[Qemu-devel] [PATCH 2/6] linux-user: Provide safe_syscall for i386

[Richard Henderson]

Signed-off-by: Richard Henderson <address@hidden>
---
 linux-user/host/i386/hostdep.h          |  34 ++++++++++
 linux-user/host/i386/safe-syscall.inc.S | 110 ++++++++++++++++++++++++++++++++
 2 files changed, 144 insertions(+)
 create mode 100644 linux-user/host/i386/hostdep.h
 create mode 100644 linux-user/host/i386/safe-syscall.inc.S

diff --git a/linux-user/host/i386/hostdep.h b/linux-user/host/i386/hostdep.h
new file mode 100644
index 0000000..9e2b4d7
--- /dev/null
+++ b/linux-user/host/i386/hostdep.h
@@ -0,0 +1,34 @@
+/*
+ * hostdep.h : things which are dependent on the host architecture
+ *
+ * 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_EIP];
+
+    if (*pcreg > (uintptr_t)safe_syscall_start
+        && *pcreg < (uintptr_t)safe_syscall_end) {
+        *pcreg = (uintptr_t)safe_syscall_start;
+    }
+}
+
+#endif /* __ASSEMBLER__ */
+
+#endif
diff --git a/linux-user/host/i386/safe-syscall.inc.S 
b/linux-user/host/i386/safe-syscall.inc.S
new file mode 100644
index 0000000..f5f0c64
--- /dev/null
+++ b/linux-user/host/i386/safe-syscall.inc.S
@@ -0,0 +1,110 @@
+/*
+ * 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
+ *
+ * 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:
+        .cfi_startproc
+        push    %ebp
+        .cfi_adjust_cfa_offset 4
+        .cfi_rel_offset ebp, 0
+       push    %esi
+        .cfi_adjust_cfa_offset 4
+        .cfi_rel_offset esi, 0
+       push    %edi
+        .cfi_adjust_cfa_offset 4
+        .cfi_rel_offset edi, 0
+       push    %ebx
+        .cfi_adjust_cfa_offset 4
+        .cfi_rel_offset ebx, 0
+
+        /* The syscall calling convention isn't the same as the C one:
+         * we enter with 0(%esp) == return address
+         *               4(%esp) == *signal_pending
+         *               8(%esp) == syscall number
+         *               12(%esp) ... 32(%esp) == syscall arguments
+         *               and return the result in eax
+         * and the syscall instruction needs
+         *               eax == syscall number
+         *               ebx, ecx, edx, esi, edi, ebp == syscall arguments
+         *               and returns the result in eax
+         * Shuffle everything around appropriately.
+        * Note the 16 bytes that we pushed to save registers.
+         */
+        mov     12+16(%esp), %ebx       /* the syscall arguments */
+        mov     16+16(%esp), %ecx
+        mov     20+16(%esp), %edx
+        mov     24+16(%esp), %esi
+        mov     28+16(%esp), %edi
+        mov     32+16(%esp), %ebp
+
+        /* 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 */
+       mov     4+16(%esp), %eax        /* signal_pending */
+        cmp    $0, (%eax)
+        mov     8+16(%esp), %eax        /* syscall number */
+        jnz     1f
+        int    $0x80
+safe_syscall_end:
+        /* code path for having successfully executed the syscall */
+        pop     %ebx
+        .cfi_remember_state
+        .cfi_def_cfa_offset 4
+        .cfi_restore ebx
+       pop     %edi
+        .cfi_def_cfa_offset 4
+        .cfi_restore edi
+       pop     %esi
+        .cfi_def_cfa_offset 4
+        .cfi_restore esi
+       pop     %ebp
+        .cfi_def_cfa_offset 4
+        .cfi_restore ebp
+        ret
+
+1:
+        /* code path when we didn't execute the syscall */
+        .cfi_restore_state
+        mov     $-TARGET_ERESTARTSYS, %eax
+        pop     %ebx
+        .cfi_remember_state
+        .cfi_def_cfa_offset 4
+        .cfi_restore ebx
+       pop     %edi
+        .cfi_def_cfa_offset 4
+        .cfi_restore edi
+       pop     %esi
+        .cfi_def_cfa_offset 4
+        .cfi_restore esi
+       pop     %ebp
+        .cfi_def_cfa_offset 4
+        .cfi_restore ebp
+        ret
+        .cfi_endproc
+
+        .size   safe_syscall_base, .-safe_syscall_base
-- 
2.5.5