aboutsummaryrefslogtreecommitdiffstats
path: root/arch/s390/mm/fault.c
diff options
context:
space:
mode:
authorGerald Schaefer <geraldsc@de.ibm.com>2007-02-05 15:18:17 -0500
committerMartin Schwidefsky <schwidefsky@de.ibm.com>2007-02-05 15:18:17 -0500
commitc1821c2e9711adc3cd298a16b7237c92a2cee78d (patch)
tree9155b089db35a37d95863125ea4c5f918bd7801b /arch/s390/mm/fault.c
parent86aa9fc2456d8a662f299a70bdb70987209170f0 (diff)
[S390] noexec protection
This provides a noexec protection on s390 hardware. Our hardware does not have any bits left in the pte for a hw noexec bit, so this is a different approach using shadow page tables and a special addressing mode that allows separate address spaces for code and data. As a special feature of our "secondary-space" addressing mode, separate page tables can be specified for the translation of data addresses (storage operands) and instruction addresses. The shadow page table is used for the instruction addresses and the standard page table for the data addresses. The shadow page table is linked to the standard page table by a pointer in page->lru.next of the struct page corresponding to the page that contains the standard page table (since page->private is not really private with the pte_lock and the page table pages are not in the LRU list). Depending on the software bits of a pte, it is either inserted into both page tables or just into the standard (data) page table. Pages of a vma that does not have the VM_EXEC bit set get mapped only in the data address space. Any try to execute code on such a page will cause a page translation exception. The standard reaction to this is a SIGSEGV with two exceptions: the two system call opcodes 0x0a77 (sys_sigreturn) and 0x0aad (sys_rt_sigreturn) are allowed. They are stored by the kernel to the signal stack frame. Unfortunately, the signal return mechanism cannot be modified to use an SA_RESTORER because the exception unwinding code depends on the system call opcode stored behind the signal stack frame. This feature requires that user space is executed in secondary-space mode and the kernel in home-space mode, which means that the addressing modes need to be switched and that the noexec protection only works for user space. After switching the addressing modes, we cannot use the mvcp/mvcs instructions anymore to copy between kernel and user space. A new mvcos instruction has been added to the z9 EC/BC hardware which allows to copy between arbitrary address spaces, but on older hardware the page tables need to be walked manually. Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Diffstat (limited to 'arch/s390/mm/fault.c')
-rw-r--r--arch/s390/mm/fault.c88
1 files changed, 86 insertions, 2 deletions
diff --git a/arch/s390/mm/fault.c b/arch/s390/mm/fault.c
index 3382e29f34a4..9ff143e87746 100644
--- a/arch/s390/mm/fault.c
+++ b/arch/s390/mm/fault.c
@@ -137,7 +137,9 @@ static int __check_access_register(struct pt_regs *regs, int error_code)
137 137
138/* 138/*
139 * Check which address space the address belongs to. 139 * Check which address space the address belongs to.
140 * Returns 1 for user space and 0 for kernel space. 140 * May return 1 or 2 for user space and 0 for kernel space.
141 * Returns 2 for user space in primary addressing mode with
142 * CONFIG_S390_EXEC_PROTECT on and kernel parameter noexec=on.
