From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arkh4mKn1ght Date: Tue, 27 May 2014 10:52:57 -0300 Subject: [ath9k-devel] Weird Atheros 9485 http connecting problems In-Reply-To: <20140527114325.GA11035@qca.qualcomm.com> References: <53795054.4050701@gmail.com> <20140519062926.GA18863@qca.qualcomm.com> <537A0346.4090300@gmail.com> <20140519144232.GA19244@qca.qualcomm.com> <537CBA30.80407@gmail.com> <20140521145902.GB9563@qca.qualcomm.com> <537CC1EB.3090500@gmail.com> <538382C1.40804@gmail.com> <20140527114325.GA11035@qca.qualcomm.com> Message-ID: <538498B9.5040902@gmail.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ath9k-devel@lists.ath9k.org Yeah i didnt try to say that the culprit was firefox, it was just an observation i made so i thought maybe the problem was related somehow to a limit or kernel parameter. Thanks for the tip about the TCP receiver buffer size, but first i have tweaked these 2 parameters that were in default low: net.core.somaxconn = 128 net.ipv4.tcp_max_syn_backlog = 128 I have both raised them to 2048. I did this because monitoring my laptop with netstat i found that the number of connections including TCP, UDP, and unix domain sockets DGRAM and STREAM were about 700 at a given time. I will see if this solves the issue. if not i will try your suggestion Greetings Oskar On 05/27/2014 08:43 AM, Rajkumar Manoharan wrote: > On Mon, May 26, 2014 at 03:06:57PM -0300, Oskar Kossuth wrote: >> Update: >> >> I did some investigation after the problem with the atheros wifi card >> happened again and i found something odd. >> The firefox process was consuming a lot of CPU, while the problem was >> ongoing, and the CPU usage decreased to normal levels after i stopped trying >> to load a website, and increased a lot after i tried to load the website >> again. >> > Oskar, > > your issue is not related to firefox as earlier you had mentioned that > with lynx also the problem occurs. isn't it? >> PS: I tested using the eth0 realtek adapter of my laptop and disabled >> wireless. The problem appeared also in this case. >> > So it is not wireless or ethernet issue. From you earlier sniffer > capture, I do see lot of TCP connection resets with window size 0. > Even TCP 3-way handshake also terminated in most of the cases. Can you > tune TCP receive buffer size. > > http://www.cyberciti.biz/faq/linux-tcp-tuning/ > > -Rajkumar > _______________________________________________ > ath9k-devel mailing list > ath9k-devel at lists.ath9k.org > https://lists.ath9k.org/mailman/listinfo/ath9k-devel > . >