Richard Yao c77d2d7415 crypto_get_ptrs() should always write to *out_data_2 ... Callers will check if it has been set to NULL before trying to access it, but never initialize it themselves. Whenever "one block spans two iovecs", `crypto_get_ptrs()` will return, without ever setting `*out_data_2 = NULL`. The caller will then do a NULL check against the uninitailized pointer and if it is not zero, pass it to `memcpy()`. The only reason this has not caused horrible runtime issues is because `memcpy()` should be told to copy zero bytes when this happens. That said, this is technically undefined behavior, so we should correct it so that future changes to the code cannot trigger it. Clang's static analyzer found this with the help of CodeChecker's CTU analysis. Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Richard Yao <richard.yao@alumni.stonybrook.edu> Closes #14043		2022-10-19 17:10:56 -07:00
..
aes	Replace dead opensolaris.org license link	2022-07-11 14:16:13 -07:00
blake3	Cleanup: Address Clang's static analyzer's unused code complaints	2022-10-14 13:37:54 -07:00
edonr	Fix -Wattribute-warning in edonr	2022-06-27 14:19:04 -07:00
modes	crypto_get_ptrs() should always write to *out_data_2	2022-10-19 17:10:56 -07:00
sha2	Remove bcopy(), bzero(), bcmp()	2022-03-15 15:13:42 -07:00
skein	Forbid b{copy,zero,cmp}(). Don't include <strings.h> for <string.h>	2022-03-15 15:13:48 -07:00