linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Catalin Marinas <catalin.marinas@arm.com>
To: Florent Revest <revest@chromium.org>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
	akpm@linux-foundation.org, anshuman.khandual@arm.com,
	joey.gouly@arm.com, mhocko@suse.com, keescook@chromium.org,
	david@redhat.com, peterx@redhat.com, izbyshev@ispras.ru,
	broonie@kernel.org, szabolcs.nagy@arm.com, kpsingh@kernel.org,
	gthelen@google.com, toiwoton@gmail.com
Subject: Re: [PATCH v2 4/5] mm: Add a NO_INHERIT flag to the PR_SET_MDWE prctl
Date: Tue, 23 May 2023 17:36:00 +0100	[thread overview]
Message-ID: <ZGzrcOeXc/slVpCY@arm.com> (raw)
In-Reply-To: <20230517150321.2890206-5-revest@chromium.org>

On Wed, May 17, 2023 at 05:03:20PM +0200, Florent Revest wrote:
> diff --git a/include/linux/sched/coredump.h b/include/linux/sched/coredump.h
> index 0ee96ea7a0e9..11f5e3dacb4e 100644
> --- a/include/linux/sched/coredump.h
> +++ b/include/linux/sched/coredump.h
> @@ -91,4 +91,14 @@ static inline int get_dumpable(struct mm_struct *mm)
>  				 MMF_DISABLE_THP_MASK | MMF_HAS_MDWE_MASK)
>  
>  #define MMF_VM_MERGE_ANY	29
> +#define MMF_HAS_MDWE_NO_INHERIT	30
> +
> +#define MMF_INIT_FLAGS(flags)	({					\
> +	unsigned long new_flags = flags;				\
> +	if (new_flags & (1UL << MMF_HAS_MDWE_NO_INHERIT))		\
> +		new_flags &= ~((1UL << MMF_HAS_MDWE) |			\
> +				(1UL << MMF_HAS_MDWE_NO_INHERIT));	\
> +	new_flags & MMF_INIT_MASK;					\
> +})

A function is better indeed, not sure who came up with this macro idea ;).

> diff --git a/kernel/sys.c b/kernel/sys.c
> index 339fee3eff6a..320eae3b12ab 100644
> --- a/kernel/sys.c
> +++ b/kernel/sys.c
> @@ -2368,9 +2368,25 @@ static inline int prctl_set_mdwe(unsigned long bits, unsigned long arg3,
>  	if (arg3 || arg4 || arg5)
>  		return -EINVAL;
>  
> -	if (bits & ~(PR_MDWE_REFUSE_EXEC_GAIN))
> +	if (bits & ~(PR_MDWE_REFUSE_EXEC_GAIN | PR_MDWE_NO_INHERIT))
>  		return -EINVAL;
>  
> +	/* NO_INHERIT only makes sense with REFUSE_EXEC_GAIN */
> +	if (bits & PR_MDWE_NO_INHERIT && !(bits & PR_MDWE_REFUSE_EXEC_GAIN))
> +		return -EINVAL;
> +
> +	/* Can't gain NO_INHERIT from !NO_INHERIT */
> +	if (bits & PR_MDWE_NO_INHERIT &&
> +	    test_bit(MMF_HAS_MDWE, &current->mm->flags) &&
> +	    !test_bit(MMF_HAS_MDWE_NO_INHERIT, &current->mm->flags))
> +		return -EPERM;
> +
> +	if (bits & PR_MDWE_NO_INHERIT)
> +		set_bit(MMF_HAS_MDWE_NO_INHERIT, &current->mm->flags);
> +	else if (test_bit(MMF_HAS_MDWE_NO_INHERIT, &current->mm->flags)
> +		 && !(bits & PR_MDWE_REFUSE_EXEC_GAIN))
> +		return -EPERM; /* Cannot unset the flag */

Is this about not unsetting the MMF_HAS_MDWE bit? We already have a
check further down that covers this case.

Related to this, do we want to allow unsetting MMF_HAS_MDWE_NO_INHERIT?
It looks like it can't be unset but no error either. The above check,
IIUC, looks more like ensuring we don't clear MMF_HAS_MDWE.

Maybe we should tighten the logic here a bit and not allow any changes
after the initial flag setting:

current->mm->flags == 0, we allow:
	bits == 0 or
	bits == PR_MDWE_REFUSE_EXEC_GAIN or
	bits == PR_MDWE_REFUSE_EXEC_GAIN | PR_MDWE_NO_INHERIT

current->mm->flags != 0 (some bits were set), we only allow the exactly
the same bit combination or -EPERM.

So basically build the flags based on the PR_* input bits and compare
them with current->mm->flags when not 0, return -EPERM if different. I
think this preserves the ABI as we only have a single bit currently and
hopefully makes the logic here easier to parse.

> +
>  	if (bits & PR_MDWE_REFUSE_EXEC_GAIN)
>  		set_bit(MMF_HAS_MDWE, &current->mm->flags);
>  	else if (test_bit(MMF_HAS_MDWE, &current->mm->flags))
> @@ -2385,8 +2401,10 @@ static inline int prctl_get_mdwe(unsigned long arg2, unsigned long arg3,
>  	if (arg2 || arg3 || arg4 || arg5)
>  		return -EINVAL;
>  
> -	return test_bit(MMF_HAS_MDWE, &current->mm->flags) ?
> -		PR_MDWE_REFUSE_EXEC_GAIN : 0;
> +	return (test_bit(MMF_HAS_MDWE, &current->mm->flags) ?
> +		PR_MDWE_REFUSE_EXEC_GAIN : 0) |
> +		(test_bit(MMF_HAS_MDWE_NO_INHERIT, &current->mm->flags) ?
> +		PR_MDWE_NO_INHERIT : 0);
>  }

Just personal preference, use explicit 'if' blocks and add bits to a
local variable variable than multiple ternary operators.

-- 
Catalin


  parent reply	other threads:[~2023-05-23 16:36 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-05-17 15:03 [PATCH v2 0/5] MDWE without inheritance Florent Revest
2023-05-17 15:03 ` [PATCH v2 1/5] kselftest: vm: Fix tabs/spaces inconsistency in the mdwe test Florent Revest
2023-05-22  8:52   ` David Hildenbrand
2023-05-17 15:03 ` [PATCH v2 2/5] kselftest: vm: Fix mdwe's mmap_FIXED test case Florent Revest
2023-05-22  8:53   ` David Hildenbrand
2023-05-17 15:03 ` [PATCH v2 3/5] mm: Make PR_MDWE_REFUSE_EXEC_GAIN an unsigned long Florent Revest
2023-05-22  8:55   ` David Hildenbrand
     [not found]     ` <884d131bbc28ebfa0b729176e6415269@ispras.ru>
2023-05-22 16:22       ` David Hildenbrand
     [not found]         ` <3c2e210b75bd56909322e8a3e5086d91@ispras.ru>
2023-05-23  9:12           ` David Hildenbrand
2023-05-23 13:07             ` Catalin Marinas
     [not found]               ` <f47d587fe5a6285f88191fbb13f367c7@ispras.ru>
2023-05-23 14:09                 ` Catalin Marinas
2023-05-23 15:01                   ` Szabolcs Nagy
     [not found]             ` <7c572622c0d8e283fc880fe3f4ffac27@ispras.ru>
2023-05-23 14:10               ` David Hildenbrand
2023-05-26 19:04                 ` Florent Revest
2023-05-26 19:02           ` Florent Revest
2023-05-23 14:11   ` Catalin Marinas
2023-05-17 15:03 ` [PATCH v2 4/5] mm: Add a NO_INHERIT flag to the PR_SET_MDWE prctl Florent Revest
2023-05-22  9:01   ` David Hildenbrand
2023-05-22 16:11     ` Florent Revest
2023-05-23 16:36   ` Catalin Marinas [this message]
2023-05-26 19:05     ` Florent Revest
2023-05-17 15:03 ` [PATCH v2 5/5] kselftest: vm: Add tests for no-inherit memory-deny-write-execute Florent Revest

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ZGzrcOeXc/slVpCY@arm.com \
    --to=catalin.marinas@arm.com \
    --cc=akpm@linux-foundation.org \
    --cc=anshuman.khandual@arm.com \
    --cc=broonie@kernel.org \
    --cc=david@redhat.com \
    --cc=gthelen@google.com \
    --cc=izbyshev@ispras.ru \
    --cc=joey.gouly@arm.com \
    --cc=keescook@chromium.org \
    --cc=kpsingh@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mhocko@suse.com \
    --cc=peterx@redhat.com \
    --cc=revest@chromium.org \
    --cc=szabolcs.nagy@arm.com \
    --cc=toiwoton@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox