsb.org> wrote:
> Hi, > > I added the new RFC https://wiki.php.net/rfc/csrandombytes which is in its > first draft. > > > I attempted to be fair with respect to the discussion that has already > taken place here. > > How does this get listed on the https://wiki.php.net/rfc page? > > Tom > >
I added it there under the In Draft category, if you have a better name or short description, just tell me, and I will update it there.
-- Ferenc Kovács @Tyr43l - http://tyrael.hu

Tom Worster

14 years ago
I have also set up a github repo with 4 files in it. It is a first hack of a function that does part of what I described in the RFC. It's based on the interface of openssl_random_pseudo_bytes() and the guts of mcrypt_create_iv(). It is provisionally named cs_random_bytes(). For now it builds and works at least this much: $ sapi/cli/php -r 'echo bin2hex(cs_random_bytes(8)) . PHP_EOL;' 4cd0965922470560 The hard work will be implementing the $is_strong_result flag in a platform independent way. You need to read the status of the entropy pool. The current code does that for Linux (maybe?). On FreeBSD you use sysctl(3) to read kern.random.sys.seeded. On OS X you ask securityd. Windows is actually easier. And what about other OSs? What is PHP normally tested on and would that be a suitable guide for cs_random_bytes()? In any case, I am no C programmer. I'm just a web dev. I don't even know how to ask if sysctl(3) is present. Tom On 1/8/12 7:42 PM, "Tom Worster" <fsb@thefsb.org> wrote:

Tom Worster

14 years ago
I forgot the URL: https://github.com/tom--/php-cs_random_bytesemo :X tom On 1/8/12 8:56 PM, "Tom Worster" <fsb@thefsb.org> wrote:

Pierre Joye

14 years ago
pls add it to the RFC, the right one as this one is a 404. On Mon, Jan 9, 2012 at 2:58 AM, Tom Worster <fsb@thefsb.org> wrote:
> I forgot the URL: https://github.com/tom--/php-cs_random_bytesemo > > :X > > tom > > On 1/8/12 8:56 PM, "Tom Worster" <fsb@thefsb.org> wrote: > >>I have also set up a github repo with 4 files in it. It is a first hack >>of >>a function that does part of what I described in the RFC. It's based on >>the interface of openssl_random_pseudo_bytes() and the guts of >>mcrypt_create_iv(). It is provisionally named cs_random_bytes(). >> >>For now it builds and works at least this much: >> >>$ sapi/cli/php -r 'echo bin2hex(cs_random_bytes(8)) . PHP_EOL;' >>4cd0965922470560 >> >> >>The hard work will be implementing the $is_strong_result flag in a >>platform independent way. You need to read the status of the entropy >>pool. >>The current code does that for Linux (maybe?). On FreeBSD you use >>sysctl(3) to read kern.random.sys.seeded. On OS X you ask securityd. >>Windows is actually easier. >> >>And what about other OSs? What is PHP normally tested on and would that >>be >>a suitable guide for cs_random_bytes()? >> >>In any case, I am no C programmer. I'm just a web dev. I don't even know >>how to ask if sysctl(3) is present. >> >> >>Tom >> >> >>On 1/8/12 7:42 PM, "Tom Worster" <fsb@thefsb.org> wrote: >> >>>I added the new RFC https://wiki.php.net/rfc/csrandombytes which is in >>>its >>>first draft. > > > > -- > PHP Internals - PHP Runtime Development Mailing List > To unsubscribe, visit: http://www.php.net/unsub.php >
-- Pierre @pierrejoye | http://blog.thepimp.net | http://www.libgd.org

Tom Worster

14 years ago
I added it at the top, I hope that's ok. the URL is https://github.com/tom--/php-cs_random_bytes Thanks for pointing that out Tom On 1/9/12 7:54 AM, "Pierre Joye" <pierre.php@gmail.com> wrote:

