#include <asm/unistd.h>
#include "entry.h"
+#include "paravirt_inst.h"
/*
* See Documentation/ia64/fsys.txt for details on fsyscalls.
.prologue
.altrp b6
.body
+ add r17=IA64_TASK_GROUP_LEADER_OFFSET,r16
+ ;;
+ ld8 r17=[r17] // r17 = current->group_leader
add r9=TI_FLAGS+IA64_TASK_SIZE,r16
;;
ld4 r9=[r9]
- add r8=IA64_TASK_TGID_OFFSET,r16
+ add r17=IA64_TASK_TGIDLINK_OFFSET,r17
;;
and r9=TIF_ALLWORK_MASK,r9
- ld4 r8=[r8] // r8 = current->tgid
+ ld8 r17=[r17] // r17 = current->group_leader->pids[PIDTYPE_PID].pid
+ ;;
+ add r8=IA64_PID_LEVEL_OFFSET,r17
+ ;;
+ ld4 r8=[r8] // r8 = pid->level
+ add r17=IA64_PID_UPID_OFFSET,r17 // r17 = &pid->numbers[0]
+ ;;
+ shl r8=r8,IA64_UPID_SHIFT
+ ;;
+ add r17=r17,r8 // r17 = &pid->numbers[pid->level]
+ ;;
+ ld4 r8=[r17] // r8 = pid->numbers[pid->level].nr
+ ;;
+ mov r17=0
;;
cmp.ne p8,p0=0,r9
(p8) br.spnt.many fsys_fallback_syscall
.altrp b6
.body
add r9=TI_FLAGS+IA64_TASK_SIZE,r16
+ add r17=IA64_TASK_TGIDLINK_OFFSET,r16
;;
ld4 r9=[r9]
tnat.z p6,p7=r32 // check argument register for being NaT
+ ld8 r17=[r17] // r17 = current->pids[PIDTYPE_PID].pid
;;
and r9=TIF_ALLWORK_MASK,r9
- add r8=IA64_TASK_PID_OFFSET,r16
+ add r8=IA64_PID_LEVEL_OFFSET,r17
add r18=IA64_TASK_CLEAR_CHILD_TID_OFFSET,r16
;;
- ld4 r8=[r8]
+ ld4 r8=[r8] // r8 = pid->level
+ add r17=IA64_PID_UPID_OFFSET,r17 // r17 = &pid->numbers[0]
+ ;;
+ shl r8=r8,IA64_UPID_SHIFT
+ ;;
+ add r17=r17,r8 // r17 = &pid->numbers[pid->level]
+ ;;
+ ld4 r8=[r17] // r8 = pid->numbers[pid->level].nr
+ ;;
cmp.ne p8,p0=0,r9
mov r17=-1
;;
// Note that instructions are optimized for McKinley. McKinley can
// process two bundles simultaneously and therefore we continuously
// try to feed the CPU two bundles and then a stop.
- //
- // Additional note that code has changed a lot. Optimization is TBD.
- // Comments begin with "?" are maybe outdated.
- tnat.nz p6,p0 = r31 // ? branch deferred to fit later bundle
- mov pr = r30,0xc000 // Set predicates according to function
+
add r2 = TI_FLAGS+IA64_TASK_SIZE,r16
+ tnat.nz p6,p0 = r31 // guard against Nat argument
+(p6) br.cond.spnt.few .fail_einval
movl r20 = fsyscall_gtod_data // load fsyscall gettimeofday data address
;;
+ ld4 r2 = [r2] // process work pending flags
movl r29 = itc_jitter_data // itc_jitter
add r22 = IA64_GTOD_WALL_TIME_OFFSET,r20 // wall_time
- ld4 r2 = [r2] // process work pending flags
- ;;
-(p15) add r22 = IA64_GTOD_MONO_TIME_OFFSET,r20 // monotonic_time
add r21 = IA64_CLKSRC_MMIO_OFFSET,r20
- add r19 = IA64_ITC_LASTCYCLE_OFFSET,r29
+ mov pr = r30,0xc000 // Set predicates according to function
+ ;;
and r2 = TIF_ALLWORK_MASK,r2
-(p6) br.cond.spnt.few .fail_einval // ? deferred branch
+ add r19 = IA64_ITC_LASTCYCLE_OFFSET,r29
+(p15) add r22 = IA64_GTOD_MONO_TIME_OFFSET,r20 // monotonic_time
;;
- add r26 = IA64_CLKSRC_CYCLE_LAST_OFFSET,r20 // clksrc_cycle_last
+ add r26 = IA64_CLKSRC_CYCLE_LAST_OFFSET,r20 // clksrc_cycle_last
cmp.ne p6, p0 = 0, r2 // Fallback if work is scheduled
-(p6) br.cond.spnt.many fsys_fallback_syscall
+(p6) br.cond.spnt.many fsys_fallback_syscall
;;
// Begin critical section
.time_redo:
(p9) cmp.eq p13,p0 = 0,r30 // if mmio_ptr, clear p13 jitter control
;;
.pred.rel.mutex p8,p9
-(p8) mov r2 = ar.itc // CPU_TIMER. 36 clocks latency!!!
