[PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[Vishal Verma]

The conversion of command sizes to unsigned missed a couple of checks
against variable size payloads during command validation, which made all
variable payload commands unconditionally fail. Add the checks back using
the new CXL_VARIABLE_PAYLOAD scheme.

Reported-by: Abhi Cs <abhi.cs@intel.com>
Fixes: 26f89535a5bb ("cxl/mbox: Use type __u32 for mailbox payload sizes")
Cc: <stable@vger.kernel.org>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
---
 drivers/cxl/core/mbox.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index 40e3ccb2bf3e..d929b89d12a7 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -355,12 +355,14 @@ static int cxl_to_mem_cmd(struct cxl_mem_command *mem_cmd,
 		return -EBUSY;
 
 	/* Check the input buffer is the expected size */
-	if (info->size_in != send_cmd->in.size)
-		return -ENOMEM;
+	if (info->size_in != CXL_VARIABLE_PAYLOAD)
+		if (info->size_in != send_cmd->in.size)
+			return -ENOMEM;
 
 	/* Check the output buffer is at least large enough */
-	if (send_cmd->out.size < info->size_out)
-		return -ENOMEM;
+	if (info->size_out != CXL_VARIABLE_PAYLOAD)
+		if (send_cmd->out.size < info->size_out)
+			return -ENOMEM;
 
 	*mem_cmd = (struct cxl_mem_command) {
 		.info = {

base-commit: 1985cf58850562e4b960e19d46f0d8f19d6c7cbd
-- 
2.36.1

Re: [PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[Alison Schofield]

On Tue, Jun 28, 2022 at 01:04:27PM -0700, Vishal Verma wrote:
> The conversion of command sizes to unsigned missed a couple of checks
> against variable size payloads during command validation, which made all
> variable payload commands unconditionally fail. Add the checks back using
> the new CXL_VARIABLE_PAYLOAD scheme.
> 
> Reported-by: Abhi Cs <abhi.cs@intel.com>
> Fixes: 26f89535a5bb ("cxl/mbox: Use type __u32 for mailbox payload sizes")
> Cc: <stable@vger.kernel.org>
> Cc: Ira Weiny <ira.weiny@intel.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Alison Schofield <alison.schofield@intel.com>
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>

with one caveat below...
Reviewed-by: Alison Schofield <alison.schofield@intel.com>


> ---
>  drivers/cxl/core/mbox.c | 10 ++++++----
>  1 file changed, 6 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> index 40e3ccb2bf3e..d929b89d12a7 100644
> --- a/drivers/cxl/core/mbox.c
> +++ b/drivers/cxl/core/mbox.c
> @@ -355,12 +355,14 @@ static int cxl_to_mem_cmd(struct cxl_mem_command *mem_cmd,
>  		return -EBUSY;
>  
>  	/* Check the input buffer is the expected size */
> -	if (info->size_in != send_cmd->in.size)
> -		return -ENOMEM;
> +	if (info->size_in != CXL_VARIABLE_PAYLOAD)
> +		if (info->size_in != send_cmd->in.size)
> +			return -ENOMEM;

We can leave it to Dan to arbitrate, but I don't think nested
if's without brackets follow kernel coding style.

However, Dan didn't like my nested if's with brackets either.
He'd prefer using &&

>  
>  	/* Check the output buffer is at least large enough */
> -	if (send_cmd->out.size < info->size_out)
> -		return -ENOMEM;
> +	if (info->size_out != CXL_VARIABLE_PAYLOAD)
> +		if (send_cmd->out.size < info->size_out)
> +			return -ENOMEM;
>  
>  	*mem_cmd = (struct cxl_mem_command) {
>  		.info = {
> 
> base-commit: 1985cf58850562e4b960e19d46f0d8f19d6c7cbd
> -- 
> 2.36.1
> 

Re: [PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[Verma, Vishal L]

On Tue, 2022-06-28 at 14:42 -0700, Alison Schofield wrote:
> On Tue, Jun 28, 2022 at 01:04:27PM -0700, Vishal Verma wrote:
> > The conversion of command sizes to unsigned missed a couple of
> > checks
> > against variable size payloads during command validation, which
> > made all
> > variable payload commands unconditionally fail. Add the checks back
> > using
> > the new CXL_VARIABLE_PAYLOAD scheme.
> > 
> > Reported-by: Abhi Cs <abhi.cs@intel.com>
> > Fixes: 26f89535a5bb ("cxl/mbox: Use type __u32 for mailbox payload
> > sizes")
> > Cc: <stable@vger.kernel.org>
> > Cc: Ira Weiny <ira.weiny@intel.com>
> > Cc: Dan Williams <dan.j.williams@intel.com>
> > Cc: Alison Schofield <alison.schofield@intel.com>
> > Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> 
> with one caveat below...
> Reviewed-by: Alison Schofield <alison.schofield@intel.com>

Thanks for the review!


