Linux CXL
 help / color / mirror / Atom feed
From: Dan Williams <dan.j.williams@intel.com>
To: Dave Jiang <dave.jiang@intel.com>, <linux-cxl@vger.kernel.org>
Cc: <dan.j.williams@intel.com>, <ira.weiny@intel.com>,
	<vishal.l.verma@intel.com>, <alison.schofield@intel.com>,
	<Jonathan.Cameron@huawei.com>, <dave@stgolabs.net>,
	<jgg@nvidia.com>, <shiju.jose@huawei.com>
Subject: Re: [PATCH v4 07/15] cxl: Add FWCTL support to CXL
Date: Fri, 7 Feb 2025 17:24:16 -0800	[thread overview]
Message-ID: <67a6b24080c9a_2d1e294ab@dwillia2-xfh.jf.intel.com.notmuch> (raw)
In-Reply-To: <20250207233914.2375110-8-dave.jiang@intel.com>

Dave Jiang wrote:
> Add fwctl support code to allow sending of CXL feature commands from
> userspace through as ioctls via FWCTL. Provide initial setup bits. The
> CXL PCI probe function will call cxl_setup_fwctl() after the cxl_memdev
> has been enumerated in order to setup FWCTL char device under the
> cxl_memdev like the existing memdev char device for issuing CXL raw
> mailbox commands from userspace via ioctls.
> 
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
> ---
>  drivers/cxl/core/features.c  | 74 ++++++++++++++++++++++++++++++++++++
>  drivers/cxl/pci.c            |  4 ++
>  include/cxl/features.h       | 14 +++++++
>  include/uapi/fwctl/fwctl.h   |  1 +
>  tools/testing/cxl/test/mem.c |  4 ++
>  5 files changed, 97 insertions(+)
> 
> diff --git a/drivers/cxl/core/features.c b/drivers/cxl/core/features.c
> index 82f21f64452a..81e8ff66c12e 100644
> --- a/drivers/cxl/core/features.c
> +++ b/drivers/cxl/core/features.c
> @@ -1,5 +1,6 @@
>  // SPDX-License-Identifier: GPL-2.0-only
>  /* Copyright(c) 2024-2025 Intel Corporation. All rights reserved. */
> +#include <linux/fwctl.h>
>  #include <linux/device.h>
>  #include <cxl/mailbox.h>
>  #include <cxl/features.h>
> @@ -167,6 +168,13 @@ static void free_cxlfs(void *_cxlfs)
>  
>  	cxlds->cxlfs = NULL;
>  	kvfree(cxlfs->entries);
> +
> +	if (cxlfs->cxl_fwctl) {
> +		struct cxl_fwctl *fwctl = cxlfs->cxl_fwctl;
> +
> +		fwctl_unregister(&fwctl->fwctl_dev);
> +		fwctl_put(&fwctl->fwctl_dev);
> +	}

I think this is going to lead to use after free bugs.

Consider that this free_cxlfs() call is a devm action that was
registered *before* devm_cxl_add_memdev(). That means that the memdev
gets destroyed before fwctl is unregistered. That sounds like a recipe
for use after free bugs because userspace could still be using the fwctl
interface after the memdev has been unregistered.

More below...

[..]
> +DEFINE_FREE(free_fwctl, struct cxl_fwctl *, if (_T) fwctl_put(&_T->fwctl_dev))
> +
> +int cxl_setup_fwctl(struct cxl_memdev *cxlmd)

This needs to be devm_...

> +{
> +	struct cxl_dev_state *cxlds = cxlmd->cxlds;
> +	struct cxl_features_state *cxlfs;
> +	int rc;
> +
> +	cxlfs = to_cxlfs(cxlds);
> +	if (!cxlfs)
> +		return -ENODEV;
> +
> +	/* No need to setup FWCTL if there are no user allowed features found */
> +	if (!cxlfs->entries->num_user_features)
> +		return -ENODEV;
> +
> +	struct cxl_fwctl *fwctl __free(free_fwctl) =
> +		fwctl_alloc_device(&cxlmd->dev, &cxlctl_ops,
> +				   struct cxl_fwctl, fwctl_dev);
> +	if (!fwctl)
> +		return -ENOMEM;
> +
> +	fwctl->cxlmd = cxlmd;
> +	rc = fwctl_register(&fwctl->fwctl_dev);
> +	if (rc)
> +		return rc;

...and a devm action needs to be registered here to make sure fwctl is
torn down before the memdev is torn down.

