Merge remote-tracking branch 'remotes/riku/tags/pull-linux-user-20160527' into staging

linux-user pull request v2 for may 2016

# gpg: Signature made Fri 27 May 2016 12:51:10 BST using RSA key ID DE3C9BC0
# gpg: Good signature from "Riku Voipio <riku.voipio@iki.fi>"
# gpg:                 aka "Riku Voipio <riku.voipio@linaro.org>"

* remotes/riku/tags/pull-linux-user-20160527: (38 commits)
  linux-user,target-ppc: fix use of MSR_LE
  linux-user/signal.c: Use s390 target space address instead of host space
  linux-user/signal.c: Use target address instead of host address for microblaze restorer
  linux-user/signal.c: Generate opcode data for restorer in setup_rt_frame
  linux-user: arm: Remove ARM_cpsr and similar #defines
  linux-user: Use direct syscalls for setuid(), etc
  linux-user: x86_64: Don't use 16-bit UIDs
  linux-user: Use g_try_malloc() in do_msgrcv()
  linux-user: Handle msgrcv error case correctly
  linux-user: Handle negative values in timespec conversion
  linux-user: Use safe_syscall for futex syscall
  linux-user: Use safe_syscall for pselect, select syscalls
  linux-user: Use safe_syscall for execve syscall
  linux-user: Use safe_syscall for wait system calls
  linux-user: Use safe_syscall for open and openat system calls
  linux-user: Use safe_syscall for read and write system calls
  linux-user: Provide safe_syscall for fixing races between signals and syscalls
  linux-user: Add debug code to exercise restarting system calls
  linux-user: Support for restarting system calls for Microblaze targets
  linux-user: Set r14 on exit from microblaze syscall
  ...
Signed-off-by: NPeter Maydell <peter.maydell@linaro.org>

Makefile.target

@@ -108,7 +108,12 @@ obj-$(CONFIG_LIBDECNUMBER) += libdecnumber/dpd/decimal128.o
 
 ifdef CONFIG_LINUX_USER
 
-QEMU_CFLAGS+=-I$(SRC_PATH)/linux-user/$(TARGET_ABI_DIR) -I$(SRC_PATH)/linux-user
+# Note that we only add linux-user/host/$ARCH if it exists, and
+# that it must come before linux-user/host/generic in the search path.
+QEMU_CFLAGS+=-I$(SRC_PATH)/linux-user/$(TARGET_ABI_DIR) \
+             $(patsubst %,-I%,$(wildcard $(SRC_PATH)/linux-user/host/$(ARCH))) \
+             -I$(SRC_PATH)/linux-user/host/generic \
+             -I$(SRC_PATH)/linux-user
 
 obj-y += linux-user/
 obj-y += gdbstub.o thunk.o user-exec.o

linux-user/Makefile.objs

 obj-y = main.o syscall.o strace.o mmap.o signal.o \
-	elfload.o linuxload.o uaccess.o uname.o
+	elfload.o linuxload.o uaccess.o uname.o \
+	safe-syscall.o
 
 obj-$(TARGET_HAS_BFLT) += flatload.o
 obj-$(TARGET_I386) += vm86.o

linux-user/alpha/target_signal.h

@@ -27,6 +27,7 @@ static inline abi_ulong get_sp_from_cpustate(CPUAlphaState *state)
     return state->ir[IR_SP];
 }
 
+
 /* From <asm/gentrap.h>.  */
 #define TARGET_GEN_INTOVF      -1      /* integer overflow */
 #define TARGET_GEN_INTDIV      -2      /* integer division by zero */

linux-user/arm/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUARMState *state)
    return state->regs[13];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/arm/target_syscall.h

@@ -4,29 +4,11 @@
 /* this struct defines the way the registers are stored on the
    stack during a system call. */
 
+/* uregs[0..15] are r0 to r15; uregs[16] is CPSR; uregs[17] is ORIG_r0 */
 struct target_pt_regs {
     abi_long uregs[18];
 };
 
-#define ARM_cpsr	uregs[16]
-#define ARM_pc		uregs[15]
-#define ARM_lr		uregs[14]
-#define ARM_sp		uregs[13]
-#define ARM_ip		uregs[12]
-#define ARM_fp		uregs[11]
-#define ARM_r10		uregs[10]
-#define ARM_r9		uregs[9]
-#define ARM_r8		uregs[8]
-#define ARM_r7		uregs[7]
-#define ARM_r6		uregs[6]
-#define ARM_r5		uregs[5]
-#define ARM_r4		uregs[4]
-#define ARM_r3		uregs[3]
-#define ARM_r2		uregs[2]
-#define ARM_r1		uregs[1]
-#define ARM_r0		uregs[0]
-#define ARM_ORIG_r0	uregs[17]
-
 #define ARM_SYSCALL_BASE	0x900000
 #define ARM_THUMB_SYSCALL	0
 

linux-user/cris/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUCRISState *state)
     return state->regs[14];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/elfload.c

