Re: [PATCH] cxl/mem: Fix an error code in cxl_mem_mbox_get()

From: Dan Williams <dan.j.williams@intel.com>
To: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Alison Schofield <alison.schofield@intel.com>,
	Ben Widawsky <ben.widawsky@intel.com>,
	Vishal Verma <vishal.l.verma@intel.com>,
	Ira Weiny <ira.weiny@intel.com>,
	linux-cxl@vger.kernel.org,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	kernel-janitors@vger.kernel.org,
	Colin King <colin.king@canonical.com>
Subject: Re: [PATCH] cxl/mem: Fix an error code in cxl_mem_mbox_get()
Date: Tue, 16 Feb 2021 12:52:46 -0800	[thread overview]
Message-ID: <CAPcyv4jKL7J0XoXE=uz-DOowYg888A5ajhQUxuFC2x8x5JoL3g@mail.gmail.com> (raw)
In-Reply-To: <YCwaIGPd9ktMoYPu@mwanda>

On Tue, Feb 16, 2021 at 11:17 AM Dan Carpenter <dan.carpenter@oracle.com> wrote:
>
> Smatch complains that sometimes when we return success we are holding
> the mutex and sometimes we have released the mutex.  It turns out that
> the bug is a missing error code if the mbox is not ready.  It should
> return -EBUSY instead of success.
>
> Fixes: cc1967ac93ab ("cxl/mem: Find device capabilities")
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> ---
>  drivers/cxl/mem.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/cxl/mem.c b/drivers/cxl/mem.c
> index 3bca8451348a..2ebc84e4d202 100644
> --- a/drivers/cxl/mem.c
> +++ b/drivers/cxl/mem.c
> @@ -383,8 +383,8 @@ static int __cxl_mem_mbox_send_cmd(struct cxl_mem *cxlm,
>  static int cxl_mem_mbox_get(struct cxl_mem *cxlm)
>  {
>         struct device *dev = &cxlm->pdev->dev;
> -       int rc = -EBUSY;
>         u64 md_status;
> +       int rc;
>
>         mutex_lock_io(&cxlm->mbox_mutex);
>
> @@ -414,6 +414,7 @@ static int cxl_mem_mbox_get(struct cxl_mem *cxlm)
>         md_status = readq(cxlm->memdev_regs + CXLMDEV_STATUS_OFFSET);
>         if (!(md_status & CXLMDEV_MBOX_IF_READY && CXLMDEV_READY(md_status))) {
>                 dev_err(dev, "mbox: reported doorbell ready, but not mbox ready\n");
> +               rc = -EBUSY;
>                 goto out;

Hey, Smatch gave a more thorough report than Coverity:

http://lore.kernel.org/r/20210215195313.90231-1-colin.king@canonical.com