man 2 sigprocmask
It seems the sigprocmask syscall takes different arguments than the libc stub for sigprocmask(2). Though the differences are easily understood by browsing source, there don't appear to be any man pages describing the syscall interface, and wonder if this is an intentional or accidental oversight. Thanks. --david
Re: man 2 sigprocmask
On Sat, Oct 6, 2012 at 4:06 PM, David Higgs hig...@gmail.com wrote: It seems the sigprocmask syscall takes different arguments than the libc stub for sigprocmask(2). Though the differences are easily understood by browsing source, there don't appear to be any man pages describing the syscall interface, and wonder if this is an intentional or accidental oversight. The only places this should possibly be documented, IMO, are kdump(1) and syscall(2), and in the latter it should be at most a warning and _not_ specific info about the affected calls. The syscall ABI is only supported for the calls that the system libraries actually use, and even there the guarantee is only for a couple releases. For example, the sigreturn() in the OpenBSD 5.0 amd64 libc will not work against a 5.2 kernel. Philip Guenther
Re: man 2 sigprocmask
On Sat, Oct 6, 2012 at 7:48 PM, Philip Guenther guent...@gmail.com wrote: On Sat, Oct 6, 2012 at 4:06 PM, David Higgs hig...@gmail.com wrote: It seems the sigprocmask syscall takes different arguments than the libc stub for sigprocmask(2). Though the differences are easily understood by browsing source, there don't appear to be any man pages describing the syscall interface, and wonder if this is an intentional or accidental oversight. The only places this should possibly be documented, IMO, are kdump(1) and syscall(2), and in the latter it should be at most a warning and _not_ specific info about the affected calls. The syscall ABI is only supported for the calls that the system libraries actually use, and even there the guarantee is only for a couple releases. For example, the sigreturn() in the OpenBSD 5.0 amd64 libc will not work against a 5.2 kernel. To reiterate for personal clarity, section 2 manpages for syscalls actually refers to the corresponding libc interfaces? That makes sense in that no userland code would be using syscalls directly due to complications like errno handling. Thanks. --david
Re: man 2 sigprocmask
On Sat, Oct 6, 2012 at 8:25 PM, David Higgs hig...@gmail.com wrote: To reiterate for personal clarity, section 2 manpages for syscalls actually refers to the corresponding libc interfaces? That makes sense in that no userland code would be using syscalls directly due to complications like errno handling. Yes, exactly. Philip Guenther