Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1625723 > unrolled thread

[PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl

Back to article view | Back to linux.kernel

Contents

  [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl Logan Gunthorpe <logang@deltatee.com> - 2017-04-19 01:40 +0200
    Re: [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl Johannes Thumshirn <jthumshirn@suse.de> - 2017-04-19 11:40 +0200
      Re: [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl Logan Gunthorpe <logang@deltatee.com> - 2017-04-19 17:50 +0200
    Re: [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl Christoph Hellwig <hch@lst.de> - 2017-04-19 21:40 +0200
    Re: [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl Sagi Grimberg <sagi@grimberg.me> - 2017-04-20 13:40 +0200

#1625723 — [PATCH] nvmet: convert from kmap to nvmet_copy_from_sgl

This is safer as it doesn't rely on the data being stored in
a single page in an sgl.

It also aids our effort to start phasing out users of sg_page. See [1].

For this we kmalloc some memory, copy to it and free at the end. Note:
we can't allocate this memory on the stack as the kbuild test robot
reports some frame size overflows on i386.

[1] https://lwn.net/Articles/720053/

Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Sagi Grimberg <sagi@grimberg.me>
---
 drivers/nvme/target/fabrics-cmd.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)

diff --git a/drivers/nvme/target/fabrics-cmd.c b/drivers/nvme/target/fabrics-cmd.c
index 8bd022af..2e0ab10 100644
--- a/drivers/nvme/target/fabrics-cmd.c
+++ b/drivers/nvme/target/fabrics-cmd.c
@@ -122,7 +122,15 @@ static void nvmet_execute_admin_connect(struct nvmet_req *req)
 	struct nvmet_ctrl *ctrl = NULL;
 	u16 status = 0;
 
-	d = kmap(sg_page(req->sg)) + req->sg->offset;
+	d = kmalloc(sizeof(*d), GFP_KERNEL);
+	if (!d) {
+		status = NVME_SC_INTERNAL;
+		goto complete;
+	}
+
+	status = nvmet_copy_from_sgl(req, 0, d, sizeof(*d));
+	if (status)
+		goto out;
 
 	/* zero out initial completion result, assign values as needed */
 	req->rsp->result.u32 = 0;
@@ -143,7 +151,7 @@ static void nvmet_execute_admin_connect(struct nvmet_req *req)
 	}
 
 	status = nvmet_alloc_ctrl(d->subsysnqn, d->hostnqn, req,
-			le32_to_cpu(c->kato), &ctrl);
+				  le32_to_cpu(c->kato), &ctrl);
 	if (status)
 		goto out;
 
@@ -158,7 +166,8 @@ static void nvmet_execute_admin_connect(struct nvmet_req *req)
 	req->rsp->result.u16 = cpu_to_le16(ctrl->cntlid);
 
 out:
-	kunmap(sg_page(req->sg));
+	kfree(d);
+complete:
 	nvmet_req_complete(req, status);
 }
 
@@ -170,7 +179,15 @@ static void nvmet_execute_io_connect(struct nvmet_req *req)
 	u16 qid = le16_to_cpu(c->qid);
 	u16 status = 0;
 
-	d = kmap(sg_page(req->sg)) + req->sg->offset;
+	d = kmalloc(sizeof(*d), GFP_KERNEL);
+	if (!d) {
+		status = NVME_SC_INTERNAL;
+		goto complete;
+	}
+
+	status = nvmet_copy_from_sgl(req, 0, d, sizeof(*d));
+	if (status)
+		goto out;
 
 	/* zero out initial completion result, assign values as needed */
 	req->rsp->result.u32 = 0;
@@ -183,8 +200,8 @@ static void nvmet_execute_io_connect(struct nvmet_req *req)
 	}
 
 	status = nvmet_ctrl_find_get(d->subsysnqn, d->hostnqn,
-			le16_to_cpu(d->cntlid),
-			req, &ctrl);
+				     le16_to_cpu(d->cntlid),
+				     req, &ctrl);
 	if (status)
 		goto out;
 
@@ -205,7 +222,8 @@ static void nvmet_execute_io_connect(struct nvmet_req *req)
 	pr_info("adding queue %d to ctrl %d.\n", qid, ctrl->cntlid);
 
 out:
-	kunmap(sg_page(req->sg));
+	kfree(d);
+complete:
 	nvmet_req_complete(req, status);
 	return;
 
-- 
2.1.4

[toc] | [next] | [standalone]

#1625973

On Wed, Apr 19, 2017 at 12:33:09PM +0300, Max Gurtovoy wrote:
> Hi Logan,
> 
> On 4/19/2017 2:32 AM, Logan Gunthorpe wrote:
> >This is safer as it doesn't rely on the data being stored in
> >a single page in an sgl.
> >
> >It also aids our effort to start phasing out users of sg_page. See [1].
> >
> >For this we kmalloc some memory, copy to it and free at the end. Note:
> >we can't allocate this memory on the stack as the kbuild test robot
> >reports some frame size overflows on i386.
> >
> >[1] https://lwn.net/Articles/720053/
> >
> >Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
> >Cc: Christoph Hellwig <hch@lst.de>
> >Cc: Sagi Grimberg <sagi@grimberg.me>
> >---
> > drivers/nvme/target/fabrics-cmd.c | 32 +++++++++++++++++++++++++-------
> > 1 file changed, 25 insertions(+), 7 deletions(-)
> >
> >diff --git a/drivers/nvme/target/fabrics-cmd.c b/drivers/nvme/target/fabrics-cmd.c
> >index 8bd022af..2e0ab10 100644
> >--- a/drivers/nvme/target/fabrics-cmd.c
> >+++ b/drivers/nvme/target/fabrics-cmd.c
> >@@ -122,7 +122,15 @@ static void nvmet_execute_admin_connect(struct nvmet_req *req)
> > 	struct nvmet_ctrl *ctrl = NULL;
> > 	u16 status = 0;
> >
> >-	d = kmap(sg_page(req->sg)) + req->sg->offset;
> >+	d = kmalloc(sizeof(*d), GFP_KERNEL);
> 
> I'd prefer removing the dynamic allocation and use d on the stack to
> simplify the code.
> Any thoughts ?

Hi Max, 

Pasting from above:

> >we can't allocate this memory on the stack as the kbuild test robot
> >reports some frame size overflows on i386.


-- 
Johannes Thumshirn                                          Storage
jthumshirn@suse.de                                +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)
Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850

[toc] | [prev] | [next] | [standalone]

#1626408

Thanks Max!

On 19/04/17 03:44 AM, Max Gurtovoy wrote:
> Looks good,
> 
> Reviewed-by: Max Gurtovoy <maxg@mellanox.com>

[toc] | [prev] | [next] | [standalone]

#1626728

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

[toc] | [prev] | [next] | [standalone]

#1627377

Reviewed-by: Sagi Grimberg <sagi@grimberg.me>

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web