Pierre Joye

14 years ago
hi, Some comments, same as I did before: cs means crypto safe. In your implementation, it is by no mean crypto safe as it relies on urandom for the cases where /dev/random did not return enough data (some OS are non blocking here), while linux is. It should detect which entropy sources are available. Be /dev/random or other like entropy daemon or similar systems providing a device. Doing so will make this function works fine on almost all systems. Cheers, On Mon, Jan 9, 2012 at 2:00 PM, Tom Worster <fsb@thefsb.org> wrote:
> I added it at the top, I hope that's ok. > > the URL is https://github.com/tom--/php-cs_random_bytes > > Thanks for pointing that out > Tom > > On 1/9/12 7:54 AM, "Pierre Joye" <pierre.php@gmail.com> wrote: > >>pls add it to the RFC, the right one as this one is a 404. >> >>On Mon, Jan 9, 2012 at 2:58 AM, Tom Worster <fsb@thefsb.org> wrote: >>> I forgot the URL: https://github.com/tom--/php-cs_random_bytesemo >>> >>> :X >>> >>> tom >>> >>> On 1/8/12 8:56 PM, "Tom Worster" <fsb@thefsb.org> wrote: >>> >>>>I have also set up a github repo with 4 files in it. It is a first hack >>>>of >>>>a function that does part of what I described in the RFC. It's based on >>>>the interface of openssl_random_pseudo_bytes() and the guts of >>>>mcrypt_create_iv(). It is provisionally named cs_random_bytes(). >>>> >>>>For now it builds and works at least this much: >>>> >>>>$ sapi/cli/php -r 'echo bin2hex(cs_random_bytes(8)) . PHP_EOL;' >>>>4cd0965922470560 >>>> >>>> >>>>The hard work will be implementing the $is_strong_result flag in a >>>>platform independent way. You need to read the status of the entropy >>>>pool. >>>>The current code does that for Linux (maybe?). On FreeBSD you use >>>>sysctl(3) to read kern.random.sys.seeded. On OS X you ask securityd. >>>>Windows is actually easier. >>>> >>>>And what about other OSs? What is PHP normally tested on and would that >>>>be >>>>a suitable guide for cs_random_bytes()? >>>> >>>>In any case, I am no C programmer. I'm just a web dev. I don't even know >>>>how to ask if sysctl(3) is present. >>>> >>>> >>>>Tom >>>> >>>> >>>>On 1/8/12 7:42 PM, "Tom Worster" <fsb@thefsb.org> wrote: >>>> >>>>>I added the new RFC https://wiki.php.net/rfc/csrandombytes which is in >>>>>its >>>>>first draft. >>> >>> >>> >>> -- >>> PHP Internals - PHP Runtime Development Mailing List >>> To unsubscribe, visit: http://www.php.net/unsub.php >>> >> >> >> >>-- >>Pierre >> >>@pierrejoye | http://blog.thepimp.net | http://www.libgd.org > >
-- Pierre @pierrejoye | http://blog.thepimp.net | http://www.libgd.org

Tom Worster

14 years ago
Pierre, 1. The proposal in the RFC is explicit: "The function should neither block nor return a failure status in the case that the systems entropy pool is depleted. However, it should allow the caller to discover if this is the case. etc..." 2. The current code implements that by setting the $is_strong_result flag if it has to resort to urandom on Linux. So, afaikt, what has been done _so_far_ is in alignment with your preferences. Correct? As I said in the email discussing implementation, setting $is_strong_result requires different code for each OS and this has not been implemented yet. Tom On 1/9/12 9:33 AM, "Pierre Joye" <pierre.php@gmail.com> wrote:

Thomas Hruska

14 years ago
On 1/9/2012 7:50 AM, Tom Worster wrote:
> Pierre, > > 1. The proposal in the RFC is explicit: "The function should neither block > nor return a failure status in the case that the systems entropy pool is > depleted. However, it should allow the caller to discover if this is the > case. etc..." > > 2. The current code implements that by setting the $is_strong_result flag > if it has to resort to urandom on Linux. > > So, afaikt, what has been done _so_far_ is in alignment with your > preferences. Correct? > > As I said in the email discussing implementation, setting > $is_strong_result requires different code for each OS and this has not > been implemented yet. > > Tom
I'm glad someone is doing something about making more sources of randomness in PHP. Instead of a 'bool', use an 'int' for $is_strong_result but more accurately call it $result_strength. Assign a minimum recommended threshold in the documentation against which application developers can act. That is, let the application developer decide what the minimum strength is that they are willing to accept. Otherwise, from an application developer's perspective, it will appear to be some arbitrary internal decision and won't really be a significant improvement over other PHP functions - it'll just be one more confusing option to pick from. As an aside, I recently developed a sufficient-for-most-needs CSPRNG in pure PHP: http://barebonescms.com/documentation/csprng/
-- Thomas Hruska CubicleSoft President Barebones CMS is a high-performance, open source content management system for web developers operating in a team environment. An open source CubicleSoft initiative. Your choice of a MIT or LGPL license. http://barebonescms.com/

Tom Worster

14 years ago
On 1/9/12 10:39 AM, "Thomas Hruska" <thruska@cubiclesoft.com> wrote:
>I'm glad someone is doing something about making more sources of >randomness in PHP.
I am pleased to hear it.
>Instead of a 'bool', use an 'int' for $is_strong_result but more >accurately call it $result_strength. Assign a minimum recommended >threshold in the documentation against which application developers can >act. That is, let the application developer decide what the minimum >strength is that they are willing to accept. Otherwise, from an >application developer's perspective, it will appear to be some arbitrary >internal decision and won't really be a significant improvement over >other PHP functions - it'll just be one more confusing option to pick >from.
The RFC has its lengthy name because this effort aims to add an API to the system CSPRNG and nothing else. The various system CSPRNGs, to the best of my knowledge, are not able to provide a randomness metric. They indicate if they are adequately seeded or not. Hence $is_strong_result is a boolean. If we limit to the scope as currently defined then I do not see how to introduce a "degree of true randomness" result.