xen: mapcache: Fix finding matching entry
PATCH 1/1 xen: mapcache: Fix finding matching entry
If we have request without lock and hit unlocked or invalid entry during the search, we remap it immediately, even if we have matching entry in next entries in bucket. This leads to duplication of mappings of the same size, and to possibility of selecting the wrong element during invalidation and underflow it's entry->lock counter
--- hw/xen/xen-mapcache.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
hw/xen/xen-mapcache.c +3 -3
@@ -376,12 +376,12 @@ tryagain: entry = &mc->entry[address_index % mc->nr_buckets]; -while (entry && (lock || entry->lock) && entry->vaddr_base && - (entry->paddr_index != address_index || entry->size != cache_size || +while (entry && (!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))) { -if (!free_entry && !entry->lock) { +if (!free_entry && (!entry->lock || !entry->vaddr_base)) { free_entry = entry; free_pentry = pentry; }
Signature · 1 line
2.39.5
Quoted 8 lines · Aleksandr Partanen
If we have request without lock and hit unlocked or invalid entry during the search, we remap it immediately, even if we have matching entry in next entries in bucket. This leads to duplication of mappings of the same size, and to possibility of selecting the wrong element during invalidation and underflow it's
entry->lockcounterSigned-off-by Aleksandr Partanen <[email protected]>
Hi Aleksandr, thanks for the patch, it looks correct to me.
Edgar, would you be able to give it a look as well to make sure?
Quoted 27 lines
--- hw/xen/xen-mapcache.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)hw/xen/xen-mapcache.c +3
-3@@ -376,12 +376,12 @@ tryagain: entry = &mc->entry[address_index % mc->nr_buckets]; -while (entry && (lock || entry->lock) && entry->vaddr_base && - (entry->paddr_index != address_index || entry->size != cache_size || +while (entry && (!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))) { -if (!free_entry && !entry->lock) { +if (!free_entry && (!entry->lock || !entry->vaddr_base)) { free_entry = entry; free_pentry = pentry; }Signature · 1 line
2.39.5
Edgar E. Iglesias
Reviewed-byQuoted 8 lines · Stefano Stabellini
Quoted 7 lines · Aleksandr Partanen
If we have request without lock and hit unlocked or invalid entry during the search, we remap it immediately, even if we have matching entry in next entries in bucket. This leads to duplication of mappings of the same size, and to possibility of selecting the wrong element during invalidation and underflow it's
entry->lockcounter
Quoted 8 lines
Quoted 1 line
Signed-off-by Aleksandr Partanen <[email protected]>Hi Aleksandr, thanks for the patch, it looks correct to me.
Reviewed-by Stefano Stabellini <[email protected]>Edgar, would you be able to give it a look as well to make sure?
Looks good to me too, thanks!
Quoted 29 lines
Quoted 29 lines
--- hw/xen/xen-mapcache.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)hw/xen/xen-mapcache.c +3
-3@@ -376,12 +376,12 @@ tryagain: entry = &mc->entry[address_index % mc->nr_buckets]; -while (entry && (lock || entry->lock) && entry->vaddr_base && - (entry->paddr_index != address_index || entry->size != cache_size || +while (entry && (!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))) { -if (!free_entry && !entry->lock) { +if (!free_entry && (!entry->lock || !entry->vaddr_base)) { free_entry = entry; free_pentry = pentry; }Signature · 1 line
2.39.5