Re: [Qemu-devel] [PATCH v3] migration: discard non-migratable RAMBlocks

[Cédric Le Goater]

On 05/11/2018 06:59 AM, Peter Xu wrote:
> On Fri, May 11, 2018 at 12:02:48AM +0200, Cédric Le Goater wrote:
>> On the POWER9 processor, the XIVE interrupt controller can control
>> interrupt sources using MMIO to trigger events, to EOI or to turn off
>> the sources. Priority management and interrupt acknowledgment is also
>> controlled by MMIO in the presenter sub-engine.
>>
>> These MMIO regions are exposed to guests in QEMU with a set of 'ram
>> device' memory mappings, similarly to VFIO, and the VMAs are populated
>> dynamically with the appropriate pages using a fault handler.
>>
>> But, these regions are an issue for migration. We need to discard the
>> associated RAMBlocks from the RAM state on the source VM and let the
>> destination VM rebuild the memory mappings on the new host in the
>> post_load() operation just before resuming the system.
>>
>> To achieve this goal, the following introduces a new RAMBlock flag
>> RAM_MIGRATABLE which is updated in the vmstate_register_ram() and
>> vmstate_unregister_ram() routines. This flag is then used by the
>> migration to identify RAMBlocks to discard on the source. Some checks
>> are also performed on the destination to make sure nothing invalid was
>> sent.
>>
>> Signed-off-by: Cédric Le Goater <address@hidden>
>> ---
>>
>>  Changes sinve v2:
>>
>>  - added an error_report() in ram_save_host_page() 
>>  - un/set the RAMBlock RAM_MIGRATABLE directly under 
>> vmstate_un/register_ram()
>>    with some new flag helpers 
>>  
>>  exec.c                    | 18 ++++++++++++++++++
>>  include/exec/cpu-common.h |  3 +++
>>  migration/ram.c           | 43 +++++++++++++++++++++++++++++++++----------
>>  migration/savevm.c        |  2 ++
>>  4 files changed, 56 insertions(+), 10 deletions(-)
>>
>> diff --git a/exec.c b/exec.c
>> index c7fcefa851b2..079c5c8bab7b 100644
>> --- a/exec.c
>> +++ b/exec.c
>> @@ -104,6 +104,9 @@ static MemoryRegion io_mem_unassigned;
>>   * (Set during postcopy)
>>   */
>>  #define RAM_UF_ZEROPAGE (1 << 3)
>> +
>> +/* RAM can be migrated */
>> +#define RAM_MIGRATABLE (1 << 4)
>>  #endif
>>  
>>  #ifdef TARGET_PAGE_BITS_VARY
>> @@ -1797,6 +1800,21 @@ void qemu_ram_set_uf_zeroable(RAMBlock *rb)
>>      rb->flags |= RAM_UF_ZEROPAGE;
>>  }
>>  
>> +bool qemu_ram_is_migratable(RAMBlock *rb)
>> +{
>> +    return rb->flags & RAM_MIGRATABLE;
>> +}
>> +
>> +void qemu_ram_set_migratable(RAMBlock *rb)
>> +{
>> +    rb->flags |= RAM_MIGRATABLE;
>> +}
>> +
>> +void qemu_ram_unset_migratable(RAMBlock *rb)
>> +{
>> +    rb->flags &= ~RAM_MIGRATABLE;
>> +}
>> +
>>  /* Called with iothread lock held.  */
>>  void qemu_ram_set_idstr(RAMBlock *new_block, const char *name, DeviceState 
>> *dev)
>>  {
>> diff --git a/include/exec/cpu-common.h b/include/exec/cpu-common.h
>> index 24d335f95d45..488288fce959 100644
>> --- a/include/exec/cpu-common.h
>> +++ b/include/exec/cpu-common.h
>> @@ -75,6 +75,9 @@ const char *qemu_ram_get_idstr(RAMBlock *rb);
>>  bool qemu_ram_is_shared(RAMBlock *rb);
>>  bool qemu_ram_is_uf_zeroable(RAMBlock *rb);
>>  void qemu_ram_set_uf_zeroable(RAMBlock *rb);
>> +bool qemu_ram_is_migratable(RAMBlock *rb);
>> +void qemu_ram_set_migratable(RAMBlock *rb);
>> +void qemu_ram_unset_migratable(RAMBlock *rb);
>>  
>>  size_t qemu_ram_pagesize(RAMBlock *block);
>>  size_t qemu_ram_pagesize_largest(void);
>> diff --git a/migration/ram.c b/migration/ram.c
>> index 912810c18e0f..dfdec78ecb03 100644
>> --- a/migration/ram.c
>> +++ b/migration/ram.c
>> @@ -187,6 +187,11 @@ void ramblock_recv_bitmap_set_range(RAMBlock *rb, void 
>> *host_addr,
>>                        nr);
>>  }
>>  
>> +/* Should be holding either ram_list.mutex, or the RCU lock. */
>> +#define RAMBLOCK_FOREACH_MIGRATABLE(block)             \
>> +    RAMBLOCK_FOREACH(block)                            \
>> +        if (!qemu_ram_is_migratable(block)) {} else
> 
> Pure question: is here an explicit use of "else" rather than:
> 
>       if (qemu_ram_is_migratable(block))
> 
> ?

