2017-08-16 11:20:04

by Ryder Lee

[permalink] [raw]
Subject: [PATCH] crypto: mediatek - Add empty messages check in GCM mode

Currently, empty messages are not supported in GCM mode, hence add
a check to prevent producing incorrect results.

Signed-off-by: Ryder Lee <[email protected]>
---
drivers/crypto/mediatek/mtk-aes.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/drivers/crypto/mediatek/mtk-aes.c b/drivers/crypto/mediatek/mtk-aes.c
index 9e845e8..efe4e63 100644
--- a/drivers/crypto/mediatek/mtk-aes.c
+++ b/drivers/crypto/mediatek/mtk-aes.c
@@ -928,8 +928,13 @@ static int mtk_aes_gcm_start(struct mtk_cryp *cryp, struct mtk_aes_rec *aes)
static int mtk_aes_gcm_crypt(struct aead_request *req, u64 mode)
{
struct mtk_aes_base_ctx *ctx = crypto_aead_ctx(crypto_aead_reqtfm(req));
+ struct mtk_aes_gcm_ctx *gctx = mtk_aes_gcm_ctx_cast(ctx);
struct mtk_aes_reqctx *rctx = aead_request_ctx(req);

+ /* Empty messages are not supported yet */
+ if (!gctx->textlen && !req->assoclen)
+ return -EINVAL
+
rctx->mode = AES_FLAGS_GCM | mode;

return mtk_aes_handle_queue(ctx->cryp, !!(mode & AES_FLAGS_ENCRYPT),
--
1.9.1


2017-08-16 12:18:06

by Stephan Müller

[permalink] [raw]
Subject: Re: [PATCH] crypto: mediatek - Add empty messages check in GCM mode

Am Mittwoch, 16. August 2017, 13:19:48 CEST schrieb Ryder Lee:

Hi Ryder,

> Currently, empty messages are not supported in GCM mode, hence add
> a check to prevent producing incorrect results.

In case the caller would provide empty messages, would there be just wrong
data or a real problem? Note, unprivileged user space can make such requests.

If there would be a real problem (like crash), this should also go to stable.

Ciao
Stephan

2017-08-16 12:50:24

by Ryder Lee

[permalink] [raw]
Subject: Re: [PATCH] crypto: mediatek - Add empty messages check in GCM mode

Hi,

On Wed, 2017-08-16 at 14:18 +0200, Stephan Mueller wrote:
> Am Mittwoch, 16. August 2017, 13:19:48 CEST schrieb Ryder Lee:
>
> Hi Ryder,
>
> > Currently, empty messages are not supported in GCM mode, hence add
> > a check to prevent producing incorrect results.
>
> In case the caller would provide empty messages, would there be just wrong
> data or a real problem? Note, unprivileged user space can make such requests.
>
> If there would be a real problem (like crash), this should also go to stable.
>
> Ciao
> Stephan

This does not cause any problem and it just returns a wrong result.

Ryder

2017-08-22 07:02:10

by Herbert Xu

[permalink] [raw]
Subject: Re: [PATCH] crypto: mediatek - Add empty messages check in GCM mode

On Wed, Aug 16, 2017 at 07:19:48PM +0800, Ryder Lee wrote:
> Currently, empty messages are not supported in GCM mode, hence add
> a check to prevent producing incorrect results.
>
> Signed-off-by: Ryder Lee <[email protected]>
> ---
> drivers/crypto/mediatek/mtk-aes.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/drivers/crypto/mediatek/mtk-aes.c b/drivers/crypto/mediatek/mtk-aes.c
> index 9e845e8..efe4e63 100644
> --- a/drivers/crypto/mediatek/mtk-aes.c
> +++ b/drivers/crypto/mediatek/mtk-aes.c
> @@ -928,8 +928,13 @@ static int mtk_aes_gcm_start(struct mtk_cryp *cryp, struct mtk_aes_rec *aes)
> static int mtk_aes_gcm_crypt(struct aead_request *req, u64 mode)
> {
> struct mtk_aes_base_ctx *ctx = crypto_aead_ctx(crypto_aead_reqtfm(req));
> + struct mtk_aes_gcm_ctx *gctx = mtk_aes_gcm_ctx_cast(ctx);
> struct mtk_aes_reqctx *rctx = aead_request_ctx(req);
>
> + /* Empty messages are not supported yet */
> + if (!gctx->textlen && !req->assoclen)
> + return -EINVAL

Your patch does not compile.
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

2017-08-22 07:25:28

by Ryder Lee

[permalink] [raw]
Subject: Re: [PATCH] crypto: mediatek - Add empty messages check in GCM mode

Hi,
On Tue, 2017-08-22 at 15:01 +0800, Herbert Xu wrote:
> On Wed, Aug 16, 2017 at 07:19:48PM +0800, Ryder Lee wrote:
> > Currently, empty messages are not supported in GCM mode, hence add
> > a check to prevent producing incorrect results.
> >
> > Signed-off-by: Ryder Lee <[email protected]>
> > ---
> > drivers/crypto/mediatek/mtk-aes.c | 5 +++++
> > 1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/crypto/mediatek/mtk-aes.c b/drivers/crypto/mediatek/mtk-aes.c
> > index 9e845e8..efe4e63 100644
> > --- a/drivers/crypto/mediatek/mtk-aes.c
> > +++ b/drivers/crypto/mediatek/mtk-aes.c
> > @@ -928,8 +928,13 @@ static int mtk_aes_gcm_start(struct mtk_cryp *cryp, struct mtk_aes_rec *aes)
> > static int mtk_aes_gcm_crypt(struct aead_request *req, u64 mode)
> > {
> > struct mtk_aes_base_ctx *ctx = crypto_aead_ctx(crypto_aead_reqtfm(req));
> > + struct mtk_aes_gcm_ctx *gctx = mtk_aes_gcm_ctx_cast(ctx);
> > struct mtk_aes_reqctx *rctx = aead_request_ctx(req);
> >
> > + /* Empty messages are not supported yet */
> > + if (!gctx->textlen && !req->assoclen)
> > + return -EINVAL
>
> Your patch does not compile.

Ohh..Somehow I delete the semicolon. I will send v1 soon, sorry about
that.

Ryder