[PATCH] uml: skas0 stubs now check system call return values

Change syscall-stub's data to include a "expected retval".

Stub now checks syscalls retval and aborts execution of syscall list, if
retval != expected retval.

run_syscall_stub prints the data of the failed syscall, using the data pointer
and retval written by the stub to the beginning of the stack.

one_syscall_stub is removed, to simplify code, because only some instructions
are saved by one_syscall_stub, no host-syscall.

Using the stub with additional data (modify_ldt via stub)
is prepared also.

Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com>
Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

3 files changed, 168 insertions, 130 deletions

diff --git a/arch/um/kernel/skas/include/skas.h b/arch/um/kernel/skas/include/skas.h
index 03c9d99e15e4..060934740f9f 100644
--- a/arch/um/kernel/skas/include/skas.h
+++ b/arch/um/kernel/skas/include/skas.h
@@ -24,14 +24,14 @@ extern void new_thread_proc(void *stack, void (*handler)(int sig));
 extern void remove_sigstack(void);
 extern void new_thread_handler(int sig);
 extern void handle_syscall(union uml_pt_regs *regs);
-extern void *map(struct mm_id * mm_idp, unsigned long virt,
-                 unsigned long len, int r, int w, int x, int phys_fd,
-                 unsigned long long offset, int done, void *data);
-extern void *unmap(struct mm_id * mm_idp, void *addr,
-                   unsigned long len, int done, void *data);
-extern void *protect(struct mm_id * mm_idp, unsigned long addr,
-                     unsigned long len, int r, int w, int x, int done,
-                     void *data);
+extern int map(struct mm_id * mm_idp, unsigned long virt,
+               unsigned long len, int r, int w, int x, int phys_fd,
+               unsigned long long offset, int done, void **data);
+extern int unmap(struct mm_id * mm_idp, void *addr, unsigned long len,
+                 int done, void **data);
+extern int protect(struct mm_id * mm_idp, unsigned long addr,
+                   unsigned long len, int r, int w, int x, int done,
+                   void **data);
 extern void user_signal(int sig, union uml_pt_regs *regs, int pid);
 extern int new_mm(int from, unsigned long stack);
 extern int start_userspace(unsigned long stub_stack);
@@ -39,16 +39,11 @@ extern int copy_context_skas0(unsigned long stack, int pid);
 extern void get_skas_faultinfo(int pid, struct faultinfo * fi);
 extern long execute_syscall_skas(void *r);
 extern unsigned long current_stub_stack(void);
+extern long run_syscall_stub(struct mm_id * mm_idp,
+                             int syscall, unsigned long *args, long expected,
+                             void **addr, int done);
+extern long syscall_stub_data(struct mm_id * mm_idp,
+                              unsigned long *data, int data_count,
+                              void **addr, void **stub_addr);
 
 #endif
-
-/*
- * Overrides for Emacs so that we follow Linus's tabbing style.
- * Emacs will notice this stuff at the end of the file and automatically
- * adjust the settings for this buffer only.  This must remain at the end
- * of the file.
- * ---------------------------------------------------------------------------
- * Local variables:
- * c-file-style: "linux"
- * End:
- */
diff --git a/arch/um/kernel/skas/mem_user.c b/arch/um/kernel/skas/mem_user.c
index c976320ebe84..1d89640bd502 100644
--- a/arch/um/kernel/skas/mem_user.c
+++ b/arch/um/kernel/skas/mem_user.c
@@ -5,13 +5,14 @@
 
 #include <signal.h>
 #include <errno.h>
+#include <string.h>
 #include <sys/mman.h>
 #include <sys/wait.h>
 #include <asm/page.h>
 #include <asm/unistd.h>
 #include "mem_user.h"
 #include "mem.h"
-#include "mm_id.h"
+#include "skas.h"
 #include "user.h"
 #include "os.h"
 #include "proc_mm.h"
@@ -23,56 +24,99 @@
 #include "uml-config.h"
 #include "sysdep/ptrace.h"
 #include "sysdep/stub.h"
-#include "skas.h"
 
-extern unsigned long syscall_stub, batch_syscall_stub, __syscall_stub_start;
+extern unsigned long batch_syscall_stub, __syscall_stub_start;
 
 extern void wait_stub_done(int pid, int sig, char * fname);
 
+static inline unsigned long *check_init_stack(struct mm_id * mm_idp,
+                                              unsigned long *stack)
+{
+        if(stack == NULL){
+                stack = (unsigned long *) mm_idp->stack + 2;
+                *stack = 0;
+        }
+        return stack;
+}
+
+extern int proc_mm;
+
 int single_count = 0;
+int multi_count = 0;
+int multi_op_count = 0;
 
