kernel_optimize_test/arch/sparc/kernel/hvtramp.S
David S. Miller ef3e035c3a sparc64: Fix register corruption in top-most kernel stack frame during boot.
Meelis Roos reported that kernels built with gcc-4.9 do not boot, we
eventually narrowed this down to only impacting machines using
UltraSPARC-III and derivitive cpus.

The crash happens right when the first user process is spawned:

[   54.451346] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000004
[   54.451346]
[   54.571516] CPU: 1 PID: 1 Comm: init Not tainted 3.16.0-rc2-00211-gd7933ab #96
[   54.666431] Call Trace:
[   54.698453]  [0000000000762f8c] panic+0xb0/0x224
[   54.759071]  [000000000045cf68] do_exit+0x948/0x960
[   54.823123]  [000000000042cbc0] fault_in_user_windows+0xe0/0x100
[   54.902036]  [0000000000404ad0] __handle_user_windows+0x0/0x10
[   54.978662] Press Stop-A (L1-A) to return to the boot prom
[   55.050713] ---[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000004

Further investigation showed that compiling only per_cpu_patch() with
an older compiler fixes the boot.

Detailed analysis showed that the function is not being miscompiled by
gcc-4.9, but it is using a different register allocation ordering.

With the gcc-4.9 compiled function, something during the code patching
causes some of the %i* input registers to get corrupted.  Perhaps
we have a TLB miss path into the firmware that is deep enough to
cause a register window spill and subsequent restore when we get
back from the TLB miss trap.

Let's plug this up by doing two things:

1) Stop using the firmware stack for client interface calls into
   the firmware.  Just use the kernel's stack.

2) As soon as we can, call into a new function "start_early_boot()"
   to put a one-register-window buffer between the firmware's
   deepest stack frame and the top-most initial kernel one.

Reported-by: Meelis Roos <mroos@linux.ee>
Tested-by: Meelis Roos <mroos@linux.ee>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-10-24 09:52:49 -07:00

137 lines
2.6 KiB
ArmAsm

/* hvtramp.S: Hypervisor start-cpu trampoline code.
*
* Copyright (C) 2007, 2008 David S. Miller <davem@davemloft.net>
*/
#include <asm/thread_info.h>
#include <asm/hypervisor.h>
#include <asm/scratchpad.h>
#include <asm/spitfire.h>
#include <asm/hvtramp.h>
#include <asm/pstate.h>
#include <asm/ptrace.h>
#include <asm/head.h>
#include <asm/asi.h>
#include <asm/pil.h>
.align 8
.globl hv_cpu_startup, hv_cpu_startup_end
/* This code executes directly out of the hypervisor
* with physical addressing (va==pa). %o0 contains
* our client argument which for Linux points to
* a descriptor data structure which defines the
* MMU entries we need to load up.
*
* After we set things up we enable the MMU and call
* into the kernel.
*
* First setup basic privileged cpu state.
*/
hv_cpu_startup:
SET_GL(0)
wrpr %g0, PIL_NORMAL_MAX, %pil
wrpr %g0, 0, %canrestore
wrpr %g0, 0, %otherwin
wrpr %g0, 6, %cansave
wrpr %g0, 6, %cleanwin
wrpr %g0, 0, %cwp
wrpr %g0, 0, %wstate
wrpr %g0, 0, %tl
sethi %hi(sparc64_ttable_tl0), %g1
wrpr %g1, %tba
mov %o0, %l0
lduw [%l0 + HVTRAMP_DESCR_CPU], %g1
mov SCRATCHPAD_CPUID, %g2
stxa %g1, [%g2] ASI_SCRATCHPAD
ldx [%l0 + HVTRAMP_DESCR_FAULT_INFO_VA], %g2
stxa %g2, [%g0] ASI_SCRATCHPAD
mov 0, %l1
lduw [%l0 + HVTRAMP_DESCR_NUM_MAPPINGS], %l2
add %l0, HVTRAMP_DESCR_MAPS, %l3
1: ldx [%l3 + HVTRAMP_MAPPING_VADDR], %o0
clr %o1
ldx [%l3 + HVTRAMP_MAPPING_TTE], %o2
mov HV_MMU_IMMU | HV_MMU_DMMU, %o3
mov HV_FAST_MMU_MAP_PERM_ADDR, %o5
ta HV_FAST_TRAP
brnz,pn %o0, 80f
nop
add %l1, 1, %l1
cmp %l1, %l2
blt,a,pt %xcc, 1b
add %l3, HVTRAMP_MAPPING_SIZE, %l3
ldx [%l0 + HVTRAMP_DESCR_FAULT_INFO_PA], %o0
mov HV_FAST_MMU_FAULT_AREA_CONF, %o5
ta HV_FAST_TRAP
brnz,pn %o0, 80f
nop
wrpr %g0, (PSTATE_PRIV | PSTATE_PEF), %pstate
ldx [%l0 + HVTRAMP_DESCR_THREAD_REG], %l6
mov 1, %o0
set 1f, %o1
mov HV_FAST_MMU_ENABLE, %o5
ta HV_FAST_TRAP
ba,pt %xcc, 80f
nop
1:
wr %g0, 0, %fprs
wr %g0, ASI_P, %asi
mov PRIMARY_CONTEXT, %g7
stxa %g0, [%g7] ASI_MMU
membar #Sync
mov SECONDARY_CONTEXT, %g7
stxa %g0, [%g7] ASI_MMU
membar #Sync
mov %l6, %g6
ldx [%g6 + TI_TASK], %g4
mov 1, %g5
sllx %g5, THREAD_SHIFT, %g5
sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
add %g6, %g5, %sp
call init_irqwork_curcpu
nop
call hard_smp_processor_id
nop
call sun4v_register_mondo_queues
nop
call init_cur_cpu_trap
mov %g6, %o0
wrpr %g0, (PSTATE_PRIV | PSTATE_PEF | PSTATE_IE), %pstate
call smp_callin
nop
call cpu_panic
nop
80: ba,pt %xcc, 80b
nop
.align 8
hv_cpu_startup_end: