Re: [Qemu-devel] [PATCH 0/3] ppc: complete the new HV mode

[Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>]

On 05/06/16 18:41, Cédric Le Goater wrote:

> Hello Mark,
> 
> On 06/03/2016 03:52 PM, Mark Cave-Ayland wrote:
>> On 03/06/16 13:11, Cédric Le Goater wrote:
>>
>>> This is follow up to complete the serie "ppc: preparing pnv landing
>>> (round 2)" plus a little fix on instruction privileges.
>>>
>>> Tested on a POWER8 pserie guest and on mac99.
>>>
>>> Benjamin Herrenschmidt (2):
>>>   ppc: Fix hreg_store_msr() so that non-HV mode cannot alter MSR:HV
>>>   ppc: Better figure out if processor has HV mode
>>>
>>> Cédric Le Goater (1):
>>>   ppc: fix hrfid, tlbia and slbia privilege
>>>
>>>  target-ppc/cpu.h            |  4 ++++
>>>  target-ppc/excp_helper.c    |  8 ++++++--
>>>  target-ppc/helper_regs.h    |  4 ++--
>>>  target-ppc/translate.c      | 10 ++++++----
>>>  target-ppc/translate_init.c | 19 +++++++++++++++----
>>>  5 files changed, 33 insertions(+), 12 deletions(-)
>>
>> Hi Cédric,
>>
>> I can confirm that this patchset fixes starting up OpenBIOS for both
>> g3beige/mac99 in my tests here. With the escc fix also applied, the only
>> outstanding issue is the removal of the tlb_flush() statements which
>> causes Darwin, MacOS X and HelenOS 0.60 to panic on boot
>>
>> My only request is if it would be possible to move patch 2 "ppc: Better
>> figure out if processor has HV mode" to the front of this patchset which
>> will make the remaining patches bisectable for the Mac machines. With that:
>>
>> Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
>>
>> Does anyone know if Ben has any ideas as to why the MMU tlb_flush
>> changes patch is causing such problems?
> 
> 
> Here is a fix I think. Could you give it a try ? 
> 
> Cheers,
> 
> C. 
> 
> From: Cédric Le Goater <clg@kaod.org>
> Subject: [PATCH] ppc: fix tlb flushes for 32bit environment
> Date: Sun, 05 Jun 2016 18:46:19 +0200
> MIME-Version: 1.0
> Content-Type: text/plain; charset=UTF-8
> Content-Transfer-Encoding: 8bit
> 
> commit cd0c6f473532 ('ppc: Do some batching of TCG tlb flushes')
> introduced an optimisation to flush TLBs only when a context
> synchronizing event is reached (interrupt, rfi). This was done for
> ppc64 but 32bit was forgotten on the way.
> 
> Tested on mac99 and g3beige with
> 
>     qemu-system-ppc -cdrom darwinppc-602.cdr -boot d
> 
> Signed-off-by: Cédric Le Goater <clg@kaod.org>
> ---
> 
>  I think the hunk in powerpc_excp() is needed if we don't generate a
>  context synchronizing event. what is best to do ?
> 
>  target-ppc/cpu.h         |    2 +-
>  target-ppc/excp_helper.c |   10 ++++++++++
>  target-ppc/helper_regs.h |    9 ++++++++-
>  target-ppc/translate.c   |    2 +-
>  4 files changed, 20 insertions(+), 3 deletions(-)
> 
> Index: qemu-dgibson-for-2.7.git/target-ppc/translate.c
> ===================================================================
> --- qemu-dgibson-for-2.7.git.orig/target-ppc/translate.c
> +++ qemu-dgibson-for-2.7.git/target-ppc/translate.c
> @@ -3290,7 +3290,7 @@ static void gen_eieio(DisasContext *ctx)
>  {
>  }
>  
> -#if !defined(CONFIG_USER_ONLY) && defined(TARGET_PPC64)
> +#if !defined(CONFIG_USER_ONLY)
>  static inline void gen_check_tlb_flush(DisasContext *ctx)
>  {
>      TCGv_i32 t = tcg_temp_new_i32();
> Index: qemu-dgibson-for-2.7.git/target-ppc/cpu.h
> ===================================================================
> --- qemu-dgibson-for-2.7.git.orig/target-ppc/cpu.h
> +++ qemu-dgibson-for-2.7.git/target-ppc/cpu.h
> @@ -958,9 +958,9 @@ struct CPUPPCState {
>      /* PowerPC 64 SLB area */
>      ppc_slb_t slb[MAX_SLB_ENTRIES];
>      int32_t slb_nr;
> +#endif
>      /* tcg TLB needs flush (deferred slb inval instruction typically) */
>      uint32_t tlb_need_flush;
> -#endif
>      /* segment registers */
>      hwaddr htab_base;
>      /* mask used to normalize hash value to PTEG index */
> Index: qemu-dgibson-for-2.7.git/target-ppc/helper_regs.h
> ===================================================================
> --- qemu-dgibson-for-2.7.git.orig/target-ppc/helper_regs.h
> +++ qemu-dgibson-for-2.7.git/target-ppc/helper_regs.h
> @@ -121,6 +121,13 @@ static inline int hreg_store_msr(CPUPPCS
>      }
>      if (((value >> MSR_IR) & 1) != msr_ir ||
>          ((value >> MSR_DR) & 1) != msr_dr) {
> +        /* A change of the instruction relocation bit in the MSR can
> +         * cause an implicit branch in the address space. This
> +         * requires a tlb flush.
> +         */
> +        if (env->mmu_model & POWERPC_MMU_32B) {
> +            env->tlb_need_flush = 1;
> +        }
>          cs->interrupt_request |= CPU_INTERRUPT_EXITTB;
>      }
>      if ((env->mmu_model & POWERPC_MMU_BOOKE) &&
> @@ -151,7 +158,7 @@ static inline int hreg_store_msr(CPUPPCS
>      return excp;
>  }
>  
> -#if !defined(CONFIG_USER_ONLY) && defined(TARGET_PPC64)
> +#if !defined(CONFIG_USER_ONLY)
>  static inline void check_tlb_flush(CPUPPCState *env)
>  {
>      CPUState *cs = CPU(ppc_env_get_cpu(env));
> Index: qemu-dgibson-for-2.7.git/target-ppc/excp_helper.c
> ===================================================================
> --- qemu-dgibson-for-2.7.git.orig/target-ppc/excp_helper.c
> +++ qemu-dgibson-for-2.7.git/target-ppc/excp_helper.c
> @@ -709,6 +709,16 @@ static inline void powerpc_excp(PowerPCC
>          }
>      }
>  #endif
> +    if (((new_msr >> MSR_IR) & 1) != msr_ir ||
> +        ((new_msr >> MSR_DR) & 1) != msr_dr) {
> +        /* A change of the instruction relocation bit in the MSR can
> +         * cause an implicit branch in the address space. This
> +         * requires a tlb flush.
> +         */
> +        if (env->mmu_model & POWERPC_MMU_32B) {
> +            env->tlb_need_flush = 1;
> +        }
> +    }
>      /* We don't use hreg_store_msr here as already have treated
>       * any special case that could occur. Just store MSR and update hflags
>       *
> 
> 

Hi Cédric,

I've just tried this patch on a selection of images here with both
g3beige and mac99 and as far as I can tell the crash has now gone away:

Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>

Thanks for getting this sorted so quickly.


ATB,

Mark.