+ MOV_FROM_ITC(p8, p6, r2, r10) // CPU_TIMER. 36 clocks latency!!!
(p9) ld8 r2 = [r30] // MMIO_TIMER. Could also have latency issues..
(p13) ld8 r25 = [r19] // get itc_lastcycle value
- ;; // ? could be removed by moving the last add upward
ld8 r9 = [r22],IA64_TIMESPEC_TV_NSEC_OFFSET // tv_sec
;;
ld8 r8 = [r22],-IA64_TIMESPEC_TV_NSEC_OFFSET // tv_nsec
EX(.fail_efault, probe.w.fault r31, 3)
xmpy.l f8 = f8,f7 // nsec_per_cyc*(counter-last_counter)
;;
- // ? simulate tbit.nz.or p7,p0 = r28,0
getf.sig r2 = f8
mf
;;
ld4 r10 = [r20] // gtod_lock.sequence
shr.u r2 = r2,r23 // shift by factor
- ;; // ? overloaded 3 bundles!
+ ;;
add r8 = r8,r2 // Add xtime.nsecs
cmp4.ne p7,p0 = r28,r10
(p7) br.cond.dpnt.few .time_redo // sequence number changed, redo
EX(.fail_efault, probe.w.fault r23, 3) // This also costs 5 cycles
(p14) xmpy.hu f8 = f8, f7 // xmpy has 5 cycles latency so use it
;;
- mov r8 = r0
(p14) getf.sig r2 = f8
;;
+ mov r8 = r0
(p14) shr.u r21 = r2, 4
;;
EX(.fail_efault, st8 [r31] = r9)
mov r17=(1 << (SIGKILL - 1)) | (1 << (SIGSTOP - 1))
;;
- rsm psr.i // mask interrupt delivery
+ RSM_PSR_I(p0, r18, r19) // mask interrupt delivery
mov ar.ccv=0
andcm r14=r14,r17 // filter out SIGKILL & SIGSTOP
#ifdef CONFIG_SMP
st4.rel [r31]=r0 // release the lock
#endif
- ssm psr.i
+ SSM_PSR_I(p0, p9, r31)
;;
srlz.d // ensure psr.i is set again
#ifdef CONFIG_SMP
st4.rel [r31]=r0 // release the lock
#endif
- ssm psr.i
+ SSM_PSR_I(p0, p9, r17)
;;
srlz.d
br.sptk.many fsys_fallback_syscall // with signal pending, do the heavy-weight syscall
#ifdef CONFIG_SMP
.lock_contention:
/* Rather than spinning here, fall back on doing a heavy-weight syscall. */
- ssm psr.i
+ SSM_PSR_I(p0, p9, r17)
;;
srlz.d
br.sptk.many fsys_fallback_syscall
adds r17=-1024,r15
movl r14=sys_call_table
;;
- rsm psr.i
+ RSM_PSR_I(p0, r26, r27)
shladd r18=r17,3,r14
;;
ld8 r18=[r18] // load normal (heavy-weight) syscall entry-point
- mov r29=psr // read psr (12 cyc load latency)
+ MOV_FROM_PSR(p0, r29, r26) // read psr (12 cyc load latency)
mov r27=ar.rsc
mov r21=ar.fpsr
mov r26=ar.pfs
END(fsys_fallback_syscall)
/* FALL THROUGH */
-GLOBAL_ENTRY(fsys_bubble_down)
+GLOBAL_ENTRY(paravirt_fsys_bubble_down)
.prologue
.altrp b6
.body
*
* PSR.BE : already is turned off in __kernel_syscall_via_epc()
* PSR.AC : don't care (kernel normally turns PSR.AC on)
- * PSR.I : already turned off by the time fsys_bubble_down gets
+ * PSR.I : already turned off by the time paravirt_fsys_bubble_down gets
* invoked
* PSR.DFL: always 0 (kernel never turns it on)
* PSR.DFH: don't care --- kernel never touches f32-f127 on its own
* PSR.DB : don't care --- kernel never enables kernel-level
* breakpoints
* PSR.TB : must be 0 already; if it wasn't zero on entry to
- * __kernel_syscall_via_epc, the branch to fsys_bubble_down
+ * __kernel_syscall_via_epc, the branch to paravirt_fsys_bubble_down
* will trigger a taken branch; the taken-trap-handler then
* converts the syscall into a break-based system-call.
