NetworkManager/shared/nm-glib-aux/nm-ref-string.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

196 lines
5.3 KiB
C
Raw Permalink Normal View History

/* SPDX-License-Identifier: LGPL-2.1+ */
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
#include "nm-default.h"
#include "nm-ref-string.h"
/*****************************************************************************/
typedef struct {
NMRefString r;
volatile int ref_count;
char str_data[];
} RefString;
G_LOCK_DEFINE_STATIC(gl_lock);
static GHashTable *gl_hash;
/* the first field of NMRefString is a pointer to the NUL terminated string.
* This also allows to compare strings with nm_pstr_equal(), although, pointer
* equality might be better. */
G_STATIC_ASSERT(G_STRUCT_OFFSET(NMRefString, str) == 0);
G_STATIC_ASSERT(G_STRUCT_OFFSET(RefString, r) == 0);
G_STATIC_ASSERT(G_STRUCT_OFFSET(RefString, r.str) == 0);
/*****************************************************************************/
static guint
_ref_string_hash(gconstpointer ptr)
{
const RefString *a = ptr;
NMHashState h;
nm_hash_init(&h, 1463435489u);
nm_hash_update(&h, a->r.str, a->r.len);
return nm_hash_complete(&h);
}
static gboolean
_ref_string_equal(gconstpointer pa, gconstpointer pb)
{
const RefString *a = pa;
const RefString *b = pb;
return a->r.len == b->r.len && memcmp(a->r.str, b->r.str, a->r.len) == 0;
}
/*****************************************************************************/
static void
_ASSERT(const RefString *rstr0)
{
#if NM_MORE_ASSERTS
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
int r;
nm_assert(rstr0);
G_LOCK(gl_lock);
r = g_atomic_int_get(&rstr0->ref_count);
nm_assert(r > 0);
nm_assert(r < G_MAXINT);
nm_assert(rstr0 == g_hash_table_lookup(gl_hash, rstr0));
G_UNLOCK(gl_lock);
#endif
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
}
/**
* nm_ref_string_new_len:
* @cstr: the string to intern. Must contain @len bytes.
* If @len is zero, @cstr may be %NULL. Note that it is
* acceptable that the string contains a NUL character
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
* within the first @len bytes. That is, the string is
* not treated as a NUL terminated string, but as binary.
* Also, contrary to strncpy(), this will read all the
* first @len bytes. It won't stop at the first NUL.
* @len: the length of the string (usually there is no NUL character
* within the first @len bytes, but that would be acceptable as well
* to add binary data).
*
* Note that the resulting NMRefString instance will always be NUL terminated
* (at position @len).
*
* Note that NMRefString are always interned/deduplicated. If such a string
* already exists, the existing instance will be referred and returned.
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
*
*
* Since all NMRefString are shared and interned, you may use
* pointer equality to compare them. Note that if a NMRefString contains
* a NUL character (meaning, if
*
* strlen (nm_ref_string_get_str (str)) != nm_ref_string_get_len (str)
*
* ), then pointer in-equality does not mean that the NUL terminated strings
* are also unequal. In other words, for strings that contain NUL characters,
*
* if (str1 != str2)
* assert (!nm_streq0 (nm_ref_string_get_str (str1), nm_ref_string_get_str (str2)));
*
* might not hold!
*
*
* NMRefString is thread-safe.
*
* Returns: (transfer full): the interned string. This is
* never %NULL, but note that %NULL is also a valid NMRefString.
* The result must be unrefed with nm_ref_string_unref().
