Reorganize syscall entry and leave handling.

Extend struct sysvec with three new elements:
sv_fetch_syscall_args - the method to fetch syscall arguments from
  usermode into struct syscall_args. The structure is machine-depended
  (this might be reconsidered after all architectures are converted).
sv_set_syscall_retval - the method to set a return value for usermode
  from the syscall. It is a generalization of
  cpu_set_syscall_retval(9) to allow ABIs to override the way to set a
  return value.
sv_syscallnames - the table of syscall names.

Use sv_set_syscall_retval in kern_sigsuspend() instead of hardcoding
the call to cpu_set_syscall_retval().

The new functions syscallenter(9) and syscallret(9) are provided that
use sv_*syscall* pointers and contain the common repeated code from
the syscall() implementations for the architecture-specific syscall
trap handlers.

Syscallenter() fetches arguments, calls syscall implementation from
ABI sysent table, and set up return frame. The end of syscall
bookkeeping is done by syscallret().

Take advantage of single place for MI syscall handling code and
implement ptrace_lwpinfo pl_flags PL_FLAG_SCE, PL_FLAG_SCX and
PL_FLAG_EXEC. The SCE and SCX flags notify the debugger that the
thread is stopped at syscall entry or return point respectively.  The
EXEC flag augments SCX and notifies debugger that the process address
space was changed by one of exec(2)-family syscalls.

The i386, amd64, sparc64, sun4v, powerpc and ia64 syscall()s are
changed to use syscallenter()/syscallret(). MIPS and arm are not
converted and use the mostly unchanged syscall() implementation.

Reviewed by:	jhb, marcel, marius, nwhitehorn, stas
Tested by:	marcel (ia64), marius (sparc64), nwhitehorn (powerpc),
	stas (mips)
MFC after:	1 month

1 files changed, 38 insertions, 122 deletions

diff --git a/sys/amd64/ia32/ia32_syscall.c b/sys/amd64/ia32/ia32_syscall.c
index aa1ae6c..6a649ae 100644
--- a/sys/amd64/ia32/ia32_syscall.c
+++ b/sys/amd64/ia32/ia32_syscall.c
@@ -56,6 +56,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/ktr.h>
 #include <sys/lock.h>
 #include <sys/mutex.h>
+#include <sys/proc.h>
 #include <sys/ptrace.h>
 #include <sys/resourcevar.h>
 #include <sys/signalvar.h>
@@ -81,62 +82,54 @@ __FBSDID("$FreeBSD$");
 #include <machine/intr_machdep.h>
 #include <machine/md_var.h>
 
+#include <compat/freebsd32/freebsd32_util.h>
+
 #define	IDTVEC(name)	__CONCAT(X,name)
 
 extern inthand_t IDTVEC(int0x80_syscall), IDTVEC(rsvd);
-extern const char *freebsd32_syscallnames[];
 
 void ia32_syscall(struct trapframe *frame);	/* Called from asm code */
 
-struct ia32_syscall_args {
-	u_int code;
-	caddr_t params;
-	struct sysent *callp;
-	u_int64_t args64[8];
-	int narg;
-};
+void
+ia32_set_syscall_retval(struct thread *td, int error)
+{
 
-static int
-fetch_ia32_syscall_args(struct thread *td, struct ia32_syscall_args *sa)
+	cpu_set_syscall_retval(td, error);
+}
+
+int
+ia32_fetch_syscall_args(struct thread *td, struct syscall_args *sa)
 {
 	struct proc *p;
 	struct trapframe *frame;
+	caddr_t params;
 	u_int32_t args[8];
 	int error, i;
 
 	p = td->td_proc;
 	frame = td->td_frame;
 
-	sa->params = (caddr_t)frame->tf_rsp + sizeof(u_int32_t);
+	params = (caddr_t)frame->tf_rsp + sizeof(u_int32_t);
 	sa->code = frame->tf_rax;
 
-	if (p->p_sysent->sv_prepsyscall) {
+	/*
+	 * Need to check if this is a 32 bit or 64 bit syscall.
+	 */
+	if (sa->code == SYS_syscall) {
 		/*
-		 * The prep code is MP aware.
+		 * Code is first argument, followed by actual args.
 		 */
-		(*p->p_sysent->sv_prepsyscall)(frame, args, &sa->code,
-		    &sa->params);
-	} else {
+		sa->code = fuword32(params);
+		params += sizeof(int);
+	} else if (sa->code == SYS___syscall) {
 		/*
-		 * Need to check if this is a 32 bit or 64 bit syscall.
-		 * fuword is MP aware.
+		 * Like syscall, but code is a quad, so as to maintain
+		 * quad alignment for the rest of the arguments.
+		 * We use a 32-bit fetch in case params is not
+		 * aligned.
 		 */
-		if (sa->code == SYS_syscall) {
-			/*
-			 * Code is first argument, followed by actual args.
-			 */
-			sa->code = fuword32(sa->params);
-			sa->params += sizeof(int);
-		} else if (sa->code == SYS___syscall) {
-			/*
-			 * Like syscall, but code is a quad, so as to maintain
-			 * quad alignment for the rest of the arguments.
-			 * We use a 32-bit fetch in case params is not
-			 * aligned.
-			 */
-			sa->code = fuword32(sa->params);
-			sa->params += sizeof(quad_t);
-		}
+		sa->code = fuword32(params);
+		params += sizeof(quad_t);
 	}
  	if (p->p_sysent->sv_mask)
  		sa->code &= p->p_sysent->sv_mask;
@@ -146,19 +139,19 @@ fetch_ia32_syscall_args(struct thread *td, struct ia32_syscall_args *sa)
  		sa->callp = &p->p_sysent->sv_table[sa->code];
 	sa->narg = sa->callp->sy_narg;
 
-	if (sa->params != NULL && sa->narg != 0)
-		error = copyin(sa->params, (caddr_t)args,
+	if (params != NULL && sa->narg != 0)
+		error = copyin(params, (caddr_t)args,
 		    (u_int)(sa->narg * sizeof(int)));
 	else
 		error = 0;
 
 	for (i = 0; i < sa->narg; i++)
-		sa->args64[i] = args[i];
+		sa->args[i] = args[i];
 
-#ifdef KTRACE
-	if (KTRPOINT(td, KTR_SYSCALL))
-		ktrsyscall(sa->code, sa->narg, sa->args64);
-#endif
+	if (error == 0) {
+		td->td_retval[0] = 0;
+		td->td_retval[1] = frame->tf_rdx;
+	}
 
 	return (error);
 }
@@ -167,58 +160,16 @@ void
 ia32_syscall(struct trapframe *frame)
 {
 	struct thread *td;
-	struct proc *p;
-	struct ia32_syscall_args sa;
+	struct syscall_args sa;
 	register_t orig_tf_rflags;
 	int error;
 	ksiginfo_t ksi;
 
-	PCPU_INC(cnt.v_syscall);
+	orig_tf_rflags = frame->tf_rflags;
 	td = curthread;
-	p = td->td_proc;
-	td->td_syscalls++;
-
-	td->td_pticks = 0;
 	td->td_frame = frame;
-	if (td->td_ucred != p->p_ucred) 
-		cred_update_thread(td);
-	orig_tf_rflags = frame->tf_rflags;
-	if (p->p_flag & P_TRACED) {
-		PROC_LOCK(p);
-		td->td_dbgflags &= ~TDB_USERWR;
-		PROC_UNLOCK(p);
-	}
-	error = fetch_ia32_syscall_args(td, &sa);
-
-	CTR4(KTR_SYSC, "syscall enter thread %p pid %d proc %s code %d", td,
-	    td->td_proc->p_pid, td->td_name, sa.code);
-
-	if (error == 0) {
-		td->td_retval[0] = 0;
-		td->td_retval[1] = frame->tf_rdx;
 
-		STOPEVENT(p, S_SCE, sa.narg);
-		PTRACESTOP_SC(p, td, S_PT_SCE);
-		if (td->td_dbgflags & TDB_USERWR) {
-			/*
-			 * Reread syscall number and arguments if
-			 * debugger modified registers or memory.
-			 */
-			error = fetch_ia32_syscall_args(td, &sa);
-			if (error != 0)
-				goto retval;
-			td->td_retval[1] = frame->tf_rdx;
-		}
-
-		AUDIT_SYSCALL_ENTER(sa.code, td);
-		error = (*sa.callp->sy_call)(td, sa.args64);
-		AUDIT_SYSCALL_EXIT(error, td);
-
-		/* Save the latest error return value. */
-		td->td_errno = error;
-	}
- retval:
-	cpu_set_syscall_retval(td, error);
+	error = syscallenter(td, &sa);
 
 	/*
 	 * Traced syscall.
@@ -232,44 +183,9 @@ ia32_syscall(struct trapframe *frame)
 		trapsignal(td, &ksi);
 	}
 
-	/*
-	 * Check for misbehavior.
-	 */
-	WITNESS_WARN(WARN_PANIC, NULL, "System call %s returning",
-	    (sa.code >= 0 && sa.code < SYS_MAXSYSCALL) ?
-	     freebsd32_syscallnames[sa.code] : "???");
-	KASSERT(td->td_critnest == 0,
-	    ("System call %s returning in a critical section",
-	    (sa.code >= 0 && sa.code < SYS_MAXSYSCALL) ?
-	     freebsd32_syscallnames[sa.code] : "???"));
-	KASSERT(td->td_locks == 0,
-	    ("System call %s returning with %d locks held",
-	    (sa.code >= 0 && sa.code < SYS_MAXSYSCALL) ?
-	     freebsd32_syscallnames[sa.code] : "???", td->td_locks));
-
-	/*
-	 * Handle reschedule and other end-of-syscall issues
-	 */
-	userret(td, frame);
-
-	CTR4(KTR_SYSC, "syscall exit thread %p pid %d proc %s code %d", td,
-	    td->td_proc->p_pid, td->td_proc->p_comm, sa.code);
-#ifdef KTRACE
-	if (KTRPOINT(td, KTR_SYSRET))
-		ktrsysret(sa.code, error, td->td_retval[0]);
-#endif
-
-	/*
-	 * This works because errno is findable through the
-	 * register set.  If we ever support an emulation where this
-	 * is not the case, this code will need to be revisited.
-	 */
-	STOPEVENT(p, S_SCX, sa.code);
- 
-	PTRACESTOP_SC(p, td, S_PT_SCX);
+	syscallret(td, error, &sa);
 }
 
-
 static void
 ia32_syscall_enable(void *dummy)
 {