Re: [RFC PATCH v1 1/1] powerpc/85xx: Wakeup kexec smp slave cpusin second kernel

[Yang,Wei]

On 08/31/2013 05:12 PM, Yu Chen wrote:
> >From 1ccf579b871dfd5938ce958f729361a203485c74 Mon Sep 17 00:00:00 2001
> From: Yu Chen <chenyu105@xxxxxxxxx>
> Date: Sat, 31 Aug 2013 23:52:31 +0800
> Subject: [PATCH]  powerpc/85xx: Wakeup kexec smp slave cpus in second kernel
>
> In current 85xx smp kexec implementation,master cpu reset slave cpus
> by mpic_reset_core,
> before jump to second kernel.In order to wake slave cpus up in second
> kernel,we debug
> this patch on p2041rdb.

What problem causes that you do the modification? I am just curious as 
kexec feature always is fine on our
P2041RDB board.:-)

Wei
> The main principle of this patch,is to get slave cpus polling for flag
> to change,
> thus waiting for master cpu to set it with non-zero cpu number(see misc_32.S).
> This flag is placed in kexec control page,so it would not be
> overlapped when copying kimage.
> The master cpu put flag's physical address in r28 as a parameter
> passed to second kernel,
> so the latter knows how to wake slave cpus up in smp_85xx_kick_cpu.
> The pseudo-code may be like:
> void slave_cpu_spin(void)
> {
>          int cpu = smp_processor_id();
>          while (*kexec_poll != cpu)
>                  ;
>          /*slave wakeup and jump*/
>          jump(*(kexec_poll+1));
> }
>
> void master_cpu_wakeup(unsigned long *kexec_poll, int cpu)
> {
>          *(kexec_poll+1) = __early_start;
>          mb();
>          *kexec_poll = cpu;
> }
>
> However,after applied this patch,we got some kernel exception during
> booting second kernel,
> I'm not sure if it's caused by improper treament of cache,or tlb,or
> other.So I put this
> patch here hoping someone can check and review it.
>
> Signed-off-by: Yu Chen <chenyu105@xxxxxxxxx>
> ---
>   arch/powerpc/kernel/head_fsl_booke.S |    7 ++
>   arch/powerpc/kernel/misc_32.S        |   66 +++++++++++++-
>   arch/powerpc/platforms/85xx/smp.c    |  166 ++++++++++++++++++++++++++++++----
>   3 files changed, 222 insertions(+), 17 deletions(-)
>   mode change 100644 => 100755 arch/powerpc/kernel/head_fsl_booke.S
>   mode change 100644 => 100755 arch/powerpc/kernel/misc_32.S
>   mode change 100644 => 100755 arch/powerpc/platforms/85xx/smp.c
>
> diff --git a/arch/powerpc/kernel/head_fsl_booke.S
> b/arch/powerpc/kernel/head_fsl_booke.S
> old mode 100644
> new mode 100755
> index d10a7ca..63c8392
> --- a/arch/powerpc/kernel/head_fsl_booke.S
> +++ b/arch/powerpc/kernel/head_fsl_booke.S
> @@ -178,6 +178,13 @@ _ENTRY(__early_start)
>        * This is where the main kernel code starts.
>        */
>
> +#if defined(CONFIG_KEXEC) && defined(CONFIG_SMP)
> +    /* r28 contain position where slave cpus spin*/
> +    lis    r1,kexec_poll_phy@h
> +    ori    r1,r1,kexec_poll_phy@l
> +    stw    r28,0(r1)
> +#endif
> +
>       /* ptr to current */
>       lis    r2,init_task@h
>       ori    r2,r2,init_task@l
> diff --git a/arch/powerpc/kernel/misc_32.S b/arch/powerpc/kernel/misc_32.S
> old mode 100644
> new mode 100755
> index e469f30..d9eefc2
> --- a/arch/powerpc/kernel/misc_32.S
> +++ b/arch/powerpc/kernel/misc_32.S
> @@ -120,7 +120,7 @@ _GLOBAL(reloc_got2)
>       addi    r4,r4,1b@l
>       subf    r0,r4,r0
>       add    r7,r0,r7
> -2:    lwz    r0,0(r7)
> +    2:    lwz    r0,0(r7)
>       add    r0,r0,r3
>       stw    r0,0(r7)
>       addi    r7,r7,4
> @@ -692,6 +692,7 @@ _GLOBAL(__main)
>       blr
>
>   #ifdef CONFIG_KEXEC
> +#define KEXEC_MAGIC 0xdeadbeef
>       /*
>        * Must be relocatable PIC code callable as a C function.
