diff options
| author | Paul Mackerras <paulus@samba.org> | 2008-01-15 09:29:33 +0300 |
|---|---|---|
| committer | Paul Mackerras <paulus@samba.org> | 2008-01-15 09:30:58 +0300 |
| commit | dfbe0d3b6be52596b5694b1bb75b19562e769021 (patch) | |
| tree | 75e3ef05b6dd8cc7065fe77a05a9cd1b3495619a /arch/powerpc/mm/slb.c | |
| parent | d262c32a4bcc3e5fda0325a64e53c25fe1e999d7 (diff) | |
| download | linux-dfbe0d3b6be52596b5694b1bb75b19562e769021.tar.xz | |
[POWERPC] Fix boot failure on POWER6
Commit 473980a99316c0e788bca50996375a2815124ce1 added a call to clear
the SLB shadow buffer before registering it. Unfortunately this means
that we clear out the entries that slb_initialize has previously set in
there. On POWER6, the hypervisor uses the SLB shadow buffer when doing
partition switches, and that means that after the next partition switch,
each non-boot CPU has no SLB entries to map the kernel text and data,
which causes it to crash.
This fixes it by reverting most of 473980a9 and instead clearing the
3rd entry explicitly in slb_initialize. This fixes the problem that
473980a9 was trying to solve, but without breaking POWER6.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Diffstat (limited to 'arch/powerpc/mm/slb.c')
| -rw-r--r-- | arch/powerpc/mm/slb.c | 10 |
1 files changed, 2 insertions, 8 deletions
diff --git a/arch/powerpc/mm/slb.c b/arch/powerpc/mm/slb.c index a282bc212e80..50d7372bc2ce 100644 --- a/arch/powerpc/mm/slb.c +++ b/arch/powerpc/mm/slb.c @@ -82,14 +82,6 @@ static inline void slb_shadow_clear(unsigned long entry) get_slb_shadow()->save_area[entry].esid = 0; } -void slb_shadow_clear_all(void) -{ - int i; - - for (i = 0; i < SLB_NUM_BOLTED; i++) - slb_shadow_clear(i); -} - static inline void create_shadowed_slbe(unsigned long ea, int ssize, unsigned long flags, unsigned long entry) @@ -300,6 +292,8 @@ void slb_initialize(void) create_shadowed_slbe(VMALLOC_START, mmu_kernel_ssize, vflags, 1); + slb_shadow_clear(2); + /* We don't bolt the stack for the time being - we're in boot, * so the stack is in the bolted segment. By the time it goes * elsewhere, we'll call _switch() which will bolt in the new |