On 21 September 2018 at 07:45, Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote: > On Thu, Sep 13, 2018 at 10:51:30AM +0200, Ondrej Mosnacek wrote: >> This patchset contains a corner-case fix and several improvements for >> the LRW template. >> >> The first patch fixes an out-of-bounds array access (and subsequently >> incorrect cipher output) when the LRW counter goes from all ones to all >> zeros. This patch should be applied to the crypto-2.6 tree and also go >> to stable. >> >> The second patch adds a test vector for lrw(aes) that covers the above >> bug. >> >> The third patch is a small optimization of the LRW tweak computation. >> >> The fourth patch is a follow-up to a similar patch for XTS (it >> simplifies away the use of dynamically allocated auxiliary buffer to >> cache the computed tweak values): >> https://patchwork.kernel.org/patch/10588775/ >> >> Patches 2-4 should be applied only to cryptodev-2.6, but they all depend >> on the first patch. >> >> Changes in v4: >> - applied various corrections/suggestions from Eric Biggers >> - added a fix for buggy behavior on counter wrap-around (+ test vector) >> >> v3: https://www.spinics.net/lists/linux-crypto/msg34946.html >> Changes in v3: >> - fix a copy-paste error >> >> v2: https://www.spinics.net/lists/linux-crypto/msg34890.html >> Changes in v2: >> - small cleanup suggested by Eric Biggers >> >> v1: https://www.spinics.net/lists/linux-crypto/msg34871.html >> >> Ondrej Mosnacek (4): >> crypto: lrw - Fix out-of bounds access on counter overflow >> crypto: testmgr - Add test for LRW counter wrap-around >> crypto: lrw - Optimize tweak computation >> crypto: lrw - Do not use auxiliary buffer >> >> crypto/lrw.c | 342 +++++++++++++---------------------------------- >> crypto/testmgr.h | 21 +++ >> 2 files changed, 112 insertions(+), 251 deletions(-) > > All applied. Thanks. I am seeing tcrypt failures with this code: alg: skcipher: Test 8 failed (invalid result) on encryption for lrw(ecb-aes-ce) 00000000: 47 90 50 f6 f4 8d 5c 7f 84 c7 83 95 2d a2 02 c0 00000010: da 7f a3 c0 88 2a 0a 50 fb c1 78 03 39 fe 1d e5 00000020: 47 90 50 f6 f4 8d 5c 7f 84 c7 83 95 2d a2 02 c0 reproduced on both arm64 and ARM (the latter in LE and BE modes) -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