-static long one_syscall_stub(struct mm_id * mm_idp, int syscall,
-                             unsigned long *args)
+static long do_syscall_stub(struct mm_id *mm_idp, void **addr)
 {
+        unsigned long regs[MAX_REG_NR];
+        unsigned long *data;
+        unsigned long *syscall;
+        long ret, offset;
         int n, pid = mm_idp->u.pid;
-        unsigned long regs[MAX_REG_NR];
+
+        if(proc_mm)
+#warning Need to look up userspace_pid by cpu
+                pid = userspace_pid[0];
+
+        multi_count++;
 
         get_safe_registers(regs);
         regs[REGS_IP_INDEX] = UML_CONFIG_STUB_CODE +
-                ((unsigned long) &syscall_stub -
+                ((unsigned long) &batch_syscall_stub -
                  (unsigned long) &__syscall_stub_start);
-        /* XXX Don't have a define for starting a syscall */
-        regs[REGS_SYSCALL_NR] = syscall;
-        regs[REGS_SYSCALL_ARG1] = args[0];
-        regs[REGS_SYSCALL_ARG2] = args[1];
-        regs[REGS_SYSCALL_ARG3] = args[2];
-        regs[REGS_SYSCALL_ARG4] = args[3];
-        regs[REGS_SYSCALL_ARG5] = args[4];
-        regs[REGS_SYSCALL_ARG6] = args[5];
-        n = ptrace_setregs(pid, regs);
-        if(n < 0){
-                printk("one_syscall_stub : PTRACE_SETREGS failed, "
-                       "errno = %d\n", n);
-                return(n);
+        n = ptrace_setregs(pid, regs);
+        if(n < 0)
+                panic("do_syscall_stub : PTRACE_SETREGS failed, errno = %d\n",
+                      n);
+
+        wait_stub_done(pid, 0, "do_syscall_stub");
+
+        /* When the stub stops, we find the following values on the
+         * beginning of the stack:
+         * (long )return_value
+         * (long )offset to failed sycall-data (0, if no error)
+         */
+        ret = *((unsigned long *) mm_idp->stack);
+        offset = *((unsigned long *) mm_idp->stack + 1);
+        if (offset) {
+                data = (unsigned long *)(mm_idp->stack +
+                                         offset - UML_CONFIG_STUB_DATA);
+                syscall = (unsigned long *)((unsigned long)data + data[0]);
+                printk("do_syscall_stub: syscall %ld failed, return value = "
+                       "0x%lx, expected return value = 0x%lx\n",
+                       syscall[0], ret, syscall[7]);
+                printk("    syscall parameters: "
+                       "0x%lx 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx\n",
+                       syscall[1], syscall[2], syscall[3],
+                       syscall[4], syscall[5], syscall[6]);
+                for(n = 1; n < data[0]/sizeof(long); n++) {
+                        if(n == 1)
+                                printk("    additional syscall data:");
+                        if(n % 4 == 1)
+                                printk("\n      ");
+                        printk("  0x%lx", data[n]);
+                }
+                if(n > 1)
+                        printk("\n");
         }
+        else ret = 0;
 
-        wait_stub_done(pid, 0, "one_syscall_stub");
+        *addr = check_init_stack(mm_idp, NULL);
 
-        return(*((unsigned long *) mm_idp->stack));
+        return ret;
 }
 
-int multi_count = 0;
-int multi_op_count = 0;
-
-static long many_syscall_stub(struct mm_id * mm_idp, int syscall,
-                              unsigned long *args, int done, void **addr_out)
+long run_syscall_stub(struct mm_id * mm_idp, int syscall,
+                      unsigned long *args, long expected, void **addr,
+                      int done)
 {
-        unsigned long regs[MAX_REG_NR], *stack;
-        int n, pid = mm_idp->u.pid;
+        unsigned long *stack = check_init_stack(mm_idp, *addr);
+
+        if(done && *addr == NULL)
+                single_count++;
+
+        *stack += sizeof(long);
+        stack += *stack / sizeof(long);
 
-        stack = *addr_out;
-        if(stack == NULL)
-                stack = (unsigned long *) current_stub_stack();
         *stack++ = syscall;
         *stack++ = args[0];
         *stack++ = args[1];
@@ -80,53 +124,55 @@ static long many_syscall_stub(struct mm_id * mm_idp, int syscall,
         *stack++ = args[3];
         *stack++ = args[4];
         *stack++ = args[5];
+        *stack++ = expected;
         *stack = 0;
         multi_op_count++;
 
         if(!done && ((((unsigned long) stack) & ~PAGE_MASK) <
-                     PAGE_SIZE - 8 * sizeof(long))){
-                *addr_out = stack;
+                     PAGE_SIZE - 10 * sizeof(long))){
+                *addr = stack;
                 return 0;
         }
 
