sysdeps/unix/sysv/linux directory structure question

Steve Ellcey sellcey@caviumnetworks.com
Thu Nov 3 18:12:00 GMT 2016


On Thu, 2016-11-03 at 18:01 +0000, Joseph Myers wrote:
> On Thu, 3 Nov 2016, Steve Ellcey wrote:
> 
> > 
> > So I have several questions.  What is the difference between
> > sysdeps/unix/sysv/linux and sysdeps/unix/sysv/linux/generic and why
> > are some files in one vs. the other vs. both?   Is generic only for
> > 64-
> generic is for newer architectures using the generic syscall ABI,
> where 
> syscalls that can be implemented in terms of newer syscalls are
> generally 
> omitted (so only statat syscalls not the older stat ones, for
> example), 
> and where structures are the same between different architectures as
> far 
> as possible (whereas various structures on older architectures
> commonly 
> tried to follow the layout used by some proprietary Unix on that 
> architecture).
> 
> > 
> > bit kernels?  Is that why it has a wordsize-32 subdirectory?  Are
> The generic syscall ABI only has 64-bit versions of various
> structures, 
> with 32-bit architectures expected to do EOVERFLOW handling in
> userspace.

So this use of 'generic' is different than the 'sysdeps/generic'
directory.  I.e. sysdeps/generics is automatically added to the end of
the sysdep directory search list, but 'sysdeps/unix/sysv/linux/generic'
is only going to be used if it appears in a targets Implies file.  Is
that correct?

Steve Ellcey
sellcey@caviumnetworks.com



More information about the Libc-alpha mailing list