]> err.no Git - linux-2.6/log
linux-2.6
18 years ago[SPARC64]: Drop %gl to 0 before re-enabling PSTATE_IE in rtrap
David S. Miller [Thu, 23 Feb 2006 00:15:45 +0000 (16:15 -0800)]
[SPARC64]: Drop %gl to 0 before re-enabling PSTATE_IE in rtrap

If we take a window fault, on SUN4V set %gl to zero before we
turn PSTATE_IE back on in %pstate.  Otherwise if we take an
interrupt we'll end up with corrupt register state.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Create a seperate kernel TSB for 4MB/256MB mappings.
David S. Miller [Wed, 22 Feb 2006 06:31:11 +0000 (22:31 -0800)]
[SPARC64]: Create a seperate kernel TSB for 4MB/256MB mappings.

It can map all of the linear kernel mappings with zero TSB hash
conflicts for systems with 16GB or less ram.  In such cases, on
SUN4V, once we load up this TSB the first time with all the
mappings, we never take a linear kernel mapping TLB miss ever
again, the hypervisor handles them all.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Make use of Niagara 256MB PTEs for kernel mappings.
David S. Miller [Wed, 22 Feb 2006 04:51:13 +0000 (20:51 -0800)]
[SPARC64]: Make use of Niagara 256MB PTEs for kernel mappings.

We use a bitmap, one bit for every 256MB of memory.  If the
bit is set we can use a 256MB PTE for linear mappings, else
we have to use a 4MB PTE.

SUN4V support is there, and we can very easily add support
for Panther cpu 256MB PTEs in the future.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use sun4v_cpu_idle() in cpu_idle() on SUN4V.
David S. Miller [Wed, 22 Feb 2006 00:55:23 +0000 (16:55 -0800)]
[SPARC64]: Use sun4v_cpu_idle() in cpu_idle() on SUN4V.

We have to turn off the "polling nrflag" bit when we sleep
the cpu like this, so that we'll get a cross-cpu interrupt
to wake the processor up from the yield.

We also have to disable PSTATE_IE in %pstate around the yield
call and recheck need_resched() in order to avoid any races.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] math-emu: Delete debugging printk left by previous commit.
David S. Miller [Wed, 22 Feb 2006 00:53:26 +0000 (16:53 -0800)]
[SPARC64] math-emu: Delete debugging printk left by previous commit.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Add sun4v_cpu_yield().
David S. Miller [Tue, 21 Feb 2006 23:42:09 +0000 (15:42 -0800)]
[SPARC64]: Add sun4v_cpu_yield().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Kill cpudata->idle_volume.
David S. Miller [Tue, 21 Feb 2006 23:41:01 +0000 (15:41 -0800)]
[SPARC64]: Kill cpudata->idle_volume.

Set, but never used.

We used to use this for dynamic IRQ retargetting, but that
code died a long time ago.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Niagara optimized memset/bzero/clear_user.
David S. Miller [Tue, 21 Feb 2006 22:29:42 +0000 (14:29 -0800)]
[SPARC64]: Niagara optimized memset/bzero/clear_user.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Pass multiple CPUs at once to hypervisor cross-call API.
David S. Miller [Tue, 21 Feb 2006 21:59:47 +0000 (13:59 -0800)]
[SPARC64]: Pass multiple CPUs at once to hypervisor cross-call API.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Args to SUNW,set-trap-table are 64-bit.
David S. Miller [Tue, 21 Feb 2006 06:56:01 +0000 (22:56 -0800)]
[SPARC64]: Args to SUNW,set-trap-table are 64-bit.

They were getting truncated to 32-bit and this is very bad
when your MMU fault status area is in physical memory above
4GB on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Handle unimplemented FPU square-root on Niagara.
David S. Miller [Tue, 21 Feb 2006 00:02:24 +0000 (16:02 -0800)]
[SPARC64]: Handle unimplemented FPU square-root on Niagara.

The math-emu code only expects unfinished fpop traps when
emulating FPU sqrt instructions on pre-Niagara chips.
On Niagara we can get unimplemented fpop, so handle that.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC] serial: Make sure sysfs nodes get named correctly.
David S. Miller [Mon, 20 Feb 2006 22:39:16 +0000 (14:39 -0800)]
[SPARC] serial: Make sure sysfs nodes get named correctly.

