1
0
Fork 0
mirror of https://github.com/git/git.git synced 2024-11-05 08:47:56 +01:00
git/name-hash.c

240 lines
5.9 KiB
C
Raw Normal View History

/*
* name-hash.c
*
* Hashing names in the index state
*
* Copyright (C) 2008 Linus Torvalds
*/
#define NO_THE_INDEX_COMPATIBILITY_MACROS
#include "cache.h"
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
struct dir_entry {
struct hashmap_entry ent;
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
struct dir_entry *parent;
int nr;
unsigned int namelen;
char name[FLEX_ARRAY];
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
};
static int dir_entry_cmp(const struct dir_entry *e1,
const struct dir_entry *e2, const char *name)
{
return e1->namelen != e2->namelen || strncasecmp(e1->name,
name ? name : e2->name, e1->namelen);
}
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
static struct dir_entry *find_dir_entry(struct index_state *istate,
const char *name, unsigned int namelen)
{
struct dir_entry key;
hashmap_entry_init(&key, memihash(name, namelen));
key.namelen = namelen;
return hashmap_get(&istate->dir_hash, &key, name);
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
}
static struct dir_entry *hash_dir_entry(struct index_state *istate,
struct cache_entry *ce, int namelen)
{
/*
* Throw each directory component in the hash for quick lookup
name-hash: stop storing trailing '/' on paths in index_state.dir_hash When 5102c617 (Add case insensitivity support for directories when using git status, 2010-10-03) added directories to the name-hash there was only a single hash table in which both real cache entries and leading directory prefixes were registered. To distinguish between the two types of entries, directories were stored with a trailing '/'. 2092678c (name-hash.c: fix endless loop with core.ignorecase=true, 2013-02-28), however, moved directories to a separate hash table (index_state.dir_hash) but retained the (now) redundant trailing '/', thus callers continue to bear the burden of ensuring the slash's presence before searching the index for a directory. Eliminate this redundancy by storing paths in the dir-hash without the trailing '/'. An important benefit of this change is that it eliminates undocumented and dangerous behavior of dir.c:directory_exists_in_index_icase() in which it assumes not only that it can validly access one character beyond the end of its incoming directory argument, but also that that character will unconditionally be a '/'. This perilous behavior was "tolerated" because the string passed in by its lone caller always had a '/' in that position, however, things broke [1] when 2eac2a4c (ls-files -k: a directory only can be killed if the index has a non-directory, 2013-08-15) added a new caller which failed to respect the undocumented assumption. [1]: http://thread.gmane.org/gmane.comp.version-control.git/232727 Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-09-17 09:06:16 +02:00
* during a git status. Directory components are stored without their
* closing slash. Despite submodules being a directory, they never
name-hash: stop storing trailing '/' on paths in index_state.dir_hash When 5102c617 (Add case insensitivity support for directories when using git status, 2010-10-03) added directories to the name-hash there was only a single hash table in which both real cache entries and leading directory prefixes were registered. To distinguish between the two types of entries, directories were stored with a trailing '/'. 2092678c (name-hash.c: fix endless loop with core.ignorecase=true, 2013-02-28), however, moved directories to a separate hash table (index_state.dir_hash) but retained the (now) redundant trailing '/', thus callers continue to bear the burden of ensuring the slash's presence before searching the index for a directory. Eliminate this redundancy by storing paths in the dir-hash without the trailing '/'. An important benefit of this change is that it eliminates undocumented and dangerous behavior of dir.c:directory_exists_in_index_icase() in which it assumes not only that it can validly access one character beyond the end of its incoming directory argument, but also that that character will unconditionally be a '/'. This perilous behavior was "tolerated" because the string passed in by its lone caller always had a '/' in that position, however, things broke [1] when 2eac2a4c (ls-files -k: a directory only can be killed if the index has a non-directory, 2013-08-15) added a new caller which failed to respect the undocumented assumption. [1]: http://thread.gmane.org/gmane.comp.version-control.git/232727 Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-09-17 09:06:16 +02:00
* reach this point, because they are stored
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
* in index_state.name_hash (as ordinary cache_entries).
