summaryrefslogtreecommitdiff
path: root/arch/parisc/mm/fault.c
diff options
context:
space:
mode:
Diffstat (limited to 'arch/parisc/mm/fault.c')
-rw-r--r--arch/parisc/mm/fault.c393
1 files changed, 319 insertions, 74 deletions
diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c
index f247a3480e8e..f1785640b049 100644
--- a/arch/parisc/mm/fault.c
+++ b/arch/parisc/mm/fault.c
@@ -13,15 +13,16 @@
#include <linux/mm.h>
#include <linux/ptrace.h>
#include <linux/sched.h>
+#include <linux/sched/debug.h>
#include <linux/interrupt.h>
-#include <linux/module.h>
+#include <linux/extable.h>
+#include <linux/uaccess.h>
+#include <linux/hugetlb.h>
+#include <linux/perf_event.h>
-#include <asm/uaccess.h>
#include <asm/traps.h>
-#define PRINT_USER_FAULTS /* (turn this on if you want user faults to be */
- /* dumped to the console via printk) */
-
+#define DEBUG_NATLB 0
/* Various important other fields */
#define bit22set(x) (x & 0x00000200)
@@ -32,12 +33,12 @@
#define BITSSET 0x1c0 /* for identifying LDCW */
-DEFINE_PER_CPU(struct exception_data, exception_data);
+int show_unhandled_signals = 1;
/*
* parisc_acctyp(unsigned int inst) --
* Given a PA-RISC memory access instruction, determine if the
- * the instruction would perform a memory read or memory write
+ * instruction would perform a memory read or memory write
* operation.
*
* This function assumes that the given instruction is a memory access
@@ -49,7 +50,7 @@ DEFINE_PER_CPU(struct exception_data, exception_data);
* VM_WRITE if write operation
* VM_EXEC if execute operation
*/
-static unsigned long
+unsigned long
parisc_acctyp(unsigned long code, unsigned int inst)
{
if (code == 6 || code == 16)
@@ -68,6 +69,7 @@ parisc_acctyp(unsigned long code, unsigned int inst)
case 0x30000000: /* coproc2 */
if (bit22set(inst))
return VM_WRITE;
+ fallthrough;
case 0x0: /* indexed/memory management */
if (bit22set(inst)) {
@@ -144,13 +146,31 @@ int fixup_exception(struct pt_regs *regs)
fix = search_exception_tables(regs->iaoq[0]);
if (fix) {
- struct exception_data *d;
- d = &__get_cpu_var(exception_data);
- d->fault_ip = regs->iaoq[0];
- d->fault_space = regs->isr;
- d->fault_addr = regs->ior;
+ /*
+ * Fix up get_user() and put_user().
+ * ASM_EXCEPTIONTABLE_ENTRY_EFAULT() sets the least-significant
+ * bit in the relative address of the fixup routine to indicate
+ * that the register encoded in the "or %r0,%r0,register"
+ * opcode should be loaded with -EFAULT to report a userspace
+ * access error.
+ */
+ if (fix->fixup & 1) {
+ int fault_error_reg = fix->err_opcode & 0x1f;
+ if (!WARN_ON(!fault_error_reg))
+ regs->gr[fault_error_reg] = -EFAULT;
+ pr_debug("Unalignment fixup of register %d at %pS\n",
+ fault_error_reg, (void*)regs->iaoq[0]);
+
+ /* zero target register for get_user() */
+ if (parisc_acctyp(0, regs->iir) == VM_READ) {
+ int treg = regs->iir & 0x1f;
+ BUG_ON(treg == 0);
+ regs->gr[treg] = 0;
+ }
+ }
- regs->iaoq[0] = ((fix->fixup) & ~3);
+ regs->iaoq[0] = (unsigned long)&fix->fixup + fix->fixup;
+ regs->iaoq[0] &= ~3;
/*
* NOTE: In some cases the faulting instruction
* may be in the delay slot of a branch. We
@@ -167,33 +187,125 @@ int fixup_exception(struct pt_regs *regs)
return 0;
}
+/*
+ * parisc hardware trap list
+ *
+ * Documented in section 3 "Addressing and Access Control" of the
+ * "PA-RISC 1.1 Architecture and Instruction Set Reference Manual"