Because we play this trick where we use ttyS? in increasing minor
numbers for different sunfoo.c drivers, we have to inform the TTY
layer of this.

Do so by setting the tty->name_base appropriately.

Probably there should be a generic way to do this in the serial core,
but for now...

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Typo in sun4v_data_access_exception log message.
David S. Miller [Mon, 20 Feb 2006 09:50:09 +0000 (01:50 -0800)]
[SPARC64]: Typo in sun4v_data_access_exception log message.

Should be "Dax" not "Iax".

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Handle zero-length map requests in pci_sun4v.c
David S. Miller [Mon, 20 Feb 2006 09:42:51 +0000 (01:42 -0800)]
[SPARC64]: Handle zero-length map requests in pci_sun4v.c

By simply changing the do-while loop into a plain
while loop.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Kill stray PGLIST_NENTS check in pci_sun4v.c
David S. Miller [Mon, 20 Feb 2006 09:09:01 +0000 (01:09 -0800)]
[SPARC64]: Kill stray PGLIST_NENTS check in pci_sun4v.c

I forgot to remove the one in pci_4v_map_sg() during the
iommu batching commit.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix typo in dump_tl1_traplog()
David S. Miller [Mon, 20 Feb 2006 08:54:09 +0000 (00:54 -0800)]
[SPARC64]: Fix typo in dump_tl1_traplog()

Actually make use of the 'limit' we compute.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Disable smp_report_regs() for now.
David S. Miller [Mon, 20 Feb 2006 08:36:57 +0000 (00:36 -0800)]
[SPARC64]: Disable smp_report_regs() for now.

It's extremely noisy and causes much grief on slow
consoles with large numbers of cpus.

We'll have to provide this some saner way in order
to re-enable this.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Remove PGLIST_NENTS PCI IOMMU mapping limitation on SUN4V.
David S. Miller [Mon, 20 Feb 2006 06:21:32 +0000 (22:21 -0800)]
[SPARC64]: Remove PGLIST_NENTS PCI IOMMU mapping limitation on SUN4V.

Use a batching queue system for IOMMU mapping setup,
with a page sized batch.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use KERN_EMERG in dump_tl1_traplog() and sun4v TLB errors.
David S. Miller [Sun, 19 Feb 2006 01:06:28 +0000 (17:06 -0800)]
[SPARC64]: Use KERN_EMERG in dump_tl1_traplog() and sun4v TLB errors.

We're about to seriously die in these cases so it is important
that the messages make it to the console.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix unaligned access winfxup handling on SUN4V.
David S. Miller [Sun, 19 Feb 2006 00:39:39 +0000 (16:39 -0800)]
[SPARC64]: Fix unaligned access winfxup handling on SUN4V.

Another case where we have to force ourselves into global register
level one.  Also make sure the arguments passed to sun4v_do_mna() are
correct.

This area actually needs some more work, for example spill fixup is
not necessarily going to do the right thing for this case.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Set %gl to 1 in kvmap_itlb_longpath on SUN4V.
David S. Miller [Sun, 19 Feb 2006 00:36:39 +0000 (16:36 -0800)]
[SPARC64]: Set %gl to 1 in kvmap_itlb_longpath on SUN4V.

Just like kvmap_dtlb_longpath we have to force the
global register level to one in order to mimick the
PSTATE_MG --> PSTATE_AG trasition done on SUN4U.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] sunhv: Fix locking in sunhv_start_tx()
David S. Miller [Sun, 19 Feb 2006 00:32:22 +0000 (16:32 -0800)]
[SPARC64] sunhv: Fix locking in sunhv_start_tx()

Caller takes the lock already.

Also, fixup the poll loop in sunhv_break_ctl().  Just
like in console write, we udelay(2) and use a loop
limit of 1000000 iterations.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Put sunhv.o earliest in the list of sparc serial drivers.
David S. Miller [Sun, 19 Feb 2006 00:30:59 +0000 (16:30 -0800)]
[SPARC64]: Put sunhv.o earliest in the list of sparc serial drivers.