>        */
> @@ -707,6 +708,16 @@ relocate_new_kernel:
>       mr    r30, r4
>       mr    r31, r5
>
> +#ifdef CONFIG_SMP
> +    bl    1f
> +1:    mflr    r8
> +    addi    r8,r8,kexec_flag-1b
> +    lis     r7,PAGE_OFFSET@h
> +    ori     r7,r7,PAGE_OFFSET@l
> +    /*r28 contain slave cpu spin physical address */
> +    subf    r28, r7, r8
> +#endif
> +
>   #define ENTRY_MAPPING_KEXEC_SETUP
>   #include "fsl_booke_entry_mapping.S"
>   #undef ENTRY_MAPPING_KEXEC_SETUP
> @@ -1172,4 +1183,57 @@ relocate_new_kernel_end:
>       .globl relocate_new_kernel_size
>   relocate_new_kernel_size:
>       .long relocate_new_kernel_end - relocate_new_kernel
> +#ifdef CONFIG_FSL_BOOKE
> +    /**
> +    * Slave cpus wait for kexec_flag to change
> +    */
> +    .globl relocate_smp_cpu_offset
> +relocate_smp_cpu_offset:
> +    .long relocate_smp_cpu_wait-relocate_new_kernel
> +
> +    .globl relocate_smp_cpu_wait
> +relocate_smp_cpu_wait:
> +
> +    bl    1f
> +1:    mflr    r5
> +    addi    r5,r5,kexec_flag-1b
> +    /*see if anyone calls me?*/
> +    mfspr   r24,SPRN_PIR
> +99:    lwz    r4,4(r5)
> +    cmpw    r4,r24
> +    msync
> +    bne        99b
> +
> +    msync
> +    /*r4 contains jump address*/
> +    lwz    r4,8(r5)
> +    msync
> +    lis    r5,MSR_KERNEL@h
> +    ori    r5,r5,MSR_KERNEL@l
> +    msync
> +    isync
> +    mtspr    SPRN_SRR1, r5
> +    mtspr    SPRN_SRR0, r4
> +    msync
> +    isync
> +    rfi
> +    isync
> +1:    b    1b
> +
> +    /**
> +    * kexec_flag indicates a kexec magic
> +    * kexec_flag+4 bytes supposed to be set with cpu number
> +    * kexec_flag+8 countain addr for slave cpu to jump into
> +    */
> +    .globl kexec_flag
> +kexec_flag:
> +    .long   KEXEC_MAGIC
> +    .long    0
> +    .long    0
> +relocate_smp_cpu_wait_end:
> +    .globl relocate_smp_cpu_size
> +relocate_smp_cpu_size:
> +    .long relocate_smp_cpu_wait_end-relocate_smp_cpu_wait
> +#endif
> +
>   #endif
> diff --git a/arch/powerpc/platforms/85xx/smp.c
> b/arch/powerpc/platforms/85xx/smp.c
> old mode 100644
> new mode 100755
> index 5ced4f5..c4f5c4c
> --- a/arch/powerpc/platforms/85xx/smp.c
> +++ b/arch/powerpc/platforms/85xx/smp.c
> @@ -140,6 +140,70 @@ static inline u32 read_spin_table_addr_l(void *spin_table)
>           (ulong)spin_table + sizeof(struct epapr_spin_table));
>       return in_be32(&((struct epapr_spin_table *)spin_table)->addr_l);
>   }
> +#ifdef CONFIG_KEXEC
> +
> +#define KEXEC_MAGIC 0xdeadbeef
> +#define KEXEC_RESERVE_LIMIT 0x10
> +unsigned long kexec_poll_phy;
> +extern void reserve_kexec_bootmem(unsigned long poll_phy, int size);
> +
> +/*
> + * Reserved bootmem for slave cpus kexec spin area.
> + */
> +void mpc85xx_smp_reserve_kexec(void)
> +{
> +    unsigned long kexec_poll_virt;
> +    unsigned long *kexec_magic_virt;
> +
> +    if (!kexec_poll_phy ||
> +            kexec_poll_phy >= __max_low_memory)
> +        return;
> +
> +    kexec_poll_virt = (unsigned long)phys_to_virt(kexec_poll_phy);
> +    kexec_magic_virt = (unsigned long *)kexec_poll_virt;
> +
> +    if (*kexec_magic_virt == KEXEC_MAGIC)
> +        reserve_kexec_bootmem(kexec_poll_phy, KEXEC_RESERVE_LIMIT);
> +}
> +
> +/*
> + * Kick slave cpus from kexec spin area.
