Re: [PATCH] ieee1394: ohci1394: add reg_flush, remove bogus barriers

2007-03-04 Thread Stefan Richter
I wrote: > @@ -1521,7 +1513,7 @@ static int ohci_iso_recv_start(struct hp > > if (sync != -1) { > /* set sync flag on first DMA descriptor */ > - struct dma_cmd *cmd = >block[recv->block_dma]; > + struct dma_cmd *cmd = >block[recv->block_dma]; >

[PATCH] ieee1394: ohci1394: add reg_flush, remove bogus barriers

2007-03-04 Thread Stefan Richter
Add a utility function to flush MMIO writes. Remove some mb()s which didn't do anything. There are still a few wmb()s in ohci1394 which look suspicious. Shouldn't there be dma_sync_*() calls instead? Signed-off-by: Stefan Richter <[EMAIL PROTECTED]> --- drivers/ieee1394/ohci1394.c | 47

[PATCH] ieee1394: ohci1394: add reg_flush, remove bogus barriers

2007-03-04 Thread Stefan Richter
Add a utility function to flush MMIO writes. Remove some mb()s which didn't do anything. There are still a few wmb()s in ohci1394 which look suspicious. Shouldn't there be dma_sync_*() calls instead? Signed-off-by: Stefan Richter [EMAIL PROTECTED] --- drivers/ieee1394/ohci1394.c | 47

Re: [PATCH] ieee1394: ohci1394: add reg_flush, remove bogus barriers

2007-03-04 Thread Stefan Richter
I wrote: @@ -1521,7 +1513,7 @@ static int ohci_iso_recv_start(struct hp if (sync != -1) { /* set sync flag on first DMA descriptor */ - struct dma_cmd *cmd = recv-block[recv-block_dma]; + struct dma_cmd *cmd = recv-block[recv-block_dma];