Mon system shows high CPU

John,

thanks for the data

Are you actually running something, which could be generating that much interrupts ?

Please post SDA> PCS SHOW TRACE/STAT summary data, not the trace data itself.

Could you put the dump on a WO:R directory on EISNER:: and send me mail (EISNER::HALLE) ?

Please consider to run a quick test with PersonalAlpha to try and verify, if these seem to be OpenVMS or emulator related problems. Much less work than copying this system disk to your real Alpha...

Volker.<hr>

John,

initial analysis of the MONITOR.DMP file indicates, that there should not have been a HPARITH trap, so this could be a floating point emulation problem in the emulator:

DBG&gt; exa/ins @pc-20:@pc+10
SHARE$MONITOR_DATA2+147FC: LDQ_U R31,(SP)
SHARE$MONITOR_DATA2+14800: LDQ R19,#XF2C8(R2)
SHARE$MONITOR_DATA2+14804: LDS F10,#XF32C(R2)
SHARE$MONITOR_DATA2+14808: STL R31,(R19)
SHARE$MONITOR_DATA2+1480C: LDS F11,#XF378(R2)
SHARE$MONITOR_DATA2+14810: LDQ R20,#XF2C0(R2)
SHARE$MONITOR_DATA2+14814: BIS R31,R4,R16
SHARE$MONITOR_DATA2+14818: BIS R31,R3,R17
SHARE$MONITOR_DATA2+1481C: LDA R27,#XFE90(R2)
SHARE$MONITOR_DATA2+14820: DIVS/SU F11,F10,F12
SHARE$MONITOR_DATA2+14824: STS F12,(R20)
SHARE$MONITOR_DATA2+14828: TRAPB
SHARE$MONITOR_DATA2+1482C: BSR R26,#X0005D8
DBG&gt; exa/s_float @r2+0fffff32c
SHARE$MONITOR_DATA0+714: 100.0000
DBG&gt; exa/s_float @r2+0fffff378
SHARE$MONITOR_DATA0+760: 40.00000
DBG&gt; exa/s_float f10
0&#92;%F10: 100.0000
DBG&gt; exa/s_float f11
0&#92;%F11: 40.00000
DBG&gt; exa/s_float f12
0&#92;%F12: 0.4000000
DBG&gt; exa f10
0&#92;%F10: 100.000003814697
DBG&gt; exa f11
0&#92;%F11: 40.0000000000000
DBG&gt; exa f12
0&#92;%F12: 0.399999999642373

Dividing 100./40. should give about 0.4 and not report an HPARITH exception, so the correct execution of the DIVS/SU F11,F10,F12 instruction is most questionable.

And if you look at the value stored in F10, it looks 'suspicious':

DBG&gt; exa/hex f10
0&#92;%F10: 4059000010000000
DBG&gt; exa f10
0&#92;%F10: 100.000003814697

The instruction LDS F10,#XF32C(R2) can NOT have picked up that bit in the low-order longword.

The instruction stream matches with the Fmask=00001000 as reported in the exception (= F12 has been modified). Also summary=01 indicates Software Completion, which matches the /S flag.

Which version of FreeAXP are you using ?

Volker.<hr>

The emulator claims:
x86 version 1.0 patch 6 build 271 (Apr 10 2010 21:04:10)

I also had a crash last night:

NETDLLERR, DECnet Datalink Layer detected a fatal error.

Looks like it occurred when the MacBook came out of power save state.

John,

could you also send me the CLUE file from the crash (CLUE$COLLECT:CLUE$node_ddmmyy_hhmm.LIS) for analysis.

Volker.

We have just posted FreeAXP 283. It corrects a floating point bug that may impact Monitor behavior. Visit the FreeAXP page for download, documentation, and forum links:

<center> www.FreeAXP.com </center>

I removed SRI IDLE and that solved the excessive CPU problem.