So that it will show up as /dev/ttyS0.  Otherwise things like
installers will try to run on whatever serial port gets probed
first.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Export a PAGE_SHARED symbol.
David S. Miller [Sat, 18 Feb 2006 20:43:16 +0000 (12:43 -0800)]
[SPARC64]: Export a PAGE_SHARED symbol.

For drivers/media/*, noticed by Fabbione.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] Fix build if CONFIG_HUGETLB_PAGE is not set
Fabio M. Di Nitto [Sat, 18 Feb 2006 08:32:31 +0000 (00:32 -0800)]
[SPARC64] Fix build if CONFIG_HUGETLB_PAGE is not set

Signed-off-by: Fabio M. Di Nitto <fabbione@ubuntu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: More TLB/TSB handling fixes.
David S. Miller [Sat, 18 Feb 2006 02:01:02 +0000 (18:01 -0800)]
[SPARC64]: More TLB/TSB handling fixes.

The SUN4V convention with non-shared TSBs is that the context
bit of the TAG is clear.  So we have to choose an "invalid"
bit and initialize new TSBs appropriately.  Otherwise a zero
TAG looks "valid".

Make sure, for the window fixup cases, that we use the right
global registers and that we don't potentially trample on
the live global registers in etrap/rtrap handling (%g2 and
%g6) and that we put the missing virtual address properly
in %g5.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Don't do anything in flush_ptrace_access() on SUN4V.
David S. Miller [Fri, 17 Feb 2006 23:07:43 +0000 (15:07 -0800)]
[SPARC64]: Don't do anything in flush_ptrace_access() on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix some SUN4V TLB handling bugs.
David S. Miller [Fri, 17 Feb 2006 22:58:02 +0000 (14:58 -0800)]
[SPARC64]: Fix some SUN4V TLB handling bugs.

1) Add error return checking for TLB load hypervisor
   calls.

2) Don't fallthru to dtlb tsb miss handler from itlb tsb
   miss handler, oops.

3) On window fixups, propagate fault information to fixup
   handler correctly.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Check for errors in hypervisor_tlb_lock().
David S. Miller [Fri, 17 Feb 2006 22:40:30 +0000 (14:40 -0800)]
[SPARC64]: Check for errors in hypervisor_tlb_lock().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Init boot cpu's trap_block[] before paging_init()
David S. Miller [Mon, 27 Feb 2006 07:32:33 +0000 (23:32 -0800)]
[SPARC64]: Init boot cpu's trap_block[] before paging_init()

It must be ready when we take over the trap table.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Define ARCH_HAS_READ_CURRENT_TIMER.
David S. Miller [Fri, 17 Feb 2006 20:33:13 +0000 (12:33 -0800)]
[SPARC64]: Define ARCH_HAS_READ_CURRENT_TIMER.

This gives more consistent bogomips and delay() semantics,
especially on sun4v.  It gives weird looking values though...

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Set associativity of kernel TSB descriptor correctly.
David S. Miller [Fri, 17 Feb 2006 20:03:20 +0000 (12:03 -0800)]
[SPARC64]: Set associativity of kernel TSB descriptor correctly.

It should be 1, not 0.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: __bzero_noasi --> __clear_user
David S. Miller [Fri, 17 Feb 2006 18:35:23 +0000 (10:35 -0800)]
[SPARC64]: __bzero_noasi --> __clear_user

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Put SUN4V ITSB miss into correct trap table entry.
David S. Miller [Fri, 17 Feb 2006 18:28:24 +0000 (10:28 -0800)]
[SPARC64]: Put SUN4V ITSB miss into correct trap table entry.

It's 0x9 not 0xb.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Add HWCAP_SPARC_BLKINIT elf capability flag for Niagara.
David S. Miller [Fri, 17 Feb 2006 18:14:38 +0000 (10:14 -0800)]
[SPARC64]: Add HWCAP_SPARC_BLKINIT elf capability flag for Niagara.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use phys tsb address in tsb_insert() in SUN4V.
David S. Miller [Fri, 17 Feb 2006 17:54:42 +0000 (09:54 -0800)]
[SPARC64]: Use phys tsb address in tsb_insert() in SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix uniprocessor IRQ targetting on SUN4V.
David S. Miller [Fri, 17 Feb 2006 16:38:06 +0000 (08:38 -0800)]
[SPARC64]: Fix uniprocessor IRQ targetting on SUN4V.

We need to use the real hardware processor ID when
targetting interrupts, not the "define to 0" thing
the uniprocessor build gives us.

Also, fill in the Node-ID and Agent-ID fields properly
on sun4u/Safari.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix PCI IRQ probing regression.
David S. Miller [Fri, 17 Feb 2006 16:20:18 +0000 (08:20 -0800)]
[SPARC64]: Fix PCI IRQ probing regression.

If the top-level cnode had multi entries in it's "reg"
property, we'd fail.  The buffer wasn't large enough in
such cases.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Get SUN4V SMP working.
David S. Miller [Fri, 17 Feb 2006 09:29:17 +0000 (01:29 -0800)]
[SPARC64]: Get SUN4V SMP working.

The sibling cpu bringup is extremely fragile.  We can only
perform the most basic calls until we take over the trap
table from the firmware/hypervisor on the new cpu.

This means no accesses to %g4, %g5, %g6 since those can't be
TLB translated without our trap handlers.

In order to achieve this:

1) Change sun4v_init_mondo_queues() so that it can operate in
   several modes.

   It can allocate the queues, or install them in the current
   processor, or both.

   The boot cpu does both in it's call early on.

   Later, the boot cpu allocates the sibling cpu queue, starts
   the sibling cpu, then the sibling cpu loads them in.

2) init_cur_cpu_trap() is changed to take the current_thread_info()
   as an argument instead of reading %g6 directly on the current
   cpu.

3) Create a trampoline stack for the sibling cpus.  We do our basic
   kernel calls using this stack, which is locked into the kernel
   image, then go to our proper thread stack after taking over the
   trap table.

4) While we are in this delicate startup state, we put 0xdeadbeef
   into %g4/%g5/%g6 in order to catch accidental accesses.

5) On the final prom_set_trap_table*() call, we put &init_thread_union
   into %g6.  This is a hack to make prom_world(0) work.  All that
   wants to do is restore the %asi register using
   get_thread_current_ds().

Longer term we should just do the OBP calls to set the trap table by
hand just like we do for everything else.  This would avoid that silly
prom_world(0) issue, then we can remove the init_thread_union hack.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Disable smp_report_regs() for now.
David S. Miller [Fri, 17 Feb 2006 09:17:21 +0000 (01:17 -0800)]
[SPARC64]: Disable smp_report_regs() for now.

For 32 cpus and a slow console, it just wedges the
machine especially with DETECT_SOFTLOCKUP enabled.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Rewrite pci_intmap_match().
David S. Miller [Fri, 17 Feb 2006 07:01:10 +0000 (23:01 -0800)]
[SPARC64]: Rewrite pci_intmap_match().

The whole algorithm was wrong.  What we need to do is:

1) Walk each PCI bus above this device on the path to the
   PCI controller nexus, and for each:
      a) If interrupt-map exists, apply it, record IRQ controller node
      b) Else, swivel interrupt number using PCI_SLOT(), use PCI bus
 parent OBP node as controller node
      c) Walk up to "controller node" until we hit the first PCI bus
 in this domain, or "controller node" is the PCI controller
 OBP node
2) If we walked to PCI controller OBP node, we're done.
3) Else, apply PCI controller interrupt-map to interrupt.

There is some stuff that needs to be checked out for ebus and
isa, but the PCI part is good to go.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Don't set interrupt state to IDLE in enable_irq().
David S. Miller [Fri, 17 Feb 2006 04:44:25 +0000 (20:44 -0800)]
[SPARC64]: Don't set interrupt state to IDLE in enable_irq().

We'll lose events that way.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix return from trap on SUN4V.
David S. Miller [Fri, 17 Feb 2006 00:23:45 +0000 (16:23 -0800)]
[SPARC64]: Fix return from trap on SUN4V.

We need to set the global register set _AND_ disable
PSTATE_IE in %pstate.  The original patch sequence was
leaving PSTATE_IE enabled when returning to kernel mode,
oops.

This fixes the random register corruption being seen
on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Add GET_GL_GLOBAL() macro for SUN4V.
David S. Miller [Fri, 17 Feb 2006 00:22:26 +0000 (16:22 -0800)]
[SPARC64]: Add GET_GL_GLOBAL() macro for SUN4V.

