Experimental Physics and
Industrial Control System

Timo Korhonen via Core-talk <core-talk at aps.anl.gov> · Tue, 21 Jun 2022 14:26:42 +0000

Colleagues,

The topic of put logging came up in our discussions at ESS again. I found this old thread and tried to scratch my memory where we ended with this. I thought this was working (with PVA as well) but I am not sure anymore, and re-reading the thread could not enlighten me either.

Access security works exactly as discussed for us, with Posix groups (PVA-only) and all that, and I assumed that the put-logging issue was resolved as a part of that implementation, but is there still something missing?

Timo

On 2019-09-06, 14:39, "core-talk-bounces at aps.anl.gov on behalf of Johnson, Andrew N. via Core-talk" <core-talk-bounces at aps.anl.gov on behalf of core-talk at aps.anl.gov> wrote:

    Hi Michael,

    > On Sep 4, 2019, at 4:45 AM, Michael Davidsaver <mdavidsaver at gmail.com> wrote:
    > 
    > https://github.com/epics-modules/caPutLog/blob/b544f92c6efe1df98506b567f3f7e5137965e2a4/caPutLogApp/caPutLogAs.c#L104
    > 
    > dbGetField() is used, which of course calls dbScanLock().
    > This is done twice, for the "before" and "after" values.
    > So 'pmessage->data' is not used.
    > 
    > I guess I can just pass data=NULL or a dummy value.

    Please look at commit c5ded30684 in Base. The asTrapWrite.h API was enhanced about 4 years ago in 3.15 and caPutLog is no longer the only external user of it. Tim Mooney's caPutRecorder module *does* rely on pmessage->data. Direct access to the data was added on his request, so he can record exactly what value was put even if the record processing subsequently modifies that field value (puts to the Motor record can fall into that category).

    Someone could modify caPutLog to make use of the data field instead of calling dbGetField() twice.

    - Andrew

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System