Rework image handling for on demand decode

Introduce a new 'fz_image' type; this type contains rudimentary
information about images (such as native, size, colorspace etc)
and a function to call to get a pixmap of that image (with a
size hint).

Instead of passing pixmaps through the device interface (and
holding pixmaps in the display list) we now pass images instead.

The rendering routines therefore call fz_image_to_pixmap to get
pixmaps to render, and fz_pixmap_drop those afterwards.

The file format handling routines therefore need to produce
images rather than pixmaps; xps and cbz currently just wrap
pixmaps as images. PDF is more involved.

The stream handling routines in PDF have been altered so that
they can recognise when the last stream entry in a filter
dictionary is an image decoding filter. Rather than applying
this filter, they read and store the parameters into a
pdf_image_params structure, and stop decoding at that point.

This allows us to read the compressed data for an image into
memory as a block. We can then restart the image decode process
later.

pdf_images therefore consist of the compressed image data for
images. When a pixmap is requested for such an image, the code
checks to see if we have one (of an appropriate size), and if
not, decodes it.

The size hint is used to determine whether it is possible to
subsample the image; currently this is only supported for
JPEGs, but we could add generic subsampling code later.

In order to handle caching the produced images, various changes
have been made to the store and the underlying hash table.

Previously the store was indexed purely by fz_obj keys; we don't
have an fz_obj key any more, so have extended the store by adding
a concept of a key 'type'. A key type is a pointer to a set of
functions that keep/drop/compare and make a hashable key from
a key pointer.

We make a pdf_store.c file that contains functions to offer the
existing fz_obj based functions, and add a new 'type' for keys
(based on the fz_image handle, and the subsample factor) in the
pdf_image.c file.

While working on this, a problem became apparent in the existing
store codel; fz_obj objects had no protection on their reference
counts, hence an interpreter thread could try to alter a ref count
at the same time as a malloc caused an eviction from the store.
This has been solved by using the alloc lock as protection. This in
turn requires some tweaks to the code to make sure we don't try
and keep/drop fz_obj's from the store code while the alloc lock is
held.

A side effect of this work is that when a hash table is created, we
inform it what lock should be used to protect its innards (if any).
If the alloc lock is used, the insert method knows to drop/retake it
to allow it to safely expand the hash table. Callers to the hash
functions have the responsibility of taking/dropping the appropriate
lock, and ensuring that they cope with the possibility that insert
might drop the alloc lock, causing race conditions.

1 files changed, 62 insertions, 68 deletions

diff --git a/fitz/res_store.c b/fitz/res_store.c
index 5eec1f0d..461ea7c8 100644
--- a/fitz/res_store.c
+++ b/fitz/res_store.c
@@ -1,21 +1,14 @@
 #include "fitz.h"
-#include "mupdf.h"
 
 struct fz_item_s
 {
-	fz_obj *key;
+	void *key;
 	fz_storable *val;
 	unsigned int size;
 	fz_item *next;
 	fz_item *prev;
 	fz_store *store;
-};
-
-struct refkey
-{
-	fz_store_free_fn *free;
-	int num;
-	int gen;
+	fz_store_type *type;
 };
 
 struct fz_store_s
@@ -43,7 +36,7 @@ fz_new_store_context(fz_context *ctx, unsigned int max)
 	store = fz_malloc_struct(ctx, fz_store);
 	fz_try(ctx)
 	{
-		store->hash = fz_new_hash_table(ctx, 4096, sizeof(struct refkey));
+		store->hash = fz_new_hash_table(ctx, 4096, sizeof(fz_store_hash), FZ_LOCK_ALLOC);
 	}
 	fz_catch(ctx)
 	{
@@ -114,20 +107,19 @@ evict(fz_context *ctx, fz_item *item)
 		store->head = item->next;
 	/* Drop a reference to the value (freeing if required) */
 	drop = (item->val->refs > 0 && --item->val->refs == 0);
-	fz_unlock(ctx, FZ_LOCK_ALLOC);
 	/* Remove from the hash table */
-	if (fz_is_indirect(item->key))
+	if (item->type->make_hash_key)
 	{
-		struct refkey refkey;
-		refkey.free = item->val->free;
-		refkey.num = fz_to_num(item->key);
-		refkey.gen = fz_to_gen(item->key);
-		fz_hash_remove(ctx, store->hash, &refkey);
+		fz_store_hash hash = { NULL };
+		hash.free = item->val->free;
+		if (item->type->make_hash_key(&hash, item->key))
+			fz_hash_remove(ctx, store->hash, &hash);
 	}
+	fz_unlock(ctx, FZ_LOCK_ALLOC);
 	if (drop)
 		item->val->free(ctx, item->val);
 	/* Always drops the key and free the item */
-	fz_drop_obj(item->key);
+	item->type->drop_key(ctx, item->key);
 	fz_free(ctx, item);
 	fz_lock(ctx, FZ_LOCK_ALLOC);
 }
@@ -194,30 +186,21 @@ ensure_space(fz_context *ctx, unsigned int tofree)
 	return count;
 }
 
