[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r1093983 - in /subversion/trunk/subversion: include/private/svn_cache.h libsvn_fs_fs/caching.c libsvn_subr/cache-inprocess.c libsvn_subr/cache-membuffer.c libsvn_subr/cache-memcache.c libsvn_subr/cache.c libsvn_subr/cache.h

From: Stefan Fuhrmann <stefanfuhrmann_at_alice-dsl.de>
Date: Sun, 15 May 2011 18:05:02 +0200

On 14.05.2011 03:35, Daniel Shahaf wrote:
> stefan2_at_apache.org wrote on Sat, Apr 16, 2011 at 13:06:07 -0000:
>> Author: stefan2
>> Date: Sat Apr 16 13:06:07 2011
>> New Revision: 1093983
>>
>> URL: http://svn.apache.org/viewvc?rev=1093983&view=rev
>> Log:
>> In symmetry to svn_cache__get_partial, introduce a partial
>> setter / modifier API function that enabled the caller to
>> modify changed items without the full deserialization -
>> modification - serialization chain. Therefore, adding entries
>> to e.g. directories becomes O(1) instead of O(n).
>>
> ...
>> +++ subversion/trunk/subversion/libsvn_subr/cache-membuffer.c Sat Apr 16 13:06:07 2011
>> @@ -1174,6 +1174,12 @@ membuffer_cache_get(svn_membuffer_t *cac
>> return deserializer(item, buffer, entry->size, pool);
>> }
>>
>> +/* Look for the cache entry identified by KEY and KEY_LEN. FOUND indicates
>> + * whether that entry exists. If not found, *ITEM will be NULL. Otherwise,
>> + * the DESERIALIZER is called with that entry and the BATON provided
>> + * and will extract the desired information. The result is set in *ITEM.
>> + * Allocations will be done in POOL.
>> + */
>> static svn_error_t *
>> membuffer_cache_get_partial(svn_membuffer_t *cache,
>> const void *key,
>> @@ -1239,6 +1245,105 @@ membuffer_cache_get_partial(svn_membuffe
>> return unlock_cache(cache, err);
>> }
>>
>> +/* Look for the cache entry identified by KEY and KEY_LEN. If no entry
>> + * has been found, the function returns without modifying the cache.
>> + * Otherwise, FUNC is called with that entry and the BATON provided
>> + * and may modify the cache entry. Allocations will be done in POOL.
>> + */
>> +static svn_error_t *
>> +membuffer_cache_set_partial(svn_membuffer_t *cache,
>> + const void *key,
>> + apr_size_t key_len,
>> + svn_cache__partial_setter_func_t func,
>> + void *baton,
>> + DEBUG_CACHE_MEMBUFFER_TAG_ARG
>> + apr_pool_t *pool)
>> +{
>> + apr_uint32_t group_index;
>> + unsigned char to_find[KEY_SIZE];
>> + entry_t *entry;
>> + svn_error_t *err = SVN_NO_ERROR;
>> +
>> + /* cache item lookup
>> + */
>> + group_index = get_group_index(&cache, key, key_len, to_find, pool);
>> +
>> + SVN_ERR(lock_cache(cache));
>> +
>> + entry = find_entry(cache, group_index, to_find, FALSE);
>> + cache->total_reads++;
>> +
>> + /* this function is a no-op if the item is not in cache
>> + */
>> + if (entry != NULL)
>> + {
>> + /* access the serialized cache item */
>> + char *data = (char*)cache->data + entry->offset;
>> + char *orig_data = data;
>> + apr_size_t size = entry->size;
>> +
>> + entry->hit_count++;
>> + cache->hit_count++;
>> + cache->total_writes++;
>> +
>> +#ifdef DEBUG_CACHE_MEMBUFFER
>> +
>> + /* Check for overlapping entries.
>> + */
>> + SVN_ERR_ASSERT(entry->next == NO_INDEX ||
>> + entry->offset + size
>> +<= get_entry(cache, entry->next)->offset);
>> +
>> + /* Compare original content, type and key (hashes)
>> + */
>> + SVN_ERR(store_content_part(tag, data, size, pool));
>> + SVN_ERR(assert_equal_tags(&entry->tag, tag));
>> +
>> +#endif
>> +
>> + /* modify it, preferrably in-situ.
>> + */
>> + err = func(&data,&size, baton, pool);
>> +
>> + /* if modification caused a re-allocation, we need to remove the old
>> + * entry and to copy the new data back into cache.
>> + */
>> + if (data != orig_data)
> Shouldn't you be checking ERR before storing the change made by FUNC?
>
You are right! Fixed in r1103391.
Thanks for the review!

-- Stefan^2.
Received on 2011-05-15 18:05:39 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.