So we can read the %gl register for debugging.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Set dummy bucket->{imap,iclr} unique on SUN4V.
David S. Miller [Thu, 16 Feb 2006 22:37:05 +0000 (14:37 -0800)]
[SPARC64]: Set dummy bucket->{imap,iclr} unique on SUN4V.

So that free_irq() disable's the IRQ correctly.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Add sun4v_cpu_qconf() hypervisor call.
David S. Miller [Thu, 16 Feb 2006 22:26:53 +0000 (14:26 -0800)]
[SPARC64]: Add sun4v_cpu_qconf() hypervisor call.

Call it from register_one_mondo().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC]: Kill off these __put_user_ret things.
David S. Miller [Mon, 6 Mar 2006 00:46:58 +0000 (16:46 -0800)]
[SPARC]: Kill off these __put_user_ret things.

They are bogus and haven't been referenced in years.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: do_fptrap needs to load the thread reg into %g6.
David S. Miller [Thu, 16 Feb 2006 10:18:49 +0000 (02:18 -0800)]
[SPARC64]: do_fptrap needs to load the thread reg into %g6.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix bogus call to sun4v_mna in winfixup code.
David S. Miller [Thu, 16 Feb 2006 09:45:49 +0000 (01:45 -0800)]
[SPARC64]: Fix bogus call to sun4v_mna in winfixup code.

The C function is named sun4v_do_mna not sun4v_mna.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix tl1 trap state capture/dump on SUN4V.
David S. Miller [Thu, 16 Feb 2006 09:41:41 +0000 (01:41 -0800)]
[SPARC64]: Fix tl1 trap state capture/dump on SUN4V.

No trap levels above 2 in privileged mode on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] sunhv: Support SYSRQ properly.
David S. Miller [Thu, 16 Feb 2006 08:42:21 +0000 (00:42 -0800)]
[SPARC64] sunhv: Support SYSRQ properly.

By calling uart_handle_break().  We'll still do the
"sun_do_break()" handling if the user gives two
breaks in a row.

We should probably do this in the other Sparc serial
drivers too.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] PCI: Size TSB correctly on SUN4V.
David S. Miller [Thu, 16 Feb 2006 06:25:27 +0000 (22:25 -0800)]
[SPARC64] PCI: Size TSB correctly on SUN4V.

Forgot to multiply by 8 * 1024, oops.  Correct the size constant when
the virtual-dma arena is 2GB in size, it should bet 256 not 128.

Finally, log some info about the TSB at probe time.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix OOPS on sunhv interrupts.
David S. Miller [Thu, 16 Feb 2006 05:56:49 +0000 (21:56 -0800)]
[SPARC64]: Fix OOPS on sunhv interrupts.

Until the uart is openned, port->info is NULL.

Also, init the port->irq properly and give a non-zero
port->membase so that the uart device reporting is done.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Don't use ASI_QUAD_LDD_PHYS on SUN4V.
David S. Miller [Thu, 16 Feb 2006 05:21:17 +0000 (21:21 -0800)]
[SPARC64]: Don't use ASI_QUAD_LDD_PHYS on SUN4V.

Need to use ASI_QUAD_LDD_PHYS_4V instead.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Do not write garbage into %pstate in tsb_context_switch().
David S. Miller [Thu, 16 Feb 2006 05:16:42 +0000 (21:16 -0800)]
[SPARC64]: Do not write garbage into %pstate in tsb_context_switch().

For SUN4V, we were clobbering %o5 to do the hypervisor call.
This clobbers the saved %pstate value and we end up writing
garbage into that register as a result.  Oops.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix flush_tsb_user() on SUN4V.
David S. Miller [Thu, 16 Feb 2006 05:01:31 +0000 (21:01 -0800)]
[SPARC64]: Fix flush_tsb_user() on SUN4V.

Needs to use physical addressing just like cheetah_plus.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix comment typo in __flush_tlb_kernel_range.
David S. Miller [Thu, 16 Feb 2006 04:35:10 +0000 (20:35 -0800)]
[SPARC64]: Fix comment typo in __flush_tlb_kernel_range.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Decode virtual-devices interrupts correctly.
David S. Miller [Thu, 16 Feb 2006 03:48:54 +0000 (19:48 -0800)]
[SPARC64]: Decode virtual-devices interrupts correctly.

Need to translate through the interrupt-map{,-mask] properties.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Add prom_{start,stop}cpu_cpuid().
David S. Miller [Wed, 15 Feb 2006 10:26:54 +0000 (02:26 -0800)]
[SPARC64]: Add prom_{start,stop}cpu_cpuid().

Use prom_startcpu_cpuid() on SUN4V instead of prom_startcpu().

We should really test for "SUNW,start-cpu-by-cpuid" presence
and use it if present even on SUN4U.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix pci_intmap_match().
David S. Miller [Wed, 15 Feb 2006 09:19:56 +0000 (01:19 -0800)]
[SPARC64]: Fix pci_intmap_match().

When crawling up the PCI bus chain, stop at the first node
that has an interrupt-map property before we hit the root.

Also, if we use a bus interrupt-{map,mask} do not forget to
update the 'intmask' pointer as we do for the 'intmap' pointer.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Two IRQ handling fixes.
David S. Miller [Wed, 15 Feb 2006 09:18:19 +0000 (01:18 -0800)]
[SPARC64]: Two IRQ handling fixes.

On SUN4V, force IRQ state to idle in enable_irq().  However,
I'm still not sure this is %100 correct.

Call add_interrupt_randomness() on SUN4V too.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fixup TSTATE layout diagram in asm/pstate.h
David S. Miller [Wed, 15 Feb 2006 08:41:47 +0000 (00:41 -0800)]
[SPARC64]: Fixup TSTATE layout diagram in asm/pstate.h

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use different cache sizing defaults on SUN4V.
David S. Miller [Wed, 15 Feb 2006 08:35:50 +0000 (00:35 -0800)]
[SPARC64]: Use different cache sizing defaults on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Make lack of interrupt-map-* a fatal error on SUN4V.
David S. Miller [Wed, 15 Feb 2006 06:20:41 +0000 (22:20 -0800)]
[SPARC64]: Make lack of interrupt-map-* a fatal error on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix sun4v_intr_setenabled() return value check in enable_irq().
David S. Miller [Wed, 15 Feb 2006 06:20:13 +0000 (22:20 -0800)]
[SPARC64]: Fix sun4v_intr_setenabled() return value check in enable_irq().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Explicitly init *nregs to 0 in find_device_prom_node().
David S. Miller [Wed, 15 Feb 2006 00:44:39 +0000 (16:44 -0800)]
[SPARC64]: Explicitly init *nregs to 0 in find_device_prom_node().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Restrict PCI bus scanning on SUN4V.
David S. Miller [Wed, 15 Feb 2006 00:42:11 +0000 (16:42 -0800)]
[SPARC64]: Restrict PCI bus scanning on SUN4V.

On the PBM's first bus number, only allow device 0, function 0, to be
poked at with PCI config space accesses.

For some reason, this single device responds to all device numbers.

Also, reduce the verbiage of the debugging log printk's for PCI cfg
space accesses in the SUN4V PCI controller driver, so that it doesn't
overwhelm the slow SUN4V hypervisor console.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix C-function name called by sun4v_mna trap code.
David S. Miller [Wed, 15 Feb 2006 00:39:22 +0000 (16:39 -0800)]
[SPARC64]: Fix C-function name called by sun4v_mna trap code.

The trap code was calling itself :-)

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Don't printk() any messaages in sun4v_build_irq().
David S. Miller [Wed, 15 Feb 2006 00:37:13 +0000 (16:37 -0800)]
[SPARC64]: Don't printk() any messaages in sun4v_build_irq().

It just clutters up the log.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: INO is never fully specified already on SUN4V.
David S. Miller [Tue, 14 Feb 2006 22:12:44 +0000 (14:12 -0800)]
[SPARC64]: INO is never fully specified already on SUN4V.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Kill sun4v_register_fault_status() on SMP.
David S. Miller [Tue, 14 Feb 2006 21:49:32 +0000 (13:49 -0800)]
[SPARC64]: Kill sun4v_register_fault_status() on SMP.

