You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apr.apache.org by "William A. Rowe, Jr." <wr...@rowe-clan.net> on 2007/01/09 07:28:17 UTC

Re: [VOTE] httpd-2.2.4 release candidate for review

Tom Donovan wrote:
> 
> I see that apr_os_sock_put() set remote_addr_unknown=1 in earlier APR
> versions too.

It's actually apr_os_sock_make() and although it set unknown=1 where there
was no remote addr, it assumed unknown from alloc_socket() was 0.

Try the attached patch please?

Bill

Re: [VOTE] httpd-2.2.4 release candidate for review

Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Tom Donovan wrote:
> 
> I'd vote to fix it in 2.2.4 rather than deal with the noise - but
> personally I'm happy either way since I now know the workaround.

We won't - 2.2.4 is done.

We could scuttle 2.2.4, but given the overwhelming improvements I'm
really loathe to do that.  Let 2.2.4 live, and pick up 2.2.5 soon.

We can't help it when MS doesn't follow their own documented API /shrug

Re: [VOTE] httpd-2.2.4 release candidate for review

Posted by Tom Donovan <do...@bellatlantic.net>.
re: "start an incident at Microsoft"

I'll try - but since Win2000 is in what MS calls "Extended support phase", only security bugs are 
accepted.  I'll describe it as creatively as I can...

re: "I don't see this as a showstopper"

True. The Win32DisableAcceptEx directive certainly works.

There could be a large number of production sites running Win2k, so there is apt to be a lot of buzz 
and forum posts about "you need to add this new directive for 2.2.4".

I'd vote to fix it in 2.2.4 rather than deal with the noise - but personally I'm happy either way 
since I now know the workaround.

Glad I could contribute to an Apache release.

-tom-

William A. Rowe, Jr. wrote:
>> The Win32DisableAcceptEx directive also works to correct the problem
>> (without the patch).
> 
> That's your bug - AcceptEx inhibits proper behavior of getpeername(),
> if you would like to start an incident at Microsoft with respect to your
> 2000 machine's IP stack and drivers.
> 
>> Applying win32sock_is_known.patch corrects the problem for win2000.
> 
> Is our bug, fixing on all apr applicable branches today, thanks for all
> of your detailed analysis and dedicated testing!
> 
> I don't see this as a showstopper, as there is a user-configurable
> workaround (Win32DisableAcceptEx) if the client in server-status, and
> the access.log/error.log entries displays as 0.0.0.0.
> 
> Bill

Re: [VOTE] httpd-2.2.4 release candidate for review

Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Tom Donovan wrote:
> re: the Windows 2000   0.0.0.0 IP address problem
> 
> Bill,
> 
> Both solutions work.
> 
> The Win32DisableAcceptEx directive also works to correct the problem
> (without the patch).

That's your bug - AcceptEx inhibits proper behavior of getpeername(),
if you would like to start an incident at Microsoft with respect to your
2000 machine's IP stack and drivers.

> Applying win32sock_is_known.patch corrects the problem for win2000.

Is our bug, fixing on all apr applicable branches today, thanks for all
of your detailed analysis and dedicated testing!

I don't see this as a showstopper, as there is a user-configurable
workaround (Win32DisableAcceptEx) if the client in server-status, and
the access.log/error.log entries displays as 0.0.0.0.

Bill


Re: [VOTE] httpd-2.2.4 release candidate for review

Posted by Tom Donovan <do...@bellatlantic.net>.
re: the Windows 2000   0.0.0.0 IP address problem

Bill,

Both solutions work.

Applying win32sock_is_known.patch corrects the problem for win2000.

The Win32DisableAcceptEx directive also works to correct the problem (without the patch).

-tom-

William A. Rowe, Jr. wrote:
> It's actually apr_os_sock_make() and although it set unknown=1 where there
> was no remote addr, it assumed unknown from alloc_socket() was 0.
> 
> Try the attached patch please?
> 
> Bill

William A. Rowe, Jr. wrote:
 > Tom, speculating here without a 2000 box close - would you try to
 > DisableWin32AcceptEx please?  Perhaps the flaw actually resides in how
 > AcceptEx and GetAcceptExSockaddrs, and how they interact with the socket?
 >
 > Bill
 >