* Urgent work ! please help @ 2005-12-07 10:25 Conio sandiago 2005-12-07 10:36 ` Bernd Petrovitsch 2005-12-07 21:35 ` Willy Tarreau 0 siblings, 2 replies; 5+ messages in thread From: Conio sandiago @ 2005-12-07 10:25 UTC (permalink / raw) To: linux-kernel Hi all i am conio, i am facing some problems. I have a embedded monta vista linux kernel running on arm processor, the linux kernel version is 2.6.x I have developed a ethernet driver for the same. But now i am facing a strange problems- 1) If i do a ftp for a very big file from one board to another pc using a switching hub then At an early stage ftp-data is normally sent to LinuxPC. Then suddenly the embedded linux asks the IP address of LinuxPC with using ARP. After this point it start taking a lot of time to transfer data with ftp command. The packet analysis shows there are errors ,"TCP CHECKSUM INCORRECT". Please help if somebody has some idea whether it can be a hardware / software bug??? Regards conio ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Urgent work ! please help 2005-12-07 10:25 Urgent work ! please help Conio sandiago @ 2005-12-07 10:36 ` Bernd Petrovitsch 2005-12-07 21:35 ` Willy Tarreau 1 sibling, 0 replies; 5+ messages in thread From: Bernd Petrovitsch @ 2005-12-07 10:36 UTC (permalink / raw) To: Conio sandiago; +Cc: linux-kernel On Wed, 2005-12-07 at 15:55 +0530, Conio sandiago wrote: > Hi all > i am conio, > i am facing some problems. > I have a embedded monta vista linux kernel running on arm processor, What says MontaVistas support to the question about MontaVistas product? [...] > The packet analysis shows there are errors ,"TCP CHECKSUM INCORRECT". Your network (i.e .at least one part of it) is broken. Bernd -- Firmix Software GmbH http://www.firmix.at/ mobil: +43 664 4416156 fax: +43 1 7890849-55 Embedded Linux Development and Services ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Urgent work ! please help 2005-12-07 10:25 Urgent work ! please help Conio sandiago 2005-12-07 10:36 ` Bernd Petrovitsch @ 2005-12-07 21:35 ` Willy Tarreau 2005-12-09 5:52 ` Conio sandiago 1 sibling, 1 reply; 5+ messages in thread From: Willy Tarreau @ 2005-12-07 21:35 UTC (permalink / raw) To: Conio sandiago; +Cc: linux-kernel On Wed, Dec 07, 2005 at 03:55:32PM +0530, Conio sandiago wrote: > Hi all > i am conio, > i am facing some problems. > I have a embedded monta vista linux kernel running on arm processor, > the linux kernel version is 2.6.x First rule if you're looking for some help : be precise. 2.6.x is not a kernel version, and ARM is more a family than a processor. > I have developed a ethernet driver for the same. >From your report below, I think that you now need to add some debug code to your driver. It seems reasonable that a new development is buggy at first, and a broken ethernet driver can easily lead to corrupted data. > But now i am facing a strange problems- > > 1) If i do a ftp for a very big file from one board to another pc > using a switching hub then what size ? what speed ? half/full duplex ? what transfer speed do you observe before the problem ? > At an early stage ftp-data is normally sent to LinuxPC. Then suddenly > the embedded linux asks the IP address of LinuxPC with using ARP. it may simply mean that it has lost access to the other one and that the ARP cache has expired. > After this point it start taking a lot of time to transfer data with > ftp command. > > The packet analysis shows there are errors ,"TCP CHECKSUM INCORRECT". before asking, have you tried : - change your switch and connect with a cross-over cable - sending icmp, udp - transfer in each direction separately to check whether it happens on transmit / receive / both - reduce MTU on one, then the other, then both ends - ping during the transfer ? > Please help > if somebody has some idea > whether it can be a hardware / software bug??? with so few information, it's hard to say. At most, we can deduce that you encountered a problem the first time you tried your fresh new driver. > Regards > conio Regards, Willy ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Urgent work ! please help 2005-12-07 21:35 ` Willy Tarreau @ 2005-12-09 5:52 ` Conio sandiago 2005-12-09 6:36 ` Willy Tarreau 0 siblings, 1 reply; 5+ messages in thread From: Conio sandiago @ 2005-12-09 5:52 UTC (permalink / raw) To: Willy Tarreau; +Cc: linux-kernel Hello Willy, Thanks a lot for reply. i have done some more investigation and now i have the following observation:- 1) When i analyse the packets using ethereal, i saw some duplicate Acks from the reciever, after which transmitter does fast -re transmission, In this transmission i saw some TCP checksum incorrect messages. 