That now gets done as a side effect of taking over the
trap table from OBP.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix gcc-3.3.x warnings.
David S. Miller [Tue, 14 Feb 2006 09:32:29 +0000 (01:32 -0800)]
[SPARC64]: Fix gcc-3.3.x warnings.

It doesn't like const variables being passed into
"i" constraing asm operations.  It's a bug, but
there is nothing we can really do but work around
it.

Based upon a report from Andrew Morton.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: arch/sparc64/kernel/trampoline.S needs asm/cpudata.h
David S. Miller [Tue, 14 Feb 2006 08:55:49 +0000 (00:55 -0800)]
[SPARC64]: arch/sparc64/kernel/trampoline.S needs asm/cpudata.h

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Make error codes available from sun4v_intr_get*().
David S. Miller [Tue, 14 Feb 2006 06:56:27 +0000 (22:56 -0800)]
[SPARC64]: Make error codes available from sun4v_intr_get*().

And check for errors at call sites.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Pass correct ino to sun4v_intr_*().
David S. Miller [Tue, 14 Feb 2006 06:37:32 +0000 (22:37 -0800)]
[SPARC64]: Pass correct ino to sun4v_intr_*().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use TRAP_LOAD_IRQ_WORK() in sun4v device mondo handler.
David S. Miller [Tue, 14 Feb 2006 06:37:07 +0000 (22:37 -0800)]
[SPARC64]: Use TRAP_LOAD_IRQ_WORK() in sun4v device mondo handler.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix IOMMU mapping on sun4v.
David S. Miller [Tue, 14 Feb 2006 05:50:27 +0000 (21:50 -0800)]
[SPARC64]: Fix IOMMU mapping on sun4v.

We should dynamically allocate the per-cpu pglist not use
an in-kernel-image datum, since __pa() does not work on
such addresses.

Also, consistently use "u32" for devhandle.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Trim down sun4v IRQ translation kernel log message.
David S. Miller [Tue, 14 Feb 2006 05:36:30 +0000 (21:36 -0800)]
[SPARC64]: Trim down sun4v IRQ translation kernel log message.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] sunhv: Bug fixes.
David S. Miller [Tue, 14 Feb 2006 05:28:40 +0000 (21:28 -0800)]
[SPARC64] sunhv: Bug fixes.

Add udelay to polling console write loop, and increment
the loop limit.

Name the device "ttyHV" and pass that to add_preferred_console()
when we're using hypervisor console.

Kill sunhv_console_setup(), it's empty.

Handle the case where we don't want to use hypervisor console.
(ie. we have a head attached to a sun4v machine)

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix comment typo in asm/hypervisor.h
David S. Miller [Tue, 14 Feb 2006 05:15:44 +0000 (21:15 -0800)]
[SPARC64]: Fix comment typo in asm/hypervisor.h

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64] sunhv: Use virtual-devices layer to get interrupt.
David S. Miller [Tue, 14 Feb 2006 04:43:02 +0000 (20:43 -0800)]
[SPARC64] sunhv: Use virtual-devices layer to get interrupt.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Probe virtual-devices root node on sun4v.
David S. Miller [Tue, 14 Feb 2006 04:42:16 +0000 (20:42 -0800)]
[SPARC64]: Probe virtual-devices root node on sun4v.

This is where we learn how to get the interrupts
for things like the hypervisor console device.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Kill spurious semicolon in sun4v_pci_init().
David S. Miller [Tue, 14 Feb 2006 04:41:11 +0000 (20:41 -0800)]
[SPARC64]: Kill spurious semicolon in sun4v_pci_init().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Prevent registering wrong serial console.
David S. Miller [Tue, 14 Feb 2006 04:09:10 +0000 (20:09 -0800)]
[SPARC64]: Prevent registering wrong serial console.

If the console is not for a particular Sun serial
controller, set the drv->cons to NULL.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Program IRQ registers correctly on sun4v.
David S. Miller [Tue, 14 Feb 2006 02:22:57 +0000 (18:22 -0800)]
[SPARC64]: Program IRQ registers correctly on sun4v.

