[ECOS] 82558B works with RedBoot, but not in ping_test (recvfrom: Operation timed out)

[Vladimir Sterjantov]

Yes, I comletely agree with You, since there were no interrupts, no incoming packets were processed.

But I'm not sure that target sends ICMP packets, because:
It's seems like after receiving configuration from DHCP-server via 
init_all_network_interfaces() call,
ICMP packets can't be sent to host (when tcpdump is running
on host, it shows that all is Ok for DHCP, but no information about ICMP 
packets) - see trace below.
When target is booted from RedBoot diskette with network support, with 
manually pings from RedBoot> prompt,
tcpdump on host says (all Ok here for me):
17:28:42.609557 arp who-has vfshost (ff:ff:0:43:0:44) tell pctarget
17:28:42.609599 arp reply vfshost is-at 0:50:4:ec:7d:48
17:28:42.613147 pctarget > vfshost: icmp: echo request (ttl 64, id 1, 
len 92)
17:28:42.613218 vfshost > pctarget: icmp: echo reply (ttl 255, id 41492, 
len 92)

Trace from tcpdump, when running ping_test:

# tcpdump -vvv -i eth1
tcpdump: listening on eth1
17:05:41.836548 arp who-has 0.0.0.0 tell 0.0.0.0
17:05:43.197972 0.0.0.0.bootpc > 255.255.255.255.bootps:  xid:0xf6fff8f8 
flags:0x8000 [|bootp] (ttl 64, id 0, len 328)
17:05:43.268626 vfshost.bootps > 255.255.255.255.bootpc:  xid:0xf6fff8f8 
flags:0x8000 Y:pctarget S:vfshost ether 0:90:27:59:f9:f6 [|bootp] (DF) 
(ttl 64, id 0, len 328)
17:05:45.696610 0.0.0.0.bootpc > 255.255.255.255.bootps:  xid:0xf6fff8f8 
flags:0x8000 [|bootp] (ttl 64, id 1, len 328)
17:05:45.696852 vfshost.bootps > 255.255.255.255.bootpc:  xid:0xf6fff8f8 
flags:0x8000 Y:pctarget S:vfshost ether 0:90:27:59:f9:f6 [|bootp] (DF) 
(ttl 64, id 0, len 328)
17:05:48.206765 0.0.0.0.bootpc > 255.255.255.255.bootps:  xid:0xf6fff9f8 
flags:0x8000 [|bootp] (ttl 64, id 2, len 335)
17:05:48.207015 vfshost.bootps > 255.255.255.255.bootpc:  xid:0xf6fff9f8 
flags:0x8000 Y:pctarget S:vfshost ether 0:90:27:59:f9:f6 [|bootp] (DF) 
(ttl 64, id 0, len 328)
17:05:50.367775 arp who-has pctarget tell pctarget
17:05:50.756534 arp who-has pctarget tell pctarget
17:05:51.166685 arp who-has vfshost tell pctarget
17:05:51.166728 arp reply vfshost is-at 0:50:4:ec:7d:48
17:05:52.497429 arp who-has vfshost tell pctarget
17:05:52.497458 arp reply vfshost is-at 0:50:4:ec:7d:48

Truely, I need any advice. Thank You for attention .
Vladimir

Gary Thomas wrote:

>The most likely difference to look into is that RedBoot polls
>the device whereas eCos expects it to generate interrupts.
>
>The flow shown below indicates that the eCos driver is being
>somewhat successful talking to the device (do you ever see any
>packets come out?), but that since there were no interrupts,
>no incoming packets were processed.
>
>On Thu, 2002-06-20 at 01:42, Vladimir Sterjantov wrote:
>
>>Novice in eCos, and any help will be greatly appreciated.
>>
>>Problem:
>>
>>- target: pc,
>>            Network card Intel EtherExpress 100 Pro, chip 82558B
>>- host: linux at i386
>>
>>Both RedBoot and ping_test (OpenBSD stack used, template "net") built from
>>eCos CVS sources, downloaded 11 jun 2002.
>>
>>Target PC boots from RedBoot floppy, without network support, but with
>>support of debugging via serial line.
>>82559 driver compiled with DEBUG_82559 turned on.
>>
>>Now run ping_test and see output:
>>
>># gdb -nw ping_test
>>
>>........SKIPPED....................................................
>>
>>
>>i82559_start 0 flg 47 Link = Down, 10 Mbps, Half Duplex
>>PING server 172.16.200.90
>>Tx 0 13af40: 1 sg's, 42 bytes, KEY 156800
>>Tx 0 13af40: Starting Engines: KEY 156800 TxCB 1bb26c
>>TxDone 0 13af40: KEY 156800 TxCB 1bb26c
>>recvfrom: Operation timed out                                  --- Problem.
>>.........
>>
>>
>>Any network test don't works too.
>>May be difference between 82558 and 82559 not solved in network driver?
>>
>>Thanks,
>>
>>Vladimir
>>




-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss