Re: [PATCH v2 5/6] sysctl/infiniband: Fixes infiniband sysctl bounds
From: Leon Romanovsky
Date: Mon Mar 03 2025 - 13:54:39 EST
On Mon, Mar 03, 2025 at 02:57:29PM +0100, Joel Granados wrote:
> On Mon, Feb 24, 2025 at 03:41:05PM +0200, Leon Romanovsky wrote:
> > On Mon, Feb 24, 2025 at 10:58:20AM +0100, nicolas.bouchinet@xxxxxxxxxxx wrote:
> > > From: Nicolas Bouchinet <nicolas.bouchinet@xxxxxxxxxxx>
> > >
> > > Bound infiniband iwcm and ucma sysctl writings between SYSCTL_ZERO
> > > and SYSCTL_INT_MAX.
> > >
> > > The proc_handler has thus been updated to proc_dointvec_minmax.
> > >
> > > Signed-off-by: Nicolas Bouchinet <nicolas.bouchinet@xxxxxxxxxxx>
> > > ---
> > > drivers/infiniband/core/iwcm.c | 4 +++-
> > > drivers/infiniband/core/ucma.c | 4 +++-
> > > 2 files changed, 6 insertions(+), 2 deletions(-)
> > >
> >
> > Acked-by: Leon Romanovsky <leon@xxxxxxxxxx>
> >
> > How do you want to proceed from here? Should I take to RDMA repository?
> >
> > Thanks
> It would be great if we push this through RDMA. Here are a few comments:
> 1. Having the upper bound be SYSCTL_INT_MAX is not necessary (as it is
> silently capped by proc_dointvec_minmax, but it is good to have as it
> gives understanding on what the spread of the values are.
>
> 2. Having the lower bound capped by SYSCTL_ZERO is needed as it will
> prevent ppl trying to assing negative values to the unsigned integers
>
> Please let me know if you will push this through RDMA, so I know to
> remove it from sysctl.
Applied to RDMA tree.
https://web.git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma.git/commit/?h=wip/leon-for-next&id=f33cd9b3fd03a791296ab37550ffd26213a90c4e
>
> Best
>
> Reviewed-by: Joel Granados <joel.granados@xxxxxxxxxx>
Thanks
>
> --
>
> Joel Granados