On Sat, Jun 23, 2012 at 02:53:21PM +0200, Peter Bex wrote:
> > User programs can override the finalizer limit using the "-:f" runtime
> > option if a large number of finalizations are expected or required in
> > specific applications.
>
> I'd like a little more time to investigate the bug. This patch can
> always be applied. It would be a shame if we can come up with a better
> fix after this makes it into the release.
Here's a patch that made one test function correctly. The test is my
own simplified version of Megane's test:
(let lp ((x (list 1 2 3)))
(set-finalizer! x (lambda _ #f))
(lp (list 1 2 3)))
This test used to fail with something like this:
Warning: in finalizer: call of non-procedure: #<unspecified>
Call history:
/tmp/test.scm:3: lp
/tmp/test.scm:2: set-finalizer!
[...]
[panic] out of memory - heap full while resizing - execution terminated
After applying the patch, it just loops endlessly (which is the correct,
expected behavior) without any errors.
About the patch:
##sys#fudge 26 returns the number of live finalizers, which (I think)
should be identical to the value of the first slot in the
##sys#pending-finalizers. In any case, the check in set-finalizer!
compares if fudge 26 is LARGER than the maximum number of pending
finalizers. If that's the case, we're actually already storing one
finalizer more than the vector can take.
Originally I overlooked it because (fx+ (fx* 2 _max_pending_finalizers) 1)
looks like it can hold one more finalizer than _max_pending_finalizers,
but then I figured that the vector is twice the size of
_max_pending_finalizers because it holds pairs of values and their
associated finalizer procedure. The +1 is for the extra slot that keeps
track of the number of finalizers.
Question: Why the duplication with live_finalizer_count? Are these
two values subtly different somehow?
Anyway, the attached patch changes fx> to fx>= which makes the error go
away with my test case. Unfortunately, the tests/finalizer-error-test
still fails. I have some hopes to get that fixed too.
Cheers,
Peter
--
http://sjamaan.ath.cx
--
"The process of preparing programs for a digital computer
is especially attractive, not only because it can be economically
and scientifically rewarding, but also because it can be an aesthetic
experience much like composing poetry or music."
-- Donald Knuth
>From 934ba6a2d16e840c08baec39855ed393104c2140 Mon Sep 17 00:00:00 2001
From: Peter Bex <[email protected]>
Date: Sat, 23 Jun 2012 19:33:11 +0200
Subject: [PATCH] Fix finalizer size check: if it's bigger, we're already
using one slot beyond the vector
---
library.scm | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/library.scm b/library.scm
index 90d22c6..f9142c3 100644
--- a/library.scm
+++ b/library.scm
@@ -4573,7 +4573,7 @@ EOF
(define set-finalizer!
(lambda (x y)
- (when (fx> (##sys#fudge 26) _max_pending_finalizers)
+ (when (fx>= (##sys#fudge 26) _max_pending_finalizers)
(if (##core#inline "C_resize_pending_finalizers" (fx* 2
_max_pending_finalizers))
(begin
(set! ##sys#pending-finalizers (##sys#grow-vector
##sys#pending-finalizers
--
1.7.9.1
_______________________________________________
Chicken-hackers mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/chicken-hackers
