[PATCH] getpid/vfork/raise fix
David Mosberger
davidm@napali.hpl.hp.com
Wed Mar 10 19:24:00 GMT 2004
>>>>> On Wed, 10 Mar 2004 17:07:58 +0100, Jakub Jelinek <jakub@redhat.com> said:
Jakub> One (plus b0) is needed. But first of all I'd like to
Jakub> understand the differences between NPTL pt-vfork.S and libc
Jakub> vfork.S on IA-64. Why does one use DO_CALL_VIA_BREAK and the
Jakub> other one DO_CALL, one has gas workarounds and one does not?
My memory is a bit fuzzy (isn't it always... ;-), but as I remember
it, new-style syscall stub support is tied to NPTL, so vfork.S is
never actually compiled when new-style syscalls stubs are in effect.
Because of that, DO_CALL_VIA_BREAK isn't _necessary_ for vfork.S.
Furthermore, the gas workarounds are needed only for the new-style
syscall stubs. If you preferred them to be in sync, I _think_ it
would be OK to make vfork.S more similar (perhaps even identical) to
pt-vfork.S.
--david
More information about the Libc-hacker
mailing list