Re: [Xen-devel] [PATCH 3/4] xen/mapcache: introduce xen_remap_cache_entry()

[Igor Druzhinin <igor.druzhinin@xxxxxxxxxx>]

On 01/07/17 01:08, Stefano Stabellini wrote:
> On Fri, 30 Jun 2017, Igor Druzhinin wrote:
>> This new call is trying to update a requested map cache entry
>> according to the changes in the physmap. The call is searching
>> for the entry, unmaps it, tries to translate the address and
>> maps again at the same place. If the mapping is dummy this call
>> will make it real.
>>
>> This function makes use of a new xenforeignmemory_map2() call
>> with extended interface that was recently introduced in
>> libxenforeignmemory [1].
>>
>> [1] https://www.mail-archive.com/xen-devel@xxxxxxxxxxxxx/msg113007.html
>>
>> Signed-off-by: Igor Druzhinin <igor.druzhinin@xxxxxxxxxx>
>> ---
>>  configure                     |  18 ++++++++
>>  hw/i386/xen/xen-mapcache.c    | 105 
>> +++++++++++++++++++++++++++++++++++++++---
>>  include/hw/xen/xen_common.h   |   7 +++
>>  include/sysemu/xen-mapcache.h |   6 +++
>>  4 files changed, 130 insertions(+), 6 deletions(-)
>>
>> diff --git a/configure b/configure
>> index c571ad1..ad6156b 100755
>> --- a/configure
>> +++ b/configure
>> @@ -2021,6 +2021,24 @@ EOF
>>      # Xen unstable
>>      elif
>>          cat > $TMPC <<EOF &&
>> +#undef XC_WANT_COMPAT_MAP_FOREIGN_API
>> +#include <xenforeignmemory.h>
>> +int main(void) {
>> +  xenforeignmemory_handle *xfmem;
>> +
>> +  xfmem = xenforeignmemory_open(0, 0);
>> +  xenforeignmemory_map2(xfmem, 0, 0, 0, 0, 0, 0, 0);
>> +
>> +  return 0;
>> +}
>> +EOF
>> +        compile_prog "" "$xen_libs -lxendevicemodel $xen_stable_libs"
>> +      then
>> +      xen_stable_libs="-lxendevicemodel $xen_stable_libs"
>> +      xen_ctrl_version=41000
>> +      xen=yes
>> +    elif
>> +        cat > $TMPC <<EOF &&
>>  #undef XC_WANT_COMPAT_DEVICEMODEL_API
>>  #define __XEN_TOOLS__
>>  #include <xendevicemodel.h>
>> diff --git a/hw/i386/xen/xen-mapcache.c b/hw/i386/xen/xen-mapcache.c
>> index 05050de..5d8d990 100644
>> --- a/hw/i386/xen/xen-mapcache.c
>> +++ b/hw/i386/xen/xen-mapcache.c
>> @@ -149,6 +149,7 @@ void xen_map_cache_init(phys_offset_to_gaddr_t f, void 
>> *opaque)
>>  }
>>  
>>  static void xen_remap_bucket(MapCacheEntry *entry,
>> +                             void *vaddr,
>>                               hwaddr size,
>>                               hwaddr address_index,
>>                               bool dummy)
>> @@ -179,11 +180,11 @@ static void xen_remap_bucket(MapCacheEntry *entry,
>>      }
>>  
>>      if (!dummy) {
>> -        vaddr_base = xenforeignmemory_map(xen_fmem, xen_domid,
>> -                                           PROT_READ|PROT_WRITE,
>> +        vaddr_base = xenforeignmemory_map2(xen_fmem, xen_domid, vaddr,
>> +                                           PROT_READ|PROT_WRITE, 0,
>>                                             nb_pfn, pfns, err);
>>          if (vaddr_base == NULL) {
>> -            perror("xenforeignmemory_map");
>> +            perror("xenforeignmemory_map2");
>>              exit(-1);
>>          }
>>      } else {
>> @@ -191,7 +192,7 @@ static void xen_remap_bucket(MapCacheEntry *entry,
>>           * We create dummy mappings where we are unable to create a foreign
>>           * mapping immediately due to certain circumstances (i.e. on resume 
>> now)
>>           */
>> -        vaddr_base = mmap(NULL, size, PROT_READ|PROT_WRITE,
>> +        vaddr_base = mmap(vaddr, size, PROT_READ|PROT_WRITE,
>>                            MAP_ANON|MAP_SHARED, -1, 0);
>>          if (vaddr_base == NULL) {
>>              perror("mmap");
>> @@ -278,14 +279,14 @@ tryagain:
>>      if (!entry) {
>>          entry = g_malloc0(sizeof (MapCacheEntry));
>>          pentry->next = entry;
>> -        xen_remap_bucket(entry, cache_size, address_index, dummy);
>> +        xen_remap_bucket(entry, NULL, cache_size, address_index, dummy);
>>      } else if (!entry->lock) {
>>          if (!entry->vaddr_base || entry->paddr_index != address_index ||
>>                  entry->size != cache_size ||
>>                  !test_bits(address_offset >> XC_PAGE_SHIFT,
>>                      test_bit_size >> XC_PAGE_SHIFT,
>>                      entry->valid_mapping)) {
>> -            xen_remap_bucket(entry, cache_size, address_index, dummy);
>> +            xen_remap_bucket(entry, NULL, cache_size, address_index, dummy);
>>          }
>>      }
>>  
>> @@ -482,3 +483,95 @@ void xen_invalidate_map_cache(void)
>>  
>>      mapcache_unlock();
>>  }
>> +
>> +static uint8_t *xen_remap_cache_entry_unlocked(hwaddr phys_addr, hwaddr 
>> size)
> 
> I think it's best if we use a more descriptive name, such as
> xen_replace_dummy_entry to avoid confusion.
> 
> 
>> +{
>> +    MapCacheEntry *entry, *pentry = NULL;
>> +    hwaddr address_index;
>> +    hwaddr address_offset;
>> +    hwaddr cache_size = size;
>> +    hwaddr test_bit_size;
>> +    void *vaddr = NULL;
>> +    uint8_t lock;
>> +
>> +    address_index  = phys_addr >> MCACHE_BUCKET_SHIFT;
>> +    address_offset = phys_addr & (MCACHE_BUCKET_SIZE - 1);
>> +
>> +    /* test_bit_size is always a multiple of XC_PAGE_SIZE */
>> +    if (size) {
> 
> There is no need to make xen_remap_cache_entry_unlocked generic: it's
> only used with explicitly sized mappings, right? We could assert(!size).
> 