@@ -274,19 +274,20 @@ static inline void init_thread(struct target_pt_regs *regs,
     abi_long stack = infop->start_stack;
     memset(regs, 0, sizeof(*regs));
 
-    regs->ARM_cpsr = 0x10;
-    if (infop->entry & 1)
-        regs->ARM_cpsr |= CPSR_T;
-    regs->ARM_pc = infop->entry & 0xfffffffe;
-    regs->ARM_sp = infop->start_stack;
+    regs->uregs[16] = ARM_CPU_MODE_USR;
+    if (infop->entry & 1) {
+        regs->uregs[16] |= CPSR_T;
+    }
+    regs->uregs[15] = infop->entry & 0xfffffffe;
+    regs->uregs[13] = infop->start_stack;
     /* FIXME - what to for failure of get_user()? */
-    get_user_ual(regs->ARM_r2, stack + 8); /* envp */
-    get_user_ual(regs->ARM_r1, stack + 4); /* envp */
+    get_user_ual(regs->uregs[2], stack + 8); /* envp */
+    get_user_ual(regs->uregs[1], stack + 4); /* envp */
     /* XXX: it seems that r0 is zeroed after ! */
-    regs->ARM_r0 = 0;
+    regs->uregs[0] = 0;
     /* For uClinux PIC binaries.  */
     /* XXX: Linux does this only on ARM with no MMU (do we care ?) */
-    regs->ARM_r10 = infop->start_data;
+    regs->uregs[10] = infop->start_data;
 }
 
 #define ELF_NREG    18

linux-user/errno_defs.h

@@ -139,3 +139,20 @@
 /* for robust mutexes */
 #define TARGET_EOWNERDEAD      130     /* Owner died */
 #define TARGET_ENOTRECOVERABLE 131     /* State not recoverable */
+
+/* QEMU internal, not visible to the guest. This is returned when a
+ * system call should be restarted, to tell the main loop that it
+ * should wind the guest PC backwards so it will re-execute the syscall
+ * after handling any pending signals. They match with the ones the guest
+ * kernel uses for the same purpose.
+ */
+#define TARGET_ERESTARTSYS     512     /* Restart system call (if SA_RESTART) */
+
+/* QEMU internal, not visible to the guest. This is returned by the
+ * do_sigreturn() code after a successful sigreturn syscall, to indicate
+ * that it has correctly set the guest registers and so the main loop
+ * should not touch them. We use the value the guest would use for
+ * ERESTART_NOINTR (which is kernel internal) to guarantee that we won't
+ * clash with a valid guest errno now or in the future.
+ */
+#define TARGET_QEMU_ESIGRETURN 513     /* Return from signal */

linux-user/host/generic/hostdep.h

+/*
+ * hostdep.h : fallback generic version of header for things
+ * which are dependent on the host architecture
+ *
+ *  * Written by Peter Maydell <peter.maydell@linaro.org>
+ *
+ * Copyright (C) 2016 Linaro Limited
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_HOSTDEP_H
+#define QEMU_HOSTDEP_H
+
+/* This is the fallback header which is only used if the host
+ * architecture doesn't provide one in linux-user/host/$ARCH.
+ */
+
+#endif

linux-user/host/x86_64/hostdep.h

+/*
+ * hostdep.h : things which are dependent on the host architecture
+ *
+ *  * Written by Peter Maydell <peter.maydell@linaro.org>
+ *
+ * Copyright (C) 2016 Linaro Limited
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_HOSTDEP_H
+#define QEMU_HOSTDEP_H
+
+/* We have a safe-syscall.inc.S */
+#define HAVE_SAFE_SYSCALL
+
+#ifndef __ASSEMBLER__
+
+/* These are defined by the safe-syscall.inc.S file */
+extern char safe_syscall_start[];
+extern char safe_syscall_end[];
+
+/* Adjust the signal context to rewind out of safe-syscall if we're in it */
+static inline void rewind_if_in_safe_syscall(void *puc)
+{
+    struct ucontext *uc = puc;
+    greg_t *pcreg = &uc->uc_mcontext.gregs[REG_RIP];
+
+    if (*pcreg > (uintptr_t)safe_syscall_start
+        && *pcreg < (uintptr_t)safe_syscall_end) {
+        *pcreg = (uintptr_t)safe_syscall_start;
+    }
+}
+
+#endif /* __ASSEMBLER__ */
+
+#endif