The above is an imperfect macro 'trick' trying to handle code that 
would have omitted {}. It provides some protection.
   
>> +
>>  /*
>>   * An outstanding page request, on the source, having been received
>>   * and queued
>> @@ -813,6 +818,10 @@ unsigned long migration_bitmap_find_dirty(RAMState *rs, 
>> RAMBlock *rb,
>>      unsigned long *bitmap = rb->bmap;
>>      unsigned long next;
>>  
>> +    if (!qemu_ram_is_migratable(rb)) {
>> +        return size;
>> +    }
> 
> (I would change the caller side at find_dirty_block(), we can just
>  skip that if not migratable.  But this is also fine to me)

Hmm. The do {} while() loop and callees in ram_find_and_save_block() 
is a bit complex to follow. I stumbled a few time when trying to 
change it.

>> +
>>      if (rs->ram_bulk_stage && start > 0) {
>>          next = start + 1;
>>      } else {
>> @@ -858,7 +867,7 @@ uint64_t ram_pagesize_summary(void)
>>      RAMBlock *block;
>>      uint64_t summary = 0;
>>  
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          summary |= block->page_size;
>>      }
>>  
>> @@ -882,7 +891,7 @@ static void migration_bitmap_sync(RAMState *rs)
>>  
>>      qemu_mutex_lock(&rs->bitmap_mutex);
>>      rcu_read_lock();
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          migration_bitmap_sync_range(rs, block, 0, block->used_length);
>>      }
>>      rcu_read_unlock();
>> @@ -1521,6 +1530,11 @@ static int ram_save_host_page(RAMState *rs, 
>> PageSearchStatus *pss,
>>      size_t pagesize_bits =
>>          qemu_ram_pagesize(pss->block) >> TARGET_PAGE_BITS;
>>  
>> +    if (!qemu_ram_is_migratable(pss->block)) {
>> +        error_report("block %s should not be migrated !", 
>> pss->block->idstr);
>> +        return 0;
>> +    }
>> +
>>      do {
>>          /* Check the pages is dirty and if it is send it */
>>          if (!migration_bitmap_clear_dirty(rs, pss->block, pss->page)) {
>> @@ -1619,7 +1633,7 @@ uint64_t ram_bytes_total(void)
>>      uint64_t total = 0;
>>  
>>      rcu_read_lock();
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          total += block->used_length;
>>      }
>>      rcu_read_unlock();
>> @@ -1674,7 +1688,7 @@ static void ram_save_cleanup(void *opaque)
>>       */
>>      memory_global_dirty_log_stop();
>>  
>> -    QLIST_FOREACH_RCU(block, &ram_list.blocks, next) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          g_free(block->bmap);
>>          block->bmap = NULL;
>>          g_free(block->unsentmap);
>> @@ -1737,7 +1751,7 @@ void 
>> ram_postcopy_migrated_memory_release(MigrationState *ms)
>>  {
>>      struct RAMBlock *block;
>>  
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          unsigned long *bitmap = block->bmap;
>>          unsigned long range = block->used_length >> TARGET_PAGE_BITS;
>>          unsigned long run_start = find_next_zero_bit(bitmap, range, 0);
>> @@ -1815,7 +1829,7 @@ static int 
>> postcopy_each_ram_send_discard(MigrationState *ms)
>>      struct RAMBlock *block;
>>      int ret;
>>  
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          PostcopyDiscardState *pds =
>>              postcopy_discard_send_init(ms, block->idstr);
>>  
>> @@ -2023,7 +2037,7 @@ int ram_postcopy_send_discard_bitmap(MigrationState 
>> *ms)
>>      rs->last_sent_block = NULL;
>>      rs->last_page = 0;
>>  
>> -    QLIST_FOREACH_RCU(block, &ram_list.blocks, next) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          unsigned long pages = block->used_length >> TARGET_PAGE_BITS;
>>          unsigned long *bitmap = block->bmap;
>>          unsigned long *unsentmap = block->unsentmap;
>> @@ -2182,7 +2196,7 @@ static void ram_list_init_bitmaps(void)
>>  
>>      /* Skip setting bitmap if there is no RAM */
>>      if (ram_bytes_total()) {
>> -        QLIST_FOREACH_RCU(block, &ram_list.blocks, next) {
>> +        RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>              pages = block->max_length >> TARGET_PAGE_BITS;
>>              block->bmap = bitmap_new(pages);
>>              bitmap_set(block->bmap, 0, pages);
>> @@ -2263,7 +2277,7 @@ static int ram_save_setup(QEMUFile *f, void *opaque)
>>  
>>      qemu_put_be64(f, ram_bytes_total() | RAM_SAVE_FLAG_MEM_SIZE);
>>  
>> -    RAMBLOCK_FOREACH(block) {
>> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
>>          qemu_put_byte(f, strlen(block->idstr));
>>          qemu_put_buffer(f, (uint8_t *)block->idstr, strlen(block->idstr));
>>          qemu_put_be64(f, block->used_length);
>> @@ -2507,6 +2521,11 @@ static inline RAMBlock 
>> *ram_block_from_stream(QEMUFile *f, int flags)
>>          return NULL;
>>      }
>>  
>> +    if (!qemu_ram_is_migratable(block)) {
>> +        error_report("block %s should not be migrated !", id);
>> +        return NULL;
>> +    }
>> +
>>      return block;
>>  }
>>  
>> @@ -3011,7 +3030,11 @@ static int ram_load(QEMUFile *f, void *opaque, int 
>> version_id)
>>                  length = qemu_get_be64(f);
>>  
>>                  block = qemu_ram_block_by_name(id);
>> -                if (block) {
>> +                if (block && !qemu_ram_is_migratable(block)) {
>> +                    error_report("block %s should not be migrated !", id);
>> +                    ret = -EINVAL;
>> +
> 
> (extra new line)

