Shao Miller wrote

A packet capture might help
See attachment, was created using this command on linux host:
tcpdump -i eth0 ether host 00:50:fc:6e:00:22 -s 0 -w dumpfile.pcap

On DOS AoE client , I typed following commands

c:\verify off
c:\echo gpxe_verify_test > 1.txt
c:\verify on
c:\echo gpxe_verify_test > 2.txt
=> General failure writing drive C.......

When I open this trace in wireshark, I see following packets
-packet 1..4 = read of root dir
-packet 5&6  = write altered root folder (zero byte file entry added)
-packet 7&8 = read part of 1st FAT copy
-packet 9&10 =write back part of altered fat , copy 1
-packet 11&12 =write back part of altered fat , copy 2
-packet 13&14 = write file contents (entire sector is written, bogus content after gpxe_verify_test string)
-packet 15&16  = write altered root folder (proper file size entered)

so far so good, Packets  17..20  are in "verify on" mode.
The directory entry is cached, and isn't reread, so packet 17 resembles 5 (alter root dir)
This write already fails somehow, there's just one retry (packet 19&20)

However, I can't figure out what's the big difference between write requests 5-17 and responses in 6-18
It even fails without trying a single disk read attempt!

Adrian Bacon

Attachment: dumpfile.pcap
Description: Binary data

_______________________________________________
gPXE mailing list
[email protected]
http://etherboot.org/mailman/listinfo/gpxe

Reply via email to