linux-user/host/x86_64/safe-syscall.inc.S

+/*
+ * safe-syscall.inc.S : host-specific assembly fragment
+ * to handle signals occurring at the same time as system calls.
+ * This is intended to be included by linux-user/safe-syscall.S
+ *
+ * Copyright (C) 2015 Timothy Edward Baldwin <T.E.Baldwin99@members.leeds.ac.uk>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+        .global safe_syscall_base
+        .global safe_syscall_start
+        .global safe_syscall_end
+        .type   safe_syscall_base, @function
+
+        /* This is the entry point for making a system call. The calling
+         * convention here is that of a C varargs function with the
+         * first argument an 'int *' to the signal_pending flag, the
+         * second one the system call number (as a 'long'), and all further
+         * arguments being syscall arguments (also 'long').
+         * We return a long which is the syscall's return value, which
+         * may be negative-errno on failure. Conversion to the
+         * -1-and-errno-set convention is done by the calling wrapper.
+         */
+safe_syscall_base:
+        /* This saves a frame pointer and aligns the stack for the syscall.
+         * (It's unclear if the syscall ABI has the same stack alignment
+         * requirements as the userspace function call ABI, but better safe than
+         * sorry. Appendix A2 of http://www.x86-64.org/documentation/abi.pdf
+         * does not list any ABI differences regarding stack alignment.)
+         */
+        push    %rbp
+
+        /* The syscall calling convention isn't the same as the
+         * C one:
+         * we enter with rdi == *signal_pending
+         *               rsi == syscall number
+         *               rdx, rcx, r8, r9, (stack), (stack) == syscall arguments
+         *               and return the result in rax
+         * and the syscall instruction needs
+         *               rax == syscall number
+         *               rdi, rsi, rdx, r10, r8, r9 == syscall arguments
+         *               and returns the result in rax
+         * Shuffle everything around appropriately.
+         * Note that syscall will trash rcx and r11.
+         */
+        mov     %rsi, %rax /* syscall number */
+        mov     %rdi, %rbp /* signal_pending pointer */
+        /* and the syscall arguments */
+        mov     %rdx, %rdi
+        mov     %rcx, %rsi
+        mov     %r8,  %rdx
+        mov     %r9,  %r10
+        mov     16(%rsp), %r8
+        mov     24(%rsp), %r9
+
+        /* This next sequence of code works in conjunction with the
+         * rewind_if_safe_syscall_function(). If a signal is taken
+         * and the interrupted PC is anywhere between 'safe_syscall_start'
+         * and 'safe_syscall_end' then we rewind it to 'safe_syscall_start'.
+         * The code sequence must therefore be able to cope with this, and
+         * the syscall instruction must be the final one in the sequence.
+         */
+safe_syscall_start:
+        /* if signal_pending is non-zero, don't do the call */
+        testl   $1, (%rbp)
+        jnz     return_ERESTARTSYS
+        syscall
+safe_syscall_end:
+        /* code path for having successfully executed the syscall */
+        pop     %rbp
+        ret
+
+return_ERESTARTSYS:
+        /* code path when we didn't execute the syscall */
+        mov     $-TARGET_ERESTARTSYS, %rax
+        pop     %rbp
+        ret
+
+        .size   safe_syscall_base, .-safe_syscall_base

linux-user/m68k/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUM68KState *state)
     return state->aregs[7];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/main.c

@@ -285,6 +285,7 @@ void cpu_loop(CPUX86State *env)
     CPUState *cs = CPU(x86_env_get_cpu(env));
     int trapnr;
     abi_ulong pc;
+    abi_ulong ret;
     target_siginfo_t info;
 
     for(;;) {
@@ -294,28 +295,38 @@ void cpu_loop(CPUX86State *env)
         switch(trapnr) {
         case 0x80:
             /* linux syscall from int $0x80 */
-            env->regs[R_EAX] = do_syscall(env,
-                                          env->regs[R_EAX],
-                                          env->regs[R_EBX],
-                                          env->regs[R_ECX],
-                                          env->regs[R_EDX],
-                                          env->regs[R_ESI],
-                                          env->regs[R_EDI],
-                                          env->regs[R_EBP],
-                                          0, 0);
+            ret = do_syscall(env,
+                             env->regs[R_EAX],
+                             env->regs[R_EBX],
+                             env->regs[R_ECX],
+                             env->regs[R_EDX],
+                             env->regs[R_ESI],
+                             env->regs[R_EDI],
+                             env->regs[R_EBP],
+                             0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->eip -= 2;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[R_EAX] = ret;
+            }
             break;
 #ifndef TARGET_ABI32
         case EXCP_SYSCALL:
             /* linux syscall from syscall instruction */
-            env->regs[R_EAX] = do_syscall(env,
-                                          env->regs[R_EAX],
-                                          env->regs[R_EDI],
-                                          env->regs[R_ESI],
-                                          env->regs[R_EDX],
-                                          env->regs[10],
-                                          env->regs[8],
-                                          env->regs[9],
-                                          0, 0);
+            ret = do_syscall(env,
+                             env->regs[R_EAX],
+                             env->regs[R_EDI],
+                             env->regs[R_ESI],
+                             env->regs[R_EDX],
+                             env->regs[10],
+                             env->regs[8],
+                             env->regs[9],
+                             0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->eip -= 2;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[R_EAX] = ret;
+            }
             break;
 #endif
         case EXCP0B_NOSEG:
@@ -716,6 +727,7 @@ void cpu_loop(CPUARMState *env)
     unsigned int n, insn;
     target_siginfo_t info;
     uint32_t addr;
+    abi_ulong ret;
 
     for(;;) {
         cpu_exec_start(cs);
@@ -854,15 +866,20 @@ void cpu_loop(CPUARMState *env)
                             break;
                         }
                     } else {
-                        env->regs[0] = do_syscall(env,
-                                                  n,
-                                                  env->regs[0],
-                                                  env->regs[1],
-                                                  env->regs[2],
-                                                  env->regs[3],
-                                                  env->regs[4],
-                                                  env->regs[5],
-                                                  0, 0);
+                        ret = do_syscall(env,
+                                         n,
+                                         env->regs[0],
+                                         env->regs[1],
+                                         env->regs[2],
+                                         env->regs[3],
+                                         env->regs[4],
+                                         env->regs[5],
+                                         0, 0);
+                        if (ret == -TARGET_ERESTARTSYS) {
+                            env->regs[15] -= env->thumb ? 2 : 4;
+                        } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                            env->regs[0] = ret;
+                        }
                     }
                 } else {
                     goto error;
@@ -1045,6 +1062,7 @@ void cpu_loop(CPUARMState *env)
 {
     CPUState *cs = CPU(arm_env_get_cpu(env));
     int trapnr, sig;
+    abi_long ret;
     target_siginfo_t info;
 
     for (;;) {
@@ -1054,15 +1072,20 @@ void cpu_loop(CPUARMState *env)
 
         switch (trapnr) {
         case EXCP_SWI:
-            env->xregs[0] = do_syscall(env,
-                                       env->xregs[8],
-                                       env->xregs[0],
-                                       env->xregs[1],
-                                       env->xregs[2],
-                                       env->xregs[3],
-                                       env->xregs[4],
-                                       env->xregs[5],
-                                       0, 0);
+            ret = do_syscall(env,
+                             env->xregs[8],
+                             env->xregs[0],
+                             env->xregs[1],
+                             env->xregs[2],
+                             env->xregs[3],
+                             env->xregs[4],
+                             env->xregs[5],
+                             0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->pc -= 4;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->xregs[0] = ret;
+            }
             break;
         case EXCP_INTERRUPT:
             /* just indicate that signals should be handled asap */
@@ -1148,7 +1171,7 @@ void cpu_loop(CPUUniCore32State *env)
                             cpu_set_tls(env, env->regs[0]);
                             env->regs[0] = 0;
                     } else {
-                        env->regs[0] = do_syscall(env,
+                        abi_long ret = do_syscall(env,
                                                   n,
                                                   env->regs[0],
                                                   env->regs[1],
@@ -1157,6 +1180,11 @@ void cpu_loop(CPUUniCore32State *env)
                                                   env->regs[4],
                                                   env->regs[5],
                                                   0, 0);
+                        if (ret == -TARGET_ERESTARTSYS) {
+                            env->regs[31] -= 4;
+                        } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                            env->regs[0] = ret;
+                        }
                     }
                 } else {
                     goto error;
@@ -1353,6 +1381,9 @@ void cpu_loop (CPUSPARCState *env)
                               env->regwptr[2], env->regwptr[3],
                               env->regwptr[4], env->regwptr[5],
                               0, 0);
+            if (ret == -TARGET_ERESTARTSYS || ret == -TARGET_QEMU_ESIGRETURN) {
+                break;
+            }
             if ((abi_ulong)ret >= (abi_ulong)(-515)) {
 #if defined(TARGET_SPARC64) && !defined(TARGET_ABI32)
                 env->xcc |= PSR_CARRY;
@@ -1964,6 +1995,10 @@ void cpu_loop(CPUPPCState *env)
             ret = do_syscall(env, env->gpr[0], env->gpr[3], env->gpr[4],
                              env->gpr[5], env->gpr[6], env->gpr[7],
                              env->gpr[8], 0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->nip -= 4;
+                break;
+            }
             if (ret == (target_ulong)(-TARGET_QEMU_ESIGRETURN)) {
                 /* Returning from a successful sigreturn syscall.
                    Avoid corrupting register state.  */
@@ -2505,6 +2540,10 @@ done_syscall:
                              env->active_tc.gpr[8], env->active_tc.gpr[9],
                              env->active_tc.gpr[10], env->active_tc.gpr[11]);
 # endif /* O32 */
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->active_tc.PC -= 4;
+                break;
+            }
             if (ret == -TARGET_QEMU_ESIGRETURN) {
                 /* Returning from a successful sigreturn syscall.
                    Avoid clobbering register state.  */
@@ -2685,6 +2724,7 @@ void cpu_loop(CPUOpenRISCState *env)
 {
     CPUState *cs = CPU(openrisc_env_get_cpu(env));
     int trapnr, gdbsig;
+    abi_long ret;
 
