This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: rebase not compilable

Dave Korn wrote:

> Regardless of how well (or poorly) the
> hash function distributes DLLS into the various buckets, there are only 1024
> of them, and we have many DLLs, many of which will occupy multiple buckets;
> collisions are inevitable.

First of all, I don't see where this 1024 comes from.  By my reading the
hash distributes over the range 0x61300000 - 0x712C0000 in 64k
increments, meaning 4092 buckets.

But what I really meant wasn't necessarily to improve the hashing
function per se but to give it more buckets, a wider range.  I realize
that higher than 0x712c0000 tends to run into the Microsoft-assigned
ranges for things in %systemroot%\system32, but it could also place
things on the other side of the Cygwin DLL as well.  Isn't that what
Reini is trying to do here anyway?

I guess my meta-point is that if we have a problem with the way we
distribute image bases, shouldn't we try to find a way to solve that
problem once in a central manner at the time the DLL is created by the
package maintainer, rather than making the end user deal with it by
brute force of resuffling everything on their system (in a way that is
non-obvious if they are not familiar with the problem.)


Unsubscribe info:
Problem reports:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]