Re: [RESEND PATCH] sys_personality: Add optional arch hook arch_check_personality

[Dominik Brodowski <linux@dominikbrodowski.net>]

On Mon, Jun 08, 2020 at 10:46:41AM +0100, Catalin Marinas wrote:
> On Mon, Jun 08, 2020 at 10:49:25AM +0800, Wang ShaoBo wrote:
> > Currently arm64 personality syscall uses wrapper __arm64_sys_personality
> > to redirect to __arm64_sys_arm64_personality, it's easily confused,
> > Whereas using an normal hook arch_check_personality() can reject
> > additional settings like this for special case of different architectures.
> > 
> > This makes code clean and easier for subsequent modification.
> 
> Do you plan to add more stuff here? Curious what triggered this patch.
> 
> > diff --git a/arch/arm64/kernel/sys.c b/arch/arm64/kernel/sys.c
> > index d5ffaaab31a7..5c01816d7a77 100644
> > --- a/arch/arm64/kernel/sys.c
> > +++ b/arch/arm64/kernel/sys.c
> > @@ -28,12 +28,13 @@ SYSCALL_DEFINE6(mmap, unsigned long, addr, unsigned long, len,
> >  	return ksys_mmap_pgoff(addr, len, prot, flags, fd, off >> PAGE_SHIFT);
> >  }
> >  
> > -SYSCALL_DEFINE1(arm64_personality, unsigned int, personality)
> > +int arch_check_personality(unsigned int personality)
> >  {
> >  	if (personality(personality) == PER_LINUX32 &&
> >  		!system_supports_32bit_el0())
> >  		return -EINVAL;
> > -	return ksys_personality(personality);
> > +
> > +	return 0;
> >  }
> 
> We use the ksys_* pattern in other places as well, so this wouldn't be
> something new.
> 
> > diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
> > index 1815065d52f3..3dbbad498027 100644
> > --- a/include/linux/syscalls.h
> > +++ b/include/linux/syscalls.h
> > @@ -1393,16 +1393,6 @@ static inline long ksys_truncate(const char __user *pathname, loff_t length)
> >  	return do_sys_truncate(pathname, length);
> >  }
> >  
> > -static inline unsigned int ksys_personality(unsigned int personality)
> > -{
> > -	unsigned int old = current->personality;
> > -
> > -	if (personality != 0xffffffff)
> > -		set_personality(personality);
> > -
> > -	return old;
> > -}
> > -
> >  /* for __ARCH_WANT_SYS_IPC */
> >  long ksys_semtimedop(int semid, struct sembuf __user *tsops,
> >  		     unsigned int nsops,
> > diff --git a/kernel/exec_domain.c b/kernel/exec_domain.c
> > index 33f07c5f2515..f3682f4bf205 100644
> > --- a/kernel/exec_domain.c
> > +++ b/kernel/exec_domain.c
> > @@ -35,9 +35,21 @@ static int __init proc_execdomains_init(void)
> >  module_init(proc_execdomains_init);
> >  #endif
> >  
> > +int __weak arch_check_personality(unsigned int personality)
> > +{
> > +	return 0;
> > +}
> > +
> >  SYSCALL_DEFINE1(personality, unsigned int, personality)
> >  {
> > -	unsigned int old = current->personality;
> > +	int err;
> > +	unsigned int old;
> > +
> > +	err = arch_check_personality(personality);
> > +	if (err)
> > +		return err;
> > +
> > +	old = current->personality;
> 
> I'm surprised that the generic sys_personality() doesn't call
> ksys_personality() directly but rather duplicates the code.

It was the other way round, and the duplication is based on a
suggestion by Christoph Hellwig IIRC,
	https://lore.kernel.org/lkml/20180514120756.GA11638@infradead.org/

Thanks,
	Dominik