Sure.

> 
>> +        test_bit_size = size + (phys_addr & (XC_PAGE_SIZE - 1));
>> +        if (test_bit_size % XC_PAGE_SIZE) {
>> +            test_bit_size += XC_PAGE_SIZE - (test_bit_size % XC_PAGE_SIZE);
>> +        }
>> +        cache_size = size + address_offset;
>> +        if (cache_size % MCACHE_BUCKET_SIZE) {
>> +            cache_size += MCACHE_BUCKET_SIZE - (cache_size % 
>> MCACHE_BUCKET_SIZE);
>> +        }
>> +    } else {
>> +        test_bit_size = XC_PAGE_SIZE;
>> +        cache_size = MCACHE_BUCKET_SIZE;
>> +    }
>> +
>> +    /* Search for the requested map cache entry to invalidate */
>> +    entry = &mapcache->entry[address_index % mapcache->nr_buckets];
>> +    while (entry && !(entry->paddr_index == address_index && entry->size == 
>> cache_size)) {
>> +        pentry = entry;
>> +        entry = entry->next;
>> +    }
>> +    if (!entry) {
>> +        DPRINTF("Trying to update an entry for %lx that is not in the 
>> mapcache!\n", phys_addr);
>> +        return NULL;
>> +    }
>> +
>> +    vaddr = entry->vaddr_base;
>> +    lock = entry->lock;
>> +    if (entry->vaddr_base) {
>> +        ram_block_notify_remove(entry->vaddr_base, entry->size);
>> +        if (munmap(entry->vaddr_base, entry->size) != 0) {
>> +            perror("unmap fails");
>> +            exit(-1);
>> +        }
>> +    }
> 
> Why are we calling ram_block_notify_remove? Isn't the ram_block about to
> be remapped with the correct underlying pages?
> 

It's here for symmetry. xen_remap_bucket is going to call
ram_block_notify_add and I didn't want to gate it there. But I probably
should.

