problem with setresuid in child process(Pink Process)

[Aviansh]

Hi,

can anybody tell me some thing about the pink Process in UNIX..

Actually I'm running a daemon process as server.
when I'm connecting to the server using non-root login it doesn't dump the core upon receiving any core dumping Signal.

I'm using setresuid inside the signal handler..
even the uid is 111, the child is able to execute setresuid(0,0,0); inside the signal handler, which it should not.

somebody told me "A process is called pink process when the process becomes higher priviledged user". And Pink process doesn't dump core...

can anybody give me some idea about the pink process.

I'm new to this..
Regards
-Avinash

[Steven E. Protter]

Shalom,

As noted in another thread, there is no process designation pink. There is no daemon or status with this designation.

This terminology must be coming from one of the process/applications being run not by the OS. See the application documentation.

SEP

[Aviansh]

Thanks for your reply..

I'm not getting the process designation as pink, rather it's concept called Pink process.

By the way my problem is that the child Daemon is not dumping core even if it executes the signal handler till the last line.
I've checked the "ulimit" and full permission is given to the directory where it is supposed to dump the core.
Even the effective and real user id's are same.

Please help me in this regard..

[Dennis Handly]

You probably should have had the moderators move your other thread here, not create a new one:
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1315595

>somebody told me

You have a URL?
The only place google finds "pink process" is you and this other forum:
ttp://www.unix.com/hp-ux/62542-core-hp-ux.html

As mentioned there, you can't get a coredump unless you attach with a debugger and ttrace(2) TT_PROC_CORE.
http://h20293.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=EnhancedCore

[Aviansh]

I tried with ttrace but it's failing with errno set to 13(EACCES). This happening because the process is already being traced...

any other way of getting the core file..??

[Don Morris_1]

"Pink" is a frankly a bit of an inside joke for the process management kernel side of things. It is a state for any given process which represents "Are you now, or have you ever been -- a process with escalated privileges." As with most inside jokes -- this really shouldn't have been shared verbatim as it is not documented externally.

The external representation of this is the PS_CHANGEDPRIV flag being present in the pst_status structure returned from pstat_proc().

If this is 11.11 (11i v1) -- when you say "ttrace is failing with EACCES" is that failing to attach or failing when you issue the TT_PROC_CORE? If the latter - you do need to have the EnhancedCore optional product installed (http://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=EnhancedCore)?

v3 should work with ttrace as-is. I'd have to look more at v2.

[James R. Ferguson]

Hi:

> Don: "Pink" is a frankly a bit of an inside joke for the process management kernel side of things. It is a state for any given process which represents "Are you now, or have you ever been -- a process with escalated privileges."

...interesting! I would have guessed that "pink" might have been someone's reference to newness or (re)birth, or at the height; paragon; pinnacle ... :-)

Regards!

...JRF...

[Rajesh K Chaurasia]

Hi Avinash:

If you are running on HP-UX 11i v3 (11.31) please check the output of 'coreadm' command. Refer to coreadm(1M) and coreadm(2) man pages to know how to enable/disable core file writing by setid processes.

Regards
-Rajesh

[Aviansh]

My product is running on all versions beginning 11.11 ia and pa onwards.

currently I'm working on 11.23 IA..

The ttrace is failing with errno set to 13(EACCES), I think this happening because the process is already being traced...