Re: [RFC v3 2/2] kernel/sysctl: support handling command line aliases

[Christian Brauner]

On Thu, Mar 26, 2020 at 01:34:08PM -0700, Kees Cook wrote:
> On Thu, Mar 26, 2020 at 07:16:06PM +0100, Vlastimil Babka wrote:
> > We can now handle sysctl parameters on kernel command line, but historically
> > some parameters introduced their own command line equivalent, which we don't
> > want to remove for compatibility reasons. We can however convert them to the
> > generic infrastructure with a table translating the legacy command line
> > parameters to their sysctl names, and removing the one-off param handlers.
> > 
> > This patch adds the support and makes the first conversion to demonstrate it,
> > on the (deprecated) numa_zonelist_order parameter.
> > 
> > Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> > ---
> > Changes in v3:
> > - constify some things according to Kees
> > - expand the comment of sysctl_aliases to note on different timings
> > 
> >  fs/proc/proc_sysctl.c | 48 ++++++++++++++++++++++++++++++++++++-------
> >  mm/page_alloc.c       |  9 --------
> >  2 files changed, 41 insertions(+), 16 deletions(-)
> > 
> > diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
> > index 8ee3273e4540..3a861e0a7c7e 100644
> > --- a/fs/proc/proc_sysctl.c
> > +++ b/fs/proc/proc_sysctl.c
> > @@ -1729,6 +1729,37 @@ int __init proc_sys_init(void)
> >  
> >  struct vfsmount *proc_mnt = NULL;
> >  
> > +struct sysctl_alias {
> > +	const char *kernel_param;
> > +	const char *sysctl_param;
> > +};
> > +
> > +/*
> > + * Historically some settings had both sysctl and a command line parameter.
> > + * With the generic sysctl. parameter support, we can handle them at a single
> > + * place and only keep the historical name for compatibility. This is not meant
> > + * to add brand new aliases. When adding existing aliases, consider whether
> > + * the possibly different moment of changing the value (e.g. from early_param
> > + * to the moment do_sysctl_args() is called) is an issue for the specific
> > + * parameter.
> > + */
> > +static const struct sysctl_alias sysctl_aliases[] = {
> > +	{"numa_zonelist_order",		"vm.numa_zonelist_order" },
> > +	{ }
> > +};
> > +
> > +const char *sysctl_find_alias(char *param)
> 
> This should be "static" too.
> 
> > +{
> > +	const struct sysctl_alias *alias;
> > +
> > +	for (alias = &sysctl_aliases[0]; alias->kernel_param != NULL; alias++) {
> > +		if (strcmp(alias->kernel_param, param) == 0)
> > +			return alias->sysctl_param;
> > +	}
> > +
> > +	return NULL;
> > +}
> > +
> >  /* Set sysctl value passed on kernel command line. */
> >  static int process_sysctl_arg(char *param, char *val,
> >  			       const char *unused, void *arg)
> > @@ -1741,15 +1772,18 @@ static int process_sysctl_arg(char *param, char *val,
> >  	loff_t pos = 0;
> >  	ssize_t wret;
> >  
> > -	if (strncmp(param, "sysctl", sizeof("sysctl") - 1))

Somewhat off-topic but in some projects I use macro-helpers to make that
more transparent. So this would become a constant expression:

#define STRLITERALLEN(x) (sizeof(""x"") - 1)
or
#define STRLEN(x) (sizeof(""x"") - 1)

But I guess that's a matter of style/taste.

Christian