-        multi_count++;
-        get_safe_registers(regs);
-        regs[REGS_IP_INDEX] = UML_CONFIG_STUB_CODE +
-                ((unsigned long) &batch_syscall_stub -
-                 (unsigned long) &__syscall_stub_start);
-        regs[REGS_SP_INDEX] = UML_CONFIG_STUB_DATA;
+        return do_syscall_stub(mm_idp, addr);
+}
 
-        n = ptrace_setregs(pid, regs);
-        if(n < 0){
-                printk("many_syscall_stub : PTRACE_SETREGS failed, "
-                       "errno = %d\n", n);
-                return(n);
-        }
+long syscall_stub_data(struct mm_id * mm_idp,
+                       unsigned long *data, int data_count,
+                       void **addr, void **stub_addr)
+{
+        unsigned long *stack;
+        int ret = 0;
+
+        /* If *addr still is uninitialized, it *must* contain NULL.
+         * Thus in this case do_syscall_stub correctly won't be called.
+         */
+        if((((unsigned long) *addr) & ~PAGE_MASK) >=
+           PAGE_SIZE - (10 + data_count) * sizeof(long)) {
+                ret = do_syscall_stub(mm_idp, addr);
+                /* in case of error, don't overwrite data on stack */
+                if(ret)
+                        return ret;
+        }
 
-        wait_stub_done(pid, 0, "many_syscall_stub");
-        stack = (unsigned long *) mm_idp->stack;
+        stack = check_init_stack(mm_idp, *addr);
+        *addr = stack;
 
-        *addr_out = stack;
-        return(*stack);
-}
+        *stack = data_count * sizeof(long);
 
-static long run_syscall_stub(struct mm_id * mm_idp, int syscall,
-                             unsigned long *args, void **addr, int done)
-{
-        long res;
+        memcpy(stack + 1, data, data_count * sizeof(long));
 
-        if((*addr == NULL) && done)
-                res = one_syscall_stub(mm_idp, syscall, args);
-        else res = many_syscall_stub(mm_idp, syscall, args, done, addr);
+        *stub_addr = (void *)(((unsigned long)(stack + 1) & ~PAGE_MASK) +
+                              UML_CONFIG_STUB_DATA);
 
-        return res;
+        return 0;
 }
 
-void *map(struct mm_id * mm_idp, unsigned long virt, unsigned long len,
-          int r, int w, int x, int phys_fd, unsigned long long offset,
-          int done, void *data)
+int map(struct mm_id * mm_idp, unsigned long virt, unsigned long len,
+        int r, int w, int x, int phys_fd, unsigned long long offset,
+        int done, void **data)
 {
-        int prot, n;
+        int prot, ret;
 
         prot = (r ? PROT_READ : 0) | (w ? PROT_WRITE : 0) |
                 (x ? PROT_EXEC : 0);
@@ -146,29 +192,27 @@ void *map(struct mm_id * mm_idp, unsigned long virt, unsigned long len,
                                                  .fd    = phys_fd,
                                                  .offset= offset
                                                } } } );
-                n = os_write_file(fd, &map, sizeof(map));
-                if(n != sizeof(map))
-                        printk("map : /proc/mm map failed, err = %d\n", -n);
+                ret = os_write_file(fd, &map, sizeof(map));
+                if(ret != sizeof(map))
+                        printk("map : /proc/mm map failed, err = %d\n", -ret);
+                else ret = 0;
         }
         else {
-                long res;
                 unsigned long args[] = { virt, len, prot,
                                          MAP_SHARED | MAP_FIXED, phys_fd,
                                          MMAP_OFFSET(offset) };
 
-                res = run_syscall_stub(mm_idp, STUB_MMAP_NR, args,
-                                       &data, done);
-                if((void *) res == MAP_FAILED)
-                        printk("mmap stub failed, errno = %d\n", res);
+                ret = run_syscall_stub(mm_idp, STUB_MMAP_NR, args, virt,
+                                       data, done);
         }
 
-        return data;
+        return ret;
 }
 
-void *unmap(struct mm_id * mm_idp, void *addr, unsigned long len, int done,
-            void *data)
+int unmap(struct mm_id * mm_idp, void *addr, unsigned long len, int done,
+          void **data)
 {
-        int n;
+        int ret;
 
         if(proc_mm){
                 struct proc_mm_op unmap;
@@ -180,29 +224,29 @@ void *unmap(struct mm_id * mm_idp, void *addr, unsigned long len, int done,
                                                  { .addr        =
                                                    (unsigned long) addr,
                                                    .len         = len } } } );
-                n = os_write_file(fd, &unmap, sizeof(unmap));
-                if(n != sizeof(unmap))
-                  printk("unmap - proc_mm write returned %d\n", n);
+                ret = os_write_file(fd, &unmap, sizeof(unmap));
+                if(ret != sizeof(unmap))
+                        printk("unmap - proc_mm write returned %d\n", ret);
+                else ret = 0;
         }
         else {
-                int res;
                 unsigned long args[] = { (unsigned long) addr, len, 0, 0, 0,
                                          0 };
 
