To Test
Peters error code was tested yesterday without errors - all ok.
Try with errors (switch on all fibres). Results below.
16h10 DSP code from /daqsoft/sct/devdsp/2005-11-03
/daqsoft/sct/SctRodDaq_0104 > MdspFlashLoad -s10 Enter binary file name, including extension (q to quit): /daqsoft/sct/devdsp/2005-11-03/mdsp120b2.bin <ipramFile>/daqsoft/sct/devdsp/2005-11-03/sdsp120b22_ipram.bin</ipramFile> <idramFile>/daqsoft/sct/devdsp/2005-11-03/sdsp120b22_idram.bin</idramFile> <extFile>/daqsoft/sct/devdsp/2005-11-03/sdsp120b22_xcode.bin</extFile>
1064-0 NMask. Aborts. ~/SctApiCrateServer0_ctatsct01_srsctdaq1.cern.ch_25761_15.out
Got some text: crate: UCID:0.0 rod slot: 10 Text TRANSFER : 264 [SDSP 1: rodRun.c, 298]:: caching on! ^@mbly language histogram function. [SDSP 1: taskManager.c, 223]:: Histogramming task started on DSP. ^@ZZZZZZJZZZZZZZZZZZZZZZZZZZZZZZZZZ^ZZZ^ZZZZZZZZZZZZZZZZZZZZZZZzZZZZZZZZJZZZZZZZZZXZZZZZZZZZZZzZZZjJZZZZZZZZZZJZZ Got some text: crate: UCID:0.0 rod slot: 10 Text INFO : 176 [MDSP: masterTasks_sct.c, 2769]:: Timeout in Trigger Loop cnt 0000eefe efbCnt 00000000 [etc]
(16:49:09) PeterP: OK, you got 93 events then it bombed (16:49:45) PeterP: the Zs are probably the output of the error decoder - not much use, eh (16:51:23) PeterP: probably = hardware error, was "recovered", but main trigger loop not restarted. Hmmm (16:59:30) PeterP: new mdsp coming your way - extra diagnostic info (17:08:45) PeterP: can almost see where it's failing, but output from the version I just sent would help
17:20
MdspFlashLoad -s10 /daqsoft/sct/devdsp/2005-11-03/mdsp120b23.bin 1066 - 0 NMask SctApiCrateServer0_ctatsct01_srsctdaq1.cern.ch_29398_15.out <pre> <pre> 17:25:27) PeterP: OK. Problem is nearly what I'd expected, but not quite (17:56:07) PeterP: try afs/cern.ch/user/p/pwp/public/mdsp120b24.bin - yet more diagnostics (18:14:12) PeterP: ok, problem is that efb event counter increments for no apparent reason after event recovery resets it (18:22:51) PeterP: guessing - register is read only.
18:44
MdspFlashLoad -s 10 /daqsoft/sct/devdsp/2005-11-03/mdsp120b25.bin 1068 - 0 nmask stalls SctApiCrateServer0_ctatsct01_srsctdaq1.cern.ch_30406_15.out
(18:54:01) PeterP: link 80 looks pretty bad (counting 0 to 95) (18:54:11) Alan: should I take it out? (18:54:35) PeterP: if you want scans to run, yes! (18:54:48) Alan: ok (19:00:05) PeterP: Ah - do try with ccode histogramming, caching may be interfering with the c code error decoder
(19:02:15) PeterP: OK, so still works if no errors. Don't understand these text buffers, though. Also try STN with 80 out, might give a clue as to cause of the failure. (19:02:53) Alan: SctApiCrateServer0_ctatsct01_srsctdaq1.cern.ch_31256_15.out
[SDSP 1: slaveTasks.c, 205]:: EVENT_ERROR bin 10 n_good_events 1 [SDSP 1: histogram_sct.c, 770]:: EVENT_ERROR: header OK L1id 9 BCid 8704 [SDSP 1: histogram_sct.c, 986]:: ^Ah n_errors 80 [SDSP 1: histogram_sct.c, 812]:: EVENT_ERROR (0x0015) Link 80 Chip 2 ABCD_ERROR: code 5: n_errors 3152 Next event 106 (bin 10) on ROD (0, 0, 0) Next bin 0+0 on ROD (0, 0, 0) Got some text: crate: UCID:0.0 rod slot: 10 Text INFO : 552 [MDSP: rodConfiguration.c, 763]:: Configuring ROD: initialization mode. [MDSP: rodConfiguration.c, 217]::
(19:03:02) PeterP: Think I know what to do next. Abandon present implementation of event recovery. Just send HR, then carry on. If have to send HR more than 10 times, abort scan. (19:03:12) Alan: ok (19:03:24) PeterP: Present code - attempts to recover each error event 10 times. NO overall limt as to # times this process repeated. (19:03:25) Alan: HR requires DDC (19:03:33) PeterP: Sorry, menat SR