+ * https://parisc.wiki.kernel.org/index.php/File:Pa11_acd.pdf
+ *
+ * For implementation see handle_interruption() in traps.c
+ */
+static const char * const trap_description[] = {
+ [1] = "High-priority machine check (HPMC)",
+ [2] = "Power failure interrupt",
+ [3] = "Recovery counter trap",
+ [5] = "Low-priority machine check",
+ [6] = "Instruction TLB miss fault",
+ [7] = "Instruction access rights / protection trap",
+ [8] = "Illegal instruction trap",
+ [9] = "Break instruction trap",
+ [10] = "Privileged operation trap",
+ [11] = "Privileged register trap",
+ [12] = "Overflow trap",
+ [13] = "Conditional trap",
+ [14] = "FP Assist Exception trap",
+ [15] = "Data TLB miss fault",
+ [16] = "Non-access ITLB miss fault",
+ [17] = "Non-access DTLB miss fault",
+ [18] = "Data memory protection/unaligned access trap",
+ [19] = "Data memory break trap",
+ [20] = "TLB dirty bit trap",
+ [21] = "Page reference trap",
+ [22] = "Assist emulation trap",
+ [25] = "Taken branch trap",
+ [26] = "Data memory access rights trap",
+ [27] = "Data memory protection ID trap",
+ [28] = "Unaligned data reference trap",
+};
+
+const char *trap_name(unsigned long code)
+{
+ const char *t = NULL;
+
+ if (code < ARRAY_SIZE(trap_description))
+ t = trap_description[code];
+
+ return t ? t : "Unknown trap";
+}
+
+/*
+ * Print out info about fatal segfaults, if the show_unhandled_signals
+ * sysctl is set:
+ */
+static inline void
+show_signal_msg(struct pt_regs *regs, unsigned long code,
+ unsigned long address, struct task_struct *tsk,
+ struct vm_area_struct *vma)
+{
+ if (!unhandled_signal(tsk, SIGSEGV))
+ return;
+
+ if (!printk_ratelimit())
+ return;
+
+ pr_warn("\n");
+ pr_warn("do_page_fault() command='%s' type=%lu address=0x%08lx",
+ tsk->comm, code, address);
+ print_vma_addr(KERN_CONT " in ", regs->iaoq[0]);
+
+ pr_cont("\ntrap #%lu: %s%c", code, trap_name(code),
+ vma ? ',':'\n');
+
+ if (vma)
+ pr_cont(" vm_start = 0x%08lx, vm_end = 0x%08lx\n",
+ vma->vm_start, vma->vm_end);
+
+ show_regs(regs);
+}
+
void do_page_fault(struct pt_regs *regs, unsigned long code,
unsigned long address)
{
struct vm_area_struct *vma, *prev_vma;
- struct task_struct *tsk = current;
- struct mm_struct *mm = tsk->mm;
+ struct task_struct *tsk;
+ struct mm_struct *mm;
unsigned long acc_type;
- int fault;
- unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
-
- if (in_atomic() || !mm)
+ vm_fault_t fault = 0;
+ unsigned int flags;
+ char *msg;
+
+ tsk = current;
+ mm = tsk->mm;
+ if (!mm) {
+ msg = "Page fault: no context";
goto no_context;
+ }
+
+ flags = FAULT_FLAG_DEFAULT;
+ if (user_mode(regs))
+ flags |= FAULT_FLAG_USER;
+ acc_type = parisc_acctyp(code, regs->iir);
+ if (acc_type & VM_WRITE)
+ flags |= FAULT_FLAG_WRITE;
+ perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
retry:
- down_read(&mm->mmap_sem);
+ mmap_read_lock(mm);
vma = find_vma_prev(mm, address, &prev_vma);
- if (!vma || address < vma->vm_start)
- goto check_expansion;
+ if (!vma || address < vma->vm_start) {
+ if (!prev_vma || !(prev_vma->vm_flags & VM_GROWSUP))
+ goto bad_area;
+ vma = expand_stack(mm, address);
+ if (!vma)
+ goto bad_area_nosemaphore;
+ }
+
/*
* Ok, we have a good vm_area for this memory access. We still need to
* check the access permissions.
*/
-good_area:
-
- acc_type = parisc_acctyp(code,regs->iir);
-
if ((vma->vm_flags & acc_type) != acc_type)
goto bad_area;
@@ -203,10 +315,18 @@ good_area:
* fault.
