2019-05-15 13:14:34

by Sascha Hauer

Subject: [PATCH] crypto: caam: print debugging hex dumps after unmapping

The debugging hex dumps in skcipher_encrypt_done() and
skcipher_decrypt_done() are printed while the request is still DMA
mapped. This results in bogus hex dumps with things like mixtures
between plain text and cipher text. Unmap first before printing.

Signed-off-by: Sascha Hauer <[email protected]>
---
drivers/crypto/caam/caamalg.c | 21 +++++++++++----------
1 file changed, 11 insertions(+), 10 deletions(-)

diff --git a/drivers/crypto/caam/caamalg.c b/drivers/crypto/caam/caamalg.c
index 3e23d4b2cce2..a992ff56fd15 100644
--- a/drivers/crypto/caam/caamalg.c
+++ b/drivers/crypto/caam/caamalg.c
@@ -1009,15 +1009,6 @@ static void skcipher_encrypt_done(struct device *jrdev, u32 *desc, u32 err,
if (err)
caam_jr_strstatus(jrdev, err);

-#ifdef DEBUG
- print_hex_dump(KERN_ERR, "dstiv @"__stringify(__LINE__)": ",
- DUMP_PREFIX_ADDRESS, 16, 4, req->iv,
- edesc->src_nents > 1 ? 100 : ivsize, 1);
-#endif
- caam_dump_sg(KERN_ERR, "dst @" __stringify(__LINE__)": ",
- DUMP_PREFIX_ADDRESS, 16, 4, req->dst,
- edesc->dst_nents > 1 ? 100 : req->cryptlen, 1);
-
skcipher_unmap(jrdev, edesc, req);

/*
@@ -1028,6 +1019,15 @@ static void skcipher_encrypt_done(struct device *jrdev, u32 *desc, u32 err,
scatterwalk_map_and_copy(req->iv, req->dst, req->cryptlen -
ivsize, ivsize, 0);

+#ifdef DEBUG
+ print_hex_dump(KERN_ERR, "dstiv @"__stringify(__LINE__)": ",
+ DUMP_PREFIX_ADDRESS, 16, 4, req->iv,
+ edesc->src_nents > 1 ? 100 : ivsize, 1);
+#endif
+ caam_dump_sg(KERN_ERR, "dst @" __stringify(__LINE__)": ",
+ DUMP_PREFIX_ADDRESS, 16, 4, req->dst,
+ edesc->dst_nents > 1 ? 100 : req->cryptlen, 1);
+
kfree(edesc);

skcipher_request_complete(req, err);
@@ -1049,6 +1049,8 @@ static void skcipher_decrypt_done(struct device *jrdev, u32 *desc, u32 err,
if (err)
caam_jr_strstatus(jrdev, err);

+ skcipher_unmap(jrdev, edesc, req);
+
#ifdef DEBUG
print_hex_dump(KERN_ERR, "dstiv @"__stringify(__LINE__)": ",
DUMP_PREFIX_ADDRESS, 16, 4, req->iv, ivsize, 1);
@@ -1057,7 +1059,6 @@ static void skcipher_decrypt_done(struct device *jrdev, u32 *desc, u32 err,
DUMP_PREFIX_ADDRESS, 16, 4, req->dst,
edesc->dst_nents > 1 ? 100 : req->cryptlen, 1);

- skcipher_unmap(jrdev, edesc, req);
kfree(edesc);

skcipher_request_complete(req, err);
--
2.20.1

2019-05-15 16:28:36

by Horia Geanta

Subject: Re: [PATCH] crypto: caam: print debugging hex dumps after unmapping

On 5/15/2019 4:13 PM, Sascha Hauer wrote:
> The debugging hex dumps in skcipher_encrypt_done() and
> skcipher_decrypt_done() are printed while the request is still DMA
> mapped. This results in bogus hex dumps with things like mixtures
> between plain text and cipher text. Unmap first before printing.
>
The description is not accurate.
req->iv is no longer DMA mapped since commit 115957bb3e59 ("crypto: caam - fix
IV DMA mapping and updating"), so this is not related to IV DMA unmapping vs.
print order.

Currently:
-for encryption, printed req->iv contains the input IV; copy of output IV into
req->iv is done further below
-for decryption, printed req->iv should be correct, since output IV is copied
into req->iv in skcipher_decrypt(), before accelerator runs

Could you please resubmit with updated message?

Thanks,
Horia

2019-05-16 14:26:49

by Sascha Hauer

Subject: Re: [PATCH] crypto: caam: print debugging hex dumps after unmapping

On Wed, May 15, 2019 at 04:27:51PM +0000, Horia Geanta wrote:
> On 5/15/2019 4:13 PM, Sascha Hauer wrote:
> > The debugging hex dumps in skcipher_encrypt_done() and
> > skcipher_decrypt_done() are printed while the request is still DMA
> > mapped. This results in bogus hex dumps with things like mixtures
> > between plain text and cipher text. Unmap first before printing.
> >
> The description is not accurate.
> req->iv is no longer DMA mapped since commit 115957bb3e59 ("crypto: caam - fix
> IV DMA mapping and updating"), so this is not related to IV DMA unmapping vs.
> print order.
>
> Currently:
> -for encryption, printed req->iv contains the input IV; copy of output IV into
> req->iv is done further below
> -for decryption, printed req->iv should be correct, since output IV is copied
> into req->iv in skcipher_decrypt(), before accelerator runs
>
> Could you please resubmit with updated message?

Just did that.

Thanks
Sascha


--
Pengutronix e.K. | |
Industrial Linux Solutions | http://www.pengutronix.de/ |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |