linux-kernel - Re: [PATCH] ucounts: add missing data type changes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20210721124301.btbciip4biph2cxd@example.org>
Date:   Wed, 21 Jul 2021 14:43:01 +0200
From:   Alexey Gladkov <legion@...nel.org>
To:     Sven Schnelle <svens@...ux.ibm.com>
Cc:     "Eric W. Biederman" <ebiederm@...ssion.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] ucounts: add missing data type changes

On Wed, Jul 21, 2021 at 02:32:38PM +0200, Sven Schnelle wrote:
> Alexey Gladkov <legion@...nel.org> writes:
> 
> > On Wed, Jul 21, 2021 at 01:58:00PM +0200, Sven Schnelle wrote:
> >> commit f9c82a4ea89c3 ("Increase size of ucounts to atomic_long_t")
> >> changed the data type of ucounts/ucounts_max to long, but missed to
> >> adjust a few other places. This is noticeable on big endian platforms
> >> from user space because the /proc/sys/user/max_*_names files all
> >> contain 0.
> >
> > I didn't increase them because I didn't want to increase the maximum
> > number of namespaces.
> >
> > Your patch not only fixes the described problem, but also increases the
> > maximum number of namespaces.
> >
> >> Fixes: f9c82a4ea89c ("Increase size of ucounts to atomic_long_t")
> >> Signed-off-by: Sven Schnelle <svens@...ux.ibm.com>
> >> ---
> >>  fs/notify/fanotify/fanotify_user.c | 11 +++++++----
> >>  kernel/ucount.c                    | 18 ++++++++++--------
> >>  kernel/user_namespace.c            |  2 +-
> >>  3 files changed, 18 insertions(+), 13 deletions(-)
> >> 
> >> diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c
> >> index 64864fb40b40..995379ddce86 100644
> >> --- a/fs/notify/fanotify/fanotify_user.c
> >> +++ b/fs/notify/fanotify/fanotify_user.c
> >> @@ -54,22 +54,25 @@ static int fanotify_max_queued_events __read_mostly;
> >>  
> >>  #include <linux/sysctl.h>
> >>  
> >> +static unsigned long long_max = LONG_MAX;
> >
> > I think this should be INT_MAX.
> 
> Thanks. I'll change the code to use SYSCTL_INT_MAX instead and send a
> v2.
> 
> >>  struct ctl_table fanotify_table[] = {
> >>  	{
> >>  		.procname	= "max_user_groups",
> >>  		.data	= &init_user_ns.ucount_max[UCOUNT_FANOTIFY_GROUPS],
> >> -		.maxlen		= sizeof(int),
> >> +		.maxlen		= sizeof(long),
> >>  		.mode		= 0644,
> >> -		.proc_handler	= proc_dointvec_minmax,
> >> +		.proc_handler	= proc_doulongvec_minmax,
> >>  		.extra1		= SYSCTL_ZERO,
> >> +		.extra2		= &long_max,
> >>  	},
> >>  	{
> >>  		.procname	= "max_user_marks",
> >>  		.data	= &init_user_ns.ucount_max[UCOUNT_FANOTIFY_MARKS],
> >> -		.maxlen		= sizeof(int),
> >> +		.maxlen		= sizeof(long),
> >>  		.mode		= 0644,
> >> -		.proc_handler	= proc_dointvec_minmax,
> >> +		.proc_handler	= proc_doulongvec_minmax,
> >>  		.extra1		= SYSCTL_ZERO,
> >> +		.extra2		= &long_max,
> >>  	},
> >>  	{
> >>  		.procname	= "max_queued_events",
> >> diff --git a/kernel/ucount.c b/kernel/ucount.c
> >> index 87799e2379bd..681ea1b1a06f 100644
> >> --- a/kernel/ucount.c
> >> +++ b/kernel/ucount.c
> >> @@ -58,14 +58,16 @@ static struct ctl_table_root set_root = {
> >>  	.permissions = set_permissions,
> >>  };
> >>  
> >> -#define UCOUNT_ENTRY(name)				\
> >> -	{						\
> >> -		.procname	= name,			\
> >> -		.maxlen		= sizeof(int),		\
> >> -		.mode		= 0644,			\
> >> -		.proc_handler	= proc_dointvec_minmax,	\
> >> -		.extra1		= SYSCTL_ZERO,		\
> >> -		.extra2		= SYSCTL_INT_MAX,	\
> >> +static unsigned long long_max = LONG_MAX;
> >> +
> >> +#define UCOUNT_ENTRY(name)					\
> >> +	{							\
> >> +		.procname	= name,				\
> >> +		.maxlen		= sizeof(long),			\
> >> +		.mode		= 0644,				\
> >> +		.proc_handler	= proc_doulongvec_minmax,	\
> >> +		.extra1		= SYSCTL_ZERO,			\
> >> +		.extra2		= &long_max,			\
> >>  	}
> >>  static struct ctl_table user_table[] = {
> >>  	UCOUNT_ENTRY("max_user_namespaces"),
> >> diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c
> >> index ef82d401dde8..d8ad4c261253 100644
> >> --- a/kernel/user_namespace.c
> >> +++ b/kernel/user_namespace.c
> >> @@ -120,7 +120,7 @@ int create_user_ns(struct cred *new)
> >>  	ns->group = group;
> >>  	INIT_WORK(&ns->work, free_user_ns);
> >>  	for (i = 0; i < MAX_PER_NAMESPACE_UCOUNTS; i++) {
> >> -		ns->ucount_max[i] = INT_MAX;
> >> +		ns->ucount_max[i] = LONG_MAX;
> 
> Guess this should also stay at INT_MAX?

Yes.

> >>  	}
> >>  	set_rlimit_ucount_max(ns, UCOUNT_RLIMIT_NPROC, rlimit(RLIMIT_NPROC));
> >>  	set_rlimit_ucount_max(ns, UCOUNT_RLIMIT_MSGQUEUE, rlimit(RLIMIT_MSGQUEUE));
> >> -- 
> >> 2.25.1
> >> 
> 

-- 
Rgrds, legion