feedc0de/wolfssl
1
0
Fork 0
forked from wolfSSL/wolfssl
Code Pull Requests Activity

Fixes for a few more fsanitize issues. Added alignment for ForceZero. Added word32 aligned acceleration for ForceZeero. Added 'NO_ALIGNED_FORCEZERO' define to allow disabling aligned ForceZero acceleration. We cast the 24 left-shifts to word32 because compiler assumes signed int type, and a runtime value with MSB set results into runtime fsanitize error.

Browse Source
This commit is contained in:
David Garske
2018-05-16 13:27:13 -07:00
parent f021375c4b
commit 52b66edf72
3 changed files with 27 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
wolfcrypt/src/misc.c
View File

@@ -199,13 +199,36 @@ STATIC INLINE void xorbuf(void* buf, const void* mask, word32 count)
STATIC INLINE void ForceZero(const void* mem, word32 len)
{
volatile byte* z = (volatile byte*)mem;
#ifndef NO_ALIGNED_FORCEZERO
#if defined(WOLFSSL_X86_64_BUILD) && defined(WORD64_AVAILABLE)
volatile word64* w;
/* align buffer */
while (len && ((word64)z % sizeof(word64)) != 0) {
*z++ = 0; len--;
}
/* do aligned force zero */
for (w = (volatile word64*)z; len >= sizeof(*w); len -= sizeof(*w))
*w++ = 0;
z = (volatile byte*)w;
#else
volatile word32* w;
/* align buffer */
while (len && ((word32)z % sizeof(word32)) != 0) {
*z++ = 0; len--;
}
/* do aligned force zero */
for (w = (volatile word32*)z; len >= sizeof(*w); len -= sizeof(*w))
*w++ = 0;
z = (volatile byte*)w;
#endif
#endif /* NO_ALIGNED_FORCEZERO */
/* do byte by byte force zero */
while (len--) *z++ = 0;
}
@@ -292,7 +315,7 @@ STATIC INLINE void ato16(const byte* c, word16* wc_u16)
/* convert opaque to 32 bit integer */
STATIC INLINE void ato32(const byte* c, word32* wc_u32)
{
*wc_u32 = (c[0] << 24) | (c[1] << 16) | (c[2] << 8) | c[3];
*wc_u32 = ((word32)c[0] << 24) | (c[1] << 16) | (c[2] << 8) | c[3];
}