-void
-fz_store_item(fz_context *ctx, fz_obj *key, void *val_, unsigned int itemsize)
+void *
+fz_store_item(fz_context *ctx, void *key, void *val_, unsigned int itemsize, fz_store_type *type)
 {
 	fz_item *item = NULL;
 	unsigned int size;
 	fz_storable *val = (fz_storable *)val_;
 	fz_store *store = ctx->store;
-	int indirect;
-	struct refkey refkey;
+	fz_store_hash hash = { NULL };
+	int use_hash = 0;
 
 	if (!store)
-		return;
+		return NULL;
 
 	fz_var(item);
 
-	/* Form the key before we take the lock */ 
-	indirect = fz_is_indirect(key);
-	if (indirect)
-	{
-		refkey.free = val->free;
-		refkey.num = fz_to_num(key);
-		refkey.gen = fz_to_gen(key);
-	}
-
 	/* If we fail for any reason, we swallow the exception and continue.
 	 * All that the above program will see is that we failed to store
 	 * the item. */
@@ -227,9 +210,16 @@ fz_store_item(fz_context *ctx, fz_obj *key, void *val_, unsigned int itemsize)
 	}
 	fz_catch(ctx)
 	{
-		return;
+		return NULL;
 	}
 
+	if (type->make_hash_key)
+	{
+		hash.free = val->free;
+		use_hash = type->make_hash_key(&hash, key);
+	}
+
+	type->keep_key(ctx, key);
 	fz_lock(ctx, FZ_LOCK_ALLOC);
 	if (store->max != FZ_STORE_UNLIMITED)
 	{
@@ -242,34 +232,41 @@ fz_store_item(fz_context *ctx, fz_obj *key, void *val_, unsigned int itemsize)
 				/* Failed to free any space */
 				fz_unlock(ctx, FZ_LOCK_ALLOC);
 				fz_free(ctx, item);
-				return;
+				type->drop_key(ctx, key);
+				return NULL;
 			}
 		}
 	}
 	store->size += itemsize;
 
-	item->key = fz_keep_obj(key);
+	item->key = key;
 	item->val = val;
 	item->size = itemsize;
 	item->next = NULL;
+	item->type = type;
 
 	/* If we can index it fast, put it into the hash table */
-	if (indirect)
+	if (use_hash)
 	{
-		fz_unlock(ctx, FZ_LOCK_ALLOC);
+		fz_pixmap *existing;
+
 		fz_try(ctx)
 		{
-			fz_hash_insert(ctx, store->hash, &refkey, item);
+			/* May drop and retake the lock */
+			existing = fz_hash_insert(ctx, store->hash, &hash, item);
 		}
 		fz_catch(ctx)
 		{
-			fz_free(ctx, item);
-			fz_lock(ctx, FZ_LOCK_ALLOC);
 			store->size -= itemsize;
 			fz_unlock(ctx, FZ_LOCK_ALLOC);
-			return;
+			fz_free(ctx, item);
+			return NULL;
+		}
+		if (existing)
+		{
+			fz_unlock(ctx, FZ_LOCK_ALLOC);
+			return existing;
 		}
-		fz_lock(ctx, FZ_LOCK_ALLOC);
 	}
 	/* Now we can never fail, bump the ref */
 	if (val->refs > 0)
@@ -283,15 +280,17 @@ fz_store_item(fz_context *ctx, fz_obj *key, void *val_, unsigned int itemsize)
 	store->head = item;
 	item->prev = NULL;
 	fz_unlock(ctx, FZ_LOCK_ALLOC);
+
+	return NULL;
 }
 