*/
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
struct dir_entry *dir;
/* get length of parent directory */
while (namelen > 0 && !is_dir_sep(ce->name[namelen - 1]))
namelen--;
if (namelen <= 0)
return NULL;
name-hash: stop storing trailing '/' on paths in index_state.dir_hash When 5102c617 (Add case insensitivity support for directories when using git status, 2010-10-03) added directories to the name-hash there was only a single hash table in which both real cache entries and leading directory prefixes were registered. To distinguish between the two types of entries, directories were stored with a trailing '/'. 2092678c (name-hash.c: fix endless loop with core.ignorecase=true, 2013-02-28), however, moved directories to a separate hash table (index_state.dir_hash) but retained the (now) redundant trailing '/', thus callers continue to bear the burden of ensuring the slash's presence before searching the index for a directory. Eliminate this redundancy by storing paths in the dir-hash without the trailing '/'. An important benefit of this change is that it eliminates undocumented and dangerous behavior of dir.c:directory_exists_in_index_icase() in which it assumes not only that it can validly access one character beyond the end of its incoming directory argument, but also that that character will unconditionally be a '/'. This perilous behavior was "tolerated" because the string passed in by its lone caller always had a '/' in that position, however, things broke [1] when 2eac2a4c (ls-files -k: a directory only can be killed if the index has a non-directory, 2013-08-15) added a new caller which failed to respect the undocumented assumption. [1]: http://thread.gmane.org/gmane.comp.version-control.git/232727 Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-09-17 09:06:16 +02:00
namelen--;
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
/* lookup existing entry for that directory */
dir = find_dir_entry(istate, ce->name, namelen);
if (!dir) {
/* not found, create it and add to hash table */
dir = xcalloc(1, sizeof(struct dir_entry) + namelen + 1);
hashmap_entry_init(dir, memihash(ce->name, namelen));
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
dir->namelen = namelen;
strncpy(dir->name, ce->name, namelen);
hashmap_add(&istate->dir_hash, dir);
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
/* recursively add missing parent directories */
name-hash: stop storing trailing '/' on paths in index_state.dir_hash When 5102c617 (Add case insensitivity support for directories when using git status, 2010-10-03) added directories to the name-hash there was only a single hash table in which both real cache entries and leading directory prefixes were registered. To distinguish between the two types of entries, directories were stored with a trailing '/'. 2092678c (name-hash.c: fix endless loop with core.ignorecase=true, 2013-02-28), however, moved directories to a separate hash table (index_state.dir_hash) but retained the (now) redundant trailing '/', thus callers continue to bear the burden of ensuring the slash's presence before searching the index for a directory. Eliminate this redundancy by storing paths in the dir-hash without the trailing '/'. An important benefit of this change is that it eliminates undocumented and dangerous behavior of dir.c:directory_exists_in_index_icase() in which it assumes not only that it can validly access one character beyond the end of its incoming directory argument, but also that that character will unconditionally be a '/'. This perilous behavior was "tolerated" because the string passed in by its lone caller always had a '/' in that position, however, things broke [1] when 2eac2a4c (ls-files -k: a directory only can be killed if the index has a non-directory, 2013-08-15) added a new caller which failed to respect the undocumented assumption. [1]: http://thread.gmane.org/gmane.comp.version-control.git/232727 Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-09-17 09:06:16 +02:00
dir->parent = hash_dir_entry(istate, ce, namelen);
}
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
return dir;
}
static void add_dir_entry(struct index_state *istate, struct cache_entry *ce)
{
/* Add reference to the directory entry (and parents if 0). */
struct dir_entry *dir = hash_dir_entry(istate, ce, ce_namelen(ce));
while (dir && !(dir->nr++))
dir = dir->parent;
}
static void remove_dir_entry(struct index_state *istate, struct cache_entry *ce)
{
/*
* Release reference to the directory entry. If 0, remove and continue
* with parent directory.
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
*/
struct dir_entry *dir = hash_dir_entry(istate, ce, ce_namelen(ce));
while (dir && !(--dir->nr)) {
struct dir_entry *parent = dir->parent;
hashmap_remove(&istate->dir_hash, dir, NULL);
free(dir);
dir = parent;
}
}
static void hash_index_entry(struct index_state *istate, struct cache_entry *ce)
{
if (ce->ce_flags & CE_HASHED)
return;
ce->ce_flags |= CE_HASHED;
hashmap_entry_init(ce, memihash(ce->name, ce_namelen(ce)));
hashmap_add(&istate->name_hash, ce);
if (ignore_case)
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
add_dir_entry(istate, ce);
}
static int cache_entry_cmp(const struct cache_entry *ce1,
const struct cache_entry *ce2, const void *remove)
{
/*
* For remove_name_hash, find the exact entry (pointer equality); for
* index_file_exists, find all entries with matching hash code and
* decide whether the entry matches in same_name.