> + */
> +int mpc85xx_smp_kick_kexec_cpus(int nr)
> +{
> +    unsigned long  kexec_poll_virt;
> +    unsigned long *kexec_flag_virt;
> +    unsigned long *kexec_magic_virt;
> +    unsigned long *kexec_jump_virt;
> +
> +    /*verify accessible*/
> +    if (!kexec_poll_phy ||
> +            kexec_poll_phy >= __max_low_memory)
> +        return -EBUSY;
> +
> +    kexec_poll_virt = (unsigned long)phys_to_virt(kexec_poll_phy);
> +
> +    kexec_magic_virt = (unsigned long *)kexec_poll_virt;
> +    kexec_flag_virt = (unsigned long *)kexec_poll_virt + 1;
> +    kexec_jump_virt = (unsigned long *)kexec_poll_virt + 2;
> +
> +    /*verify a valid kexec kick*/
> +    if (*kexec_magic_virt == KEXEC_MAGIC) {
> +        flush_dcache_range((ulong)kexec_poll_virt,
> +        (ulong)kexec_poll_virt + L1_CACHE_BYTES-1);
> +        *kexec_jump_virt = (unsigned long)__early_start;
> +        mb();
> +        /*kick cpu[nr] up*/
> +        *kexec_flag_virt = nr;
> +        mb();
> +        flush_dcache_range((ulong)kexec_poll_virt,
> +        (ulong)kexec_poll_virt + L1_CACHE_BYTES-1);
> +
> +        return 0;
> +    }
> +    return -EBUSY;
> +}
> +#endif
>
>   static int smp_85xx_kick_cpu(int nr)
>   {
> @@ -181,6 +245,10 @@ static int smp_85xx_kick_cpu(int nr)
>
>       local_irq_save(flags);
>   #ifdef CONFIG_PPC32
> +#ifdef CONFIG_KEXEC
> +    if (!mpc85xx_smp_kick_kexec_cpus(nr))
> +        goto kexec_kick_done;
> +#endif
>   #ifdef CONFIG_HOTPLUG_CPU
>       /* Corresponding to generic_set_cpu_dead() */
>       generic_set_cpu_up(nr);
> @@ -225,7 +293,9 @@ static int smp_85xx_kick_cpu(int nr)
>       out_be32(&spin_table->pir, hw_cpu);
>       out_be32(&spin_table->addr_l, __pa(__early_start));
>       flush_spin_table(spin_table);
> -
> +#ifdef CONFIG_KEXEC
> +kexec_kick_done:
> +#endif
>       /* Wait a bit for the CPU to ack. */
>       if (!spin_event_timeout(__secondary_hold_acknowledge == hw_cpu,
>                       10000, 100)) {
> @@ -266,7 +336,13 @@ struct smp_ops_t smp_85xx_ops = {
>   };
>
>   #ifdef CONFIG_KEXEC
> +
>   atomic_t kexec_down_cpus = ATOMIC_INIT(0);
> +atomic_t kexec_ready_to_reboot = ATOMIC_INIT(0);
> +atomic_t kexec_slave_finish = ATOMIC_INIT(0);
> +unsigned long wait_code_buffer;
> +static struct kimage *save_image;
> +extern const unsigned int relocate_smp_cpu_size;
>
>   void mpc85xx_smp_kexec_cpu_down(int crash_shutdown, int secondary)
>   {
> @@ -274,8 +350,29 @@ void mpc85xx_smp_kexec_cpu_down(int
> crash_shutdown, int secondary)
>
>       if (secondary) {
>           atomic_inc(&kexec_down_cpus);
> -        /* loop forever */
> -        while (1);
> +        mb();
> +
> +        if (crash_shutdown) {
> +            /* loop forever */
> +            while (1)
> +                ;
> +        } else {
> +            while (!atomic_read(&kexec_ready_to_reboot))
> +                cpu_relax();
> +            /*flush destination*/
> +            if (save_image)
> +                mpc85xx_smp_flush_dcache_kexec(save_image, 1);
> +
> +            flush_icache_range(wait_code_buffer,
> +                wait_code_buffer + relocate_smp_cpu_size);
> +            flush_dcache_range(wait_code_buffer,
> +                wait_code_buffer + relocate_smp_cpu_size);
> +
> +            atomic_inc(&kexec_slave_finish);
> +
> +            ((void (*)(void)) wait_code_buffer)();
> +            /* NOTREACHED */
> +        }
>       }
>   }
>
> @@ -285,13 +382,23 @@ static void mpc85xx_smp_kexec_down(void *arg)
>           ppc_md.kexec_cpu_down(0,1);
>   }
>
> -static void map_and_flush(unsigned long paddr)
> +static void map_and_flush(unsigned long paddr, int atomic)
>   {
>       struct page *page = pfn_to_page(paddr >> PAGE_SHIFT);
> -    unsigned long kaddr  = (unsigned long)kmap(page);
> +    unsigned long kaddr;
> +
> +    if (atomic)
> +        kaddr  = (unsigned long)kmap_atomic(page);
> +    else
> +        kaddr  = (unsigned long)kmap(page);
>
>       flush_dcache_range(kaddr, kaddr + PAGE_SIZE);
> -    kunmap(page);
> +    flush_icache_range(kaddr, kaddr + PAGE_SIZE);
> +
> +    if (atomic)
> +        kunmap_atomic((void *)kaddr);
> +    else
> +        kunmap(page);
>   }
>
>   /**
> @@ -300,7 +407,7 @@ static void map_and_flush(unsigned long paddr)
>    * are performed out of an overabundance of caution as interrupts are not
>    * disabled yet and we can switch cores
>    */
> -static void mpc85xx_smp_flush_dcache_kexec(struct kimage *image)
> +static void mpc85xx_smp_flush_dcache_kexec(struct kimage *image, int atomic)
>   {
>       kimage_entry_t *ptr, entry;
>       unsigned long paddr;
> @@ -312,18 +419,18 @@ static void
> mpc85xx_smp_flush_dcache_kexec(struct kimage *image)
>                ptr = (entry & IND_INDIRECTION) ?
