Hello,
I am running linux kernal 2.4.2 .
My query is concerning the manner in which the TCP code
handles the initial window size value ( RWIN ) in the TCP header.
I observed that RWIN that the client
announces is 5840 bytes in the <SYN> packet irrespective of what value
I set using the setsockopt .
Is there any particular reason for this ?
It seems to me as a too small a value.
Secondly, any way to get around this ?
I checked the /proc/sys/net/core/* where the defaults are set.
rmem_default is set to 65535 . Similarly wmem_default.
Further more, when I checked the complete packet trace using tcpdump for
the FTP data connection ( for a large file transfer ) , it seems that
though TCP starts with a RWIN of 5840 in the <SYN> segement but keeps on
increasing the RWIN as the connection progresses upto 47784 .
Next, I modified FTP code to set the initial window field to 500 K bytes.
The window field in the <SYN> packet should have been 64 K and wscale = 3.
But again the window field was 5840 bytes and wscale = 1.
After some packets were transferred , the window field increased upto
47784 but not beyond that.
Now the RFC 1323 says that the receiver will left-shift window field
of every incoming segemet by wscale bits.
Left shifting 47784 by 1 bit would in no way offer me th desired window
( 500 K bytes ).
Is there some thing that I am doing wrong or is there something wrong in
the TCP implementation ?
Please CC any replies to my email id.
Thanks,
Manish
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Jul 07 2001 - 21:00:17 EST