boringssl

History

Adam Langley 82aa28fa81 Make \|BUF_memdup\| look for zero length, not NULL. BUF_memdup tries to avoid mallocing zero bytes (and thus unduly returning an error for a NULL return value) by testing whether the input buffer is NULL. This goes back to the original OpenSSL code. However, when \|ext_npn_parse_serverhello\| tries to use \|BUF_memdup\| to copy an NPN value returned by a callback, some callbacks just set the output /length/ to zero to indicate an empty value. Thus, when \|BUF_memdup\| tests the pointer, it's an uninitialised value and MSan throws an error. Since passing a NULL pointer to \|BUF_memdup\| better imply that the length is zero, while the reverse empirically isn't true, testing the length seems safer. Change-Id: I06626f7dfb761de631fd997bda60057b76b8da94	2015-10-06 18:11:33 -07:00
..
buf.c	Make \|BUF_memdup\| look for zero length, not NULL.	2015-10-06 18:11:33 -07:00
CMakeLists.txt

Adam Langley 82aa28fa81 Make |BUF_memdup| look for zero length, not NULL.

BUF_memdup tries to avoid mallocing zero bytes (and thus unduly
returning an error for a NULL return value) by testing whether the input
buffer is NULL. This goes back to the original OpenSSL code.

However, when |ext_npn_parse_serverhello| tries to use |BUF_memdup| to
copy an NPN value returned by a callback, some callbacks just set the
output /length/ to zero to indicate an empty value. Thus, when
|BUF_memdup| tests the pointer, it's an uninitialised value and MSan
throws an error.

Since passing a NULL pointer to |BUF_memdup| better imply that the
length is zero, while the reverse empirically isn't true, testing the
length seems safer.

Change-Id: I06626f7dfb761de631fd997bda60057b76b8da94

2015-10-06 18:11:33 -07:00

buf.c

Make |BUF_memdup| look for zero length, not NULL.

2015-10-06 18:11:33 -07:00

CMakeLists.txt