Re: ATA errors.

[Johannes Hofmann <Johannes.Hofmann@xxxxxx>]

Hi,

the patch works for me. I now get 

root@blob:/home/hofmann >atacontrol mode 0
Master = UDMA33 
Slave  = PIO4

whereas before it always switched back to BIOSPIO after a while.
I don't see any warning messages though.

Thanks,
Johannes

PS: I had to add a line: 

typedef struct malloc_pipe* malloc_pipe_t;

to sys/mpipe.h, to make things compile.


Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx> wrote:
> 
>    Ok, here is a test patch.  Those people with the ATA problem please run
>    this patch and see if it fixes it.
> 
>    The issue is that M_NOWAIT can cause malloc() to return NULL in DragonFly
>    even if memory is available, due to DFly not being able to obtain the
>    kernel_map lock in some instances.
> 
>    I have not been able to fully test this patch because I don't seem to
>    have any drives with tags enabled.  I artifically set the pipeline limit
>    fairly low (to 4) but still can't seem to hit it.  If someone can get
>    the patch to generate some of those warning printfs that would be great
>    :-)
> 
>                                                -Matt
>