This replaces the "loop over all streams" with a simple hash lookup. It
makes the cost of checking for already tokenized streams basically go
away (it could be up to 5% of CPU time, almost entirely due to the
"strcmp()" of the name).
Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
---
That "up to 5%" is probably debatable, and will depend on just how many
includes you have etc etc. And on the CPU and library issues. But 3-4% is
definitely the case for my kernel C=2 build on my current machine.
So it's real, and worth it.
pre-process.c | 8 +++++---
1 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/pre-process.c b/pre-process.c
index 603cc00c999f..6d12f173146d 100644
--- a/pre-process.c
+++ b/pre-process.c
@@ -655,10 +655,12 @@ static const char *token_name_sequence(struct token *token, int endop, struct to
static int already_tokenized(const char *path)
{
- int i;
- struct stream *s = input_streams;
+ int stream, next;
+
+ for (stream = *hash_stream(path); stream >= 0 ; stream = next) {
+ struct stream *s = input_streams + stream;
- for (i = input_stream_nr; --i >= 0; s++) {
+ next = s->next_stream;
if (s->constant != CONSTANT_FILE_YES)
continue;
if (strcmp(path, s->name))
--
1.7.5.rc1.1.gc3f61.dirty
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
