-
Notifications
You must be signed in to change notification settings - Fork 29.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
src: fix mismatched delete[] in src/node_file.cc #1092
Conversation
No open issues about it yet. |
inline const char* dest() const { return dest_; } | ||
inline unsigned int dest_len() const { return dest_len_; } | ||
inline void dest_len(unsigned int dest_len) { dest_len_ = dest_len; } | ||
void* operator new(size_t size) = delete; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How does this work?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fix a bad delete of a pointer that was allocated with placement new. Casting the pointer was not the right solution because there was at least one non-placement new constructor call. This commit rewrites FSReqWrap to be more explicit about ownership of the auxiliary data and removes a number of egregious const_casts. The ASYNC_DEST_CALL macro also gets significantly slimmed down. PR-URL: nodejs#1092 Reviewed-By: Fedor Indutny <fedor@indutny.com>
The SYNC_CALL macro returns on error, bypassing the delete[] call. Mea culpa, it looks like I introduced this memory leak back in 2013, in commit d2b80b8 ("src: clean up FSReqWrap"). PR-URL: nodejs#1092 Reviewed-By: Fedor Indutny <fedor@indutny.com>
LGTM |
@indutny I pushed another commit to fix a memory leak in the fs.writeSync() error path. Can you PTAL? |
Those SYNC_CALL and ASYNC_CALL macros are a special kind of evil. One of the first rules of C macros is that you don't push control flow into them and what do they do? Exactly. |
return args.GetReturnValue().Set(SYNC_RESULT); | ||
} | ||
|
||
FSReqWrap* req_wrap = | ||
new FSReqWrap(env, req.As<Object>(), "write", must_free ? buf : nullptr); | ||
FSReqWrap::New(env, req.As<Object>(), "write", buf, ownership); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So you pas buf
here, where does it die?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Anyway it seems that the buf
should be deallocated somewhere.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's freed by FSReqWrap's destructor because ownership == MOVE
.
One comment, otherwise LGTM |
FSReqWrap* that; | ||
char* const storage = new char[sizeof(*that) + size]; | ||
that = new(storage) FSReqWrap(env, req, syscall, data); | ||
if (copy) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh, I guess it becomes parts of FSReqWrap?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Exactly. :-)
6f9e503
to
528d878
Compare
Fix a bad delete of a pointer that was allocated with placement new.
Casting the pointer was not the right solution because there was at
least one non-placement new constructor call.
This commit rewrites FSReqWrap to be more explicit about ownership of
the auxiliary data and removes a number of egregious const_casts.
The ASYNC_DEST_CALL macro also gets significantly slimmed down.
R=@indutny
Is there a bug number I should reference in the commit log?
https://jenkins-iojs.nodesource.com/view/iojs/job/iojs+any-pr+multi/257/