[PATCH 1/1] staging: media: atomisp: Disallow all private IOCTLSs but S_PARAMETERS

[PATCH 1/1] staging: media: atomisp: Disallow all private IOCTLSs but S_PARAMETERS

[Sakari Ailus]

Disallow all private IOCTLs except ATOMISP_IOC_S_PARAMETERS. These aren't
quite as safe as one could assume of IOCTL handlers; disable them for now.
Instead of removing the code, simply check the IOCTL matches with
S_PARAMETERS and bail out otherwise, in order to ease backporting.

In particular, ATOMISP_IOC_S_ISP_FPN_TABLE is disabled by this patch.

Reported-by: Soufiane Dani <soufianeda@tutanota.com>
Closes: https://lore.kernel.org/linux-staging/20260210-atomisp-fix-v1-1-024429cbff31@tutanota.com/
Cc: stable@vger.kernel.org
Fixes: ad85094b293e ("Revert "media: staging: atomisp: Remove driver"")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 drivers/staging/media/atomisp/pci/atomisp_ioctl.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/staging/media/atomisp/pci/atomisp_ioctl.c b/drivers/staging/media/atomisp/pci/atomisp_ioctl.c
index a7f22de1c889..78e14c9aeed8 100644
--- a/drivers/staging/media/atomisp/pci/atomisp_ioctl.c
+++ b/drivers/staging/media/atomisp/pci/atomisp_ioctl.c
@@ -1356,6 +1356,10 @@ static int atomisp_s_parm(struct file *file, void *fh,
 static long atomisp_vidioc_default(struct file *file, void *fh,
 				   bool valid_prio, unsigned int cmd, void *arg)
 {
+	/* Only allow ATOMISP_IOC_S_PARAMETERS for now! */
+	if (cmd != ATOMISP_IOC_S_PARAMETERS)
+		return -EINVAL;
+
 	struct video_device *vdev = video_devdata(file);
 	struct atomisp_sub_device *asd = atomisp_to_video_pipe(vdev)->asd;
 	int err;
-- 
2.47.3

Re: [PATCH 1/1] staging: media: atomisp: Disallow all private IOCTLSs but S_PARAMETERS

[Andy Shevchenko]

On Fri, Feb 27, 2026 at 11:23:27AM +0200, Sakari Ailus wrote:
> Disallow all private IOCTLs except ATOMISP_IOC_S_PARAMETERS. These aren't
> quite as safe as one could assume of IOCTL handlers; disable them for now.
> Instead of removing the code, simply check the IOCTL matches with
> S_PARAMETERS and bail out otherwise, in order to ease backporting.

> In particular, ATOMISP_IOC_S_ISP_FPN_TABLE is disabled by this patch.

Why not removing the code?
It's still and will be in the Git index, it won't disappear.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 1/1] staging: media: atomisp: Disallow all private IOCTLSs but S_PARAMETERS

[Sakari Ailus]

Hi Andy,

On Fri, Feb 27, 2026 at 11:31:29AM +0200, Andy Shevchenko wrote:
> On Fri, Feb 27, 2026 at 11:23:27AM +0200, Sakari Ailus wrote:
> > Disallow all private IOCTLs except ATOMISP_IOC_S_PARAMETERS. These aren't
> > quite as safe as one could assume of IOCTL handlers; disable them for now.
> > Instead of removing the code, simply check the IOCTL matches with
> > S_PARAMETERS and bail out otherwise, in order to ease backporting.
> 
> > In particular, ATOMISP_IOC_S_ISP_FPN_TABLE is disabled by this patch.
> 
> Why not removing the code?
> It's still and will be in the Git index, it won't disappear.

There have been changes in the IOCTL handler over the supported kernel
versions. This patch can be backported all the way down to v5.10.

I'm fine with removing the now-redundant code, but I'd rather do it
separtely from fixing the issue at hand.

-- 
Regards,

Sakari Ailus

Re: [PATCH 1/1] staging: media: atomisp: Disallow all private IOCTLSs but S_PARAMETERS

[Andy Shevchenko]

On Fri, Feb 27, 2026 at 12:54:26PM +0200, Sakari Ailus wrote:
> On Fri, Feb 27, 2026 at 11:31:29AM +0200, Andy Shevchenko wrote:
> > On Fri, Feb 27, 2026 at 11:23:27AM +0200, Sakari Ailus wrote:
> > > Disallow all private IOCTLs except ATOMISP_IOC_S_PARAMETERS. These aren't
> > > quite as safe as one could assume of IOCTL handlers; disable them for now.
> > > Instead of removing the code, simply check the IOCTL matches with
> > > S_PARAMETERS and bail out otherwise, in order to ease backporting.
> > 
> > > In particular, ATOMISP_IOC_S_ISP_FPN_TABLE is disabled by this patch.
> > 
> > Why not removing the code?
> > It's still and will be in the Git index, it won't disappear.
> 
> There have been changes in the IOCTL handler over the supported kernel
> versions. This patch can be backported all the way down to v5.10.
> 
> I'm fine with removing the now-redundant code, but I'd rather do it
> separtely from fixing the issue at hand.

Fair enough,
Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com>

-- 
With Best Regards,
Andy Shevchenko