[PATCH 3/5] um: Do a double clone to disable rseq

Benjamin Berg benjamin at sipsolutions.net
Tue May 28 03:30:31 PDT 2024


Hi Tiwei,

On Tue, 2024-05-28 at 18:16 +0800, Tiwei Bie wrote:
> On 5/28/24 4:54 PM, benjamin at sipsolutions.net wrote:
> > From: Benjamin Berg <benjamin.berg at intel.com>
> > 
> > Newer glibc versions are enabling rseq support by default. This remains
> > enabled in the cloned child process, potentially causing the host kernel
> > to write/read memory in the child.
> > 
> > It appears that this was purely not an issue because the used memory
> > area happened to be above TASK_SIZE and remains mapped.
> 
> I also encountered this issue. In my case, with "Force a static link"
> (CONFIG_STATIC_LINK) enabled, UML will crash immediately every time
> it starts up. I worked around this by setting the glibc.pthread.rseq
> tunable via GLIBC_TUNABLES [1] before launching UML.
> 
> So another easy way to work around this issue without introducing runtime
> overhead might be to add the GLIBC_TUNABLES=glibc.pthread.rseq=0 environment
> variable and exec /proc/self/exe in UML on startup.

I am not really worried about the overhead, but I agree that setting
GLIBC_TUNABLES is also a reasonable solution to the problem.

Doing the memfd/execveat dance with an embedded static binary would
still be best in my view, but either this or GLIBC_TUNABLES seem fine
in the meantime.

Do you want to submit the patch? Should I re-roll the patchset with
GLIBC_TUNABLES?

Benjamin

> [1] https://www.gnu.org/software/libc/manual/html_node/Tunables.html
> 
> Regards,
> Tiwei
> 
> > 
> > Note that a better approach would be to exec a small static binary that
> > does not link with other libraries. Using a memfd and execveat the
> > binary could be embedded into UML itself and it would result in an
> > entirely clean execution environment for userspace.
> > 
> > Signed-off-by: Benjamin Berg <benjamin.berg at intel.com>
> > ---
> >  arch/um/os-Linux/skas/process.c | 54 ++++++++++++++++++++++++++++++---
> >  1 file changed, 50 insertions(+), 4 deletions(-)
> > 
> > diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
> > index 41a288dcfc34..ee332a2aeea6 100644
> > --- a/arch/um/os-Linux/skas/process.c
> > +++ b/arch/um/os-Linux/skas/process.c
> > @@ -255,6 +255,31 @@ static int userspace_tramp(void *stack)
> >  int userspace_pid[NR_CPUS];
> >  int kill_userspace_mm[NR_CPUS];
> >  
> > +struct tramp_data {
> > +	int pid;
> > +	void *clone_sp;
> > +	void *stack;
> > +};
> > +
> > +static int userspace_tramp_clone_vm(void *data)
> > +{
> > +	struct tramp_data *tramp_data = data;
> > +
> > +	/*
> > +	 * This helper exist to do a double-clone. First with CLONE_VM which
> > +	 * effectively disables things like rseq, and then the second one to
> > +	 * get a new memory space.
> > +	 */
> > +
> > +	tramp_data->pid = clone(userspace_tramp, tramp_data->clone_sp,
> > +				CLONE_PARENT | CLONE_FILES | SIGCHLD,
> > +				tramp_data->stack);
> > +	if (tramp_data->pid < 0)
> > +		tramp_data->pid = -errno;
> > +
> > +	exit(0);
> > +}
> > +
> >  /**
> >   * start_userspace() - prepare a new userspace process
> >   * @stub_stack:	pointer to the stub stack.
> > @@ -268,9 +293,10 @@ int kill_userspace_mm[NR_CPUS];
> >   */
> >  int start_userspace(unsigned long stub_stack)
> >  {
> > +	struct tramp_data tramp_data;
> >  	void *stack;
> >  	unsigned long sp;
> > -	int pid, status, n, flags, err;
> > +	int pid, status, n, err;
> >  
> >  	/* setup a temporary stack page */
> >  	stack = mmap(NULL, UM_KERN_PAGE_SIZE,
> > @@ -286,10 +312,13 @@ int start_userspace(unsigned long stub_stack)
> >  	/* set stack pointer to the end of the stack page, so it can grow downwards */
> >  	sp = (unsigned long)stack + UM_KERN_PAGE_SIZE;
> >  
> > -	flags = CLONE_FILES | SIGCHLD;
> > +	tramp_data.stack = (void *) stub_stack;
> > +	tramp_data.clone_sp = (void *) sp;
> > +	tramp_data.pid = -EINVAL;
> >  
> >  	/* clone into new userspace process */
> > -	pid = clone(userspace_tramp, (void *) sp, flags, (void *) stub_stack);
> > +	pid = clone(userspace_tramp_clone_vm, (void *) sp,
> > +		    CLONE_VM | CLONE_FILES | SIGCHLD, &tramp_data);
> >  	if (pid < 0) {
> >  		err = -errno;
> >  		printk(UM_KERN_ERR "%s : clone failed, errno = %d\n",
> > @@ -305,7 +334,24 @@ int start_userspace(unsigned long stub_stack)
> >  			       __func__, errno);
> >  			goto out_kill;
> >  		}
> > -	} while (WIFSTOPPED(status) && (WSTOPSIG(status) == SIGALRM));
> > +	} while (!WIFEXITED(status));
> > +
> > +	pid = tramp_data.pid;
> > +	if (pid < 0) {
> > +		printk(UM_KERN_ERR "%s : second clone failed, errno = %d\n",
> > +		       __func__, -pid);
> > +		return pid;
> > +	}
> > +
> > +	do {
> > +		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED | __WALL));
> > +		if (n < 0) {
> > +			err = -errno;
> > +			printk(UM_KERN_ERR "%s : wait failed, errno = %d\n",
> > +			       __func__, errno);
> > +			goto out_kill;
> > +		}
> > +	} while (WIFEXITED(status) && (WSTOPSIG(status) == SIGALRM));
> >  
> >  	if (!WIFSTOPPED(status) || (WSTOPSIG(status) != SIGSTOP)) {
> >  		err = -EINVAL;
> 
> 




More information about the linux-um mailing list