[Charset iso-8859-1 unsupported, filtering to ASCII...]
> On Wed, 10 May 2000, Bruce Momjian wrote:
>
> > It is a nifty BSD one. If you assign argv[0] in the program to a
> > string, it shows in ps.
> >
> > argv[0] = "new ps string";
> >
> > The Linux method is:
>
> Maybe I should add that this could more accurately be called the `SysV
> method' and also works on SysV-derived models. Heubrid (a.k.a. hogwash)
> systems such as Solaris and HPUX may support both or none, depending on
> the time of day.
OK.
>
> >
> > strcpy(argv[0], "new ps string");
> >
> > In the second case, you are actually writing into the environment area
> > use to store args. Not real great, but it works on Linux.
>
> You just copy the environment somewhere else before you do that. Or don't
> use the environment. Not a big deal.
But do they really do that? The scary part about the Linux code in
pg_status.h is that is just zeros out all the argv bytes and starts
writing, and that is environment memory.
Now, I do some tricks in postmaster.c so I know I have at least 5
elements to argv[], but I never do anything that makes sure I have
enough environment space to start copying strings in there.
argv[0] = "string"strcpy(argv[0], "string");
The first makes argv[0] point into user-space memory, while the second
writes into environment memory. (Allowing ps to dynamically read
argv[0] memory that is pointing to user-space is a kvm() trick.)
Linux people seem to be happy with their version, but frankly, I would
not allow it on BSD. I would _at_ _least_ put something in postmaster.c
so I _knew_ that argv[0] had a reasonable size for me to copy into it.
(Massimo write the Linux code, I believe.)
Guess it is more a philosophical issue. Linux folks like it because it
works, while I don't because it is not bullet-proof.
-- Bruce Momjian | http://www.op.net/~candle pgman@candle.pha.pa.us | (610)
853-3000+ If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill,
Pennsylvania19026
