ltc: ctr: update pt and ct after acceleration

[vchong]

Problem occurs in the condition of the following case:

1st decryption:
Decrypt a ciphertext whose length is a multiple of the block size (16B)
(len = n * block_size)
2nd decryption:
Decrypt the continuing ciphertext whose length is not a multiple of the
block size
(len = m * block_size + l)

In this case accel_ctr_encrypt() is firstly used at the 2nd decryption.
If pt and ct are not updated, the top (l = len % block_size) bytes of
decryption result are sometimes destroyed.

Reported-by: Tetsuya Yoshizaki yoshizaki.tetsuya@socionext.com
Signed-off-by: Tetsuya Yoshizaki yoshizaki.tetsuya@socionext.com
Signed-off-by: Victor Chong victor.chong@linaro.org

[vchong]

Is the Reported-by needed in this case (reporter reported issue and also provided a patch submitted by someone else on his behalf)?

[jforissier]

TL;DR: please just replace Reported-by: ... by From: ...

I suggest using a From: tag as the first entry in the tag section. Keep the original author's Signed-off-by:. Then I will merge the patch manually, removing the From: and passing the --author option to git.

Note, the Linux kernel has the same convention when it comes to sending patches to the mailing lists, except that the From: line should be the first in the message (see here). The From: does not appear in the final change log, instead it is picked up automatically by git am. This would probably not work for us because checkpatch considers the first line of the commit log to be the subject, so we might have issues with CI. That's why I propose to move the From: to the bottom.

[Update: it looks like the presence of the From: line is enough for GitHub to attribute the commit to the original author, which is quite convenient].

[jforissier]

@sjaeckel FYI.

[vchong]

@jforissier Updated. Thanks!

[jenswi-linaro]

Here's a test case OP-TEE/optee_test#248

[sjaeckel]

thx! c.f. libtom/libtomcrypt@d1d3ae2

[jforissier]

@jenswi-linaro you beat me to it ;-)
It seems to me that the commit description in this PR is somewhat misleading. I understand you need to call the encrypt function only two times where in fact the error occurs on the third time if I'm not mistaken.

[jforissier]

Tested-by: Jerome Forissier <jerome.forissier@linaro.org> (HiKey960)

[jforissier]

Merged manually. Thanks guys.

@jenswi-linaro I have converted your "LGTM" (that you gave via email) into an Acked-by:
I have also added a link to the upstream commit.