mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
To: Dmitry Vyukov <dvyukov@google.com>,
	peterz@infradead.org, boqun.feng@gmail.com, tglx@linutronix.de,
	mingo@redhat.com, bp@alien8.de, dave.hansen@linux.intel.com,
	hpa@zytor.com, aruna.ramakrishna@oracle.com, elver@google.com
Cc: "Paul E. McKenney" <paulmck@kernel.org>,
	x86@kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4 4/4] selftests/rseq: Add test for rseq+pkeys
Date: Mon, 24 Feb 2025 14:48:55 -0500	[thread overview]
Message-ID: <cc8b653e-29ad-4cfb-91c3-1e001a811c4c@efficios.com> (raw)
In-Reply-To: <356f1e7b2c3c1dfee74f030bca603ffc9549ffb4.1740403209.git.dvyukov@google.com>

On 2025-02-24 08:20, Dmitry Vyukov wrote:
> Add a test that ensures that PKEY-protected struct rseq_cs
> works and does not lead to process kills.
> 
> Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: "Paul E. McKenney" <paulmck@kernel.org>
> Cc: Boqun Feng <boqun.feng@gmail.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Borislav Petkov <bp@alien8.de>
> Cc: Dave Hansen <dave.hansen@linux.intel.com>
> Cc: "H. Peter Anvin" <hpa@zytor.com>
> Cc: Aruna Ramakrishna <aruna.ramakrishna@oracle.com>
> Cc: x86@kernel.org
> Cc: linux-kernel@vger.kernel.org
> Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
> Fixes: d7822b1e24f2 ("rseq: Introduce restartable sequences system call")
> 
> ---
> Changes in v4:
>   - Added Fixes tag
> 
> Changes in v3:
>   - added Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
>   - rework the test to work when only pkey 0 is supported for rseq
> 
> Changes in v2:
>   - change test to install protected rseq_cs instead of rseq
> ---
>   tools/testing/selftests/rseq/Makefile    |  2 +-
>   tools/testing/selftests/rseq/pkey_test.c | 99 ++++++++++++++++++++++++
>   tools/testing/selftests/rseq/rseq.h      |  1 +
>   3 files changed, 101 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/rseq/Makefile b/tools/testing/selftests/rseq/Makefile
> index 5a3432fceb586..9111d25fea3af 100644
> --- a/tools/testing/selftests/rseq/Makefile
> +++ b/tools/testing/selftests/rseq/Makefile
> @@ -16,7 +16,7 @@ OVERRIDE_TARGETS = 1
>   
>   TEST_GEN_PROGS = basic_test basic_percpu_ops_test basic_percpu_ops_mm_cid_test param_test \
>   		param_test_benchmark param_test_compare_twice param_test_mm_cid \
> -		param_test_mm_cid_benchmark param_test_mm_cid_compare_twice
> +		param_test_mm_cid_benchmark param_test_mm_cid_compare_twice pkey_test
>   
>   TEST_GEN_PROGS_EXTENDED = librseq.so
>   
> diff --git a/tools/testing/selftests/rseq/pkey_test.c b/tools/testing/selftests/rseq/pkey_test.c
> new file mode 100644
> index 0000000000000..8752ecea21ba8
> --- /dev/null
> +++ b/tools/testing/selftests/rseq/pkey_test.c
> @@ -0,0 +1,99 @@
> +// SPDX-License-Identifier: LGPL-2.1
> +/*
> + * Ensure that rseq works when rseq data is inaccessible due to PKEYs.
> + */
> +
> +#define _GNU_SOURCE
> +#include <err.h>
> +#include <errno.h>
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +#include <sys/mman.h>
> +#include <sys/syscall.h>
> +#include <ucontext.h>
> +#include <unistd.h>
> +
> +#include "rseq.h"
> +#include "rseq-abi.h"
> +
> +int pkey;
> +ucontext_t ucp0, ucp1;

Why use an external linkage entity rather than static ?

> +
> +void coroutine(void)
> +{
> +	int i, orig_pk0, old_pk0, old_pk1, pk0, pk1;
> +	/*
> +	 * When we disable access to pkey 0, globals and TLS become
> +	 * inaccessible too, so we need to tread carefully.
> +	 * Pkey is global so we need to copy it to onto stack.

to onto -> onto the ?

> +	 * If ts is not volatile, then compiler may try to init it
> +	 * by loading a global 16-byte value.
> +	 */
> +	volatile int pk = pkey;
> +	volatile struct timespec ts;

I think you are looking for RSEQ_READ_ONCE() when loading from the
global variables to prevent re-fetch. AFAIU the volatile on the stack
variables are not what you are looking for.

> +
> +	orig_pk0 = pkey_get(0);
> +	if (pkey_set(0, PKEY_DISABLE_ACCESS))

AFAIU the pkey_set() call needs to act as a memory clobber. Therefore
having RSEQ_READ_ONCE() before the clobber to copy the global variables
onto the stack should be OK.