     for (;;) {
         cpu_exec_start(cs);
@@ -2730,14 +2770,19 @@ void cpu_loop(CPUOpenRISCState *env)
             break;
         case EXCP_SYSCALL:
             env->pc += 4;   /* 0xc00; */
-            env->gpr[11] = do_syscall(env,
-                                      env->gpr[11], /* return value       */
-                                      env->gpr[3],  /* r3 - r7 are params */
-                                      env->gpr[4],
-                                      env->gpr[5],
-                                      env->gpr[6],
-                                      env->gpr[7],
-                                      env->gpr[8], 0, 0);
+            ret = do_syscall(env,
+                             env->gpr[11], /* return value       */
+                             env->gpr[3],  /* r3 - r7 are params */
+                             env->gpr[4],
+                             env->gpr[5],
+                             env->gpr[6],
+                             env->gpr[7],
+                             env->gpr[8], 0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->pc -= 4;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->gpr[11] = ret;
+            }
             break;
         case EXCP_FPE:
             qemu_log_mask(CPU_LOG_INT, "\nFloating point error\n");
@@ -2792,7 +2837,11 @@ void cpu_loop(CPUSH4State *env)
                              env->gregs[0],
                              env->gregs[1],
                              0, 0);
-            env->gregs[0] = ret;
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->pc -= 2;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->gregs[0] = ret;
+            }
             break;
         case EXCP_INTERRUPT:
             /* just indicate that signals should be handled asap */
@@ -2865,7 +2914,11 @@ void cpu_loop(CPUCRISState *env)
                              env->pregs[7], 
                              env->pregs[11],
                              0, 0);
-            env->regs[10] = ret;
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->pc -= 2;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[10] = ret;
+            }
             break;
         case EXCP_DEBUG:
             {
@@ -2929,7 +2982,19 @@ void cpu_loop(CPUMBState *env)
                              env->regs[9], 
                              env->regs[10],
                              0, 0);
-            env->regs[3] = ret;
+            if (ret == -TARGET_ERESTARTSYS) {
+                /* Wind back to before the syscall. */
+                env->sregs[SR_PC] -= 4;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[3] = ret;
+            }
+            /* All syscall exits result in guest r14 being equal to the
+             * PC we return to, because the kernel syscall exit "rtbd" does
+             * this. (This is true even for sigreturn(); note that r14 is
+             * not a userspace-usable register, as the kernel may clobber it
+             * at any point.)
+             */
+            env->regs[14] = env->sregs[SR_PC];
             break;
         case EXCP_HW_EXCP:
             env->regs[17] = env->sregs[SR_PC] + 4;
@@ -3037,18 +3102,24 @@ void cpu_loop(CPUM68KState *env)
             break;
         case EXCP_TRAP0:
             {
+                abi_long ret;
                 ts->sim_syscalls = 0;
                 n = env->dregs[0];
                 env->pc += 2;
-                env->dregs[0] = do_syscall(env,
-                                          n,
-                                          env->dregs[1],
-                                          env->dregs[2],
-                                          env->dregs[3],
-                                          env->dregs[4],
-                                          env->dregs[5],
-                                          env->aregs[0],
-                                          0, 0);
+                ret = do_syscall(env,
+                                 n,
+                                 env->dregs[1],
+                                 env->dregs[2],
+                                 env->dregs[3],
+                                 env->dregs[4],
+                                 env->dregs[5],
+                                 env->aregs[0],
+                                 0, 0);
+                if (ret == -TARGET_ERESTARTSYS) {
+                    env->pc -= 2;
+                } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                    env->dregs[0] = ret;
+                }
             }
             break;
         case EXCP_INTERRUPT:
@@ -3229,8 +3300,11 @@ void cpu_loop(CPUAlphaState *env)
                                     env->ir[IR_A2], env->ir[IR_A3],
                                     env->ir[IR_A4], env->ir[IR_A5],
                                     0, 0);
-                if (trapnr == TARGET_NR_sigreturn
-                    || trapnr == TARGET_NR_rt_sigreturn) {
+                if (sysret == -TARGET_ERESTARTSYS) {
+                    env->pc -= 4;
+                    break;
+                }
+                if (sysret == -TARGET_QEMU_ESIGRETURN) {
                     break;
                 }
                 /* Syscall writes 0 to V0 to bypass error check, similar
@@ -3327,6 +3401,7 @@ void cpu_loop(CPUS390XState *env)
     int trapnr, n, sig;
     target_siginfo_t info;
     target_ulong addr;
+    abi_long ret;
 
