[PATCH] fast vdso implementation for CLOCK_THREAD_CPUTIME_ID

The extract cpu time instruction (ectg) instruction allows the user
process to get the current thread cputime without calling into the
kernel. The code that uses the instruction needs to switch to the
access registers mode to get access to the per-cpu info page that
contains the two base values that are needed to calculate the current
cputime from the CPU timer with the ectg instruction.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This commit is contained in:
Martin Schwidefsky
2008-12-31 15:11:42 +01:00
parent 9cfb9b3c3a
commit c742b31c03
10 changed files with 222 additions and 32 deletions

View File

@@ -22,8 +22,10 @@ __kernel_clock_gettime:
larl %r5,_vdso_data
cghi %r2,CLOCK_REALTIME
je 4f
cghi %r2,-2 /* CLOCK_THREAD_CPUTIME_ID for this thread */
je 9f
cghi %r2,CLOCK_MONOTONIC
jne 9f
jne 12f
/* CLOCK_MONOTONIC */
ltgr %r3,%r3
@@ -42,7 +44,7 @@ __kernel_clock_gettime:
alg %r0,__VDSO_WTOM_SEC(%r5)
clg %r4,__VDSO_UPD_COUNT(%r5) /* check update counter */
jne 0b
larl %r5,10f
larl %r5,13f
1: clg %r1,0(%r5)
jl 2f
slg %r1,0(%r5)
@@ -68,7 +70,7 @@ __kernel_clock_gettime:
lg %r0,__VDSO_XTIME_SEC(%r5)
clg %r4,__VDSO_UPD_COUNT(%r5) /* check update counter */
jne 5b
larl %r5,10f
larl %r5,13f
6: clg %r1,0(%r5)
jl 7f
slg %r1,0(%r5)
@@ -79,11 +81,38 @@ __kernel_clock_gettime:
8: lghi %r2,0
br %r14
/* CLOCK_THREAD_CPUTIME_ID for this thread */
9: icm %r0,15,__VDSO_ECTG_OK(%r5)
jz 12f
ear %r2,%a4
llilh %r4,0x0100
sar %a4,%r4
lghi %r4,0
sacf 512 /* Magic ectg instruction */
.insn ssf,0xc80100000000,__VDSO_ECTG_BASE(4),__VDSO_ECTG_USER(4),4
sacf 0
sar %a4,%r2
algr %r1,%r0 /* r1 = cputime as TOD value */
mghi %r1,1000 /* convert to nanoseconds */
srlg %r1,%r1,12 /* r1 = cputime in nanosec */
lgr %r4,%r1
larl %r5,13f
srlg %r1,%r1,9 /* divide by 1000000000 */
mlg %r0,8(%r5)
srlg %r0,%r0,11 /* r0 = tv_sec */
stg %r0,0(%r3)
msg %r0,0(%r5) /* calculate tv_nsec */
slgr %r4,%r0 /* r4 = tv_nsec */
stg %r4,8(%r3)
lghi %r2,0
br %r14
/* Fallback to system call */
9: lghi %r1,__NR_clock_gettime
12: lghi %r1,__NR_clock_gettime
svc 0
br %r14
10: .quad 1000000000
13: .quad 1000000000
14: .quad 19342813113834067
.cfi_endproc
.size __kernel_clock_gettime,.-__kernel_clock_gettime