Bug 63313

Summary:

The return value of SharedBuffer::createWithContentsOfFile must have valid m_size.

Product:

WebKit

Reporter:

Kim, kyusun <maniagoon>

Component:

Platform

Assignee:

Nobody <webkit-unassigned>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

andersca, darin, eric, skyul, webkit.review.bot

Priority:

Version:

528+ (Nightly build)

Hardware:

Unspecified

OS:

Linux

Attachments:

Description	Flags
Patch	none
Patch	none

Description Kim, kyusun 2011-06-23 22:48:57 PDT

SharedBufferPOSIX.cpp - SharedBuffer::m_size has to be set

Comment 1 Kim, kyusun 2011-06-23 22:54:03 PDT

Created attachment 98460 [details]
Patch

Comment 2 Eric Seidel (no email) 2011-07-05 17:57:35 PDT

Comment on attachment 98460 [details]
Patch

How do we test this?  Is there a public function we should call instead of setting m_size directly?

Comment 3 Eric Seidel (no email) 2011-07-05 18:01:24 PDT

Comment on attachment 98460 [details]
Patch

In general this looks fine though.

Comment 4 Eric Seidel (no email) 2011-07-05 18:01:50 PDT

How did you detect that this was broken?

Comment 5 Eric Seidel (no email) 2011-07-05 18:04:51 PDT

CCing anders who originally created the SharedBuffer class.  It's strange to me that m_size is not used by most ports.

Comment 6 Eric Seidel (no email) 2011-07-05 18:06:10 PDT

Comment on attachment 98460 [details]
Patch

Clearing cq+.

It seems to me it would be better for this code to read into a Vector and then create the SharedBuffer from the Vector.  SharedBuffer doesn't seem to be designed to be read into directly, it's meant to wrap other data storage mechanisms.  Since this POSIX implementation doesn't have any "native" data storage, seems we should use Vector.

Comment 7 Eric Seidel (no email) 2011-07-05 18:07:39 PDT

We already have a adoptVector(Vector) which gets all this right, including setting m_size.  I worry that there are other things that might be missing from this implementation in the future.  re-using the adoptVector(Vector) method may be more future-proof.

Comment 8 Kim, kyusun 2011-07-05 22:13:39 PDT

(In reply to comment #2)
> (From update of attachment 98460 [details])
> How do we test this?

Currently, there is no port in upstream that uses this file. So, we cannot test it for now.

> Is there a public function we should call instead of setting m_size directly?

There is no setter for m_size.

Comment 9 Kim, kyusun 2011-07-05 22:14:13 PDT

(In reply to comment #4)
> How did you detect that this was broken?

When I used the linux port that is mentioned in https://bugs.webkit.org/show_bug.cgi?id=39348#c6, I found this bug.

Comment 10 Kim, kyusun 2011-07-05 22:16:13 PDT

(In reply to comment #7)
> We already have a adoptVector(Vector) which gets all this right, including setting m_size.  I worry that there are other things that might be missing from this implementation in the future.  re-using the adoptVector(Vector) method may be more future-proof.

Ok. After modifying it, I'll submit new patch.

Comment 11 Kim, kyusun 2011-07-07 02:23:56 PDT

Created attachment 99961 [details]
Patch

Comment 12 Kim, kyusun 2011-07-07 02:42:21 PDT

Changed to use adoptVector(Vector).

Comment 13 Eric Seidel (no email) 2011-07-07 10:18:10 PDT

Comment on attachment 99961 [details]
Patch

Nice!

Comment 14 WebKit Review Bot 2011-07-07 10:29:32 PDT

Comment on attachment 99961 [details]
Patch

Clearing flags on attachment: 99961

Committed r90574: <http://trac.webkit.org/changeset/90574>

Comment 15 WebKit Review Bot 2011-07-07 10:29:37 PDT

All reviewed patches have been landed.  Closing bug.