Re: FDs & linux again.

I'v detected a "look like" problem in another program i was developing some
weeks ago, please verify if the correct includes are beeing used, in my
case, some .h's were "#undef ing" some +256FD needed values and redefining
them with wrong/old system values.

Insert the fallowing code near the "select" code in squid.

    assert( sizeof(fd_set)*8 >= FD_SETSIZE );
    assert( FD_SETSIZE >= NR_OPEN );

And see if the assert's fail!

Good luck.

[]---------------------------------------------------------------[]
  Pedro Ribeiro
  Online: http://www.isel.pt/~pribeiro/
  IRC(PTnet) Nick: PAntMaR
  e-Mail: Personal: pribeiro@isel.pt
          Admin: admin@isel.pt
[]---------------------------------------------------------------[]

-----Original Message-----
From: Pawel Pierscionek <pawel@astercity.net>
To: Squid List <squid-users@ircache.net>
Date: Segunda-feira, 11 de Janeiro de 1999 18:13
Subject: FDs & linux again.

>Hi,
>
>I have the FD problem and I'm fighting with it a long time.
>I've patched the linux 2.0.35 kernel just as it was suggested some
>time ago on the list.
>No matter if I use select or pool I have problems like unable to fork
>messages.
>This is what squid says before I kill it :
>comm_poll: poll failure: (12) Cannot allocate memory
>All this happening long before 256FDs magic number is hit.
>
>Is there any solution ?
>
>TIA,
>Pawel Pierscionek
>
>/*
> Money is the root of all evil. Send $30 for more info.
>*/
>
Received on Mon Jan 11 1999 - 11:30:54 MST