     while (1) {
         cpu_exec_start(cs);
@@ -3344,9 +3419,14 @@ void cpu_loop(CPUS390XState *env)
                 n = env->regs[1];
             }
             env->psw.addr += env->int_svc_ilen;
-            env->regs[2] = do_syscall(env, n, env->regs[2], env->regs[3],
-                                      env->regs[4], env->regs[5],
-                                      env->regs[6], env->regs[7], 0, 0);
+            ret = do_syscall(env, n, env->regs[2], env->regs[3],
+                             env->regs[4], env->regs[5],
+                             env->regs[6], env->regs[7], 0, 0);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->psw.addr -= env->int_svc_ilen;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[2] = ret;
+            }
             break;
 
         case EXCP_DEBUG:
@@ -3638,15 +3718,20 @@ void cpu_loop(CPUTLGState *env)
         cpu_exec_end(cs);
         switch (trapnr) {
         case TILEGX_EXCP_SYSCALL:
-            env->regs[TILEGX_R_RE] = do_syscall(env, env->regs[TILEGX_R_NR],
-                                                env->regs[0], env->regs[1],
-                                                env->regs[2], env->regs[3],
-                                                env->regs[4], env->regs[5],
-                                                env->regs[6], env->regs[7]);
-            env->regs[TILEGX_R_ERR] = TILEGX_IS_ERRNO(env->regs[TILEGX_R_RE])
-                                                      ? - env->regs[TILEGX_R_RE]
-                                                      : 0;
+        {
+            abi_ulong ret = do_syscall(env, env->regs[TILEGX_R_NR],
+                                       env->regs[0], env->regs[1],
+                                       env->regs[2], env->regs[3],
+                                       env->regs[4], env->regs[5],
+                                       env->regs[6], env->regs[7]);
+            if (ret == -TARGET_ERESTARTSYS) {
+                env->pc -= 8;
+            } else if (ret != -TARGET_QEMU_ESIGRETURN) {
+                env->regs[TILEGX_R_RE] = ret;
+                env->regs[TILEGX_R_ERR] = TILEGX_IS_ERRNO(ret) ? -ret : 0;
+            }
             break;
+        }
         case TILEGX_EXCP_OPCODE_EXCH:
             do_exch(env, true, false);
             break;

linux-user/microblaze/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUMBState *state)
     return state->regs[14];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/mips/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUMIPSState *state)
     return state->active_tc.gpr[29];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/mips/target_syscall.h

@@ -222,10 +222,6 @@ struct target_pt_regs {
 #define TARGET_ENOTRECOVERABLE 166     /* State not recoverable */
 
 
-
-/* Nasty hack: define a fake errno value for use by sigreturn.  */
-#define TARGET_QEMU_ESIGRETURN 255
-
 #define UNAME_MACHINE "mips"
 #define UNAME_MINIMUM_RELEASE "2.6.32"
 

linux-user/mips64/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUMIPSState *state)
     return state->active_tc.gpr[29];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/mips64/target_syscall.h

@@ -219,10 +219,6 @@ struct target_pt_regs {
 #define TARGET_ENOTRECOVERABLE 166     /* State not recoverable */
 