*/
- fault = handle_mm_fault(mm, vma, address,
- flags | ((acc_type & VM_WRITE) ? FAULT_FLAG_WRITE : 0));
+ fault = handle_mm_fault(vma, address, flags, regs);
+
+ if (fault_signal_pending(fault, regs)) {
+ if (!user_mode(regs)) {
+ msg = "Page fault: fault signal on kernel memory";
+ goto no_context;
+ }
+ return;
+ }
- if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+ /* The fault is fully completed (including releasing mmap lock) */
+ if (fault & VM_FAULT_COMPLETED)
return;
if (unlikely(fault & VM_FAULT_ERROR)) {
@@ -217,62 +337,97 @@ good_area:
*/
if (fault & VM_FAULT_OOM)
goto out_of_memory;
- else if (fault & VM_FAULT_SIGBUS)
+ else if (fault & VM_FAULT_SIGSEGV)
+ goto bad_area;
+ else if (fault & (VM_FAULT_SIGBUS|VM_FAULT_HWPOISON|
+ VM_FAULT_HWPOISON_LARGE))
goto bad_area;
BUG();
}
- if (flags & FAULT_FLAG_ALLOW_RETRY) {
- if (fault & VM_FAULT_MAJOR)
- current->maj_flt++;
- else
- current->min_flt++;
- if (fault & VM_FAULT_RETRY) {
- flags &= ~FAULT_FLAG_ALLOW_RETRY;
-
- /*
- * No need to up_read(&mm->mmap_sem) as we would
- * have already released it in __lock_page_or_retry
- * in mm/filemap.c.
- */
-
- goto retry;
- }
+ if (fault & VM_FAULT_RETRY) {
+ /*
+ * No need to mmap_read_unlock(mm) as we would
+ * have already released it in __lock_page_or_retry
+ * in mm/filemap.c.
+ */
+ flags |= FAULT_FLAG_TRIED;
+ goto retry;
}
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
return;
-check_expansion:
- vma = prev_vma;
- if (vma && (expand_stack(vma, address) == 0))
- goto good_area;
-
/*
* Something tried to access memory that isn't in our memory map..
*/
bad_area:
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
+
+bad_area_nosemaphore:
+ if (!user_mode(regs) && fixup_exception(regs)) {
+ return;
+ }
if (user_mode(regs)) {
- struct siginfo si;
-
-#ifdef PRINT_USER_FAULTS
- printk(KERN_DEBUG "\n");
- printk(KERN_DEBUG "do_page_fault() pid=%d command='%s' type=%lu address=0x%08lx\n",
- task_pid_nr(tsk), tsk->comm, code, address);
- if (vma) {
- printk(KERN_DEBUG "vm_start = 0x%08lx, vm_end = 0x%08lx\n",
- vma->vm_start, vma->vm_end);
+ int signo, si_code;
+
+ switch (code) {
+ case 15: /* Data TLB miss fault/Data page fault */
+ /* send SIGSEGV when outside of vma */
+ if (!vma ||
+ address < vma->vm_start || address >= vma->vm_end) {
+ signo = SIGSEGV;
+ si_code = SEGV_MAPERR;
+ break;
+ }
+
+ /* send SIGSEGV for wrong permissions */
+ if ((vma->vm_flags & acc_type) != acc_type) {
+ signo = SIGSEGV;
+ si_code = SEGV_ACCERR;
+ break;
+ }
+
+ /* probably address is outside of mapped file */
+ fallthrough;
+ case 17: /* NA data TLB miss / page fault */
+ case 18: /* Unaligned access - PCXS only */
+ signo = SIGBUS;
+ si_code = (code == 18) ? BUS_ADRALN : BUS_ADRERR;
+ break;
+ case 16: /* Non-access instruction TLB miss fault */
+ case 26: /* PCXL: Data memory access rights trap */
+ default:
+ signo = SIGSEGV;
+ si_code = (code == 26) ? SEGV_ACCERR : SEGV_MAPERR;
+ break;
+ }
+#ifdef CONFIG_MEMORY_FAILURE
+ if (fault & (VM_FAULT_HWPOISON|VM_FAULT_HWPOISON_LARGE)) {
+ unsigned int lsb = 0;
+ printk(KERN_ERR
+ "MCE: Killing %s:%d due to hardware memory corruption fault at %08lx\n",
+ tsk->comm, tsk->pid, address);
+ /*
+ * Either small page or large page may be poisoned.