 void *
-fz_find_item(fz_context *ctx, fz_store_free_fn *free, fz_obj *key)
+fz_find_item(fz_context *ctx, fz_store_free_fn *free, void *key, fz_store_type *type)
 {
-	struct refkey refkey;
 	fz_item *item;
 	fz_store *store = ctx->store;
-	int indirect;
+	fz_store_hash hash = { NULL };
+	int use_hash = 0;
 
 	if (!store)
 		return NULL;
@@ -299,27 +298,24 @@ fz_find_item(fz_context *ctx, fz_store_free_fn *free, fz_obj *key)
 	if (!key)
 		return NULL;
 
-	/* Form the key before we take the lock */
-	indirect = fz_is_indirect(key);
-	if (indirect)
+	if (type->make_hash_key)
 	{
-		refkey.free = free;
-		refkey.num = fz_to_num(key);
-		refkey.gen = fz_to_gen(key);
+		hash.free = free;
+		use_hash = type->make_hash_key(&hash, key);
 	}
 
 	fz_lock(ctx, FZ_LOCK_ALLOC);
-	if (indirect)
+	if (use_hash)
 	{
 		/* We can find objects keyed on indirected objects quickly */
-		item = fz_hash_find(ctx, store->hash, &refkey);
+		item = fz_hash_find(ctx, store->hash, &hash);
 	}
 	else
 	{
 		/* Others we have to hunt for slowly */
 		for (item = store->head; item; item = item->next)
 		{
-			if (item->val->free == free && !fz_objcmp(item->key, key))
+			if (item->val->free == free && !type->cmp_key(item->key, key))
 				break;
 		}
 	}
@@ -355,35 +351,33 @@ fz_find_item(fz_context *ctx, fz_store_free_fn *free, fz_obj *key)
 }
 
 void
-fz_remove_item(fz_context *ctx, fz_store_free_fn *free, fz_obj *key)
+fz_remove_item(fz_context *ctx, fz_store_free_fn *free, void *key, fz_store_type *type)
 {
-	struct refkey refkey;
 	fz_item *item;
 	fz_store *store = ctx->store;
-	int drop, indirect;
+	int drop;
+	fz_store_hash hash;
+	int use_hash = 0;
 
-	/* Form the key before we take the lock */
-	indirect = fz_is_indirect(key);
-	if (indirect)
+	if (type->make_hash_key)
 	{
-		refkey.free = free;
-		refkey.num = fz_to_num(key);
-		refkey.gen = fz_to_gen(key);
+		hash.free = free;
+		use_hash = type->make_hash_key(&hash, key);
 	}
 
 	fz_lock(ctx, FZ_LOCK_ALLOC);
-	if (indirect)
+	if (use_hash)
 	{
 		/* We can find objects keyed on indirect objects quickly */
-		item = fz_hash_find(ctx, store->hash, &refkey);
+		item = fz_hash_find(ctx, store->hash, &hash);
 		if (item)
-			fz_hash_remove(ctx, store->hash, &refkey);
+			fz_hash_remove(ctx, store->hash, &hash);
 	}
 	else
 	{
 		/* Others we have to hunt for slowly */
 		for (item = store->head; item; item = item->next)
-			if (item->val->free == free && !fz_objcmp(item->key, key))
+			if (item->val->free == free && !type->cmp_key(item->key, key))
 				break;
 	}
 	if (item)
@@ -400,7 +394,7 @@ fz_remove_item(fz_context *ctx, fz_store_free_fn *free, fz_obj *key)
 		fz_unlock(ctx, FZ_LOCK_ALLOC);
 		if (drop)
 			item->val->free(ctx, item->val);
-		fz_drop_obj(item->key);
+		type->drop_key(ctx, item->key);
 		fz_free(ctx, item);
 	}
 	else