 
-
-/* Nasty hack: define a fake errno value for use by sigreturn. */
-#define TARGET_QEMU_ESIGRETURN 255
-
 #define UNAME_MACHINE "mips64"
 #define UNAME_MINIMUM_RELEASE "2.6.32"
 

linux-user/openrisc/target_signal.h

@@ -23,4 +23,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUOpenRISCState *state)
     return state->gpr[1];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/ppc/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUPPCState *state)
     return state->gpr[1];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/ppc/target_syscall.h

@@ -53,8 +53,6 @@ struct target_revectored_struct {
 	abi_ulong __map[8];			/* 256 bits */
 };
 
-/* Nasty hack: define a fake errno value for use by sigreturn.  */
-#define TARGET_QEMU_ESIGRETURN 255
 
 /*
  * flags masks

linux-user/qemu.h

 #ifndef QEMU_H
 #define QEMU_H
 
-
+#include "hostdep.h"
 #include "cpu.h"
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
@@ -205,6 +205,131 @@ unsigned long init_guest_space(unsigned long host_start,
 
 #include "qemu/log.h"
 
+/* safe_syscall.S */
+
+/**
+ * safe_syscall:
+ * @int number: number of system call to make
+ * ...: arguments to the system call
+ *
+ * Call a system call if guest signal not pending.
+ * This has the same API as the libc syscall() function, except that it
+ * may return -1 with errno == TARGET_ERESTARTSYS if a signal was pending.
+ *
+ * Returns: the system call result, or -1 with an error code in errno
+ * (Errnos are host errnos; we rely on TARGET_ERESTARTSYS not clashing
+ * with any of the host errno values.)
+ */
+
+/* A guide to using safe_syscall() to handle interactions between guest
+ * syscalls and guest signals:
+ *
+ * Guest syscalls come in two flavours:
+ *
+ * (1) Non-interruptible syscalls
+ *
+ * These are guest syscalls that never get interrupted by signals and
+ * so never return EINTR. They can be implemented straightforwardly in
+ * QEMU: just make sure that if the implementation code has to make any
+ * blocking calls that those calls are retried if they return EINTR.
+ * It's also OK to implement these with safe_syscall, though it will be
+ * a little less efficient if a signal is delivered at the 'wrong' moment.
+ *
+ * (2) Interruptible syscalls
+ *
+ * These are guest syscalls that can be interrupted by signals and
+ * for which we need to either return EINTR or arrange for the guest
+ * syscall to be restarted. This category includes both syscalls which
+ * always restart (and in the kernel return -ERESTARTNOINTR), ones
+ * which only restart if there is no handler (kernel returns -ERESTARTNOHAND
+ * or -ERESTART_RESTARTBLOCK), and the most common kind which restart
+ * if the handler was registered with SA_RESTART (kernel returns
+ * -ERESTARTSYS). System calls which are only interruptible in some
+ * situations (like 'open') also need to be handled this way.
+ *
+ * Here it is important that the host syscall is made
+ * via this safe_syscall() function, and *not* via the host libc.
+ * If the host libc is used then the implementation will appear to work
+ * most of the time, but there will be a race condition where a
+ * signal could arrive just before we make the host syscall inside libc,
+ * and then then guest syscall will not correctly be interrupted.
+ * Instead the implementation of the guest syscall can use the safe_syscall
+ * function but otherwise just return the result or errno in the usual
+ * way; the main loop code will take care of restarting the syscall
+ * if appropriate.
+ *
+ * (If the implementation needs to make multiple host syscalls this is
+ * OK; any which might really block must be via safe_syscall(); for those
+ * which are only technically blocking (ie which we know in practice won't
+ * stay in the host kernel indefinitely) it's OK to use libc if necessary.
+ * You must be able to cope with backing out correctly if some safe_syscall
+ * you make in the implementation returns either -TARGET_ERESTARTSYS or
+ * EINTR though.)
+ *
+ *
+ * How and why the safe_syscall implementation works:
+ *
+ * The basic setup is that we make the host syscall via a known
+ * section of host native assembly. If a signal occurs, our signal
+ * handler checks the interrupted host PC against the addresse of that
+ * known section. If the PC is before or at the address of the syscall
+ * instruction then we change the PC to point at a "return
+ * -TARGET_ERESTARTSYS" code path instead, and then exit the signal handler
+ * (causing the safe_syscall() call to immediately return that value).
+ * Then in the main.c loop if we see this magic return value we adjust
+ * the guest PC to wind it back to before the system call, and invoke
+ * the guest signal handler as usual.
+ *
+ * This winding-back will happen in two cases:
+ * (1) signal came in just before we took the host syscall (a race);
+ *   in this case we'll take the guest signal and have another go
+ *   at the syscall afterwards, and this is indistinguishable for the
+ *   guest from the timing having been different such that the guest
+ *   signal really did win the race
+ * (2) signal came in while the host syscall was blocking, and the
+ *   host kernel decided the syscall should be restarted;
+ *   in this case we want to restart the guest syscall also, and so
+ *   rewinding is the right thing. (Note that "restart" semantics mean
+ *   "first call the signal handler, then reattempt the syscall".)
+ * The other situation to consider is when a signal came in while the
+ * host syscall was blocking, and the host kernel decided that the syscall
+ * should not be restarted; in this case QEMU's host signal handler will
+ * be invoked with the PC pointing just after the syscall instruction,
+ * with registers indicating an EINTR return; the special code in the
+ * handler will not kick in, and we will return EINTR to the guest as
+ * we should.
+ *
+ * Notice that we can leave the host kernel to make the decision for
+ * us about whether to do a restart of the syscall or not; we do not
+ * need to check SA_RESTART flags in QEMU or distinguish the various
+ * kinds of restartability.
+ */
+#ifdef HAVE_SAFE_SYSCALL
+/* The core part of this function is implemented in assembly */
+extern long safe_syscall_base(int *pending, long number, ...);
+
+#define safe_syscall(...)                                               \
+    ({                                                                  \
+        long ret_;                                                      \
+        int *psp_ = &((TaskState *)thread_cpu->opaque)->signal_pending; \
+        ret_ = safe_syscall_base(psp_, __VA_ARGS__);                    \
+        if (is_error(ret_)) {                                           \
+            errno = -ret_;                                              \
+            ret_ = -1;                                                  \
+        }                                                               \
+        ret_;                                                           \
+    })
+
+#else
+
+/* Fallback for architectures which don't yet provide a safe-syscall assembly
+ * fragment; note that this is racy!
+ * This should go away when all host architectures have been updated.
+ */
+#define safe_syscall syscall
+
+#endif
+
 /* syscall.c */
 int host_to_target_waitstatus(int status);
 

