Hi Gene, Zitat von Gene Heskett <ghesk...@shentel.net>
Greetings all; This scanner in an MFC-9620DW works well with the brother drivers if one is not in a hurry. I investigated the delays and found the networking portion of the Brother Drivers are sending the first 6 packets, at 1 second intervals, with bad TCP checksum according to tcpdump. The scanner of course rightfully ignores them.
it might be you're barking up the wrong tree: The bad checksums reported by tcpdump may likely be caused by offloading CRC calculation to the network card. Of course, if you were capturing at the scanner's end, you'd be right in your assumption.
What type of packets are those six? TCP session establishment? ARP? Other? Can you tell what the printer sends (probably to some totally different host, i.e. DNS server - you'd need to capture at the scanner's port) when receiving these packets?
Then their driver seems to get it all in the same sock for the rest of the scan. I have not traced a multisheet scan as yet, but this 7 seconds delay is present regardless of whether the sheet is on the glass, or in the in hopper of the ASF.
I've had a close look at brsane4 network traffic when debugging a problem with a networked Brother scanner and I don't remember seeing these unanswered packets at the start of the scan. That was a remote scanner though, crossing routers and a VPN tunnel to connect. And connected via WLAN at the scanner's site.
We have older on-site Brother MFCs used as printers and scanners, connected via local Ethernet (most likely what you call "cat5 interface"). These have shown not to be instantaneous during operation, but I'd name the slow processors in those boxes as the culprit.
This has been reported to Brother twice now, but no one has responded with a fix.
Getting to the right person can be troublesome, indeed. Been there, but later met the right person and got the brsane4 driver fixed, based on our findings.
Maybe it's some different network problem, after all, or related to a device wake-up upon seeing the first packets?
Regards, J. -- sane-devel mailing list: sane-devel@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/sane-devel Unsubscribe: Send mail with subject "unsubscribe your_password" to sane-devel-requ...@lists.alioth.debian.org