*/
return remove ? !(ce1 == ce2) : 0;
}
static void lazy_init_name_hash(struct index_state *istate)
{
int nr;
if (istate->name_hash_initialized)
return;
hashmap_init(&istate->name_hash, (hashmap_cmp_fn) cache_entry_cmp,
istate->cache_nr);
hashmap_init(&istate->dir_hash, (hashmap_cmp_fn) dir_entry_cmp, 0);
for (nr = 0; nr < istate->cache_nr; nr++)
hash_index_entry(istate, istate->cache[nr]);
istate->name_hash_initialized = 1;
}
void add_name_hash(struct index_state *istate, struct cache_entry *ce)
{
if (istate->name_hash_initialized)
hash_index_entry(istate, ce);
}
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
void remove_name_hash(struct index_state *istate, struct cache_entry *ce)
{
if (!istate->name_hash_initialized || !(ce->ce_flags & CE_HASHED))
return;
ce->ce_flags &= ~CE_HASHED;
hashmap_remove(&istate->name_hash, ce, ce);
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
if (ignore_case)
remove_dir_entry(istate, ce);
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
}
static int slow_same_name(const char *name1, int len1, const char *name2, int len2)
{
if (len1 != len2)
return 0;
while (len1) {
unsigned char c1 = *name1++;
unsigned char c2 = *name2++;
len1--;
if (c1 != c2) {
c1 = toupper(c1);
c2 = toupper(c2);
if (c1 != c2)
return 0;
}
}
return 1;
}
static int same_name(const struct cache_entry *ce, const char *name, int namelen, int icase)
{
int len = ce_namelen(ce);
/*
* Always do exact compare, even if we want a case-ignoring comparison;
* we do the quick exact one first, because it will be the common case.
*/
if (len == namelen && !memcmp(name, ce->name, len))
return 1;
if (!icase)
return 0;
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
return slow_same_name(name, namelen, ce->name, len);
}
int index_dir_exists(struct index_state *istate, const char *name, int namelen)
{
struct dir_entry *dir;
lazy_init_name_hash(istate);
dir = find_dir_entry(istate, name, namelen);
return dir && dir->nr;
}
void adjust_dirname_case(struct index_state *istate, char *name)
{
const char *startPtr = name;
const char *ptr = startPtr;
lazy_init_name_hash(istate);
while (*ptr) {
while (*ptr && *ptr != '/')
ptr++;
if (*ptr == '/') {
struct dir_entry *dir;
ptr++;
dir = find_dir_entry(istate, name, ptr - name + 1);
if (dir) {
memcpy((void *)startPtr, dir->name + (startPtr - name), ptr - startPtr);
startPtr = ptr;
}
}
}
}
struct cache_entry *index_file_exists(struct index_state *istate, const char *name, int namelen, int icase)
{
struct cache_entry *ce;
lazy_init_name_hash(istate);
ce = hashmap_get_from_hash(&istate->name_hash,
memihash(name, namelen), NULL);
while (ce) {
if (same_name(ce, name, namelen, icase))
return ce;
ce = hashmap_get_next(&istate->name_hash, ce);
}
return NULL;
}
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
void free_name_hash(struct index_state *istate)
{
if (!istate->name_hash_initialized)
return;
istate->name_hash_initialized = 0;
hashmap_free(&istate->name_hash, 0);
hashmap_free(&istate->dir_hash, 1);
name-hash.c: fix endless loop with core.ignorecase=true With core.ignorecase=true, name-hash.c builds a case insensitive index of all tracked directories. Currently, the existing cache entry structures are added multiple times to the same hashtable (with different name lengths and hash codes). However, there's only one dir_next pointer, which gets completely messed up in case of hash collisions. In the worst case, this causes an endless loop if ce == ce->dir_next (see t7062). Use a separate hashtable and separate structures for the directory index so that each directory entry has its own next pointer. Use reference counting to track which directory entry contains files. There are only slight changes to the name-hash.c API: - new free_name_hash() used by read_cache.c::discard_index() - remove_name_hash() takes an additional index_state parameter - index_name_exists() for a directory (trailing '/') may return a cache entry that has been removed (CE_UNHASHED). This is not a problem as the return value is only used to check if the directory exists (dir.c) or to normalize casing of directory names (read-cache.c). Getting rid of cache_entry.dir_next reduces memory consumption, especially with core.ignorecase=false (which doesn't use that member at all). With core.ignorecase=true, building the directory index is slightly faster as we add / check the parent directory first (instead of going through all directory levels for each file in the index). E.g. with WebKit (~200k files, ~7k dirs), time spent in lazy_init_name_hash is reduced from 176ms to 130ms. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-02-28 00:57:48 +01:00
}