linux-user/s390x/target_signal.h

@@ -23,4 +23,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUS390XState *state)
    return state->regs[15];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/safe-syscall.S

+/*
+ * safe-syscall.S : include the host-specific assembly fragment
+ * to handle signals occurring at the same time as system calls.
+ *
+ * Written by Peter Maydell <peter.maydell@linaro.org>
+ *
+ * Copyright (C) 2016 Linaro Limited
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "hostdep.h"
+#include "errno_defs.h"
+
+/* We have the correct host directory on our include path
+ * so that this will pull in the right fragment for the architecture.
+ */
+#ifdef HAVE_SAFE_SYSCALL
+#include "safe-syscall.inc.S"
+#endif
+
+/* We must specifically say that we're happy for the stack to not be
+ * executable, otherwise the toolchain will default to assuming our
+ * assembly needs an executable stack and the whole QEMU binary will
+ * needlessly end up with one. This should be the last thing in this file.
+ */
+#if defined(__linux__) && defined(__ELF__)
+.section        .note.GNU-stack, "", %progbits
+#endif

linux-user/sh4/target_signal.h

@@ -26,4 +26,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUSH4State *state)
     return state->gregs[15];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/sparc/target_signal.h

@@ -33,4 +33,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUSPARCState *state)
     return state->regwptr[UREG_FP];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/sparc64/target_signal.h

@@ -33,4 +33,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUSPARCState *state)
     return state->regwptr[UREG_FP];
 }
 
+
 #endif /* TARGET_SIGNAL_H */

linux-user/syscall_defs.h

@@ -55,7 +55,8 @@
 #define TARGET_IOC_NRBITS	8
 #define TARGET_IOC_TYPEBITS	8
 
-#if defined(TARGET_I386) || (defined(TARGET_ARM) && defined(TARGET_ABI32)) \
+#if (defined(TARGET_I386) && defined(TARGET_ABI32)) \
+    || (defined(TARGET_ARM) && defined(TARGET_ABI32)) \
     || defined(TARGET_SPARC) \
     || defined(TARGET_M68K) || defined(TARGET_SH4) || defined(TARGET_CRIS)
     /* 16 bit uid wrappers emulation */

linux-user/tilegx/target_signal.h

@@ -25,4 +25,5 @@ static inline abi_ulong get_sp_from_cpustate(CPUTLGState *state)
     return state->regs[TILEGX_R_SP];
 }
 
+
 #endif /* TARGET_SIGNAL_H */