[PATCH] D56146: [Sanitizer] enable arc4random api on Darwin
Conrad Meyer via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Sat Dec 29 10:07:31 PST 2018
cem requested changes to this revision.
cem added inline comments.
This revision now requires changes to proceed.
================
Comment at: lib/sanitizer_common/sanitizer_mac.cc:1120
return false;
- // arc4random never fails.
- arc4random_buf(buffer, length);
+ uptr rnd = getentropy(buffer, length);
+ int rverrno = 0;
----------------
I don't think this change makes much sense.
# arc4random_buf is a perfectly good userspace random API and has a long history; getentropy is only present on the latest FreeBSD. This breaks backwards compatibility with FreeBSD 11.x, for example.
# arc4random_buf maintains a userspace Chacha keystream seeded periodically by kernel getrandom(2) syscalls; getentropy(3) invokes a syscall every time it is used. This change makes GetRandom() significantly more expensive per call, without benefit.
# The previous path was failure-free; this version has to implement additional error handling, and does so questionably. See below.
================
Comment at: lib/sanitizer_common/sanitizer_mac.cc:1122-1123
+ int rverrno = 0;
+ if (internal_iserror(rnd, &rverrno) && rverrno == EINVAL)
+ return false;
return true;
----------------
# Why would getentropy produce an `EINVAL` code? The only documented errors are EFAULT (depends on `buffer`) and EIO (`length` checks above should prevent). Linux documents an additional ENOSYS code if getrandom is absent, but not EINVAL, AFAIK.
- https://man.openbsd.org/getentropy.2
- http://man7.org/linux/man-pages/man3/getentropy.3.html
- https://www.freebsd.org/cgi/man.cgi?query=getentropy&sektion=3&manpath=freebsd-release-ports
(I am unfamiliar with `internal_iserror`; please forgive me if it is expected to squash other errors to `EINVAL` for some reason. :-))
# Why should errors other than `EINVAL` be treated as success conditions?
Repository:
rCRT Compiler Runtime
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D56146/new/
https://reviews.llvm.org/D56146
More information about the llvm-commits
mailing list