> 
>> +    entry->vaddr_base = NULL;
>> +    entry->lock = 0;
> 
> Why can't we just keep using this entry as is?
> 
> 
>> +    if (mapcache->phys_offset_to_gaddr) {
>> +        phys_addr = mapcache->phys_offset_to_gaddr(phys_addr, size, 
>> mapcache->opaque);
>> +
>> +        address_index  = phys_addr >> MCACHE_BUCKET_SHIFT;
>> +        address_offset = phys_addr & (MCACHE_BUCKET_SIZE - 1);
>> +    }
> 
> Instead of having this check to find the new address, why don't we just
> pass it to xen_remap_cache_entry_unlocked as an argument?
> 

Agree.

> 
>> +    /* Address may have changed so we need to repeat the search */
>> +    entry = &mapcache->entry[address_index % mapcache->nr_buckets];
>> +    while (entry && entry->lock && entry->vaddr_base) {
>> +        pentry = entry;
>> +        entry = entry->next;
>> +    }
>> +    if (!entry) {
>> +        entry = g_malloc0(sizeof (MapCacheEntry));
>> +        pentry->next = entry;
>> +    }
> 
> Is it really possible to already have a mapcache entry for the new
> phys_addr, which has never been used before? It doesn't look like it.
> Also, why are we creating a new entry instead of simply reusing the
> previous one?
> 

I wasn't completely sure it's safe to reuse the previous entry since
it's indexed by its previous physical address. But it looks like that
QEMU is going to use only its old address for referencing. That'll make
the code much simpler.

> 
>> +    entry->lock = 0;
>> +    xen_remap_bucket(entry, vaddr, cache_size, address_index, false);
>> +    if(!test_bits(address_offset >> XC_PAGE_SHIFT,
>> +                test_bit_size >> XC_PAGE_SHIFT,
>> +                entry->valid_mapping)) {
>> +        DPRINTF("Unable to update an entry for %lx in the mapcache!\n", 
>> phys_addr);
>> +        return NULL;
>> +    }
>> +
>> +    entry->lock = lock;
>> +    return entry->vaddr_base + address_offset;
>> +}
>> +
>> +uint8_t *xen_remap_cache_entry(hwaddr phys_addr, hwaddr size)
>> +{
>> +    uint8_t *p;
>> +
>> +    mapcache_lock();
>> +    p = xen_remap_cache_entry_unlocked(phys_addr, size);
>> +    mapcache_unlock();
>> +    return p;
>> +}
>> diff --git a/include/hw/xen/xen_common.h b/include/hw/xen/xen_common.h
>> index e00ddd7..70a5cad 100644
>> --- a/include/hw/xen/xen_common.h
>> +++ b/include/hw/xen/xen_common.h
>> @@ -78,6 +78,13 @@ static inline void *xenforeignmemory_map(xc_interface *h, 
>> uint32_t dom,
>>  
>>  extern xenforeignmemory_handle *xen_fmem;
>>  
>> +#if CONFIG_XEN_CTRL_INTERFACE_VERSION < 41000
>> +
>> +#define xenforeignmemory_map2(h, d, a, p, f, ps, ar, e) \
>> +    xenforeignmemory_map(h, d, p, ps, ar, e)
>> +
>> +#endif
>> +
>>  #if CONFIG_XEN_CTRL_INTERFACE_VERSION < 40900
>>  
>>  typedef xc_interface xendevicemodel_handle;
>> diff --git a/include/sysemu/xen-mapcache.h b/include/sysemu/xen-mapcache.h
>> index 01daaad..8c140d0 100644
>> --- a/include/sysemu/xen-mapcache.h
>> +++ b/include/sysemu/xen-mapcache.h
>> @@ -21,6 +21,7 @@ uint8_t *xen_map_cache(hwaddr phys_addr, hwaddr size,
>>  ram_addr_t xen_ram_addr_from_mapcache(void *ptr);
>>  void xen_invalidate_map_cache_entry(uint8_t *buffer);
>>  void xen_invalidate_map_cache(void);
>> +uint8_t *xen_remap_cache_entry(hwaddr phys_addr, hwaddr size);
>>  
>>  #else
>>  
>> @@ -50,6 +51,11 @@ static inline void xen_invalidate_map_cache(void)
>>  {
>>  }
>>  
>> +static inline uint8_t *xen_remap_cache_entry(hwaddr phys_addr, hwaddr size)
>> +{
>> +    abort();
>> +}
>> +
>>  #endif
>>  
>>  #endif /* XEN_MAPCACHE_H */
>> -- 
>> 2.7.4
>>

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel