On Mon, 07 Jan 2008 15:25:36 +0200
Boaz Harrosh <[EMAIL PROTECTED]> wrote:
> On Mon, Jan 07 2008 at 8:53 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> wrote:
> > On Sun, 23 Dec 2007 13:09:05 +0200
> > Boaz Harrosh <[EMAIL PROTECTED]> wrote:
> >
> >> On Fri, Dec 21 2007 at 4:30 +0200, Benjamin
On Mon, Jan 07 2008 at 8:53 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> wrote:
> On Sun, 23 Dec 2007 13:09:05 +0200
> Boaz Harrosh <[EMAIL PROTECTED]> wrote:
>
>> On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt <[EMAIL
>> PROTECTED]> wrote:
>>> The sense buffer ins scsi_cmnd can
On Mon, Jan 07 2008 at 8:53 +0200, FUJITA Tomonori [EMAIL PROTECTED] wrote:
On Sun, 23 Dec 2007 13:09:05 +0200
Boaz Harrosh [EMAIL PROTECTED] wrote:
On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt [EMAIL
PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into
On Mon, 07 Jan 2008 15:25:36 +0200
Boaz Harrosh [EMAIL PROTECTED] wrote:
On Mon, Jan 07 2008 at 8:53 +0200, FUJITA Tomonori [EMAIL PROTECTED] wrote:
On Sun, 23 Dec 2007 13:09:05 +0200
Boaz Harrosh [EMAIL PROTECTED] wrote:
On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt [EMAIL
On Sun, 23 Dec 2007 13:09:05 +0200
Boaz Harrosh <[EMAIL PROTECTED]> wrote:
> On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt <[EMAIL PROTECTED]>
> wrote:
> > The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> > by some low level drivers (that typically happens with
On Sun, 23 Dec 2007 13:09:05 +0200
Boaz Harrosh [EMAIL PROTECTED] wrote:
On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt [EMAIL PROTECTED]
wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
> This has the potential of leaving a big fat ugly hole in the middle of
> scsi_cmnd. I would suggest of *just* moving the sense_buffer array to be
> the *first member* of struct scsi_cmnd. The command itself is already cache
> aligned, allocated by the proper flags to it's slab. And put a fat
On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt <[EMAIL PROTECTED]>
wrote:
> The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> by some low level drivers (that typically happens with USB mass
> storage).
>
> This is a problem on non cache coherent architectures such
On Fri, Dec 21 2007 at 4:30 +0200, Benjamin Herrenschmidt [EMAIL PROTECTED]
wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
This is a problem on non cache coherent architectures such as
This has the potential of leaving a big fat ugly hole in the middle of
scsi_cmnd. I would suggest of *just* moving the sense_buffer array to be
the *first member* of struct scsi_cmnd. The command itself is already cache
aligned, allocated by the proper flags to it's slab. And put a fat
On Fri, 2007-12-21 at 10:33 +, Alan Cox wrote:
> On Fri, 21 Dec 2007 13:30:08 +1100
> Benjamin Herrenschmidt <[EMAIL PROTECTED]> wrote:
>
> > The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> > by some low level drivers (that typically happens with USB mass
> > storage).
On Fri, 2007-12-21 at 06:16 -0700, Matthew Wilcox wrote:
> On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
> > On Fri, 21 Dec 2007 13:30:08 +1100
> > Benjamin Herrenschmidt <[EMAIL PROTECTED]> wrote:
> >
> > > The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> > > by
> It's also incomplete as a fix because I don't see what guarantees the
> buffer size will always exceed cache line size
There's a macro trick that adds a pad member after the buffer too, so
that it gets rounded up to the cacheline size:
> +#define __dma_aligned
>
On Fri, Dec 21, 2007 at 07:00:25AM -0700, Matthew Wilcox wrote:
> On Fri, Dec 21, 2007 at 02:30:28PM +0100, Thomas Bogendoerfer wrote:
> > there are SCSI host drivers, which also DMA to the sense buffer like
> > sgiwd93.c for example.
>
> Yes ... and there are others which don't, for example
On Fri, Dec 21, 2007 at 02:30:28PM +0100, Thomas Bogendoerfer wrote:
> there are SCSI host drivers, which also DMA to the sense buffer like
> sgiwd93.c for example.
Yes ... and there are others which don't, for example qla2xxx and
sym53c8xx.
--
Intel are signing my paycheques ... these opinions
On Fri, Dec 21, 2007 at 06:16:41AM -0700, Matthew Wilcox wrote:
> On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
> > On Fri, 21 Dec 2007 13:30:08 +1100
> > Benjamin Herrenschmidt <[EMAIL PROTECTED]> wrote:
> >
> > > The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
>
On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
> On Fri, 21 Dec 2007 13:30:08 +1100
> Benjamin Herrenschmidt <[EMAIL PROTECTED]> wrote:
>
> > The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> > by some low level drivers (that typically happens with USB mass
> >
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt <[EMAIL PROTECTED]> wrote:
> The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
> by some low level drivers (that typically happens with USB mass
> storage).
Should that not be fixed in USB storage by using
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt [EMAIL PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
Should that not be fixed in USB storage by using pci_alloc_coherent
On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt [EMAIL PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
On Fri, Dec 21, 2007 at 02:30:28PM +0100, Thomas Bogendoerfer wrote:
there are SCSI host drivers, which also DMA to the sense buffer like
sgiwd93.c for example.
Yes ... and there are others which don't, for example qla2xxx and
sym53c8xx.
--
Intel are signing my paycheques ... these opinions
On Fri, Dec 21, 2007 at 06:16:41AM -0700, Matthew Wilcox wrote:
On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt [EMAIL PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low
On Fri, Dec 21, 2007 at 07:00:25AM -0700, Matthew Wilcox wrote:
On Fri, Dec 21, 2007 at 02:30:28PM +0100, Thomas Bogendoerfer wrote:
there are SCSI host drivers, which also DMA to the sense buffer like
sgiwd93.c for example.
Yes ... and there are others which don't, for example qla2xxx and
It's also incomplete as a fix because I don't see what guarantees the
buffer size will always exceed cache line size
There's a macro trick that adds a pad member after the buffer too, so
that it gets rounded up to the cacheline size:
+#define __dma_aligned
On Fri, 2007-12-21 at 06:16 -0700, Matthew Wilcox wrote:
On Fri, Dec 21, 2007 at 10:33:26AM +, Alan Cox wrote:
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt [EMAIL PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level
On Fri, 2007-12-21 at 10:33 +, Alan Cox wrote:
On Fri, 21 Dec 2007 13:30:08 +1100
Benjamin Herrenschmidt [EMAIL PROTECTED] wrote:
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
Should
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
This is a problem on non cache coherent architectures such as
embedded PowerPCs where the sense buffer can share cache lines with
other structure members,
The sense buffer ins scsi_cmnd can nowadays be DMA'ed into directly
by some low level drivers (that typically happens with USB mass
storage).
This is a problem on non cache coherent architectures such as
embedded PowerPCs where the sense buffer can share cache lines with
other structure members,
28 matches
Mail list logo