Re: kqueue a alternative poll/select?

[Ranier]

> Bill wrote:
> The apr_pollset API in APR would be a suitable wrapper for a kqueue or 
> dev/poll implementation, but no one has spent the time and effort to do 
> it yet. I have a prototype httpd implementation (patch posted to this 
> list months ago) that implements event driven network reads and Ian 
> Holsman wrote up a design for doing event driven writes. That's about it.

Thanks for your reply, I would like to help more.
However, still I do not have capacity to contribute with complex codes.

I have hopes to improve.

Thanks.

Ranier.

Re: 2.0.47: crash with autoindexed download of *.msi files

[Günter Knauf]

Sorry, all back - seems it comes from a third party module which I 
forgot to deactivate with the tests!!

> I've just found a very weired behavior of Apache2:
> if I try to download a *.msi file from an autoindex-generated page
> Apache2 crashes! I've tested this on NetWare and Win32, and both
> behave the same! Not tested yet on Linux, but seems to be a gernal
> problem - can someone please try to recreate the problem?
> oh, when I rename the same file, f.e. add a .zip then all works fine...
> I've also tested with a text file with only a few bytes, renamed it
> to test.msi and Apache2 crashes...

> Guenter.

2.0.47: crash with autoindexed download of *.msi files

[Günter Knauf]

Hi all,
I've just found a very weired behavior of Apache2:
if I try to download a *.msi file from an autoindex-generated page
Apache2 crashes! I've tested this on NetWare and Win32, and both
behave the same! Not tested yet on Linux, but seems to be a gernal
problem - can someone please try to recreate the problem?
oh, when I rename the same file, f.e. add a .zip then all works fine...
I've also tested with a text file with only a few bytes, renamed it
to test.msi and Apache2 crashes...

Guenter.

Message: winnt_accept: Asynchronous AcceptEx failed.

[Alex Bereznyi]

Hi,

 

We see a lot of WARNING messages in Windows Event Log:

 

winnt_accept: Asynchronous AcceptEx failed. 

 

The actual error
is always 995: ERROR_OPERATION_ABORTED - The I/O operation has been aborted because of either a
thread exit or an application request.

 

It's coming from server\mpm\winnt\child.c

The call:

 

GetOverlappedResult((HANDLE)context->accept_socket, 


&context->Overlapped, 


&BytesRead, FALSE)

 

GetLastError()  is not called there, instead:

 

   
closesocket(context->accept_socket);

   
context->accept_socket
= INVALID_SOCKET;

 

It all seems like normal condition to me -
why log WARNING ? Is it OK to
turn it off?

 

 

Thanks,

Alex