Re: On the logic of the file checking

On Apr 24, 2006, at 10:39, Echaozh Zhang wrote:
> Hi everyone,
>
> We've been so happy using libtorrent, having everything going smooth
> and good, before we suddenly bumped into some strange problem.
>
> Before we met the problem, we just download each file... then one day
> the downloading is quite ok to us, so we try to seed the file. Seeding
> is quite ok before we close the program and restart it. When we
> restart the program with no resume data, the program crashes due to
> the failure of an assertion.
>
>             // if we're not in compact mode, make sure the
>             // pieces are spread out and placed at their
>             // final position.
>             assert(!m_unallocated_slots.empty());   // Line 1401,
> storage.cpp, version 0.9.1
>
> We tracked the code and found that when the "do the full check part"
> is about to end, it changes the state to allocating. Then the
> check_files() function is again called, the assertion fails quite
> reasonably: we have already downloaded the file, there is nothing
> unallocated! It seems to me a flaw in the logic, though I still don't
> quite follow it since I'm not really reading the compact mode code.
> BTW, we're not using the compact mode.
>
> I diff'ed the storage.cpp's of the two versions, 0.9.1 and 0.9.2, the
> logic seems the same so we didn't try the newer version.
>
> We'll be so grateful if anyone solves the myth for us,

Thanks for the report. I will try to reproduce this later today, and  
hopefully have a fix by tomorrow.

--
Arvid Norberg




-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642