Need to use hypervisor calls instead of direct register
accesses.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Generic sun4v_build_irq().
David S. Miller [Tue, 14 Feb 2006 02:16:10 +0000 (18:16 -0800)]
[SPARC64]: Generic sun4v_build_irq().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: More SUN4V PCI work.
David S. Miller [Tue, 14 Feb 2006 02:09:44 +0000 (18:09 -0800)]
[SPARC64]: More SUN4V PCI work.

Get bus range from child of PCI controller root nexus.
This is actually a hack, but the PCI-E bridge sitting
at the top of the PCI tree responds to PCI config cycles
for every device number, so best to just ignore it for now.

Preliminary PCI irq routing, needs lots of work.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Log faulting vaddr when bogus kernel PC detected.
David S. Miller [Tue, 14 Feb 2006 02:07:45 +0000 (18:07 -0800)]
[SPARC64]: Log faulting vaddr when bogus kernel PC detected.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Implement rest of generic interrupt hypervisor calls.
David S. Miller [Mon, 13 Feb 2006 08:23:32 +0000 (00:23 -0800)]
[SPARC64]: Implement rest of generic interrupt hypervisor calls.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Move devino_to_sysino out of pci_sun4v_asm.S
David S. Miller [Mon, 13 Feb 2006 08:02:16 +0000 (00:02 -0800)]
[SPARC64]: Move devino_to_sysino out of pci_sun4v_asm.S

It is not PCI specific, it is for all system interrupts.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Range check bus number in SUN4V PCI controller driver.
David S. Miller [Mon, 13 Feb 2006 07:49:18 +0000 (23:49 -0800)]
[SPARC64]: Range check bus number in SUN4V PCI controller driver.

It has to be somewhere in the range from pbm->pci_first_busno to
pbm->pci_last_busno, inclusive.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Missing 'return' statement in sun4v_pci_init().
David S. Miller [Mon, 13 Feb 2006 06:29:36 +0000 (22:29 -0800)]
[SPARC64]: Missing 'return' statement in sun4v_pci_init().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Implement basic pci_sun4v_scan_bus().
David S. Miller [Mon, 13 Feb 2006 06:18:52 +0000 (22:18 -0800)]
[SPARC64]: Implement basic pci_sun4v_scan_bus().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: PCI-SUN4V fixes.
David S. Miller [Mon, 13 Feb 2006 06:06:53 +0000 (22:06 -0800)]
[SPARC64]: PCI-SUN4V fixes.

Clear top 8-bits of physical addresses in "ranges" property.
This gives the actual physical address.

Detect PBM-A vs. PBM-B by checking bit 0x40 of the devhandle.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Use inline patching for critical PTE operations.
David S. Miller [Mon, 13 Feb 2006 05:10:07 +0000 (21:10 -0800)]
[SPARC64]: Use inline patching for critical PTE operations.

This handles the SUN4U vs SUN4V PTE layout differences
with near zero performance cost.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Move PTE field definitions back into asm/pgtable.h
David S. Miller [Mon, 13 Feb 2006 01:07:51 +0000 (17:07 -0800)]
[SPARC64]: Move PTE field definitions back into asm/pgtable.h

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Don't expect cfg space in PCI PBM ranges on SUN4V.
David S. Miller [Sun, 12 Feb 2006 07:38:00 +0000 (23:38 -0800)]
[SPARC64]: Don't expect cfg space in PCI PBM ranges on SUN4V.

PCI cfg space is accessed transparently through the Hypervisor and not
through direct cpu PIO operations.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Fix branch signedness bug in all code patching.
David S. Miller [Sun, 12 Feb 2006 07:28:40 +0000 (23:28 -0800)]
[SPARC64]: Fix branch signedness bug in all code patching.

The bug that hit SUN4V TLB patching exists elsewhere.
Make sure we cure all such cases.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Recognize "virtual-console" as input and output console device.
David S. Miller [Sun, 12 Feb 2006 07:24:30 +0000 (23:24 -0800)]
[SPARC64]: Recognize "virtual-console" as input and output console device.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC64]: Do not try to synchronize %stick registers on SUN4V.
David S. Miller [Sun, 12 Feb 2006 07:22:47 +0000 (23:22 -0800)]
[SPARC64]: Do not try to synchronize %stick registers on SUN4V.

Writes by privileged code are not allowed.

Signed-off-by: David S. Miller <davem@davemloft.net>