> +
> +	cxlfs->cxl_fwctl = no_free_ptr(fwctl);
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_NS_GPL(cxl_setup_fwctl, "CXL");
> +
> +MODULE_IMPORT_NS("FWCTL");
> diff --git a/drivers/cxl/pci.c b/drivers/cxl/pci.c
> index 3e666ec51580..b093cb16de3e 100644
> --- a/drivers/cxl/pci.c
> +++ b/drivers/cxl/pci.c
> @@ -1013,6 +1013,10 @@ static int cxl_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>  	if (rc)
>  		return rc;
>  
> +	rc = cxl_setup_fwctl(cxlmd);
> +	if (rc)
> +		dev_dbg(&pdev->dev, "No CXL FWCTL setup\n");
> +
>  	pmu_count = cxl_count_regblock(pdev, CXL_REGLOC_RBI_PMU);
>  	if (pmu_count < 0)
>  		return pmu_count;
> diff --git a/include/cxl/features.h b/include/cxl/features.h
> index 1ab97e676c03..d0c94756e452 100644
> --- a/include/cxl/features.h
> +++ b/include/cxl/features.h
> @@ -4,6 +4,7 @@
>  #define __CXL_FEATURES_H__
>  
>  #include <linux/uuid.h>
> +#include <linux/fwctl.h>
>  
>  /* Feature UUIDs used by the kernel */
>  #define CXL_FEAT_PATROL_SCRUB_UUID						\
> @@ -158,6 +159,9 @@ enum cxl_set_feat_flag_data_transfer {
>   * @entries: CXl feature entry context
>   *	@num_features: total Features supported by the device
>   *	@ent: Flex array of Feature detail entries from the device
> + * @fwctl: CXL Firmware Control context
> + *	@fwctl_dev: Firmware Control device
> + *	@cxlfs: Pointer to CXL Features state
>   */
>  struct cxl_features_state {
>  	struct cxl_dev_state *cxlds;
> @@ -166,12 +170,17 @@ struct cxl_features_state {
>  		int num_user_features;
>  		struct cxl_feat_entry ent[] __counted_by(num_features);
>  	} *entries;
> +	struct cxl_fwctl {
> +		struct fwctl_device fwctl_dev;
> +		struct cxl_memdev *cxlmd;

Is this backpointer necessary?

Just look it up from the fwctl dev.

static inline struct cxl_memdev *fwctl_to_memdev(struct fwctl_device *fwctl_dev)
{
	return to_cxl_memdev(fwctl_dev->dev.parent);
}

...then no need for 'struct cxl_fwctl' definition.

  reply	other threads:[~2025-02-08  1:24 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-02-07 23:37 [PATCH v4 00/15] cxl: Add CXL feature commands support via fwctl Dave Jiang
2025-02-07 23:37 ` [PATCH v4 01/15] cxl: Enumerate feature commands Dave Jiang
2025-02-07 23:47   ` Dan Williams
2025-02-08  6:00   ` Li Ming
2025-02-11 17:05   ` Jonathan Cameron
2025-02-07 23:37 ` [PATCH v4 02/15] cxl: Add Get Supported Features command for kernel usage Dave Jiang
2025-02-07 23:56   ` Dan Williams
2025-02-10 17:03     ` Dave Jiang
2025-02-08  6:13   ` Li Ming
2025-02-10 17:06     ` Dave Jiang
2025-02-07 23:37 ` [PATCH v4 03/15] cxl/test: Add Get Supported Features mailbox command support Dave Jiang
2025-02-07 23:37 ` [PATCH v4 04/15] cxl/mbox: Add GET_FEATURE mailbox command Dave Jiang
2025-02-08  0:12   ` Dan Williams
2025-02-07 23:37 ` [PATCH v4 05/15] cxl/mbox: Add SET_FEATURE " Dave Jiang
2025-02-08  0:21   ` Dan Williams
2025-02-08  6:17   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 06/15] cxl: Setup exclusive CXL features that are reserved for the kernel Dave Jiang
2025-02-08  1:10   ` Dan Williams
2025-02-08  1:16     ` Dave Jiang
2025-02-08  6:22   ` Li Ming
2025-02-10 17:40     ` Dave Jiang
2025-02-10 17:50     ` Dave Jiang
2025-02-11  5:46       ` Li Ming
2025-02-11 16:13         ` Dave Jiang
2025-02-07 23:37 ` [PATCH v4 07/15] cxl: Add FWCTL support to CXL Dave Jiang
2025-02-08  1:24   ` Dan Williams [this message]
2025-02-07 23:37 ` [PATCH v4 08/15] cxl: Add support for FWCTL get driver information callback Dave Jiang
2025-02-08  1:25   ` Dan Williams
2025-02-10  6:21   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 09/15] cxl: Move cxl feature command structs to user header Dave Jiang
2025-02-07 23:37 ` [PATCH v4 10/15] cxl: Add support for fwctl RPC command to enable CXL feature commands Dave Jiang
2025-02-08  1:29   ` Dan Williams
2025-02-08  1:34   ` Dan Williams
2025-02-08  7:09   ` Li Ming
2025-02-10 22:31     ` Dave Jiang
2025-02-07 23:37 ` [PATCH v4 11/15] cxl: Add support to handle user feature commands for get feature Dave Jiang
2025-02-10  6:41   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 12/15] cxl: Add support to handle user feature commands for set feature Dave Jiang
2025-02-10  6:43   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 13/15] cxl/test: Add Get Feature support to cxl_test Dave Jiang
2025-02-10  6:44   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 14/15] cxl/test: Add Set " Dave Jiang
2025-02-10  6:44   ` Li Ming
2025-02-07 23:37 ` [PATCH v4 15/15] fwctl/cxl: Add documentation to FWCTL CXL Dave Jiang
2025-02-08  1:38   ` Dan Williams

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=67a6b24080c9a_2d1e294ab@dwillia2-xfh.jf.intel.com.notmuch \
    --to=dan.j.williams@intel.com \
    --cc=Jonathan.Cameron@huawei.com \
    --cc=alison.schofield@intel.com \
    --cc=dave.jiang@intel.com \
    --cc=dave@stgolabs.net \
    --cc=ira.weiny@intel.com \
    --cc=jgg@nvidia.com \
    --cc=linux-cxl@vger.kernel.org \
    --cc=shiju.jose@huawei.com \
    --cc=vishal.l.verma@intel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox