[IA64] clean up ptrace corner cases

Patch from yanmin.zhang@intel.com to fix up some corner cases
in ptrace.  Many thanks to davidm for reviewing and improving.
Signed-off-by: Tony Luck <tony.luck@intel.com>

arch/ia64/kernel/entry.S

@@ -51,8 +51,11 @@
 	 * setup a null register window frame.
 	 */
 ENTRY(ia64_execve)
-	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(3)
-	alloc loc1=ar.pfs,3,2,4,0
+	/*
+	 * Allocate 8 input registers since ptrace() may clobber them
+	 */
+	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
+	alloc loc1=ar.pfs,8,2,4,0
 	mov loc0=rp
 	.body
 	mov out0=in0			// filename
@@ -113,8 +116,11 @@ END(ia64_execve)
  *	      u64 tls)
  */
 GLOBAL_ENTRY(sys_clone2)
-	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(6)
-	alloc r16=ar.pfs,6,2,6,0
+	/*
+	 * Allocate 8 input registers since ptrace() may clobber them
+	 */
+	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
+	alloc r16=ar.pfs,8,2,6,0
 	DO_SAVE_SWITCH_STACK
 	adds r2=PT(R16)+IA64_SWITCH_STACK_SIZE+16,sp
 	mov loc0=rp
@@ -142,8 +148,11 @@ END(sys_clone2)
  *	Deprecated.  Use sys_clone2() instead.
  */
 GLOBAL_ENTRY(sys_clone)
-	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(5)
-	alloc r16=ar.pfs,5,2,6,0
+	/*
+	 * Allocate 8 input registers since ptrace() may clobber them
+	 */
+	.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
+	alloc r16=ar.pfs,8,2,6,0
 	DO_SAVE_SWITCH_STACK
 	adds r2=PT(R16)+IA64_SWITCH_STACK_SIZE+16,sp
 	mov loc0=rp
@@ -1212,7 +1221,10 @@ END(sys_rt_sigsuspend)
 
 ENTRY(sys_rt_sigreturn)
 	PT_REGS_UNWIND_INFO(0)
-	alloc r2=ar.pfs,0,0,1,0
+	/*
+	 * Allocate 8 input registers since ptrace() may clobber them
+	 */
+	alloc r2=ar.pfs,8,0,1,0
 	.prologue
 	PT_REGS_SAVES(16)
 	adds sp=-16,sp

arch/ia64/kernel/fsys.S

@@ -612,7 +612,8 @@ GLOBAL_ENTRY(fsys_bubble_down)
 	;;
 	mov rp=r2				// set the real return addr
 	tbit.z p8,p0=r3,TIF_SYSCALL_TRACE
-
+	;;
+(p10)	br.cond.spnt.many ia64_ret_from_syscall	// p10==true means out registers are more than 8
 (p8)	br.call.sptk.many b6=b6		// ignore this return addr
 	br.cond.sptk ia64_trace_syscall
 END(fsys_bubble_down)

arch/ia64/kernel/gate.S

@@ -81,6 +81,7 @@ GLOBAL_ENTRY(__kernel_syscall_via_epc)
 	LOAD_FSYSCALL_TABLE(r14)
 
 	mov r16=IA64_KR(CURRENT)		// 12 cycle read latency
+	tnat.nz p10,p9=r15
 	mov r19=NR_syscalls-1
 	;;
 	shladd r18=r17,3,r14
@@ -119,7 +120,8 @@ GLOBAL_ENTRY(__kernel_syscall_via_epc)
 #endif
 
 	mov r10=-1
-	mov r8=ENOSYS
+(p10)	mov r8=EINVAL
+(p9)	mov r8=ENOSYS
 	FSYS_RETURN
 END(__kernel_syscall_via_epc)
 

arch/ia64/kernel/ivt.S

@@ -51,6 +51,7 @@
 #include <asm/system.h>
 #include <asm/thread_info.h>
 #include <asm/unistd.h>
+#include <asm/errno.h>
 
 #if 1
 # define PSR_DEFAULT_BITS	psr.ac
@@ -732,10 +733,12 @@ ENTRY(break_fault)
 	ssm psr.ic | PSR_DEFAULT_BITS
 	;;
 	srlz.i					// guarantee that interruption collection is on
+	mov r3=NR_syscalls - 1
 	;;
 (p15)	ssm psr.i				// restore psr.i
+	// p10==true means out registers are more than 8 or r15's Nat is true
+(p10)	br.cond.spnt.many ia64_ret_from_syscall
 	;;
-	mov r3=NR_syscalls - 1
 	movl r16=sys_call_table
 
 	adds r15=-1024,r15			// r15 contains the syscall number---subtract 1024
@@ -836,8 +839,11 @@ END(interrupt)
 	 * On exit:
 	 *	- executing on bank 1 registers
 	 *	- psr.ic enabled, interrupts restored
+	 *	-  p10: TRUE if syscall is invoked with more than 8 out
+	 *		registers or r15's Nat is true
 	 *	-  r1: kernel's gp
 	 *	-  r3: preserved (same as on entry)
+	 *	-  r8: -EINVAL if p10 is true
 	 *	- r12: points to kernel stack
 	 *	- r13: points to current task
 	 *	- p15: TRUE if interrupts need to be re-enabled
@@ -871,12 +877,17 @@ GLOBAL_ENTRY(ia64_syscall_setup)
 	;;
 
 	st8 [r16]=r19,PT(AR_RNAT)-PT(CR_IFS)	// store ar.pfs.pfm in cr.ifs
+	extr.u r11=r19,7,7	// I0		// get sol of ar.pfs
+	and r8=0x7f,r19		// A		// get sof of ar.pfs
+
 	st8 [r17]=r27,PT(AR_BSPSTORE)-PT(AR_RSC)// save ar.rsc
+	tbit.nz p15,p0=r29,IA64_PSR_I_BIT // I0
 (p9)	mov in1=-1
+	;;
 
 (pUStk) sub r18=r18,r22				// r18=RSE.ndirty*8
-	tbit.nz p15,p0=r29,IA64_PSR_I_BIT
 	tnat.nz p10,p0=in2
+	add r11=8,r11
 	;;
 (pKStk) adds r16=PT(PR)-PT(AR_RNAT),r16		// skip over ar_rnat field
 (pKStk) adds r17=PT(B0)-PT(AR_BSPSTORE),r17	// skip over ar_bspstore field
@@ -904,25 +915,29 @@ GLOBAL_ENTRY(ia64_syscall_setup)
 (p13)	mov in5=-1
 	;;
 	st8 [r16]=r21,PT(R8)-PT(AR_FPSR)	// save ar.fpsr
-	st8.spill [r17]=r15			// save r15
 	tnat.nz p14,p0=in6
+	cmp.lt p10,p9=r11,r8	// frame size can't be more than local+8
 	;;
 	stf8 [r16]=f1		// ensure pt_regs.r8 != 0 (see handle_syscall_error)
+(p9)	tnat.nz p10,p0=r15
 	adds r12=-16,r1		// switch to kernel memory stack (with 16 bytes of scratch)
+
+	st8.spill [r17]=r15			// save r15
 	tnat.nz p8,p0=in7
+	nop.i 0
 
 	mov r13=r2				// establish `current'
 	movl r1=__gp				// establish kernel global pointer
 	;;
 (p14)	mov in6=-1
 (p8)	mov in7=-1
-	tnat.nz p9,p0=r15
+	nop.i 0
 
 	cmp.eq pSys,pNonSys=r0,r0		// set pSys=1, pNonSys=0
 	movl r17=FPSR_DEFAULT
 	;;
 	mov.m ar.fpsr=r17			// set ar.fpsr to kernel default value
-(p9)	mov r15=-1
+(p10)	mov r8=-EINVAL
 	br.ret.sptk.many b7
 END(ia64_syscall_setup)
 

arch/ia64/kernel/process.c

@@ -632,7 +632,7 @@ dump_fpu (struct pt_regs *pt, elf_fpregset_t dst)
 	return 1;	/* f0-f31 are always valid so we always return 1 */
 }
 
-asmlinkage long
+long
 sys_execve (char __user *filename, char __user * __user *argv, char __user * __user *envp,
 	    struct pt_regs *regs)
 {

include/asm-ia64/unistd.h

@@ -374,7 +374,7 @@ asmlinkage unsigned long sys_mmap2(
 				int fd, long pgoff);
 struct pt_regs;
 struct sigaction;
-asmlinkage long sys_execve(char __user *filename, char __user * __user *argv,
+long sys_execve(char __user *filename, char __user * __user *argv,
 			   char __user * __user *envp, struct pt_regs *regs);
 asmlinkage long sys_pipe(long arg0, long arg1, long arg2, long arg3,
 			long arg4, long arg5, long arg6, long arg7, long stack);