Closed
Bug 31912
Opened 24 years ago
Closed 23 years ago
Leaking mEnv_HandlePool
Categories
(MailNews Core :: Database, defect, P2)
MailNews Core
Database
Tracking
(Not tracked)
VERIFIED
FIXED
Future
People
(Reporter: beard, Assigned: Bienvenu)
References
()
Details
(Keywords: memory-leak)
Attachments
(1 file)
533 bytes,
patch
|
Details | Diff | Splinter Review |
Leak detector indicates that morkEnv::mEnv_HandlePool is leaking. The URL above shows the point of allocation that leaks.
Keywords: mlk
Comment 6•24 years ago
|
||
Can we afford for this to remain futured? If nothing else, apart from the mail/news usage, mork does get used in global history, impacting the rest of the product.
Comment 8•24 years ago
|
||
David, How big of a leak is this? Will the whole db or folder get leaked because of this?
Assignee | ||
Comment 9•24 years ago
|
||
it's a very small leak.
Comment 10•24 years ago
|
||
marking nsbeta1- due to the fact that it's a small leak.
Keywords: nsbeta1-
Comment 11•23 years ago
|
||
Comment 12•23 years ago
|
||
I'll let bienvenu do the review on this, since it is mork. (I'll do the sr=) the patch seems ok to me. hwaara, how did you test it?
Comment 13•23 years ago
|
||
I compiled with the change and then ran mailnews for a good while. I didn't really know how to test, but I deleted and moved lots of mails... bienvenu?
Assignee | ||
Comment 14•23 years ago
|
||
No, I'm not sure this is right. mEnvHandles come from a pool, and are returned to the pool when CloseEnv is called. So just out and out deleting them is not the way this was designed to work, as far as I can tell.
Comment 15•23 years ago
|
||
So why is this bug filed in the first place if it ain't leaking? If it is leaking, then do you have any suggestions on approaches to fix this?
Assignee | ||
Comment 16•23 years ago
|
||
I meant that your fix was wrong, not that there wasn't a leak. It is, however, a tiny, tiny, leak. We have bigger fish to fry. For one thing, CloseEnv sets mEnv_HandlePool to null, so your fix will not do anything - mEnv_HandlePool will always be null by the time the environment destructor gets called. If I knew what the correct fix was, I would fix it. This is *not* code that I wrote; the person who wrote this code is long gone. So it requires investigation. Sorry I can't be more helpful.
Comment 17•23 years ago
|
||
Bienvenu, isn't it as easy as replacing the set-to-null code with a nullcheck and then delete?
Assignee | ||
Comment 18•23 years ago
|
||
no, we pool the handles. This means we allocate them from a pool, and return them to a pool when done. At least, that's the idea.
Comment 19•23 years ago
|
||
then I guess this bug will have to wait some more for a fix :(
QA Contact: lchiang → stephend
Summary: [MLK] Leaking mEnv_HandlePool → Leaking mEnv_HandlePool
Assignee | ||
Comment 20•23 years ago
|
||
plussing; will be fixed when mork changes land.
Keywords: nsbeta1+
Assignee | ||
Comment 21•23 years ago
|
||
should be fixed now.
Status: ASSIGNED → RESOLVED
Closed: 23 years ago
Resolution: --- → FIXED
After David's big checkin, I no longer see this when launching mail, reading an IMAP message, loading a folder or deleting messages. Verified FIXED, using Purify.
Status: RESOLVED → VERIFIED
Updated•20 years ago
|
Product: MailNews → Core
Updated•16 years ago
|
Product: Core → MailNews Core
You need to log in
before you can comment on or make changes to this bug.
Description
•