*/
NMRefString *
nm_ref_string_new_len(const char *cstr, gsize len)
{
RefString *rstr0;
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
G_LOCK(gl_lock);
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
if (G_UNLIKELY(!gl_hash)) {
gl_hash = g_hash_table_new_full(_ref_string_hash, _ref_string_equal, g_free, NULL);
rstr0 = NULL;
} else {
NMRefString rr_lookup = {
.len = len,
.str = cstr,
};
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
rstr0 = g_hash_table_lookup(gl_hash, &rr_lookup);
}
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
if (rstr0) {
nm_assert(({
int r = g_atomic_int_get(&rstr0->ref_count);
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
(r >= 0 && r < G_MAXINT);
}));
g_atomic_int_inc(&rstr0->ref_count);
} else {
rstr0 = g_malloc(sizeof(RefString) + 1 + len);
rstr0->ref_count = 1;
*((gsize *) &rstr0->r.len) = len;
*((const char **) &rstr0->r.str) = rstr0->str_data;
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
if (len > 0)
memcpy(rstr0->str_data, cstr, len);
rstr0->str_data[len] = '\0';
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
if (!g_hash_table_add(gl_hash, rstr0))
nm_assert_not_reached();
}
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
G_UNLOCK(gl_lock);
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
return &rstr0->r;
}
NMRefString *
nm_ref_string_ref(NMRefString *rstr)
{
RefString *const rstr0 = (RefString *) rstr;
if (!rstr)
return NULL;
_ASSERT(rstr0);
g_atomic_int_inc(&rstr0->ref_count);
return &rstr0->r;
}
void
_nm_ref_string_unref_non_null(NMRefString *rstr)
{
RefString *const rstr0 = (RefString *) rstr;
int r;
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
_ASSERT(rstr0);
/* fast-path: first try to decrement the ref-count without bringing it
* to zero. */
r = rstr0->ref_count;
if (G_LIKELY(r > 1 && g_atomic_int_compare_and_exchange(&rstr0->ref_count, r, r - 1)))
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
return;
/* We apparently are about to return the last reference. Take a lock. */
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
G_LOCK(gl_lock);
nm_assert(g_hash_table_lookup(gl_hash, rstr0) == rstr0);
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
if (G_LIKELY(g_atomic_int_dec_and_test(&rstr0->ref_count))) {
if (!g_hash_table_remove(gl_hash, rstr0))
nm_assert_not_reached();
}
shared: add NMRefString I'd like to refactor libnm's caching. Note that cached D-Bus objects have repeated strings all over the place. For example every object will have a set of D-Bus interfaces (strings) and properties (strings) and an object path (which is referenced by other objects). We can save a lot of redundant strings by deduplicating/interning them. Also, by interning them, we can compare them using pointer equality. Add a NMRefString implementation for this. Maybe an alternative name would be NMInternedString or NMDedupString, because this string gets always interned. There is no way to create a NMRefString that is not interned. Still, NMRefString name sounds better. It is ref-counted after all. Notes: - glib has GQuark and g_intern_string(). However, such strings cannot be unrefered and are leaked indefinitely. It is thus unsuited for anything but a fixed set of well-known strings. - glib 2.58 adds GRefString, but we cannot use that because we currently still use glib 2.40. There are some differences: - GRefString is just a typedef to char. That means, the glib API exposes GRefString like regular character strings. NMRefString intentionally does that not. This makes it slightly less convenient to pass it to API that expects "const char *". But it makes it clear to the reader, that an instance is in fact a NMRefString, which means it indicates that the string is interned and can be referenced without additional copy. - GRefString can be optionally interned. That means you can only use pointer equality for comparing values if you know that the GRefString was created with g_ref_string_new_intern(). So, GRefString looks like a "const char *" pointer and even if you know it's a GRefString, you might not know whether it is interned. NMRefString is always interned, and you can always compare it using pointer equality. - In the past I already proposed a different implementation for a ref-string. That made different choices. For example NMRefString then was a typedef to "const char *", it did not support interning but deduplication (without a global cache), ref/unref was not thread safe (but then there was no global cache so that two threads could still use the API independently). The point is, there are various choices to make. GRefString, the previous NMRefString implementation and the one here, all have pros and cons. I think for the purpose where I intend NMRefString (dedup and efficient comparison), it is a preferable implementation. Ah, and of course NMRefString is an immutable string, which is a nice property.
2019-09-02 07:54:28 +02:00
G_UNLOCK(gl_lock);
}
/*****************************************************************************/