oss-security - CVE-2022-32981: Linux kernel for powerpc 32-bit, buffer overflow in ptrace PEEKUSER/POKEUSER

Follow @Openwall on Twitter for new release announcements and other news

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

Message-ID: <87bkuwc5cr.fsf@mpe.ellerman.id.au>
Date: Tue, 14 Jun 2022 13:05:08 +1000
From: Michael Ellerman <mpe@...erman.id.au>
To: oss-security@...ts.openwall.com
Subject: CVE-2022-32981: Linux kernel for powerpc 32-bit, buffer overflow in
 ptrace PEEKUSER/POKEUSER

The Linux kernel for powerpc 32-bit has a buffer overflow in the handling of ptrace
PEEKUSER/POKEUSER when accessing floating point registers.

The fix for mainline is:
  https://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux.git/commit/?id=8e1278444446fc97778a5e5c99bca1ce0bbc5ec9

Which is included in v5.19-rc2.

Stable backports have been posted.

A test case is included below, it will report if the system is correctly
patched, it is safe to run on an unpatched system.

cheers

---
#undef NDEBUG
#include <assert.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/ptrace.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <sys/syscall.h>
#include <unistd.h>
#include <sys/ipc.h>
#include <sys/shm.h>

static double expected = 0.123456L;

static int child(int shm_id)
{
	int *cptr = shmat(shm_id, NULL, 0);

	asm volatile (
		"lfd	 %%f0, 0(%0)	;"
		"lfd	 %%f1, 0(%0)	;"
		"li 	 %%r9, 1	;"
		"stw	 %%r9, 0(%1)	;"
		"1:"
		"lwz	 %%r9, 0(%2)	;"
		"cmpwi	 %%r9, 0	;"
		"beq	   1b		;"
		: // outputs
		: // inputs
		  "b" (&expected), "b" (&cptr[1]), "b" (&cptr[0])
		: // clobbers
		  "memory", "r9", "fr0", "fr1"
	);

	return 0;
}

int start_trace(pid_t child)
{
	int ret;

	ret = ptrace(PTRACE_ATTACH, child, NULL, NULL);
	if (ret) {
		perror("ptrace(PTRACE_ATTACH) failed");
		return -1;
	}
	ret = waitpid(child, NULL, 0);
	if (ret != child) {
		perror("waitpid() failed");
		return -1;
	}
	return 0;
}

int stop_trace(pid_t child)
{
	int ret;

	ret = ptrace(PTRACE_DETACH, child, NULL, NULL);
	if (ret) {
		perror("ptrace(PTRACE_DETACH) failed");
		return -1;
	}
	return 0;
}

long raw_ptrace(enum __ptrace_request request, pid_t pid, unsigned long addr, void *data)
{
	return syscall(__NR_ptrace, request, pid, (void *)addr, data);
}

#define PEEKS_PER_FPR	(sizeof(__u64) / sizeof(unsigned long))

int peek_fpr(pid_t child, int frnum, __u64 *fpr)
{
	unsigned long *p, addr;
	int i, fpindex;
	long ret;

	fpindex = PEEKS_PER_FPR * frnum;

	p = (unsigned long *)fpr;
	for (i = 0; i < PEEKS_PER_FPR; i++, p++) {
		addr = sizeof(unsigned long) * (PT_FPR0 + fpindex + i);
		ret = raw_ptrace(PTRACE_PEEKUSER, child, addr, p);
		if (ret) {
			perror("ptrace(PTRACE_PEEKUSR) failed");
			return -1;
		}
	}

	return 0;
}

int parent(pid_t child)
{
	double f0, f1;

	assert(start_trace(child) == 0);

	assert(peek_fpr(child, 0, (__u64 *)&f0) == 0);
	assert(peek_fpr(child, 1, (__u64 *)&f1) == 0);

	assert(stop_trace(child) == 0);

	printf("expected = %e\n", f0);
	printf("f0       = %e\n", f0);
	printf("f1       = %e\n", f1);

	if (f0 != expected || f1 != expected) {
		printf("FAIL - values don't match! Kernel is buggy.\n");
		return -1;
	}

	printf("OK - values match\n");

	return 0;
}

int main(void)
{
	int shm_id, ret, status, *pptr;
	pid_t pid;

	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
	assert(shm_id != -1);

	pid = fork();
	assert(pid >= 0);

	if (pid == 0)
		exit(child(shm_id));

	pptr = shmat(shm_id, NULL, 0);

	// Wait for child to signal us to continue
	while (!pptr[1])
		asm volatile("" : : : "memory");

	ret = parent(pid);
	if (ret) {
		kill(pid, SIGTERM);
		shmdt((void *)pptr);
		shmctl(shm_id, IPC_RMID, NULL);
		return -1;
	}

	// Signal child to exit
	pptr[0] = 1;
	shmdt((void *)pptr);

	ret = wait(&status);
	shmctl(shm_id, IPC_RMID, NULL);

	assert(ret != -1 && WIFEXITED(status) && !WEXITSTATUS(status));

	return 0;
}

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.