Re: dibusb-common.c FE_HAS_LOCK problem

[Patrick Boettcher <pboettcher@kernellabs.com>]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 969 bytes --]

On Mon, 23 Nov 2009, Patrick Boettcher wrote:

> On Mon, 23 Nov 2009, grafgrimm77@gmx.de wrote:
>> [..]
>> ----- hello stupid I2C access ----
>> Pid: 255, comm: khubd Tainted: P       A   2.6.31.6 #1
>> Call Trace:
>> [<ffffffffa0042292>] ? dibusb_i2c_xfer+0xe2/0x130 [dvb_usb_dibusb_common]
>> [<ffffffff81341dc1>] ? i2c_transfer+0x91/0xe0
>> [<ffffffffa0059081>] ? dib3000_write_reg+0x51/0x70 [dib3000mb]
>> [<ffffffffa00855c9>] ? dvb_pll_attach+0xa9/0x238 [dvb_pll]
>> [..]
>
> Voila.
>
> This is the access with makes the dvb-pll-driver not create the tuner driver.
>
> This is (I forgot the correct name) read-without-write-i2caccess. It is bad 
> handled by the dibusb-driver and it can destroy the eeprom on the USB side.
>
> Please try whether the attached patch fixes the whole situation for you.
>
> If so, please send back a line like this:
>
> Tested-by: Your name <email>

The patch attached.

--

Patrick Boettcher - Kernel Labs
http://www.kernellabs.com/

[-- Attachment #2: Type: TEXT/PLAIN, Size: 776 bytes --]

diff -r 52da57b5e800 linux/drivers/media/dvb/dvb-usb/dibusb-common.c
--- a/linux/drivers/media/dvb/dvb-usb/dibusb-common.c	Thu Nov 19 17:15:37 2009 +0100
+++ b/linux/drivers/media/dvb/dvb-usb/dibusb-common.c	Mon Nov 23 13:20:10 2009 +0100
@@ -142,8 +142,13 @@
 		} else if ((msg[i].flags & I2C_M_RD) == 0) {
 			if (dibusb_i2c_msg(d, msg[i].addr, msg[i].buf,msg[i].len,NULL,0) < 0)
 				break;
-		} else
-			break;
+		} else if (msg[i].addr != 0x50) {
+			/* 0x50 is the address of the eeprom - we need to protect it
+			 * from dibusb's bad i2c implementation: reads without
+			 * writing the offset before are forbidden */
+			if (dibusb_i2c_msg(d, msg[i].addr, NULL, 0, msg[i].buf, msg[i].len) < 0)
+				break;
+		}
 	}
 
 	mutex_unlock(&d->i2c_mutex);