From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Barada Subject: Problem using UART3 on Logic Torpedo w/2.6.32 Date: Thu, 18 Nov 2010 16:05:27 -0500 Message-ID: <4CE59517.7060408@logicpd.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from 174-46-170-154.static.twtelecom.net ([174.46.170.154]:44397 "EHLO edprlnx06.logicpd.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1760406Ab0KRVOD (ORCPT ); Thu, 18 Nov 2010 16:14:03 -0500 Sender: linux-omap-owner@vger.kernel.org List-Id: linux-omap@vger.kernel.org To: "linux-omap@vger.kernel.org" Cc: Peter Barada All, I have a 2.6.32 kernel based on the L23.i3.3 kernel (2.6.32) from TI, and I've run into an interesting problem with UART3 (maps to /dev/ttyS1 on the Torpedo board). On the host I have it hooked up to /dev/ttyS1, so I turn on CTS/RTS and set the baudrate by: host$ stty 115200 crtscts < /dev/ttyS1 And the same on the target: OMAP-35x$ stty 115200 crtscts < /dev/ttyS1 To force the board to throttle the serial I slowed down the console to 19200: OMAP-35x$ stty 19200 < /dev/ttyS0 On the target console I capture the data from the host (and show it on the serial port): OMAP-35x$ cat < /dev/ttyS1 | tee /tmp/x and on the host send the data: host$ dd if=/dev/urandom count=100 bs=1024 | hexdump -C > /tmp/x host$ cat /tmp/x > /dev/ttyS1 I'd expect to see something like the following on the target: 00000000 35 2c a6 97 23 35 95 9a fb 91 6d 6e ce f8 32 d7 |5,..#5....mn..2.| 00000010 6b 0b cf 66 e6 3a b4 55 91 5f 86 8f 41 c9 49 76 |k..f.:.U._..A.Iv| 00000020 e6 cf fa 0a 0e db 69 0c db 14 6b 76 62 4c 5b 9e |......i...kvbL[.| 00000030 98 1b 5f 30 16 d6 ed 96 dc d7 f1 3b 59 a0 ec ac |.._0.......;Y...| 00000040 86 8c 9b 28 21 b8 a0 98 ed cf 96 39 15 a4 7b 9e |...(!......9..{.| 00000050 bf 01 aa 09 1a 12 1f c3 49 b3 92 73 00 84 52 de |........I..s..R.| 00000060 c0 d3 4b 1d ca 84 04 ea 60 ef 7f b2 63 36 eb 5e |..K.....`...c6.^| 00000070 28 0c 20 2a 86 a2 36 bb c7 c0 27 da 87 c8 8b 1e |(. *..6...'.....| 00000080 5d b7 04 b3 2c 0b 29 f4 8d 4f 5f fe 90 b0 1b 96 |]...,.)..O_.....| 00000090 b8 82 94 ae 92 37 31 03 dc 1b c3 c0 38 b1 16 77 |.....71.....8..w| Instead I see: OMAP-35x# head /tmp/x 00000000 35 2c a6 97 23 6b 0b ca fb 91 6d 6e ce f8 32 d6 8f 41.#5....mn..2.| 00000010 .Iv| 00f 66 e6 3a b4 55 91 5f 8669 0c c9 49 76 |k..f.:.U._..A.|.....000020 e6 cf fa 0a 0e db 1b 5f db 14 6b 76 62 4c 5b 9e b 59 a.i...kvbL[.| 00000030 98.| 000030 16 d6 ed 96 dc d7 f1 3b 59 e0 ec ac |.._0.......;Y.....(!..0040 86 8c 9b 28 21 b8 a0 98 09d cf 96 39 15 a4 7b 9e |.00 84 ....9..{.| 00000050 bf 01000000 1a 12 1f c3 49 b3 92 73 a 60 52 de |........I..s..R.| .....`60 c0 d3 4b 1d ca 84 04 e0 2a 8ef 7f b2 63 36 eb 5e |..7 c8 8b...c6.^| 00000070 28 0c 20000806 a2 36 bb c7 c0 27 da 87 8d 4f 1e |(. *..6...'.....| 00.)..O_ 5d b7 04 b3 2c 0b 29 f4 ae 92 5f fe 90 b0 1b 96 |]...,.).6 7.....| 00000090 b8 82 94 00a0 37 31 03 dc 1b c3 c0 38 b1 7c 77 |.....71.....8..w| Looking deeper: OMAP-35x# hexdump -C /tmp/x | head 00000000 30 30 30 30 30 30 30 30 20 20 33 35 20 32 63 20 |00000000 35 2c | 00000010 61 36 20 39 37 20 32 33 20 36 62 20 30 62 20 63 |a6 97 23 6b 0b c| 00000020 61 20 20 66 62 20 39 31 20 36 64 20 36 65 20 63 |a fb 91 6d 6e c| 00000030 65 20 66 38 20 33 32 20 64 36 20 38 66 20 34 31 |e f8 32 d6 8f 41| 00000040 2e 23 35 2e 2e 2e 2e 6d 6e 2e 2e 32 2e 7c 0a 30 |.#5....mn..2.|.0| 00000050 30 30 30 30 30 31 30 20 20 2e 49 76 7c 0a 30 30 |0000010 .Iv|.00| 00000060 66 20 36 36 20 65 36 20 33 61 20 62 34 20 35 35 |f 66 e6 3a b4 55| 00000070 20 20 39 31 20 35 66 20 38 36 36 39 20 30 63 20 | 91 5f 8669 0c | 00000080 20 63 39 20 34 39 20 37 36 20 20 7c 6b 2e 2e 66 | c9 49 76 |k..f| 00000090 2e 3a 2e 55 2e 5f 2e 2e 41 2e 7c 2e 2e 2e 2e 2e |.:.U._..A.|.....| However if I instead use UART2 and replicate the above steps, all is well. I've looked at the serial signals and CTS/RTS are responding as expected. I tried switching drivers from the 8250 to the OMAP_SERIAL, and the results look the same. Any ideas what is happening? -- Peter Barada peterb@logicpd.com