After all, using the i2c irq wasn't a good idea. After long hours of

testing why the !#+?§"1 TS stream stops and "debi oops" messages appear,
I found out that this is caused by the i2c irq handler. Don't ask me
why, but returning to the old code solved this mystery. Apparently, the
additional i2c interrupts caused some timing problems or the saa7146
is simply too f*cked up.

I left the code in however, it works for the analog MXB driver, so the
extension should decide if it should be used.

Removed some additional debug messages, which were commented out anyway.

1 files changed, 6 insertions, 5 deletions

diff --git a/linux/drivers/media/common/saa7146_core.c b/linux/drivers/media/common/saa7146_core.c
index bd079b0e5..f064b49d0 100644
--- a/linux/drivers/media/common/saa7146_core.c
+++ b/linux/drivers/media/common/saa7146_core.c
@@ -180,12 +180,12 @@ static void interrupt_hw(int irq, void *dev_id, struct pt_regs *regs)
 				DEB_I2C(("irq: i2c, status: 0x%08x, psr:0x%02x, ssr:0x%02x).\n",status,psr,ssr));
 				dev->i2c_op = 0;
 				wake_up(&dev->i2c_wq);
-			}
-			else
+			} else {
 				DEB_I2C(("unexpected irq: i2c, status: 0x%08x, isr %#x\n",status, isr));
-		}
-		else
+			}
+		} else {
 			DEB_I2C(("unhandled irq: i2c, status: 0x%08x, isr %#x\n",status, isr));
+		}
 		isr &= ~(MASK_16|MASK_17);
 	}
 	if( 0 != isr ) {
@@ -229,14 +229,15 @@ void try_attach_extension_and_device(struct saa7146_dev *dev, struct saa7146_ext
 		return;
 	}
 
+	dev->ext = ext;
 	if( 0 != ext->probe) {
 		if( 0 != ext->probe(dev, dev->pci->subsystem_vendor, dev->pci->subsystem_device) ) {
 			DEB_D(("ext->probe() failed for %p. skipping device.\n",dev));
+			dev->ext = NULL;
 			return;
 		}
 	}
 
-	dev->ext = ext;
 	if( 0 != ext->attach(dev, &ext->devices[i]) ) {
 		DEB_D(("ext->attach() failed for %p. skipping device.\n",dev));
 		dev->ext = NULL;