> > ---
> >  drivers/cxl/core/mbox.c | 10 ++++++----
> >  1 file changed, 6 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> > index 40e3ccb2bf3e..d929b89d12a7 100644
> > --- a/drivers/cxl/core/mbox.c
> > +++ b/drivers/cxl/core/mbox.c
> > @@ -355,12 +355,14 @@ static int cxl_to_mem_cmd(struct
> > cxl_mem_command *mem_cmd,
> >                 return -EBUSY;
> >  
> >         /* Check the input buffer is the expected size */
> > -       if (info->size_in != send_cmd->in.size)
> > -               return -ENOMEM;
> > +       if (info->size_in != CXL_VARIABLE_PAYLOAD)
> > +               if (info->size_in != send_cmd->in.size)
> > +                       return -ENOMEM;
> 
> We can leave it to Dan to arbitrate, but I don't think nested
> if's without brackets follow kernel coding style.
> 
> However, Dan didn't like my nested if's with brackets either.
> He'd prefer using &&

Ha funny - I had && originally, but then I spotted nested if () a few
lines above in the same file in cxl_mbox_send_cmd(), and switched to
the same style :)

I'd be happy to change to &&.

> 
> >  
> >         /* Check the output buffer is at least large enough */
> > -       if (send_cmd->out.size < info->size_out)
> > -               return -ENOMEM;
> > +       if (info->size_out != CXL_VARIABLE_PAYLOAD)
> > +               if (send_cmd->out.size < info->size_out)
> > +                       return -ENOMEM;
> >  
> >         *mem_cmd = (struct cxl_mem_command) {
> >                 .info = {
> > 
> > base-commit: 1985cf58850562e4b960e19d46f0d8f19d6c7cbd
> > -- 
> > 2.36.1
> > 

RE: [PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[Dan Williams]

Vishal Verma wrote:
> The conversion of command sizes to unsigned missed a couple of checks
> against variable size payloads during command validation, which made all
> variable payload commands unconditionally fail. Add the checks back using
> the new CXL_VARIABLE_PAYLOAD scheme.

Ah, looks good. Need to get label read/write regression test into
cxl-cli post-haste.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>

Re: [PATCH] cxl/mbox: Fix missing variable payload checks in cmd size validation

[Dan Williams]

Verma, Vishal L wrote:
> On Tue, 2022-06-28 at 14:42 -0700, Alison Schofield wrote:
> > On Tue, Jun 28, 2022 at 01:04:27PM -0700, Vishal Verma wrote:
> > > The conversion of command sizes to unsigned missed a couple of
> > > checks
> > > against variable size payloads during command validation, which
> > > made all
> > > variable payload commands unconditionally fail. Add the checks back
> > > using
> > > the new CXL_VARIABLE_PAYLOAD scheme.
> > > 
> > > Reported-by: Abhi Cs <abhi.cs@intel.com>
> > > Fixes: 26f89535a5bb ("cxl/mbox: Use type __u32 for mailbox payload
> > > sizes")
> > > Cc: <stable@vger.kernel.org>
> > > Cc: Ira Weiny <ira.weiny@intel.com>
> > > Cc: Dan Williams <dan.j.williams@intel.com>
> > > Cc: Alison Schofield <alison.schofield@intel.com>
> > > Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> > 
> > with one caveat below...
> > Reviewed-by: Alison Schofield <alison.schofield@intel.com>
> 
> Thanks for the review!
> 
> 
> > > ---
> > >  drivers/cxl/core/mbox.c | 10 ++++++----
> > >  1 file changed, 6 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> > > index 40e3ccb2bf3e..d929b89d12a7 100644
> > > --- a/drivers/cxl/core/mbox.c
> > > +++ b/drivers/cxl/core/mbox.c
> > > @@ -355,12 +355,14 @@ static int cxl_to_mem_cmd(struct
> > > cxl_mem_command *mem_cmd,
> > >                 return -EBUSY;
> > >  
> > >         /* Check the input buffer is the expected size */
> > > -       if (info->size_in != send_cmd->in.size)
> > > -               return -ENOMEM;
> > > +       if (info->size_in != CXL_VARIABLE_PAYLOAD)
> > > +               if (info->size_in != send_cmd->in.size)
> > > +                       return -ENOMEM;
> > 
> > We can leave it to Dan to arbitrate, but I don't think nested
> > if's without brackets follow kernel coding style.
> > 
> > However, Dan didn't like my nested if's with brackets either.
> > He'd prefer using &&
> 
> Ha funny - I had && originally, but then I spotted nested if () a few
> lines above in the same file in cxl_mbox_send_cmd(), and switched to
> the same style :)
> 
> I'd be happy to change to &&.

Yeah, lets go that route.

/me thanks Alison for catching it.