*/
nop.i 0
;;
mov ar.rsc=0 // M2 set enforced lazy mode, pl 0, LE, loadrs=0
+#ifdef CONFIG_VIRT_CPU_ACCOUNTING
+ MOV_FROM_ITC(p0, p6, r30, r23) // M get cycle for accounting
+#else
nop.m 0
+#endif
nop.i 0
;;
mov r23=ar.bspstore // M2 (12 cyc) save ar.bspstore
cmp.ne pKStk,pUStk=r0,r0 // A set pKStk <- 0, pUStk <- 1
br.call.sptk.many b7=ia64_syscall_setup // B
;;
+#ifdef CONFIG_VIRT_CPU_ACCOUNTING
+ // mov.m r30=ar.itc is called in advance
+ add r16=TI_AC_STAMP+IA64_TASK_SIZE,r2
+ add r17=TI_AC_LEAVE+IA64_TASK_SIZE,r2
+ ;;
+ ld8 r18=[r16],TI_AC_STIME-TI_AC_STAMP // time at last check in kernel
+ ld8 r19=[r17],TI_AC_UTIME-TI_AC_LEAVE // time at leave kernel
+ ;;
+ ld8 r20=[r16],TI_AC_STAMP-TI_AC_STIME // cumulated stime
+ ld8 r21=[r17] // cumulated utime
+ sub r22=r19,r18 // stime before leave kernel
+ ;;
+ st8 [r16]=r30,TI_AC_STIME-TI_AC_STAMP // update stamp
+ sub r18=r30,r19 // elapsed time in user mode
+ ;;
+ add r20=r20,r22 // sum stime
+ add r21=r21,r18 // sum utime
+ ;;
+ st8 [r16]=r20 // update stime
+ st8 [r17]=r21 // update utime
+ ;;
+#endif
mov ar.rsc=0x3 // M2 set eager mode, pl 0, LE, loadrs=0
mov rp=r14 // I0 set the real return addr
and r3=_TIF_SYSCALL_TRACEAUDIT,r3 // A
;;
- ssm psr.i // M2 we're on kernel stacks now, reenable irqs
+ SSM_PSR_I(p0, p6, r22) // M2 we're on kernel stacks now, reenable irqs
cmp.eq p8,p0=r3,r0 // A
(p10) br.cond.spnt.many ia64_ret_from_syscall // B return if bad call-frame or r15 is a NaT
nop.m 0
(p8) br.call.sptk.many b6=b6 // B (ignore return address)
br.cond.spnt ia64_trace_syscall // B
-END(fsys_bubble_down)
+END(paravirt_fsys_bubble_down)
.rodata
.align 8
- .globl fsyscall_table
+ .globl paravirt_fsyscall_table
- data8 fsys_bubble_down
-fsyscall_table:
+ data8 paravirt_fsys_bubble_down
+paravirt_fsyscall_table:
data8 fsys_ni_syscall
data8 0 // exit // 1025
data8 0 // read
// fill in zeros for the remaining entries
.zero:
- .space fsyscall_table + 8*NR_syscalls - .zero, 0
+ .space paravirt_fsyscall_table + 8*NR_syscalls - .zero, 0
projects / linux-2.6-omap-h63xx.git / blobdiff