On Thu, 28 Feb 2013 06:00:21 -0500 Akshay Saraswat akshay.s@samsung.com wrote:
Changes since v1:
- Moved code to drivers/crypto.
- Fixed few other nits.
be more specific wrt changes between revisions.
- if (buf_len == 0) {
/* ACE H/W cannot compute hash value for empty string */if (hash_type == ACE_SHA_TYPE_SHA1)memcpy(pout, sha1_digest_emptymsg, SHA1_SUM_LEN);elsememcpy(pout, sha256_digest_emptymsg, SHA256_SUM_LEN);return 0;- } else if (buf_len > BUF_LIMIT) {
debug("%s: Buffer length beyond limit\n", __func__);return -1;- }
erroring out just because the user asked for a hash on something larger than 8MB is unacceptable: any hash implementation that replaces a corresponding software implementation must not introduce new limitations and regressions.
If the hardware supports hash updating - which is likely - use that to perform hashes >8MB. If not, revert to software. And for the 0 length case.
- start = get_timer(0);
- while ((readl(&ace_sha_reg->hash_status) & ACE_HASH_MSGDONE_MASK) ==
ACE_HASH_MSGDONE_OFF) {
alignment - the 'A' in ACE_H*_OFF should fall directly below the '(' in 'readl('.
if (get_timer(start) > TIMEOUT_MS) {debug("%s: Timeout waiting for ACE\n", __func__);
see my other email to v1 of this patch wrt timing-out.
- len = (hash_type == ACE_SHA_TYPE_SHA1) ? SHA1_SUM_LEN / 4
: SHA256_SUM_LEN / 4;
alignment
a more recent version of checkpatch finds other such problems:
CHECK: No space is necessary after a cast #610: FILE: drivers/crypto/ace_sha.c:52: + struct exynos_ace_sfr *ace_sha_reg = + (struct exynos_ace_sfr *) samsung_get_base_ace_sfr();
CHECK: Alignment should match open parenthesis #630: FILE: drivers/crypto/ace_sha.c:72: + writel(ACE_HASH_SWAPDI_ON | ACE_HASH_SWAPDO_ON | ACE_HASH_SWAPIV_ON, + &ace_sha_reg->hash_byteswap);
CHECK: Blank lines aren't necessary after an open brace '{' #658: FILE: drivers/crypto/ace_sha.c:100: + ACE_HASH_MSGDONE_OFF) { +
CHECK: space prohibited before semicolon #673: FILE: drivers/crypto/ace_sha.c:115: + for (i = 0; i < len ; i++)
Kim