HP-UX(11.23) pstat_getproc() EOVERFLOW

[角谷寿哉]

Hi everyone.

I have a trouble on HP-UX(11.23 64bit)

Some application(32bit) have occurred EOVERFLOW at pstat_getproc().

I don't know what condition of application in cae of this error.

I'd like to know the reappearance condition of this error and ask the programmer to fix bug.

[Dennis Handly]

It depends, there are several that return EOVERFLOW. But probably it is saying the values are too big to return and the application should be recompiled with -D_PSTAT64.
See pstat(2).

[角谷寿哉]

I have read pstat(2) and understand how to solve the problem with -D_PSTAT64 option.

But I need to find the condition of this issue.

while this issue don't occurr at Single-processor server, it occurrs several times at 2-processor server.

[Dennis Handly]

>I need to find the condition of this issue.

Why? The solution is to always use -D_PSTAT64.
It is probably not worth your or the programmer's time to find out why.
There are status flags to indicate exactly which field doesn't fit, pst_valid. There seems to be only three fields that can have problems:
PST_ADDR PST_WCHAN PST_MAXRSS

Basically the value returned doesn't fit in either 2 G or 4 G, 32 bits.
Look at all of the fields in struct pst_status.

The 11.31 man page has lots of details:
http://docs.hp.com/en/B2355-60130/pstat.2.html

[角谷寿哉]

> Look at all of the fields in struct pst_status.

Thank you for your suggestion.

I'll try it next Monday.

[角谷寿哉]

Hi Dennis

I try to run the application with -D_PSTAT64 option compiled and I've got the solution with confidence.

From 9:00 am to 7:00 pm , the fixed application didn't occurred EOVERFLOW at pstat_getproc()

although the old application got an error per 5times.

Thank you for your advise.

[Dennis Handly · Accepted solution]

>although the old application got an error per 5times.

Somehow some of these were too big: PST_ADDR PST_WCHAN PST_MAXRSS

[角谷寿哉]

Thanks a lot.

at last

I sleep well tonight.