+ * In other words, VM_FAULT_HWPOISON_LARGE and
+ * VM_FAULT_HWPOISON are mutually exclusive.
+ */
+ if (fault & VM_FAULT_HWPOISON_LARGE)
+ lsb = hstate_index_to_shift(VM_FAULT_GET_HINDEX(fault));
+ else if (fault & VM_FAULT_HWPOISON)
+ lsb = PAGE_SHIFT;
+
+ force_sig_mceerr(BUS_MCEERR_AR, (void __user *) address,
+ lsb);
+ return;
}
- show_regs(regs);
#endif
- /* FIXME: actually we need to get the signo and code correct */
- si.si_signo = SIGSEGV;
- si.si_errno = 0;
- si.si_code = SEGV_MAPERR;
- si.si_addr = (void __user *) address;
- force_sig_info(SIGSEGV, &si, current);
+ show_signal_msg(regs, code, address, tsk, vma);
+
+ force_sig_fault(signo, si_code, (void __user *) address);
return;
}
+ msg = "Page fault: bad address";
no_context:
@@ -280,11 +435,101 @@ no_context:
return;
}
- parisc_terminate("Bad Address (null pointer deref?)", regs, code, address);
+ parisc_terminate(msg, regs, code, address);
- out_of_memory:
- up_read(&mm->mmap_sem);
- if (!user_mode(regs))
+out_of_memory:
+ mmap_read_unlock(mm);
+ if (!user_mode(regs)) {
+ msg = "Page fault: out of memory";
goto no_context;
+ }
pagefault_out_of_memory();
}
+
+/* Handle non-access data TLB miss faults.
+ *
+ * For probe instructions, accesses to userspace are considered allowed
+ * if they lie in a valid VMA and the access type matches. We are not
+ * allowed to handle MM faults here so there may be situations where an
+ * actual access would fail even though a probe was successful.
+ */
+int
+handle_nadtlb_fault(struct pt_regs *regs)
+{
+ unsigned long insn = regs->iir;
+ int breg, treg, xreg, val = 0;
+ struct vm_area_struct *vma;
+ struct task_struct *tsk;
+ struct mm_struct *mm;
+ unsigned long address;
+ unsigned long acc_type;
+
+ switch (insn & 0x380) {
+ case 0x280:
+ /* FDC instruction */
+ fallthrough;
+ case 0x380:
+ /* PDC and FIC instructions */
+ if (DEBUG_NATLB && printk_ratelimit()) {
+ pr_warn("WARNING: nullifying cache flush/purge instruction\n");
+ show_regs(regs);
+ }
+ if (insn & 0x20) {
+ /* Base modification */
+ breg = (insn >> 21) & 0x1f;
+ xreg = (insn >> 16) & 0x1f;
+ if (breg && xreg)
+ regs->gr[breg] += regs->gr[xreg];
+ }
+ regs->gr[0] |= PSW_N;
+ return 1;
+
+ case 0x180:
+ /* PROBE instruction */
+ treg = insn & 0x1f;
+ if (regs->isr) {
+ tsk = current;
+ mm = tsk->mm;
+ if (mm) {
+ /* Search for VMA */
+ address = regs->ior;
+ mmap_read_lock(mm);
+ vma = vma_lookup(mm, address);
+ mmap_read_unlock(mm);
+
+ /*
+ * Check if access to the VMA is okay.
+ * We don't allow for stack expansion.
+ */
+ acc_type = (insn & 0x40) ? VM_WRITE : VM_READ;
+ if (vma
+ && (vma->vm_flags & acc_type) == acc_type)
+ val = 1;
+ }
+ }
+ if (treg)
+ regs->gr[treg] = val;
+ regs->gr[0] |= PSW_N;
+ return 1;
+
+ case 0x300:
+ /* LPA instruction */
+ if (insn & 0x20) {
+ /* Base modification */
+ breg = (insn >> 21) & 0x1f;
+ xreg = (insn >> 16) & 0x1f;
+ if (breg && xreg)
+ regs->gr[breg] += regs->gr[xreg];
+ }
+ treg = insn & 0x1f;
+ if (treg)
+ regs->gr[treg] = 0;
+ regs->gr[0] |= PSW_N;
+ return 1;
+
+ default:
+ break;
+ }
+
+ return 0;
+}
generated by cgit (git 2.34.1) at 2025-12-21 21:33:26 +0000