2) Now if i disbale the DCACHE in the system , And then do the ftp, everything works well. No my doubt is a) When does a transmitte do Fast Retransmission,? b) if i disbale the cache then system performance goes very slow, thats why everything goes well, c) what is the case of collsion in the ethernet driver?? Please help Thanks in advance conio On 12/8/05, Willy Tarreau <willy@w.ods.org> wrote: > On Wed, Dec 07, 2005 at 03:55:32PM +0530, Conio sandiago wrote: > > Hi all > > i am conio, > > i am facing some problems. > > I have a embedded monta vista linux kernel running on arm processor, > > the linux kernel version is 2.6.x > > First rule if you're looking for some help : be precise. 2.6.x is not > a kernel version, and ARM is more a family than a processor. > > > I have developed a ethernet driver for the same. > > From your report below, I think that you now need to add some debug > code to your driver. It seems reasonable that a new development is > buggy at first, and a broken ethernet driver can easily lead to > corrupted data. > > > But now i am facing a strange problems- > > > > 1) If i do a ftp for a very big file from one board to another pc > > using a switching hub then > > what size ? what speed ? half/full duplex ? what transfer speed do > you observe before the problem ? > > > At an early stage ftp-data is normally sent to LinuxPC. Then suddenly > > the embedded linux asks the IP address of LinuxPC with using ARP. > > it may simply mean that it has lost access to the other one and that > the ARP cache has expired. > > > After this point it start taking a lot of time to transfer data with > > ftp command. > > > > The packet analysis shows there are errors ,"TCP CHECKSUM INCORRECT". > > before asking, have you tried : > - change your switch and connect with a cross-over cable > - sending icmp, udp > - transfer in each direction separately to check whether it happens > on transmit / receive / both > - reduce MTU on one, then the other, then both ends > - ping during the transfer > > ? > > > Please help > > if somebody has some idea > > whether it can be a hardware / software bug??? > > with so few information, it's hard to say. At most, we can deduce that > you encountered a problem the first time you tried your fresh new driver. > > > Regards > > conio > > Regards, > Willy > > ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Urgent work ! please help 2005-12-09 5:52 ` Conio sandiago @ 2005-12-09 6:36 ` Willy Tarreau 0 siblings, 0 replies; 5+ messages in thread From: Willy Tarreau @ 2005-12-09 6:36 UTC (permalink / raw) To: Conio sandiago; +Cc: linux-kernel On Fri, Dec 09, 2005 at 11:22:47AM +0530, Conio sandiago wrote: > Hello Willy, > Thanks a lot for reply. > i have done some more investigation and now i have the following observation:- > > > 1) When i analyse the packets using ethereal, > i saw some duplicate Acks from the reciever, after which > transmitter does fast -re transmission, > In this transmission i saw some TCP checksum incorrect messages. > > > > 2) Now if i disbale the DCACHE in the system , And then do the ftp, > everything works well. > > > > No my doubt is > > a) When does a transmitte do Fast Retransmission,? when there is a high packet loss rate > b) if i disbale the cache then system performance goes very slow, > thats why everything goes well, so probably you have a timeing problem in your driver. Maybe you has to wait milliseconds somewhere and you wait microseconds instead, or something like this. > c) what is the case of collsion in the ethernet driver?? I don't understand your question. It's not the driver which makes ethernet collisions. Collisions happen when several ethernet transmitters send simultaneously on the same bus (hub). > Please help > Thanks in advance > conio Willy ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2005-12-09 6:36 UTC | newest] Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2005-12-07 10:25 Urgent work ! please help Conio sandiago 2005-12-07 10:36 ` Bernd Petrovitsch 2005-12-07 21:35 ` Willy Tarreau 2005-12-09 5:52 ` Conio sandiago 2005-12-09 6:36 ` Willy Tarreau
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox
Powered by JetHome