home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.linux
- Path: sparky!uunet!munnari.oz.au!metro!ipso!runxtsa!bde
- From: bde@runx.oz.au (Bruce Evans)
- Subject: Re: sigset
- Message-ID: <1993Jan21.075324.21795@runx.oz.au>
- Organization: RUNX Un*x Timeshare. Sydney, Australia.
- References: <9301182200.AA11372@deepthought.cs.utexas.edu> <1993Jan19.235103.23941@cbnewsi.cb.att.com>
- Date: Thu, 21 Jan 93 07:53:24 GMT
- Lines: 22
-
- In article <1993Jan19.235103.23941@cbnewsi.cb.att.com> Kenneth Almquist (ka@hrojr.hr.att.com) writes:
- >When reliable signals were added to System V Release 3, the developers
- >added a new system call named "sigset" rather than changing the
- >semantics of "signal". There may be minor differences between the
- >behavior of "sigset" and the Linux "signal" call, but in most cases
- >compiling with "-Dsigset=signal" should work fine.
-
- I have sigset() for Minix. The differences seem to be as follows:
-
- 1. It gives reliable signals (sa_mask = 0).
- 2. If the signal is blocked, then sigset() returns SIG_HOLD instead of the
- old handler.
- 3. sigset(sig, SIG_HOLD) masks the signal.
-
- So, -Dsigset=signal probably won't work if the application uses SIG_HOLD,
- but Linux does not define SIG_HOLD (?) so the application will still not
- compile.
-
- -Dsigset=signal probably won't work if the application wants reliable
- signals, either. Linux signal() gives unreliable signals :-(.
- --
- Bruce Evans (bde@runx.oz.au)
-