141 */ 143 */
142static inline int check_user_space(struct pt_regs *regs, int error_code) 144static inline int check_user_space(struct pt_regs *regs, int error_code)
143{ 145{
@@ -154,7 +156,7 @@ static inline int check_user_space(struct pt_regs *regs, int error_code)
154 return __check_access_register(regs, error_code); 156 return __check_access_register(regs, error_code);
155 if (descriptor == 2) 157 if (descriptor == 2)
156 return current->thread.mm_segment.ar4; 158 return current->thread.mm_segment.ar4;
157 return descriptor != 0; 159 return ((descriptor != 0) ^ (switch_amode)) << s390_noexec;
158} 160}
159 161
160/* 162/*
@@ -183,6 +185,77 @@ static void do_sigsegv(struct pt_regs *regs, unsigned long error_code,
183 force_sig_info(SIGSEGV, &si, current); 185 force_sig_info(SIGSEGV, &si, current);
184} 186}
185 187
188#ifdef CONFIG_S390_EXEC_PROTECT
189extern long sys_sigreturn(struct pt_regs *regs);
190extern long sys_rt_sigreturn(struct pt_regs *regs);
191extern long sys32_sigreturn(struct pt_regs *regs);
192extern long sys32_rt_sigreturn(struct pt_regs *regs);
193
194static inline void do_sigreturn(struct mm_struct *mm, struct pt_regs *regs,
195 int rt)
196{
197 up_read(&mm->mmap_sem);
198 clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
199#ifdef CONFIG_COMPAT
200 if (test_tsk_thread_flag(current, TIF_31BIT)) {
201 if (rt)
202 sys32_rt_sigreturn(regs);
203 else
204 sys32_sigreturn(regs);
205 return;
206 }
207#endif /* CONFIG_COMPAT */
208 if (rt)
209 sys_rt_sigreturn(regs);
210 else
211 sys_sigreturn(regs);
212 return;
213}
214
215static int signal_return(struct mm_struct *mm, struct pt_regs *regs,
216 unsigned long address, unsigned long error_code)
217{
218 pgd_t *pgd;
219 pmd_t *pmd;
220 pte_t *pte;
221 u16 *instruction;
222 unsigned long pfn, uaddr = regs->psw.addr;
223
224 spin_lock(&mm->page_table_lock);
225 pgd = pgd_offset(mm, uaddr);
226 if (pgd_none(*pgd) || unlikely(pgd_bad(*pgd)))
227 goto out_fault;
228 pmd = pmd_offset(pgd, uaddr);
229 if (pmd_none(*pmd) || unlikely(pmd_bad(*pmd)))
230 goto out_fault;
231 pte = pte_offset_map(pmd_offset(pgd_offset(mm, uaddr), uaddr), uaddr);
232 if (!pte || !pte_present(*pte))
233 goto out_fault;
234 pfn = pte_pfn(*pte);
235 if (!pfn_valid(pfn))
236 goto out_fault;
237 spin_unlock(&mm->page_table_lock);
238
239 instruction = (u16 *) ((pfn << PAGE_SHIFT) + (uaddr & (PAGE_SIZE-1)));
240 if (*instruction == 0x0a77)
241 do_sigreturn(mm, regs, 0);
242 else if (*instruction == 0x0aad)
243 do_sigreturn(mm, regs, 1);
244 else {
245 printk("- XXX - do_exception: task = %s, primary, NO EXEC "
246 "-> SIGSEGV\n", current->comm);
247 up_read(&mm->mmap_sem);
248 current->thread.prot_addr = address;
249 current->thread.trap_no = error_code;
250 do_sigsegv(regs, error_code, SEGV_MAPERR, address);
251 }
252 return 0;
253out_fault:
254 spin_unlock(&mm->page_table_lock);
255 return -EFAULT;
256}
257#endif /* CONFIG_S390_EXEC_PROTECT */
258
186/* 259/*
187 * This routine handles page faults. It determines the address, 260 * This routine handles page faults. It determines the address,
188 * and the problem, and then passes it off to one of the appropriate 261 * and the problem, and then passes it off to one of the appropriate
@@ -260,6 +333,17 @@ do_exception(struct pt_regs *regs, unsigned long error_code, int is_protection)
260 vma = find_vma(mm, address); 333 vma = find_vma(mm, address);
261 if (!vma) 334 if (!vma)
262 goto bad_area; 335 goto bad_area;
336
337#ifdef CONFIG_S390_EXEC_PROTECT
338 if (unlikely((user_address == 2) && !(vma->vm_flags & VM_EXEC)))
339 if (!signal_return(mm, regs, address, error_code))
340 /*
341 * signal_return() has done an up_read(&mm->mmap_sem)
342 * if it returns 0.
343 */
344 return;
345#endif
346
263 if (vma->vm_start <= address) 347 if (vma->vm_start <= address)
264 goto good_area; 348 goto good_area;
265 if (!(vma->vm_flags & VM_GROWSDOWN)) 349 if (!(vma->vm_flags & VM_GROWSDOWN))