From: "Gautham R. Shenoy" <[email protected]>
On 64-bit servers, SPRN_SPRG3 and its userspace read-only mirror
SPRN_USPRG3 are used as userspace VDSO write and read registers
respectively.
SPRN_SPRG3 is lost when we enter stop4 and above, and is currently not
restored. As a result, any read from SPRN_USPRG3 returns zero on an
exit from stop4 and above.
Thus in this situation, on POWER9, any call from sched_getcpu() always
returns zero, as on powerpc, we call __kernel_getcpu() which relies
upon SPRN_USPRG3 to report the CPU and NUMA node information.
Fix this by restoring SPRN_SPRG3 on wake up from a deep stop state
with the sprg_vdso value that is cached in PACA.
Fixes: e1c1cfed5432 ("powerpc/powernv: Save/Restore additional SPRs
for stop4 cpuidle")
Reported-by: Florian Weimer <[email protected]>
Cc: <[email protected]> # 4.14
Cc: Oleg Nesterov <[email protected]>
Cc: Michael Neuling <[email protected]>
Cc: Michael Ellerman <[email protected]>
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Vaidyanathan Srinivasan <[email protected]>
Signed-off-by: Gautham R. Shenoy <[email protected]>
---
Change from v1:
Restoring the SPRG3 from paca->sprg_vdso instead of saving
it separately during stop-entry, as suggested by Mikey.
arch/powerpc/kernel/idle_book3s.S | 2 ++
1 file changed, 2 insertions(+)
diff --git a/arch/powerpc/kernel/idle_book3s.S b/arch/powerpc/kernel/idle_book3s.S
index d85d551..672ead8 100644
--- a/arch/powerpc/kernel/idle_book3s.S
+++ b/arch/powerpc/kernel/idle_book3s.S
@@ -144,7 +144,9 @@ power9_restore_additional_sprs:
mtspr SPRN_MMCR1, r4
ld r3, STOP_MMCR2(r13)
+ ld r4, PACA_SPRG_VDSO(r13)
mtspr SPRN_MMCR2, r3
+ mtspr SPRN_SPRG3, r4
blr
/*
--
1.8.3.1
On Wed, 2018-07-18 at 08:33:16 UTC, "Gautham R. Shenoy" wrote:
> From: "Gautham R. Shenoy" <[email protected]>
>
> On 64-bit servers, SPRN_SPRG3 and its userspace read-only mirror
> SPRN_USPRG3 are used as userspace VDSO write and read registers
> respectively.
>
> SPRN_SPRG3 is lost when we enter stop4 and above, and is currently not
> restored. As a result, any read from SPRN_USPRG3 returns zero on an
> exit from stop4 and above.
>
> Thus in this situation, on POWER9, any call from sched_getcpu() always
> returns zero, as on powerpc, we call __kernel_getcpu() which relies
> upon SPRN_USPRG3 to report the CPU and NUMA node information.
>
> Fix this by restoring SPRN_SPRG3 on wake up from a deep stop state
> with the sprg_vdso value that is cached in PACA.
>
> Fixes: e1c1cfed5432 ("powerpc/powernv: Save/Restore additional SPRs
> for stop4 cpuidle")
>
> Reported-by: Florian Weimer <[email protected]>
> Cc: <[email protected]> # 4.14
> Cc: Oleg Nesterov <[email protected]>
> Cc: Michael Neuling <[email protected]>
> Cc: Michael Ellerman <[email protected]>
> Cc: Benjamin Herrenschmidt <[email protected]>
> Cc: Vaidyanathan Srinivasan <[email protected]>
> Signed-off-by: Gautham R. Shenoy <[email protected]>
Applied to powerpc fixes, thanks.
https://git.kernel.org/powerpc/c/b03897cf318dfc47de33a7ecbc7655
cheers