>                   phys_to_virt(entry & PAGE_MASK) : ptr + 1) {
>               if (!(entry & IND_DESTINATION)) {
> -                map_and_flush(entry);
> +                map_and_flush(entry, atomic);
>               }
>           }
>           /* flush out last IND_DONE page */
> -        map_and_flush(entry);
> +        map_and_flush(entry, atomic);
>       } else {
>           /* crash type kexec images are copied to the crash region */
>           for (i = 0; i < image->nr_segments; i++) {
>               struct kexec_segment *seg = &image->segment[i];
>               for (paddr = seg->mem; paddr < seg->mem + seg->memsz;
>                    paddr += PAGE_SIZE) {
> -                map_and_flush(paddr);
> +                map_and_flush(paddr, atomic);
>               }
>           }
>       }
> @@ -335,13 +442,18 @@ static void
> mpc85xx_smp_flush_dcache_kexec(struct kimage *image)
>
>   static void mpc85xx_smp_machine_kexec(struct kimage *image)
>   {
> +    extern const unsigned char  relocate_smp_cpu_wait[];
> +    extern const unsigned int relocate_smp_cpu_offset;
>       int timeout = INT_MAX;
>       int i, num_cpus = num_present_cpus();
>
>       mpc85xx_smp_flush_dcache_kexec(image);
>
> -    if (image->type == KEXEC_TYPE_DEFAULT)
> +    if (image->type == KEXEC_TYPE_DEFAULT) {
> +        save_image = image;
> +        mb();
>           smp_call_function(mpc85xx_smp_kexec_down, NULL, 0);
> +    }
>
>       while ( (atomic_read(&kexec_down_cpus) != (num_cpus - 1)) &&
>           ( timeout > 0 ) )
> @@ -352,12 +464,34 @@ static void mpc85xx_smp_machine_kexec(struct
> kimage *image)
>       if ( !timeout )
>           printk(KERN_ERR "Unable to bring down secondary cpu(s)");
>
> -    for_each_online_cpu(i)
> -    {
> -        if ( i == smp_processor_id() ) continue;
> -        mpic_reset_core(i);
> -    }
> +    if (image->type == KEXEC_TYPE_DEFAULT) {
>
> +        wait_code_buffer =
> +        (unsigned long)page_address(image->control_code_page)+
> +                relocate_smp_cpu_offset;
> +
> +        /* copy slave cpu spin code to the control code page */
> +        memcpy((void *)wait_code_buffer, relocate_smp_cpu_wait,
> +                        relocate_smp_cpu_size);
> +        atomic_set(&kexec_ready_to_reboot, 1);
> +        mb();
> +        timeout = INT_MAX;
> +
> +        while ((atomic_read(&kexec_slave_finish) != (num_cpus-1)) &&
> +            (timeout > 0))
> +            timeout--;
> +
> +        if (!timeout)
> +            printk(KERN_ERR "Unable to wait for secondary cpu(s) to
> flush caches\n");
> +
> +        } else {
> +        for_each_online_cpu(i)
> +        {
> +            if (i == smp_processor_id())
> +                continue;
> +            mpic_reset_core(i);
> +        }
> +    }
>       default_machine_kexec(image);
>   }
>   #endif /* CONFIG_KEXEC */

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/