Thanks,

Mathieu


> +		err(1, "pkey_set failed");
> +	old_pk0 = pkey_get(0);
> +	old_pk1 = pkey_get(pk);
> +
> +	/*
> +	 * If the kernel misbehaves, context switches in the following loop
> +	 * will terminate the process with SIGSEGV.
> +	 */
> +	ts.tv_sec = 0;
> +	ts.tv_nsec = 10 * 1000;
> +	/*
> +	 * Trigger preemption w/o accessing TLS.
> +	 * Note that glibc's usleep touches errno always.
> +	 */
> +	for (i = 0; i < 10; i++)
> +		syscall(SYS_clock_nanosleep, CLOCK_MONOTONIC, 0, &ts, NULL);
> +
> +	pk0 = pkey_get(0);
> +	pk1 = pkey_get(pk);
> +	if (pkey_set(0, orig_pk0))
> +		err(1, "pkey_set failed");
> +
> +	/*
> +	 * Ensure that the kernel has restored the previous value of pkeys
> +	 * register after changing them.
> +	 */
> +	if (old_pk0 != pk0)
> +		errx(1, "pkey 0 changed %d->%d", old_pk0, pk0);
> +	if (old_pk1 != pk1)
> +		errx(1, "pkey 1 changed %d->%d", old_pk1, pk1);
> +
> +	swapcontext(&ucp1, &ucp0);
> +	abort();
> +}
> +
> +int main(int argc, char **argv)
> +{
> +	pkey = pkey_alloc(0, 0);
> +	if (pkey == -1) {
> +		printf("[SKIP]\tKernel does not support PKEYs: %s\n",
> +			strerror(errno));
> +		return 0;
> +	}
> +
> +	if (rseq_register_current_thread())
> +		err(1, "rseq_register_current_thread failed");
> +
> +	if (getcontext(&ucp1))
> +		err(1, "getcontext failed");
> +	ucp1.uc_stack.ss_size = getpagesize() * 4;
> +	ucp1.uc_stack.ss_sp = mmap(NULL, ucp1.uc_stack.ss_size,
> +		PROT_READ | PROT_WRITE, MAP_ANON | MAP_PRIVATE, -1, 0);
> +	if (ucp1.uc_stack.ss_sp == MAP_FAILED)
> +		err(1, "mmap failed");
> +	if (pkey_mprotect(ucp1.uc_stack.ss_sp, ucp1.uc_stack.ss_size,
> +			PROT_READ | PROT_WRITE, pkey))
> +		err(1, "pkey_mprotect failed");
> +	makecontext(&ucp1, coroutine, 0);
> +	if (swapcontext(&ucp0, &ucp1))
> +		err(1, "swapcontext failed");
> +	return 0;
> +}
> diff --git a/tools/testing/selftests/rseq/rseq.h b/tools/testing/selftests/rseq/rseq.h
> index ba424ce80a719..65da4a727c550 100644
> --- a/tools/testing/selftests/rseq/rseq.h
> +++ b/tools/testing/selftests/rseq/rseq.h
> @@ -8,6 +8,7 @@
>   #ifndef RSEQ_H
>   #define RSEQ_H
>   
> +#include <assert.h>
>   #include <stdint.h>
>   #include <stdbool.h>
>   #include <pthread.h>


-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

  reply	other threads:[~2025-02-24 19:48 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <cover.1740403209.git.dvyukov@google.com>
2025-02-24 13:20 ` [PATCH v4 1/4] pkeys: add API to switch to permissive/zero pkey register Dmitry Vyukov
2025-02-24 19:04   ` Mathieu Desnoyers
2025-02-25 13:54     ` Dmitry Vyukov
2025-02-24 13:20 ` [PATCH v4 2/4] x86/signal: Use write_permissive_pkey_val() helper Dmitry Vyukov
2025-02-24 19:11   ` Mathieu Desnoyers
2025-02-24 13:20 ` [PATCH v4 3/4] rseq: Make rseq work with protection keys Dmitry Vyukov
2025-02-24 19:18   ` Mathieu Desnoyers
2025-02-25 14:07     ` Dmitry Vyukov
2025-02-25 14:28       ` Mathieu Desnoyers
2025-02-25 14:51         ` Dmitry Vyukov
2025-02-25 14:53           ` Mathieu Desnoyers
2025-02-27 14:03             ` Dmitry Vyukov
2025-02-24 13:20 ` [PATCH v4 4/4] selftests/rseq: Add test for rseq+pkeys Dmitry Vyukov
2025-02-24 19:48   ` Mathieu Desnoyers [this message]
2025-02-25 13:55     ` Dmitry Vyukov
2025-02-24 13:28 ` [PATCH v4 0/4] rseq: Make rseq work with protection keys Dmitry Vyukov

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=cc8b653e-29ad-4cfb-91c3-1e001a811c4c@efficios.com \
    --to=mathieu.desnoyers@efficios.com \
    --cc=aruna.ramakrishna@oracle.com \
    --cc=boqun.feng@gmail.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=dvyukov@google.com \
    --cc=elver@google.com \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=paulmck@kernel.org \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.org \
    /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

all inboxes | Powered by JetHome®