| commit | 4891d713fb1949fe5c869974b94018380d110363 | |
| author | Luke Howard <lukeh@padl.com> | |
| Sat, 15 Jan 2022 02:38:18 +0000 (15 13:38 +1100) | ||
| committer | Luke Howard <lukeh@padl.com> | |
| Sat, 15 Jan 2022 02:38:18 +0000 (15 13:38 +1100) | ||
| tree | b7ac85661a19c9a9bba3b73acbcaf41997b4d3da | treesnapshot (tar.gz zip) |
| parent | 046675ce0ec45da5fe9cf5080f664c0a42d22dd0 | commitdiff |
bn_s_mp_rand_platform.c: s_read_urandom: correctly handle split read
s_read_urandom has a while loop to handle read() that returns less
than the full buffer (either due to EINTR or because more than the
atomic guarantee from urandom was requested). However, the target
of the read was always the base pointer p instead of the updated
pointer q, so in the end less than the requested randomness is returned.
Use q instead of p in the read() call.
Upstream pull request:
https://github.com/libtom/libtommath/pull/512
Signed-off-by: Arnout Vandecappelle <arnout@mind.be>
s_read_urandom has a while loop to handle read() that returns less
than the full buffer (either due to EINTR or because more than the
atomic guarantee from urandom was requested). However, the target
of the read was always the base pointer p instead of the updated
pointer q, so in the end less than the requested randomness is returned.
Use q instead of p in the read() call.
Upstream pull request:
https://github.com/libtom/libtommath/pull/512
Signed-off-by: Arnout Vandecappelle <arnout@mind.be>
| lib/hcrypto/libtommath/bn_s_mp_rand_platform.c | diffblobblamehistory |