On Wed, Jun 01, 2022 at 08:19:14PM +0100, Matthew Wilcox wrote:
> On Wed, Jun 01, 2022 at 03:20:29PM +0200, Alexey Gladkov wrote:
> > +struct ctl_fops {
> > + int (*open) (struct ctl_context *, struct inode *, struct file *);
> > + int (*release) (struct ctl_context *, struct inode *, struct file *);
> > + ssize_t (*read) (struct ctl_context *, struct file *, char *, size_t *, loff_t *);
> > + ssize_t (*write) (struct ctl_context *, struct file *, char *, size_t *, loff_t *);
> > +};
>
> Why not pass the iocb in ->read and ->write? We're still regretting not
> doing that with file_operations.
Because buf and len can be modified in BPF_CGROUP_RUN_PROG_SYSCTL.
We need to pass the result of this hook to read/write callback.
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/proc/proc_sysctl.c#n605
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/bpf/cgroup.c#n1441
--
Rgrds, legion
