Date: Thu, 21 Jul 2016 01:44:02 +0300 From: Solar Designer <solar@...nwall.com> To: owl-dev@...ts.openwall.com Subject: Re: passwdqc code quality Hi Dmitry, Thank you for your review and comments! On Thu, Jul 21, 2016 at 01:09:54AM +0300, Dmitry V. Levin wrote: > In this particular case, _passwdqc_memzero doesn't look like a good > addition to libpasswdqc ABI, Yes. The underscore suggests it's internal and may be gone later. > so I'd rather not export it from libpasswdqc at all, e.g. Your changes mean that there will be separate _passwdqc_memzero pointers in libpasswdqc, pam_passwdqc, and both programs, and that libpasswdqc's _passwdqc_memzero couldn't be overridden at runtime by a program using this library. This introduces the likelihood of a linker figuring out that _passwdqc_memzero is always memset, and optimizing it out. With my code, the (static) linker may only figure out that _passwdqc_memzero is often memset, so it can theoretically introduce code to handle this special case quicker, but that's less likely. If you still prefer to make these changes despite of what I wrote above, please feel free to commit them tonight. I haven't yet made my changes public, so you do not need to bump the version number. I'll probably release this tomorrow. Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.