ah. yes indeed.

> I saw quite a few of such checks in the patch that dumps errors.
> Frankly speaking I would still prefer assertions if we are sure it's
> programming errors, and I believe incorrect marking of RAMBlocks are.
> I think I understand the concern here that we are just careful with
> migration codes to avoid VM crash but at least here we are doing
> precopy loading and we won't have problem to use assertion since it
> won't crash the source (and after all we are returning -EINVAL so I
> suppose the VM will quit too, just like a crash here).

It is true that QEMU would be receiving invalid state from the source
in this cases but asserting would stop migration on the target quite 
brutally. It seems excessive for bad input.

Thanks,

C.

> 
> I don't have strong opinion on this and current patch actually looks
> good to me; I only raise this question up.
> 
> Thanks,
> 
>> +                } else if (block) {
>>                      if (length != block->used_length) {
>>                          Error *local_err = NULL;
>>  
>> diff --git a/migration/savevm.c b/migration/savevm.c
>> index e2be02afe42c..9ebfba738ea4 100644
>> --- a/migration/savevm.c
>> +++ b/migration/savevm.c
>> @@ -2501,11 +2501,13 @@ void vmstate_register_ram(MemoryRegion *mr, 
>> DeviceState *dev)
>>  {
>>      qemu_ram_set_idstr(mr->ram_block,
>>                         memory_region_name(mr), dev);
>> +    qemu_ram_set_migratable(mr->ram_block);
>>  }
>>  
>>  void vmstate_unregister_ram(MemoryRegion *mr, DeviceState *dev)
>>  {
>>      qemu_ram_unset_idstr(mr->ram_block);
>> +    qemu_ram_unset_migratable(mr->ram_block);
>>  }
>>  
>>  void vmstate_register_ram_global(MemoryRegion *mr)
>> -- 
>> 2.13.6
>>
>>
>