On Fri, May 11, 2012 at 9:10 AM, Daniel Veillard <[email protected]> wrote:
> Hi Conrad,
>
> that's interesting ! I was initially afraid of a sudden explosion of
> memory allocations for building a tree since by default buffers tend to
> "waste" memory by using doubling allocations, but that's not the case.
> xmllint --noout doc/libxml2-api.xml
> when compiled with memory debug produce
>
> paphio:~/XML -> cat .memdump
> MEMORY ALLOCATED : 0, MAX was 12756699
>
> and without your patch 12755657, i.e. the increase is minimal.
Heh, I thought that too. Actually you're looking at the result with
XML_ALLOC_EXACT! This
is because EXACT adds 10bytes "spare" on each alloc, and that interestingly
wastes about the
same amount of space as XML_ALLOC_DOUBLEIT on this example (see below).
So it turns out that the default realloc() on my system actually handles this
case really
well — and I guess that all the time in xmlRealloc() was actually in xmlStrlen,
not the
underlying realloc() after all (sorry for misleading you). If you replace the
realloc()
with a bad one (like valgrind's), then the performance degrades severely.
This patch implements a HYBRID allocator which has the behaviour you describe
(it's
like EXACT to start with, though without the spare 10 bytes; and switches to
DOUBLEIT
after 4kb) — that gets the memory back down to 12755657, with no noticeable
impact on the
performance of the synthetic pathological example under valgrind.
In summary:
max_memory on ./xmllint --noout doc/libxml2-api.xml,
valgrind time on https://gist.github.com/2656940
max_memory valgrind time
before | 12755657 | 29:18.2
EXACT | 12756699 | 2:58.6 <-- this is the state after the first patch.
DOUBLEIT | 12756727 | 0:02.7
HYBRID | 12755754 | 0:02.7 <-- this is the state with both patches.
>
> There is also the cost of creating the buffers all the time.
> I need to read the code and check but I may be interested in an hybrid
> approach where we switch to buffer only when the text node starts to
> become too big (4k would remove nearly all usuall types of "document"
> usage, i.e. not blocks of data)
I tried to avoid too much buffer creation by introducing the xmlBufferDetach
function,
which allows re-using one buffer to construct many strings. It's maybe a bit of
a "hack"
in API terms though I thought the gains would be worth it.
Conrad
------8<------
To keep memory usage tight in normal conditions it's desirable to only
allocate as much space as is needed. Unfortunately this can lead to
problems when constructing a long string out of small chunks, because
every chunk you add will need to resize the buffer.
To fix this XML_ALLOC_HYBRID will switch (when the buffer is 4kb big)
from using exact allocations to doubling buffer size every time it is
full. This limits the number of buffer resizes to O(log n) (down from
O(n)), and thus greatly increases the performance of constructing very
large strings in this manner.
---
include/libxml/tree.h | 3 ++-
tree.c | 24 +++++++++++++++++++++++-
2 files changed, 25 insertions(+), 2 deletions(-)
diff --git a/include/libxml/tree.h b/include/libxml/tree.h
index b522f61..2196f8d 100644
--- a/include/libxml/tree.h
+++ b/include/libxml/tree.h
@@ -74,7 +74,8 @@ typedef enum {
XML_BUFFER_ALLOC_DOUBLEIT, /* double each time one need to grow */
XML_BUFFER_ALLOC_EXACT, /* grow only to the minimal size */
XML_BUFFER_ALLOC_IMMUTABLE, /* immutable buffer */
- XML_BUFFER_ALLOC_IO /* special allocation scheme used for
I/O */
+ XML_BUFFER_ALLOC_IO, /* special allocation scheme used for I/O */
+ XML_BUFFER_ALLOC_HYBRID /* exact up to a threshold, and doubleit
thereafter */
} xmlBufferAllocationScheme;
/**
diff --git a/tree.c b/tree.c
index d8fafc0..47b906c 100644
--- a/tree.c
+++ b/tree.c
@@ -682,7 +682,8 @@ try_complex:
void
xmlSetBufferAllocationScheme(xmlBufferAllocationScheme scheme) {
if ((scheme == XML_BUFFER_ALLOC_EXACT) ||
- (scheme == XML_BUFFER_ALLOC_DOUBLEIT))
+ (scheme == XML_BUFFER_ALLOC_DOUBLEIT) ||
+ (scheme == XML_BUFFER_ALLOC_HYBRID))
xmlBufferAllocScheme = scheme;
}
@@ -693,6 +694,9 @@ xmlSetBufferAllocationScheme(xmlBufferAllocationScheme
scheme) {
* XML_BUFFER_ALLOC_EXACT - use exact sizes, keeps memory usage down
* XML_BUFFER_ALLOC_DOUBLEIT - double buffer when extra needed,
* improves performance
+ * XML_BUFFER_ALLOC_HYBRID - use exact sizes on small strings to keep memory
usage tight
+ * in normal usage, and doubleit on large strings
to avoid
+ * pathological performance.
*
* Returns the current allocation scheme
*/
@@ -1267,6 +1271,7 @@ xmlStringLenGetNodeList(xmlDocPtr doc, const xmlChar
*value, int len) {
buffer = xmlBufferCreateSize(0);
if (buffer == NULL) return(NULL);
+ xmlBufferSetAllocationScheme(buffer, XML_BUFFER_ALLOC_HYBRID);
q = cur;
while ((cur < end) && (*cur != 0)) {
@@ -1474,6 +1479,7 @@ xmlStringGetNodeList(xmlDocPtr doc, const xmlChar *value)
{
buffer = xmlBufferCreateSize(0);
if (buffer == NULL) return(NULL);
+ xmlBufferSetAllocationScheme(buffer, XML_BUFFER_ALLOC_HYBRID);
q = cur;
while (*cur != 0) {
@@ -7000,6 +7006,7 @@ xmlBufferSetAllocationScheme(xmlBufferPtr buf,
(buf->alloc == XML_BUFFER_ALLOC_IO)) return;
if ((scheme == XML_BUFFER_ALLOC_DOUBLEIT) ||
(scheme == XML_BUFFER_ALLOC_EXACT) ||
+ (scheme == XML_BUFFER_ALLOC_HYBRID) ||
(scheme == XML_BUFFER_ALLOC_IMMUTABLE))
buf->alloc = scheme;
}
@@ -7267,6 +7274,21 @@ xmlBufferResize(xmlBufferPtr buf, unsigned int size)
case XML_BUFFER_ALLOC_EXACT:
newSize = size+10;
break;
+ case XML_BUFFER_ALLOC_HYBRID:
+ if (buf->use < BASE_BUFFER_SIZE)
+ newSize = size;
+ else {
+ newSize = buf->size * 2;
+ while (size > newSize) {
+ if (newSize > UINT_MAX / 2) {
+ xmlTreeErrMemory("growing buffer");
+ return 0;
+ }
+ newSize *= 2;
+ }
+ }
+ break;
+
default:
newSize = size+10;
break;
--
1.7.10.207.g0bb2e
_______________________________________________
xml mailing list, project page http://xmlsoft.org/
[email protected]
http://mail.gnome.org/mailman/listinfo/xml
