The mainline signal handling code for OpenRISC has been buggy since day
one with respect to syscall restart.  This patch significantly reworks
the signal handling code:

i)   Move the "work pending" loop to C code (borrowed from ARM arch)

ii)  Allow a tracer to muck about with the IP and skip syscall restart
     in that case (again, borrowed from ARM)

iii) Make signal handling WRT syscall restart actually work

v)   Make the signal handling code look more like that of other
     architectures so that it's easier for others to follow
Reported-by: NAnders Nystrom <anders@southpole.se>
Signed-off-by: NJonas Bonn <jonas@southpole.se>

Commit 287ad220 tried to set up the argument
to schedule_tail, but ended up using TI_STACK which isn't a defined symbol.
Sadly, the old openrisc compiler silently ignores this fact and it was first
discovered now when building with an updated toolchain.
Reported-by: NChristian Svensson <blue@cmd.nu>
Signed-off-by: NJonas Bonn <jonas@southpole.se>

vmalloc'ed pages are faulted into a process' page tables on demand.  In
order to facilitate this, do_page_fault needs to know whether it was
called via a page fault exception or a TLB-miss exception.

This patch adds a wrapper around the _x_page_fault_handler entry points
that the TLB-miss exceptions can call into in order to have the relevant
parameter set to satisfy do_page_fault.

This fixes a bug and is "good enough" for now.  That said, this whole
handling of vmalloc needs to be audited for correctness at some point.
Signed-off-by: NJonas Bonn <jonas@southpole.se>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NJonas Bonn <jonas@southpole.se>

Signed-off-by: NJonas Bonn <jonas@southpole.se>

schedule_tail() requires that the 'prev' task be passed as an argument
to it.  This arg is set in _switch, just before 'returning' to one of
the ret_* functions where schedule_tail is invoked.
Signed-off-by: NJonas Bonn <jonas@southpole.se>

The function sys_or1k_atomic was using call-saved registers without
restoring their value before returning.  This is a faux pas: either
we need to restore their values or use scratch regs; the latter is
less code so that's the route this patch takes.

Thanks to David Hennerström for doing most of the heavy-lifting in
tracking this one down.
Reported-by: NDavd Hennerström <david.hennerstrom@aacmicrotec.com>
Signed-off-by: NJonas Bonn <jonas@southpole.se>

The pt_regs struct had both a 'syscallno' field and an 'orig_gpr11' field
and it wasn't really clear how these were supposed to be used.  This patch
removes the syscallno field altogether and makes orig_gpr11 work more
like other architectures: keep track of syscall number in progress or
hold -1 for non-syscall exceptions.
Signed-off-by: NJonas Bonn <jonas@southpole.se>

Architecture code and early setup routines for booting Linux.
Signed-off-by: NJonas Bonn <jonas@southpole.se>
Reviewed-by: NArnd Bergmann <arnd@arndb.de>