All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jason Gunthorpe <jgg@ziepe.ca>
To: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Lijun Ou <oulijun@huawei.com>,
	"Wei Hu(Xavier)" <xavier.huwei@huawei.com>,
	Doug Ledford <dledford@redhat.com>,
	linux-rdma@vger.kernel.org, kernel-janitors@vger.kernel.org
Subject: Re: [PATCH] RDMA/hns: prevent undefined behavior in hns_roce_set_user_sq_size()
Date: Thu, 24 Oct 2019 13:37:49 -0300	[thread overview]
Message-ID: <20191024163749.GX23952@ziepe.ca> (raw)
In-Reply-To: <20191007121821.GI21515@kadam>

On Mon, Oct 07, 2019 at 03:18:22PM +0300, Dan Carpenter wrote:
> This one still needs to be applied.
> 
> regards,
> dan carpenter

Weird, it is marked changes requested in patchworks. An email must
have been lost??

I think I probably wanted to say that:

> >  	/* Sanity check SQ size before proceeding */
> > -	if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes ||
> > +	if (ucmd->log_sq_bb_count > 31 ||
> > +	    (u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes
> > ||

Overall should probably be coded using check_shl_overflow(), as this
later shift:

	hr_qp->sq.wqe_cnt = 1 << ucmd->log_sq_bb_count;

Is storing it into an int and hardwring '31' because it magically
matches that lower shift is pretty fragile.

More like this?

diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index bec48f2593f405..6aa27d6ea3a600 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -332,9 +332,8 @@ static int check_sq_size_with_integrity(struct hns_roce_dev *hr_dev,
 	u8 max_sq_stride = ilog2(roundup_sq_stride);
 
 	/* Sanity check SQ size before proceeding */
-	if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes ||
-	     ucmd->log_sq_stride > max_sq_stride ||
-	     ucmd->log_sq_stride < HNS_ROCE_IB_MIN_SQ_STRIDE) {
+	if (ucmd->log_sq_stride > max_sq_stride ||
+	    ucmd->log_sq_stride < HNS_ROCE_IB_MIN_SQ_STRIDE) {
 		ibdev_err(&hr_dev->ib_dev, "check SQ size error!\n");
 		return -EINVAL;
 	}
@@ -358,13 +357,16 @@ static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
 	u32 max_cnt;
 	int ret;
 
+	if (check_shl_overflow(1, ucmd->log_sq_bb_count, &hr_qp->sq.wqe_cnt) ||
+	    hr_qp->sq.wqe_cnt > hr_dev->caps.max_wqes)
+		return -EINVAL;
+
 	ret = check_sq_size_with_integrity(hr_dev, cap, ucmd);
 	if (ret) {
 		ibdev_err(&hr_dev->ib_dev, "Sanity check sq size failed\n");
 		return ret;
 	}
 
-	hr_qp->sq.wqe_cnt = 1 << ucmd->log_sq_bb_count;
 	hr_qp->sq.wqe_shift = ucmd->log_sq_stride;
 
 	max_cnt = max(1U, cap->max_send_sge);

WARNING: multiple messages have this Message-ID (diff)
From: Jason Gunthorpe <jgg@ziepe.ca>
To: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Lijun Ou <oulijun@huawei.com>,
	"Wei Hu(Xavier)" <xavier.huwei@huawei.com>,
	Doug Ledford <dledford@redhat.com>,
	linux-rdma@vger.kernel.org, kernel-janitors@vger.kernel.org
Subject: Re: [PATCH] RDMA/hns: prevent undefined behavior in hns_roce_set_user_sq_size()
Date: Thu, 24 Oct 2019 16:37:49 +0000	[thread overview]
Message-ID: <20191024163749.GX23952@ziepe.ca> (raw)
In-Reply-To: <20191007121821.GI21515@kadam>

On Mon, Oct 07, 2019 at 03:18:22PM +0300, Dan Carpenter wrote:
> This one still needs to be applied.
> 
> regards,
> dan carpenter

Weird, it is marked changes requested in patchworks. An email must
have been lost??

I think I probably wanted to say that:

> >  	/* Sanity check SQ size before proceeding */
> > -	if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes ||
> > +	if (ucmd->log_sq_bb_count > 31 ||
> > +	    (u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes
> > ||

Overall should probably be coded using check_shl_overflow(), as this
later shift:

	hr_qp->sq.wqe_cnt = 1 << ucmd->log_sq_bb_count;

Is storing it into an int and hardwring '31' because it magically
matches that lower shift is pretty fragile.

More like this?

diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index bec48f2593f405..6aa27d6ea3a600 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -332,9 +332,8 @@ static int check_sq_size_with_integrity(struct hns_roce_dev *hr_dev,
 	u8 max_sq_stride = ilog2(roundup_sq_stride);
 
 	/* Sanity check SQ size before proceeding */
-	if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes ||
-	     ucmd->log_sq_stride > max_sq_stride ||
-	     ucmd->log_sq_stride < HNS_ROCE_IB_MIN_SQ_STRIDE) {
+	if (ucmd->log_sq_stride > max_sq_stride ||
+	    ucmd->log_sq_stride < HNS_ROCE_IB_MIN_SQ_STRIDE) {
 		ibdev_err(&hr_dev->ib_dev, "check SQ size error!\n");
 		return -EINVAL;
 	}
@@ -358,13 +357,16 @@ static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev,
 	u32 max_cnt;
 	int ret;
 
+	if (check_shl_overflow(1, ucmd->log_sq_bb_count, &hr_qp->sq.wqe_cnt) ||
+	    hr_qp->sq.wqe_cnt > hr_dev->caps.max_wqes)
+		return -EINVAL;
+
 	ret = check_sq_size_with_integrity(hr_dev, cap, ucmd);
 	if (ret) {
 		ibdev_err(&hr_dev->ib_dev, "Sanity check sq size failed\n");
 		return ret;
 	}
 
-	hr_qp->sq.wqe_cnt = 1 << ucmd->log_sq_bb_count;
 	hr_qp->sq.wqe_shift = ucmd->log_sq_stride;
 
 	max_cnt = max(1U, cap->max_send_sge);

  reply	other threads:[~2019-10-24 16:37 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-08  9:25 [PATCH] RDMA/hns: prevent undefined behavior in hns_roce_set_user_sq_size() Dan Carpenter
2019-10-07 12:18 ` Dan Carpenter
2019-10-07 12:18   ` Dan Carpenter
2019-10-24 16:37   ` Jason Gunthorpe [this message]
2019-10-24 16:37     ` Jason Gunthorpe
2019-10-24 18:20     ` Dan Carpenter
2019-10-24 18:20       ` Dan Carpenter
2019-10-28 14:23       ` Jason Gunthorpe
2019-10-28 14:23         ` Jason Gunthorpe

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191024163749.GX23952@ziepe.ca \
    --to=jgg@ziepe.ca \
    --cc=dan.carpenter@oracle.com \
    --cc=dledford@redhat.com \
    --cc=kernel-janitors@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=oulijun@huawei.com \
    --cc=xavier.huwei@huawei.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.