-                res = run_syscall_stub(mm_idp, __NR_munmap, args,
-                                       &data, done);
-                if(res < 0)
-                        printk("munmap stub failed, errno = %d\n", res);
+                ret = run_syscall_stub(mm_idp, __NR_munmap, args, 0,
+                                       data, done);
+                if(ret < 0)
+                        printk("munmap stub failed, errno = %d\n", ret);
         }
 
-        return data;
+        return ret;
 }
 
-void *protect(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
-              int r, int w, int x, int done, void *data)
+int protect(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
+            int r, int w, int x, int done, void **data)
 {
         struct proc_mm_op protect;
-        int prot, n;
+        int prot, ret;
 
         prot = (r ? PROT_READ : 0) | (w ? PROT_WRITE : 0) |
                 (x ? PROT_EXEC : 0);
@@ -217,21 +261,19 @@ void *protect(struct mm_id * mm_idp, unsigned long addr, unsigned long len,
                                                      .len       = len,
                                                      .prot      = prot } } } );
 
-                n = os_write_file(fd, &protect, sizeof(protect));
-                if(n != sizeof(protect))
-                        panic("protect failed, err = %d", -n);
+                ret = os_write_file(fd, &protect, sizeof(protect));
+                if(ret != sizeof(protect))
+                        printk("protect failed, err = %d", -ret);
+                else ret = 0;
         }
         else {
-                int res;
                 unsigned long args[] = { addr, len, prot, 0, 0, 0 };
 
-                res = run_syscall_stub(mm_idp, __NR_mprotect, args,
-                                       &data, done);
-                if(res < 0)
-                        panic("mprotect stub failed, errno = %d\n", res);
+                ret = run_syscall_stub(mm_idp, __NR_mprotect, args, 0,
+                                       data, done);
         }
 
-        return data;
+        return ret;
 }
 
 void before_mem_skas(unsigned long unused)
diff --git a/arch/um/kernel/skas/tlb.c b/arch/um/kernel/skas/tlb.c
index 4b5fd2049547..6e84963dfc29 100644
--- a/arch/um/kernel/skas/tlb.c
+++ b/arch/um/kernel/skas/tlb.c
@@ -18,30 +18,31 @@
 #include "os.h"
 #include "tlb.h"
 
-static void *do_ops(union mm_context *mmu, struct host_vm_op *ops, int last,
-                    int finished, void *flush)
+static int do_ops(union mm_context *mmu, struct host_vm_op *ops, int last,
+                  int finished, void **flush)
 {
         struct host_vm_op *op;
-        int i;
+        int i, ret = 0;
 
-        for(i = 0; i <= last; i++){
+        for(i = 0; i <= last && !ret; i++){
                 op = &ops[i];
                 switch(op->type){
                 case MMAP:
-                        flush = map(&mmu->skas.id, op->u.mmap.addr,
-                                    op->u.mmap.len, op->u.mmap.r, op->u.mmap.w,
-                                    op->u.mmap.x, op->u.mmap.fd,
-                                    op->u.mmap.offset, finished, flush);
+                        ret = map(&mmu->skas.id, op->u.mmap.addr,
+                                  op->u.mmap.len, op->u.mmap.r, op->u.mmap.w,
+                                  op->u.mmap.x, op->u.mmap.fd,
+                                  op->u.mmap.offset, finished, flush);
                         break;
                 case MUNMAP:
-                        flush = unmap(&mmu->skas.id, (void *) op->u.munmap.addr,
-                                      op->u.munmap.len, finished, flush);
+                        ret = unmap(&mmu->skas.id,
+                                    (void *) op->u.munmap.addr,
+                                    op->u.munmap.len, finished, flush);
                         break;
                 case MPROTECT:
-                        flush = protect(&mmu->skas.id, op->u.mprotect.addr,
-                                        op->u.mprotect.len, op->u.mprotect.r,
-                                        op->u.mprotect.w, op->u.mprotect.x,
-                                        finished, flush);
+                        ret = protect(&mmu->skas.id, op->u.mprotect.addr,
+                                      op->u.mprotect.len, op->u.mprotect.r,
+                                      op->u.mprotect.w, op->u.mprotect.x,
+                                      finished, flush);
                         break;
                 default:
                         printk("Unknown op type %d in do_ops\n", op->type);
@@ -49,7 +50,7 @@ static void *do_ops(union mm_context *mmu, struct host_vm_op *ops, int last,
                 }
         }
 
-        return flush;
+        return ret;
 }
 
 extern int proc_mm;