2019-08-08 12:50:06

by Christophe Leroy

Subject: [PATCH] powerpc/64e: drop stale call to smp_processor_id() which hangs SMP startup

Santa commit ebb9d30a6a74 ("powerpc/mm: any thread in one core can be
the first to setup TLB1") removed the need to know the cpu_id in
early_init_this_mmu(), but the call to smp_processor_id() which was
marked __maybe_used remained.

Since commit ed1cd6deb013 ("powerpc: Activate
CONFIG_THREAD_INFO_IN_TASK") thread_info cannot be reached before mmu
is properly set up.

Drop this stale call to smp_processor_id() which make SMP hang
when CONFIG_PREEMPT is set.

Reported-by: Chris Packham <[email protected]>
Fixes: ebb9d30a6a74 ("powerpc/mm: any thread in one core can be the first to setup TLB1")
Link: https://github.com/linuxppc/issues/issues/264
Signed-off-by: Christophe Leroy <[email protected]>
Cc: [email protected]
---
arch/powerpc/mm/nohash/tlb.c | 1 -
1 file changed, 1 deletion(-)

diff --git a/arch/powerpc/mm/nohash/tlb.c b/arch/powerpc/mm/nohash/tlb.c
index d4acf6fa0596..bf60983a58c7 100644
--- a/arch/powerpc/mm/nohash/tlb.c
+++ b/arch/powerpc/mm/nohash/tlb.c
@@ -630,7 +630,6 @@ static void early_init_this_mmu(void)
#ifdef CONFIG_PPC_FSL_BOOK3E
if (mmu_has_feature(MMU_FTR_TYPE_FSL_E)) {
unsigned int num_cams;
- int __maybe_unused cpu = smp_processor_id();
bool map = true;

/* use a quarter of the TLBCAM for bolted linear map */
--
2.13.3

2019-08-08 21:18:47

by Chris Packham

Subject: Re: [PATCH] powerpc/64e: drop stale call to smp_processor_id() which hangs SMP startup

Hi Christophe,

On Thu, 2019-08-08 at 12:48 +0000, Christophe Leroy wrote:
> Santa commit ebb9d30a6a74 ("powerpc/mm: any thread in one core can be
> the first to setup TLB1") removed the need to know the cpu_id in
> early_init_this_mmu(), but the call to smp_processor_id() which was
> marked __maybe_used remained.
>
> Since commit ed1cd6deb013 ("powerpc: Activate
> CONFIG_THREAD_INFO_IN_TASK") thread_info cannot be reached before mmu
> is properly set up.
>
> Drop this stale call to smp_processor_id() which make SMP hang
> when CONFIG_PREEMPT is set.
>
> Reported-by: Chris Packham <[email protected]>
> Fixes: ebb9d30a6a74 ("powerpc/mm: any thread in one core can be the
> first to setup TLB1")
> Link: https://github.com/linuxppc/issues/issues/264
> Signed-off-by: Christophe Leroy <[email protected]>
> Cc: [email protected]

Many thanks for your help.

Tested-by: Chris Packham <[email protected]>

> ---
>  arch/powerpc/mm/nohash/tlb.c | 1 -
>  1 file changed, 1 deletion(-)
>
> diff --git a/arch/powerpc/mm/nohash/tlb.c
> b/arch/powerpc/mm/nohash/tlb.c
> index d4acf6fa0596..bf60983a58c7 100644
> --- a/arch/powerpc/mm/nohash/tlb.c
> +++ b/arch/powerpc/mm/nohash/tlb.c
> @@ -630,7 +630,6 @@ static void early_init_this_mmu(void)
>  #ifdef CONFIG_PPC_FSL_BOOK3E
>   if (mmu_has_feature(MMU_FTR_TYPE_FSL_E)) {
>   unsigned int num_cams;
> - int __maybe_unused cpu = smp_processor_id();
>   bool map = true;
>  
>   /* use a quarter of the TLBCAM for bolted linear map
> */

2019-08-12 13:16:19

by Michael Ellerman

Subject: Re: [PATCH] powerpc/64e: drop stale call to smp_processor_id() which hangs SMP startup

On Thu, 2019-08-08 at 12:48:26 UTC, Christophe Leroy wrote:
> Santa commit ebb9d30a6a74 ("powerpc/mm: any thread in one core can be
> the first to setup TLB1") removed the need to know the cpu_id in
> early_init_this_mmu(), but the call to smp_processor_id() which was
> marked __maybe_used remained.
>
> Since commit ed1cd6deb013 ("powerpc: Activate
> CONFIG_THREAD_INFO_IN_TASK") thread_info cannot be reached before mmu
> is properly set up.
>
> Drop this stale call to smp_processor_id() which make SMP hang
> when CONFIG_PREEMPT is set.
>
> Reported-by: Chris Packham <[email protected]>
> Fixes: ebb9d30a6a74 ("powerpc/mm: any thread in one core can be the first to setup TLB1")
> Link: https://github.com/linuxppc/issues/issues/264
> Signed-off-by: Christophe Leroy <[email protected]>
> Cc: [email protected]

Applied to powerpc fixes, thanks.

https://git.kernel.org/powerpc/c/b9ee5e04fd77898208c51b1395fa0b5e8536f9b6

cheers