Bug #5861 for Net-Ping: icmp failures under Windows

Wed Mar 31 15:56:31 2004 Guest - Ticket created

Subject:

icmp failures under Windows

Occasionally my script using Net::Ping will abort from within a sockaddr_in() call. I've traced it to the ($from_port, $from_ip) = sockaddr_in($from_saddr); call. Windows apparently is sending back nulls for "from_saddr", and perl is aborting the script due to a length error. I'm including a patch that will protect the sockaddr_in() call with a length comparison. I've also added retries with exponential backoffs to the ping_icmp code, and corrected the exponential backoff code in ping_udp. I think I've fixed bug report #4170 by passing the destination IP address in the ICMP packet data and verifying in the returned packet. I think I've fixed bug report #5269 by changing the ICMP checksumming code. I've checked it on HP-UX 10.20, Linux on Intel, and Windows 2003. All three function correctly with both even and odd buffer sizes.

Message body is not shown because it is too large.

Mon Aug 06 09:23:26 2007 SMPETERS [...] cpan.org - Reference to http://public.activestate.com/cgi-bin/perlbrowse/p/24242 added

Mon Aug 06 09:23:59 2007 SMPETERS [...] cpan.org - Reference to http://rt.perl.org/rt3/Ticket/Display.html?id=33664 added

Mon Aug 06 09:44:31 2007 SMPETERS [...] cpan.org - Correspondence added

On Wed Mar 31 15:56:31 2004, guest wrote: Show quoted text

> Occasionally my script using Net::Ping will abort from within a > sockaddr_in() call. I've traced it to the > ($from_port, $from_ip) = sockaddr_in($from_saddr); > call. Windows apparently is sending back nulls for "from_saddr", and > perl is aborting the script due to a length error. I'm including a > patch that will protect the sockaddr_in() call with a length > comparison. > > I've also added retries with exponential backoffs to the ping_icmp > code, and corrected the exponential backoff code in ping_udp. > > I think I've fixed bug report #4170 by passing the destination IP > address in the ICMP packet data and verifying in the returned packet. > > I think I've fixed bug report #5269 by changing the ICMP checksumming > code. I've checked it on HP-UX 10.20, Linux on Intel, and Windows > 2003. All three function correctly with both even and odd buffer > sizes.

Thanks for the patches. I've used a different patch for a fix to bug report #4170. I'm also looking into your fix for RT #5269. If it checks out OK, I'll include it in the next release. Right now, I'm wondering if the problem you are seeing is related to Perl core RT #33664. Do you happen to remember if you were using Windows 2003 or XP SP2?

Mon Aug 06 09:44:34 2007 The RT System itself - Status changed from 'new' to 'open'

Mon Aug 06 09:45:26 2007 SMPETERS [...] cpan.org - Reference to ticket #4170 added

Mon Aug 06 09:45:27 2007 SMPETERS [...] cpan.org - Reference to ticket #5269 added

Mon Aug 06 10:01:27 2007 Mike.Mitchell [...] sas.com - Correspondence added

Subject:	RE: [rt.cpan.org #5861] icmp failures under Windows
Date:	Mon, 6 Aug 2007 10:01:07 -0400
To:	<bug-Net-Ping [...] rt.cpan.org>
From:	"Mike Mitchell" <Mike.Mitchell [...] sas.com>

I'm fairly certain I was running it on Windows 2003. I'm glad to see that someone has started maintaining the code. Thanks for your efforts! -- Mike Mitchell Show quoted text

-----Original Message----- From: Steve Peters via RT [mailto:bug-Net-Ping@rt.cpan.org] Sent: Monday, August 06, 2007 9:45 AM To: Mike Mitchell Subject: [rt.cpan.org #5861] icmp failures under Windows <URL: http://rt.cpan.org/Ticket/Display.html?id=5861 > On Wed Mar 31 15:56:31 2004, guest wrote:

> Occasionally my script using Net::Ping will abort from within a > sockaddr_in() call. I've traced it to the > ($from_port, $from_ip) = sockaddr_in($from_saddr); > call. Windows apparently is sending back nulls for "from_saddr", and > perl is aborting the script due to a length error. I'm including a > patch that will protect the sockaddr_in() call with a length > comparison. > > I've also added retries with exponential backoffs to the ping_icmp > code, and corrected the exponential backoff code in ping_udp. > > I think I've fixed bug report #4170 by passing the destination IP > address in the ICMP packet data and verifying in the returned packet. > > I think I've fixed bug report #5269 by changing the ICMP checksumming > code. I've checked it on HP-UX 10.20, Linux on Intel, and Windows > 2003. All three function correctly with both even and odd buffer > sizes.

Thanks for the patches. I've used a different patch for a fix to bug report #4170. I'm also looking into your fix for RT #5269. If it checks out OK, I'll include it in the next release. Right now, I'm wondering if the problem you are seeing is related to Perl core RT #33664. Do you happen to remember if you were using Windows 2003 or XP SP2?

Wed Mar 07 06:56:56 2018 RURBAN [...] cpan.org - Correspondence added

This was fixed in 2.64 by CIORNII by adding the NIx_NOSERV flag.

Wed Mar 07 06:56:57 2018 RURBAN [...] cpan.org - Status changed from 'open' to 'resolved'